University of Groningen Improving patient selection ...

University of Groningen

Improving patient selection towards personalized treatment decisions in esophageal cancerHulshoff Jan

IMPORTANT NOTE You are advised to consult the publishers version (publishers PDF) if you wish to cite fromit Please check the document version below

Document VersionPublishers PDF also known as Version of record

Publication date2017

Link to publication in University of GroningenUMCG research database

Citation for published version (APA)Hulshoff J (2017) Improving patient selection towards personalized treatment decisions in esophagealcancer Rijksuniversiteit Groningen

CopyrightOther than for strictly personal use it is not permitted to download or to forwarddistribute the text or part of it without the consent of theauthor(s) andor copyright holder(s) unless the work is under an open content license (like Creative Commons)

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act indicated by the ldquoTavernerdquo licenseMore information can be found on the University of Groningen website httpswwwrugnllibraryopen-accessself-archiving-puretaverne-amendment

Take-down policyIf you believe that this document breaches copyright please contact us providing details and we will remove access to the work immediatelyand investigate your claim

Downloaded from the University of GroningenUMCG research database (Pure) httpwwwrugnlresearchportal For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum

Download date 12-11-2021

510604-L-bw-Hulshoff510604-L-bw-Hulshoff510604-L-bw-Hulshoff510604-L-bw-HulshoffProcessed on 13-6-2017Processed on 13-6-2017Processed on 13-6-2017Processed on 13-6-2017 PDF page 1PDF page 1PDF page 1PDF page 1

IMPROVING PATIENT SELECTION TOWARDS

PERSONALIZED TREATMENT DECISIONS IN

ESOPHAGEAL CANCER

Jan Binne Hulshoff


Hulshoff JB

Improving patient selection towards personalized treatment decisions in esopha-geal cancer

PhD dissertation University of Groningen The Netherlands

ISBN 978-90-367-9889-1ISBN (electronic version) 978-90-367-9888-4

copy Copyright 2017 Jan Binne Hulshoff Groningen The NetherlandsAll rights reserved No part of this thesis may be reproduced stored in a retrieval system or transmitted in any form or by any means without prior permission of the author or when appropiate of the publishers of the published articles

Graphic design Rens Dommerholt Persoonlijk Proefschrift wwwpersoonlijk-proefschriftnlPrinted by Ipskamp Printing Enschede wwwipskampprintingnl

The publication of this thesis was financially supported by the Junior Scientific Masterclass (JSM) University of Groningen and University Medical Center Groningen


Improving patient selection towardspersonalized treatment decisions in

esophageal cancer

Proefschrift

ter verkrijging van de graad van doctor aan deRijksuniversiteit Groningen

op gezag van derector magnificus prof dr E Sterken

en volgens besluit van het College voor Promoties

De openbare verdediging zal plaatsvinden op

woensdag 12 juli 2017 om 1245 uur

door

Jan Binne Hulshoffgeboren op 11 mei 1988

te Smallingerland


PromotorProf dr JTM Plukker

CopromotoresDr T KortewegDr HM van Dullemen

BeoordelingscommissieProf dr GAP HospersProf dr OS HoekstraProf dr RGH Beets-Tan


7

21

23

37

55

71

89

109

111

131

151

165

167175182183185

TABLE OF CONTENT

Chapter 1 General introduction

Part I Improving staging and neoadjuvant chemoradiotherapy outcome in esophageal cancer patients

Chapter 2 Impact of endoscopic ultrasonography on 18F-FDG PETCT upfront towards patient specific esophageal cancer treatmentChapter 3 Impact of pretreatment nodal staging on response evaluation to neoadjuvant chemoradiotherapy and prognosis in esophageal cancerChapter 4 Evaluation of progression prior to surgery after neoadjuvant chemoradiotherapy with computed tomography in esophageal cancer patients Chapter 5 Prognostic value of the circumferential resection margin in esophageal cancer patients after neoadjuvant chemoradiotherapy Chapter 6 Effect of extending the original eligibility criteria for the ldquoCROSSrdquo neoadjuvant chemoradiotherapy on toxicity and survival in esophageal cancer

Part II Improving prediction of response to neoadjuvant chemoradiotherapy

Chapter 7 Predicting response to neoadjuvant chemoradiotherapy in esophageal cancer with textural features derived from pre-treatment 18F-FDG PETCT imagingChapter 8 Prediction of response to neoadjuvant chemoradiotherapy with spatio-temporal 18F-FDG PET image biomarkers in esophageal cancer patientsChapter 9 Prediction of tumor response to neoadjuvant chemo-radiotherapy with DW-MRI and 18F-FDG PETCT in esophageal cancer

Summary

Chapter 10 SummaryNederlandse samenvattingPublications in PubMed DankwoordCurriculum vitea



General introduction

Jan Binne Hulshoff

CHAPTER 1


INTRODUCTION

Esophageal cancer is the 8th most commonly diagnosed malignancy worldwide [1] The incidence especially that of esophageal adenocarcinoma has increased markedly in most Western countries during the last three decades [2-4] In the Netherlands the number of newly diagnosed patients has increased from around 800 to 2500 yearly during the period 1990 to 2015 [5] At the time of diagnosis approximately 50 of the patients already have distant metastases or irresectable disease and are not amenable for cure resulting in a low overall 5-year survival rate between 15 and 25 [67]In patients with potentially curable esophageal cancer surgery is the main therapeutic option However at the time of diagnosis most patients already have locally advanced disease with lymph node metastases In these patients tumor recurrence is common even after a curative intended surgical resection and lymphadenectomy The overall survival is therefore low between 25 and 35 [8-11] In order to improve the outcome in patients with locally advanced esophageal cancer the CROSS (ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study) trial was initiated in which patients were treated with neoadjuvant chemoradiotherapy prior to surgery [12] The CROSS treatment schedule which consists of carboplatin and paclitaxel with concurrent radiotherapy (414 Gy given in 23 fractions of 18 Gy) followed by esophagectomy increased the 5-year overall survival with 13 to 47 [12] In many European countries including the Netherlands neoadjuvant chemoradiotherapy (nCRT) according to the original CROSS regimen became standard treatment for locally advanced esophageal cancer However approximately 20-25 of these patients do not benefit from the administered nCRT and may suffer from harmful side-effects It should be mentioned that the side-effects are relatively mild in the CROSS approach compared to those in the less frequently applied Cisplatin-5-Fluorouracil (Cis-5FU) combination [12-14] On the other hand between 25 and 42 of the patients treated with nCRT have a pathologic complete response [121516] In the future patients who are highly suspected for a pathologic complete response to nCRT might benefit from a personalized treatment approach with refraining from surgery followed by extensive follow-up and salvage surgery in case of isolated loco-regional recurrence or local residual disease For this approach to be worthwhile prediction of complete response with medical imaging techniques should match pathologic outcome However even in patients with a pathologic complete response early tumor recurrences may occur within 6 months which might be explained by inadequate pre-treatment staging and biological aggressive behaviour [17]

8


STAGING IN ESOPHAGEAL CANCER

The standard staging work-up in patients with esophageal cancer consists of endoscopy with biopsies of suspect lesions a diagnostic computed tomography (CT) scan a 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) or integrated 18F-FDG PETCT scan and an endoscopic ultrasound (EUS) [18-20] Anatomic staging methods such as CT and EUS enable detection of disrupted anatomical structures whereas the 18F-FDG PET is a metabolic imaging technique that depicts abnormal uptake of radioactive 18F-labeled fluorodeoxyglucose (18F-FDG) After staging the outcome of these sophisticated staging techniques are scored according to the TNM classification of the American Joint Committee on Cancer in which the T-stage stands for the local tumor depth of growth the N-stage for the lymph node status and the M-stage for distant metastases [21] Thereafter patients are discussed in a multidisciplinary tumor board meeting to determine the optimal treatment based on the TNM stage presence of comorbidities patients physical and mental condition and patient preferences In order to optimize treatment decision making and to advance towards a more personalized treatment approach adequate pre-treatment staging and restaging after nCRT is of great importanceOf all staging techniques EUS is the best method to determine the growth depth of the primary tumor (T-stage) and to detect lymph node metastases (N-stage) with a sensitivity between 71-924 and 80-847 and a specificity of 846-994 and 70-846 [2223] The advantage of EUS is the possibility to obtain fine-needle aspiration cytology from suspect lymph nodes which increases the sensitivity and specificity in detecting metastatic lymph nodes [2324] Downsides of EUS are the relative high patient burden and the risks of bleeding aspiration and perforation [25] Other downsides of the EUS include that the accuracy of EUS is highly dependent on the performers skill and that passage of the primary tumor is not possible in 20-36 of the patients because of tumor stenosis [26-29] Preventing an unnecessary EUS should therefore be pursued The best staging technique to detect distant metastases (M-stage) is the 18F-FDG PET with a sensitivity and specificity of 71 and 93 respectively [22] Because of the high sensitivity of 18F-FDG PETCT in detecting distant metastases an upfront 18F-FDG PETCT is the best predictor for curative resectability in potentially curable esophageal cancer and is the most cost-effective staging approach [30 31] With the current availability of integrated 18F-FDG PETCT scans the detection of lymph node metastases is also a feasible goal The additional value of a standard EUS after 18F-FDG PETCT might therefore be limited So far no study has been performed to determine the value of EUS after integrated 18F-FDG PETCT scanning but several studies reported contradictory

9

1


results about the additional value of EUS after CT-alone [32-34] One of the additional advantages of the 18F-FDG PETCT upfront staging sequence is that fine-needle aspiration can be performed with the EUS to determine if 18F-FDG PETCT suspect lymph nodes contain tumor cells Even with the combined information of all current optimal staging methods the accuracy of detecting lymph node metastases remains low In patients treated with surgery-alone the lymph node stage is incorrect in approximately 25 [35] Inaccurate nodal staging might lead to either (1) overstaging causing overtreatment with neoadjuvant chemoradiotherapy while surgery-alone was more suitable or (2) understaging causing treatment with surgery-alone while neoadjuvant chemoradiotherapy followed by surgery would be more appropriate [35-37] Moreover determining the exact localization of suspect lymph nodes is also important in current treatment paradigm especially for accurate radiotherapy target volume delineation In patients with a pathologic complete response of the primary tumor metastatic lymph nodes may still be left behind which might have been a consequence of inadequate lymph node staging [38-42] Therefore the definition of a complete response to nCRT both clinical and pathological should be based on response of the primary tumor and regional lymph nodes as well [4344]Furthermore after extensive staging with the current optimal staging techniques between 8 and 17 of all patients have interval metastases in the period between neoadjuvant treatment and surgery [45-48] For a personalized treatment approach progressive disease and interval metastases should be detected to prevent futile surgery and to change the treatment to a suitable palliative treatment Even with the high number of patients with progressive disease after neoadjuvant chemoradiotherapy restaging was not considered standard of care in many medical centers [20] The use of 18F-FDG PETCT scans is preferable as it detected about 3 out of 4 patients with progressive disease compared to 2 of 4 patients with CT [45-47]

TREATMENT OUTCOME AFTER NEOADJUVANT CHEMORADIOTHERAPY

To improve treatment selection and the treatment outcome of patients with esophageal cancer it is important to determine which patients benefit less from neoadjuvant chemoradiotherapy One of the major advantages of neoadjuvant chemoradiotherapy is tumor downsizing which results in a significantly higher number of microscopic tumor free resection margins (R0 resection) compared to surgery-alone approach [12] Although the circumferential resection margin is a well-established prognostic factor in patients treated with surgery-alone its value after neoadjuvant chemoradiotherapy remains unclear [49-55]

10


Moreover even in patients with tumor-free resection margins the number of patients with tumor recurrence is high emphasizing the importance of adequate pre-treatment staging and treatment selection Based on the good results and the relative high pathologic complete response rate of 29 in the CROSS trial patients with potentially curable esophageal cancer who did not meet the initial CROSS eligibility criteria were also treated accordingly This raises the question whether these patients benefit equally from this regimen

RESPONSE PREDICTION IN ESOPHAGEAL CANCER PATIENTS TREATED WITH NEOADJUVANT CHEMORADIOTHERAPY

Pathologic complete response which is achieved in 25-42 of the patients leads to a significantly better outcome [121516] In the future patients with a pathologic complete response might even benefit from refrainment of surgery The most commonly applied method to predict response to neoadjuvant chemoradiotherapy is to measure the decrease in the maximal value of the standardized uptake value (SUVmax) on 18F-FDG PET scans However this yields an insufficient sensitivity and specificity of 67 and 68 respectively [56] A major shortcoming of the SUVmax is the susceptibility to noise artifacts as it is based on a single voxelCurrently radiomics (tumor imaging features) are becoming increasingly important in the prediction of response to neoadjuvant therapies [57] Textural features derived from 18F-FDG PET scans depict the spatial correlation of 18F-FDG uptake between voxels and therefore display heterogeneity in uptake [58-63] So far several studies found textural features to be better predictors of response to neoadjuvant chemoradiotherapy than the SUVmax although this benefit is still insufficient for clinical decision making [58] Moreover predicting response by comparing pre- and post-neoadjuvant chemoradiotherapy 18F-FDG uptake histograms showed promising results as well longitudinal patterns in 18F-FDG uptake provided useful information in detecting response [64] Recently diffusion-weighted magnetic resonance imaging (DW-MRI) has been introduced to assess response to neoadjuvant chemoradiotherapy [65-69] DW-MRI is sensitive to water mobility (diffusion) and can be used to measure the apparent diffusion coefficient (ADC) ADC changes might represent alterations in tissue (micro)structures So far several studies showed ADC changes to be predictive for tumor response in EC [67 68] However further studies are necessary for clinical application of DW-MRI in response prediction In conclusion neoadjuvant chemoradiotherapy has increased the survival of locally advanced esophageal cancer but not all patients will benefit from the trimodality treatment In order to increase the outcome in general the patient selection needs to be improved towards a more personalized treatment in locally advanced esophageal cancer

11

1


OUTLINE OF THIS THESIS

The aim of this thesis is to improve the patient selection with the focus on improving staging strategies and treatment outcome and on the predicting response to neoadjuvant chemoradiotherapy followed by surgery The different chapters address circumstances towards a more personalized treatment by increasing staging accuracy and improving treatment related outcome (part I) and increasing prediction of pathologic complete response to nCRT (part II)

PART I IMPROVING STAGING AND NEOADJUVANT CHEMORADIOTHERAPY OUTCOME IN ESOPHAGEAL CANCER

The additional value of endoscopic ultrasound (EUS) after an upfront 18F-FDG PETCT remains unknown especially in regard to its impact on the given treatment In Chapter 2 the staging approach was optimized by determining the impact of EUS and fine needle aspiration on the given treatment in a 18F-FDG PETCT upfront model in a multicenter studyImproving the detection of lymph node metastases is of vital importance in the era of neoadjuvant chemoradiotherapy for accurate radiotherapy delineation In Chapter 3 we examined the effectiveness of combined CT 18F-FDG PETCT and EUS in the detection of lymph node metastases The relatively high rate of progressive disease after neoadjuvant chemoradiotherapy dissembled the benefit of surgery indicating the importance in detecting interval metastases before esophageal resection In Chapter 4 we therefore determined the value of CT restaging in detecting tumor progression after neoadjuvant chemoradiotherapy To improve treatment outcome of patients with esophageal cancer we should determine which patients will benefit from neoadjuvant chemoradiotherapy In Chapter 5 we assessed the prognostic value of circumferential resection margins after neoadjuvant chemoradiotherapyTo optimize selection of patients that may benefit from neoadjuvant chemoradiotherapy we determined the effect of extending the CROSS eligibility criteria for neoadjuvant chemoradiotherapy on the toxicity rate and survival in Chapter 6

PART II IMPROVING PREDICTION OF RESPONSE TO NEOADJUVANT CHEMORADIOTHERAPY

Prediction of response to neoadjuvant chemoradiotherapy is a basic step towards a personalized treatment approach in patients with esophageal cancer In part II the focus is on non-invasive methods to predict response to neoadjuvant chemoradiotherapy

12


In Chapter 7 we assessed the impact of textural features derived from pre-treatment 18F-FDG PETCT imaging in predicting response to neoadjuvant chemoradiotherapyIn Chapter 8 we determined the value of changes in textural features and histogram distances after neoadjuvant chemoradiotherapy in detecting response Finally in Chapter 9 we report the outcome of a pilot study about the additional role of diffusion-weighted magnetic resonance imaging (DW-MRI) to 18F-FDG PETCT scans in identifying tumor response to neoadjuvant chemoradiotherapy

13

1


REFERENCES

[1] Jemal A Bray F Center MM Ferlay J Ward E Forman D Global cancer statistics CA Cancer J Clin 2011 6169-90[2] Blot WJ Devesa SS Kneller RW Fraumeni JFJr Rising incidence of adenocarcinoma of the esophagus and gastric cardia JAMA 1991 2651287-1289[3] Cen P Banki F Cheng L et al Changes in age stage distribution and survival of patients with esophageal adenocarcinoma over three decades in the United States Ann Surg Oncol 2012 191685-1691[4] Bollschweiler E Wolfgarten E Gutschow C Holscher AH Demographic variations in the rising incidence of esophageal adenocarcinoma in white males Cancer 2001 92549-555[5] Netherlands Comprehensive Cancer Organization (IKNL) No Title 2016 p httpwwwcijfersoverkankernl[6] Enzinger PC Mayer RJ Esophageal cancer N Engl J Med 2003 3492241-2252[7] Pennathur A Gibson MK Jobe BA Luketich JD Oesophageal carcinoma Lancet 2013 381400-412[8] Sihvo EI Luostarinen ME Salo JA Fate of patients with adenocarcinoma of the esophagus and the esophagogastric junction a population-based analysis Am J Gastroenterol 2004 99419-424[9] Smit JK Pultrum BB van Dullemen HM Van Dam GM Groen H Plukker JT Prognostic factors and patterns of recurrence in esophageal cancer assert arguments for extended two-field transthoracic esophagectomy Am J Surg 2010 200446-453[10] Omloo JM Lagarde SM Hulscher JB et al Extended transthoracic resection compared with limited transhiatal resection for adenocarcinoma of the middistal esophagus five-year survival of a randomized clinical trial Ann Surg 2007 246992-1000 discussion 1000-1[11] Allum WH Stenning SP Bancewicz J Clark PI Langley RE Long-term results of a randomized trial of surgery with or without preoperative chemotherapy in esophageal cancer J Clin Oncol 2009 275062-5067[12] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[13] van Meerten E Muller K Tilanus HW et al Neoadjuvant concurrent chemoradiation with weekly paclitaxel and carboplatin for patients with oesophageal cancer a phase II study Br J Cancer 2006 941389-1394[14] Walsh TN Noonan N Hollywood D Kelly A Keeling N Hennessy TP A comparison of multimodal therapy and surgery for esophageal adenocarcinoma N Engl J Med 1996 335462-467

14


[15] Smit JK Guler S Beukema JC Mul VE Burgerhof JG Hospers GA Plukker JT Different recurrence pattern after neoadjuvant chemoradiotherapy compared to surgery alone in esophageal cancer patients Ann Surg Oncol 2013 204008-4015[16] Zanoni A Verlato G Giacopuzzi S et al Neoadjuvant concurrent chemoradiotherapy for locally advanced esophageal cancer in a single high-volume center Ann Surg Oncol 2013 201993-1999[17] Honing J Pavlov KV Mul VE et al CD44 SHH and SOX2 as novel biomarkers in esophageal cancer patients treated with neoadjuvant chemoradiotherapy Radiother Oncol 2015 117152-158[18] Berry MF Esophageal cancer staging system and guidelines for staging and treatment J Thorac Dis 2014 6 Suppl 3S289-97[19] Parry K Haverkamp L Bruijnen RC Siersema PD Offerhaus GJ Ruurda JP van Hillegersberg R Staging of adenocarcinoma of the gastroesophageal junction Eur J Surg Oncol 2016 42400-406[20] Oesofaguscarcinoom Landelijke richtlijn versie 31 2015 Available at httponcolinenlrichtlijndocindexphptype=pdaamprichtlijn_id=962 [21] American Joint Committee on Cancer AJCC cancer staging manual 7th ed New York Springer 2009[22] van Vliet EP Heijenbrok-Kal MH Hunink MG Kuipers EJ Siersema PD Staging investigations for oesophageal cancer a meta-analysis Br J Cancer 2008 98547-557[23] Puli SR Reddy JB Bechtold ML Antillon D Ibdah JA Antillon MR Staging accuracy of esophageal cancer by endoscopic ultrasound a meta-analysis and systematic review World J Gastroenterol 2008 141479-1490[24] Kelly S Harris KM Berry E et al A systematic review of the staging performance of endoscopic ultrasound in gastro-oesophageal carcinoma Gut 2001 49534-539[25] Westerterp M van Westreenen HL Deutekom M et al Patients perception of diagnostic tests in the preoperative assessment of esophageal cancer Patient Prefer Adherence 2008 2157-162[26] Lee WC Lee TH Jang JY et al Staging accuracy of endoscopic ultrasound performed by nonexpert endosonographers in patients with resectable esophageal squamous cell carcinoma is it possible Dis Esophagus 2015 28574-578[27] Wani S Cote GA Keswani R et al Learning curves for EUS by using cumulative sum analysis implications for American Society for Gastrointestinal Endoscopy recommendations for training Gastrointest Endosc 2013 77558-565[28] van Vliet EP Eijkemans MJ Poley JW Steyerberg EW Kuipers EJ Siersema PD Staging of esophageal carcinoma in a low-volume EUS center compared with reported results from high-volume centers Gastrointest Endosc 2006 63938-947

15

1


[29] Catalano MF Van Dam J Sivak MVJr Malignant esophageal strictures staging accuracy of endoscopic ultrasonography Gastrointest Endosc 1995 41535-539[30] Wallace MB Nietert PJ Earle C Krasna MJ Hawes RH Hoffman BJ Reed CE An analysis of multiple staging management strategies for carcinoma of the esophagus computed tomography endoscopic ultrasound positron emission tomography and thoracoscopylaparoscopy Ann Thorac Surg 2002 741026-1032[31] Schreurs LM Janssens AC Groen H et al Value of EUS in Determining Curative Resectability in Reference to CT and FDG PET The Optimal Sequence in Preoperative Staging of Esophageal Cancer Ann Surg Oncol 2016 231021-1028[32] Mortensen MB Edwin B Hunerbein M Liedman B Nielsen HO Hovendal C Impact of endoscopic ultrasonography (EUS) on surgical decision-making in upper gastrointestinal tract cancer an international multicenter study Surg Endosc 2007 21431-438[33] Findlay JM Bradley KM Maile EJ et al Pragmatic staging of oesophageal cancer using decision theory involving selective endoscopic ultrasonography PET and laparoscopy Br J Surg 2015 1021488-1499[34] Gines A Fernandez-Esparrach G Pellise M Llach-Osendino J Mata A Bordas JM Impact of endoscopic ultrasonography (EUS) and EUS-guided fine-needle aspiration (EUS-FNA) in the management of patients with esophageal cancer A critical review of the literature Gastroenterol Hepatol 2006 29314-319[35] Crabtree TD Kosinski AS Puri V et al Evaluation of the reliability of clinical staging of T2 N0 esophageal cancer a review of the Society of Thoracic Surgeons database Ann Thorac Surg 2013 96382-390[36] Samson P Puri V Robinson C et al Clinical T2N0 Esophageal Cancer Identifying Pretreatment Characteristics Associated With Pathologic Upstaging and the Potential Role for Induction Therapy Ann Thorac Surg 2016 1012102-2111[37] Stiles BM Mirza F Coppolino A Port JL Lee PC Paul S Altorki NK Clinical T2-T3N0M0 esophageal cancer the risk of node positive disease Ann Thorac Surg 2011 92491-6 discussion 496-8[38] Cho HJ Kim YH Kim HR Kim DK Park SI Kim JH Kim SB Oncologic Outcomes According to Remnant Lymph Node Metastases in Pathologic T0 (ypT0) Esophageal Squamous Cell Carcinoma Following Prospective Neoadjuvant Therapy and Surgery Ann Surg Oncol 2015 221851-1857[39] Blackham AU Yue B Almhanna K et al The prognostic value of residual nodal disease following neoadjuvant chemoradiation for esophageal cancer in patients with complete primary tumor response J Surg Oncol 2015 112597-602

16


[40] Kim MP Correa AM Lee J et al Pathologic T0N1 esophageal cancer after neoadjuvant therapy and surgery an orphan status Ann Thorac Surg 2010 90884-90 discussion 890-1[41] Wang Q Yu S Xiao Z et al Residual lymph node status is an independent prognostic factor in esophageal squamous cell Carcinoma with pathologic T0 after preoperative radiotherapy Radiat Oncol 2015 10142-015-0450-4[42] Chao YK Chen HS Wang BY Hsu PK Liu CC Wu SC Prognosis of Patients With Pathologic T0 N+ Esophageal Squamous Cell Carcinoma After Chemoradiotherapy and Surgical Resection Results From a Nationwide Study Ann Thorac Surg 2016 1011897-1902[43] Bollschweiler E Metzger R Drebber U Baldus S Vallbohmer D Kocher M Holscher AH Histological type of esophageal cancer might affect response to neo-adjuvant radiochemotherapy and subsequent prognosis Ann Oncol 2009 20231-238[44] Bollschweiler E Holscher AH Metzger R Histologic tumor type and the rate of complete response after neoadjuvant therapy for esophageal cancer Future Oncol 2010 625-35[45] Bruzzi JF Swisher SG Truong MT et al Detection of interval distant metastases clinical utility of integrated CT-PET imaging in patients with esophageal carcinoma after neoadjuvant therapy Cancer 2007 109125-134[46] Blom RL Schreurs WM Belgers HJ Oostenbrug LE Vliegen RF Sosef MN The value of post-neoadjuvant therapy PET-CT in the detection of interval metastases in esophageal carcinoma Eur J Surg Oncol 2011 37774-778[47] Cerfolio RJ Bryant AS Ohja B Bartolucci AA Eloubeidi MA The accuracy of endoscopic ultrasonography with fine-needle aspiration integrated positron emission tomography with computed tomography and computed tomography in restaging patients with esophageal cancer after neoadjuvant chemoradiotherapy J Thorac Cardiovasc Surg 2005 1291232-1241[48] Flamen P Van Cutsem E Lerut A et al Positron emission tomography for assessment of the response to induction radiochemotherapy in locally advanced oesophageal cancer Ann Oncol 2002 13361-368[49] Pultrum BB Honing J Smit JK et al A critical appraisal of circumferential resection margins in esophageal carcinoma Ann Surg Oncol 2010 17812-820[50] Mirnezami R Rohatgi A Sutcliffe RP Hamouda A Chandrakumaran K Botha A Mason RC Multivariate analysis of clinicopathological factors influencing survival following esophagectomy for cancer Int J Surg 2010 858-63[51] Rao VS Yeung MM Cooke J Salim E Jain PK Comparison of circumferential resection margin clearance criteria with survival after surgery for cancer of esophagus J Surg Oncol 2012 105745-749

17

1


[52] Deeter M Dorer R Kuppusamy MK Koehler RP Low DE Assessment of criteria and clinical significance of circumferential resection margins in esophageal cancer Arch Surg 2009 144618-624[53] Dexter SP Sue-Ling H McMahon MJ Quirke P Mapstone N Martin IG Circumferential resection margin involvement an independent predictor of survival following surgery for oesophageal cancer Gut 2001 48667-670[54] Griffiths EA Brummell Z Gorthi G Pritchard SA Welch IM The prognostic value of circumferential resection margin involvement in oesophageal malignancy Eur J Surg Oncol 2006 32413-419[55] Khan OA Fitzgerald JJ Soomro I Beggs FD Morgan WE Duffy JP Prognostic significance of circumferential resection margin involvement following oesophagectomy for cancer Br J Cancer 2003 881549-1552[56] Kwee RM Prediction of tumor response to neoadjuvant therapy in patients with esophageal cancer with use of 18F FDG PET a systematic review Radiology 2010 254707-717[57] van Rossum P Xu C Fried D Goense L Court L LinS The emerging field of radiomics in esophageal cancer current evidence and future potential Translational Cancer Research 2016 5[58] van Rossum PS Fried DV Zhang L et al The Incremental Value of Subjective and Quantitative Assessment of 18F-FDG PET for the Prediction of Pathologic Complete Response to Preoperative Chemoradiotherapy in Esophageal Cancer J Nucl Med 2016 57691-700[59] Yip SS Coroller TP Sanford NN Mamon H Aerts HJ Berbeco RI Relationship between the Temporal Changes in Positron-Emission-Tomography-Imaging-Based Textural Features and Pathologic Response and Survival in Esophageal Cancer Patients Front Oncol 2016 672[60] Hatt M Majdoub M Vallieres M et al 18F-FDG PET uptake characterization through texture analysis investigating the complementary nature of heterogeneity and functional tumor volume in a multi-cancer site patient cohort J Nucl Med 2015 5638-44[61] Tixier F Le Rest CC Hatt M et al Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitant radiochemotherapy in esophageal cancer J Nucl Med 2011 52369-378[62] Tixier F Hatt M Le Rest CC Le Pogam A Corcos L Visvikis D Reproducibility of tumor uptake heterogeneity characterization through textural feature analysis in 18F-FDG PET J Nucl Med 2012 53693-700[63] Hatt M Tixier F Cheze Le Rest C Pradier O Visvikis D Robustness of intratumour (1)(8)F-FDG PET uptake heterogeneity quantification for therapy response prediction in oesophageal carcinoma Eur J Nucl Med Mol Imaging 2013 401662-1671

18


[64] Tan S Zhang H Zhang Y Chen W DSouza WD Lu W Predicting pathologic tumor response to chemoradiotherapy with histogram distances characterizing longitudinal changes in 18F-FDG uptake patterns Med Phys 2013 40101707[65] van Rossum PS van Hillegersberg R Lever FM et al Imaging strategies in the management of oesophageal cancer whats the role of MRI Eur Radiol 2013 231753-1765 [66] van Rossum PS van Lier AL Lips IM et al Imaging of oesophageal cancer with FDG PETCT and MRI Clin Radiol 2015 7081-95[67] van Rossum PS van Lier AL van Vulpen M et al Diffusion-weighted mag-netic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer Radiother Oncol 2015 115163-170[68] De Cobelli F Giganti F Orsenigo E et al Apparent diffusion coefficient mo-difications in assessing gastro-oesophageal cancer response to neoadjuvant treat-ment comparison with tumour regression grade at histology Eur Radiol 2013 232165-2174[69] Kwee RM Dik AK Sosef MN et al Interobserver reproducibility of dif-fusion-weighted MRI in monitoring tumor response to neoadjuvant therapy in esophageal cancer PLoS One 2014 9e92211

19

1



Improving staging and neoadjuvant chemoradiotherapy outcome in esophageal cancer patients

PART I



Impact of endoscopic ultrasonography on 18F-FDG PETCT upfront towards patient

specific esophageal cancer treatment

Jan Binne Hulshoff Veronique EM Mul Hiske EM de Boer

Walter Noordzij Tijmen Korteweg

Hendrik M van Dullemen Wouter B Nagengast

Vera Oppedijk Jean-Pierre EN Pierie

John ThM Plukker

Annals of Surgical Oncology 2017 March 16

CHAPTER 2


ABSTRACT

IntroductionIn patients with potentially resectable esophageal cancer (EC) the value of endoscopic ultrasonography (EUS) after fluorine-18 labeled fluorodeoxyglucose positron emission tomography with computed tomography (18F-FDG PETCT) is questionable Retrospectively we assessed the impact of EUS after PETCT on the given treatment in EC patients

MethodsDuring the period 2009-2015 318 EC patients were staged as T1-4aN0-3M0 with hybrid 18F-FDG PETCT or 18F-FDG PET with CT and EUS if applicable in a non-specific order We determined the impact of EUS on the given treatment in 279 patients who were also staged with EUS EUS had clinical consequences if it changed curability extent of radiation fields or lymph node resection (AJCC stations 2-5) and when the performed fine-needle aspiration (FNA) provided conclusive information of suspicious lymph node

ResultsEUS had an impact in 80 (287) patients it changed the radiation field in 63 (226) curability in 5 (18) lymphadenectomy in 48 (172) and FNA was additional in 21 (75) In patients treated with nCRT (n=194) EUS influenced treatment in 53 (273) patients in 38 (196) the radiation field changed in 3 (15) the curability in 35 (180) the lymphadenectomy and in 17 (88) FNA was additional EUS influenced both the extent of radiation field and nodal resection in 31 (160) nCRT patients

ConclusionEUS had an impact on the given treatment in approximately 29 In most patients the magnitude of EUS found expression in the extent of radiotherapy target volume delineation to upperhigh mediastinal lymph nodes

24


BACKGROUND

Accurate staging is essential for oncological outcome in patients with potentially curable esophageal cancer for determining the optimal treatment approach and adequate target volume delineation (gross tumor volume of primary tumor and malignant nodes and hence clinical target volume) when radiotherapy is indicated Fluorine-18 labeled fluorodeoxyglucose positron emission tomography with computed tomography (18F-FDG PETCT) is currently the best method to detect distant metastases whereas endoscopic ultrasonography (EUS) with or without fine-needle aspiration (FNA) is the method of choice to determine the extent and depth-growth of the primary tumor (T-stage) and detection of lymph node metastases (N-stage) [1-3] Contradictory results have been reported regarding the additional role of EUS after 18F-FDG PET 18F-FDG PETCT or CT alone [4-8] Mortensen et al found that EUS influenced treatment decisions in 34 of the EC patients after CT alone whereas Findlay et al found that the risks outweighed the potential benefit of EUS in patients with T2-4a disease on CT [4 5] Some studies showed upfront 18F-FDG PET followed by EUS as the best predictor of curative resectability and the most cost-effective staging sequence [7 8] However most studies determined the influence of EUS with questionnaires completed by involved medical specialists and not with the clinical impact or diagnostic position of EUS after 18F-FDG PETCT [4 6] Moreover these studies did not assess the magnitude of EUS on radiotherapy target volume delineation nowadays most patients receive radiotherapy in a combined curative treatment approachIn addition EUS has several limitations including the invasive character of the procedure with risk of aspiration perforation and bleeding whereas tumor stenosis which occurs between 20-36 of the patients limits its clinical usage [3 9-11] Precluding unnecessary EUS is therefore patient-friendly in case of a strong suspicion of distant metastases or in case of localized disease on 18F-FDG PETCT and reduces the risks of complications and eventually costs With a 18F-FDG PETCT upfront staging sequence unnecessary primary endoscopic procedures may be prevented by 18F-FDG PETCT guiding of concurrent EUS with fine needle aspiration (FNA) of pathological (non)regional lymph nodes The purpose of this study was to determine the clinical impact of EUS after upfront 18F-FDG PETCT staging on the given treatment by determining if EUS had changed radiotherapy target volume delineation curability extent of lymph node resection and whether FNA was of additive value We also investigated the additional value of EUS on nodal updownstaging on the number of lymph nodes suspected for metastasis and station-specific nodal (N) status

25

2


PATIENTS AND METHODS

PatientsA total of 318 patients were included in this retrospective study which was performed according to the guidelines of the local Ethical Boards and national rules for retrospective studies All esophageal cancer patients (n=388) scheduled for an in onset curative intended treatment (T1-4aN0-3M0) diagnosed at the University Medical Center Groningen (UMCG) or Medical Center Leeuwarden between 2009 and 2015 were eligible for inclusion All consecutive patients had a pathologically proven adeno- or squamous cell carcinoma located in the mid or distal esophagus or at the gastroesophageal junction Excluded were patients treated with endoscopic mucosal resection before (n=7) or after (n=3) staging without 18F-FDG PET or 18F-FDG PETCT scanning (n=17) without a visible tumor on 18F-FDG PETCT (n=15) or with missing staging data (n=28)

MethodsFirst we determined the number of patients who had a complete or incomplete EUS and those without EUS To answer the primary research question all patients with a complete or incomplete EUS were selected We then assessed if EUS had an impact on the given treatment Compared with the non-invasive 18F-FDG PETCT EUS had influenced the given treatment if EUS changed the extent of radiation target volumes (smaller larger or both) changed the curability (from incurable based on 18F-FDG PETCT to potentially curable) influenced the extent of lymph node dissection (ie changed suspicion for lymph node metastasis at AJCC lymph node stations 2-5) orand when the performed fine-needle aspiration (FNA) provided conclusive information of suspicious lymph node [12] Thereafter we assessed the clinical impact of EUS on the given treatment in the total patient group (all treatments with curative intent) followed by the neoadjuvant chemoradiotherapy (nCRT) the surgery-alone and a combined (chemoradiotherapy and definitive radiotherapy (dCRTdRT) group The researchers (JBH amp VEMM) assessed the radiotherapy tumor volume (TV) delineations and determined if the EUS findings had changed the radiotherapy target volumes EUS enlarged the radiotherapy target volume delineation if EUS-FNA had identified new lymph node suspicious for metastases that were located outside of the clinical target volume (CTV treatment section) based on the 18F-FDG PETCT Lymph nodes were scored by size shape echoic pattern and sharpness of the border EUS decreased the radiotherapy target volume delineation if initially suspected lymph nodes on the 18F-FDG PETCT located outside of the CTV were reassessed as negative based on EUS findings andor confirmed cytologically as FNA negative

26


Secondary outcomes included the number of patients with an altered N-stage with a difference in the number of lymph nodes suspected for metastases and changes in the localization of suspected lymph nodes

Staging Patients were staged according to the 7th TNM classification of the American Joint Committee on Cancer (AJCC) with diagnostic gastroscopy and biopsy a diagnostic CT thoraxabdomen 18F-FDG PET or 18F-FDG PETCT and EUS if possible [13] Depending on the tumor-specific situation EUS was performed with either a high-frequency (12-20 MHz) linear or radial probes Esophageal dilatation was not performed routinely in patients with stenosis The results of all staging methods were discussed in a multidisciplinary tumor board Abnormalities and non-regional lesions relatively suspect for being a distant metastasis were either proven pathologically under imaging guided biopsies or assessed with additional imaging techniques ie magnetic resonance imaging (MRI) or endobronchial ultrasoundFNA

TreatmentnCRT followed by surgery consisted of carboplatin and paclitaxel according to the CROSS regimen combined with radiotherapy (414 Gy23 fractions) [14] dCRT consisted of either carboplatinpaclitaxel or cisplatin and fluorouracil (Cis-5FU) combined with radiotherapy (504 Gy) [15] dRT commonly consisted of 60 Gy in 30 fractions In all patients who received external beam radiotherapy the gross tumor volume (GTV) was delineated on a planning CT scan using all additional staging information The GTV contained both the primary tumor and adjacent lymph nodes The clinical target volume (CTV) is a margin of 3-35 cm in cranial and caudal direction and 1-2 cm transversally The CTV was adjusted to normal tissue Suspicious lymph nodes located outside the CTV were radiated separately with 1 cm margin Surgery consisted of either a transthoracic transhiatal or minimally invasive radical esophagectomy with a two-field lymph node dissection and was normally performed 6-10 weeks after the end of nCRT

Statistical analysisAll data were displayed as numbers (percentages) Normally distributed data were displayed as mean (standard deviation) non-normally distributed data were displayed as median [interquartile range (IQR)] Differences in categorical variables were assessed using logistic regression Chi square test or likelihood ratio All data were analyzed using SPSS statistical software version 22 (SPSS inc Chigago IL USA)

27

2


RESULTS

Table 1 displays the patientsrsquo characteristics tumor-related and treatment-related data of the 318 included patients Of these patients 250 (786) had an adenocarcinoma and 68 (214) a squamous cell carcinoma The localization of the primary tumor was generally in the distal esophagus (n=221 695) followed by the gastroesophageal junction (n=52 164) and mid esophagus (n=45 142) Most patients were treated with curative intent with surgery after nCRT (n=212 667) followed by surgery-alone (n=50 157) dCRT (n=29 91) and dRT (n=27 85)Table 2 displays the endoscopic ultrasonography-related data EUS was not performed in 39 (123) patients because of stenosis (n=3094) patient related reasons (n=4) and for unknown reasons (n=5) In total EUS could be performed in 279 patients 200 (629) had a complete and 79 (248) an incomplete EUS because of stenosis (n=78245) and patients related reason (n=1)

Influence of EUS on the given treatmentTable 3 displays the additional information that EUS provided compared with upfront 18F-FDG PETCT EUS had influenced the primary treatment in 80 (287) patients In most patients (n=63 226) EUS influenced the extent of radiation target volumes by increasing the radiotherapy target volume delineation in 45 (161) patients decreasing these delineation in 17 (61) and both extending and decreasing the target volume in 1 (04) patient EUS had changed the curability (incurable to potentially curable) in 5 (18) patients influenced the extent of LN resection (at node level AJCC Lymph node stations 2-5) in 48 (172) and affected treatment decision-making with FNA in 21 patients (75) Thereafter we found that histologic tumor type did not influence the effect of EUS on the given treatment (P=0614) The location of the primary tumor also influenced the effect of EUS on the given treatment (P=0043) midesophageal-located cancers interfered with the treatment in 16 (471) distal EC in 50 (253) and gastroesophageal junction tumors in 14 (298) patients

28


Table 1 Patient tumor and treatment characteristicsall patients (n=318) n ()

Patients with a complete EUS (n=279) n ()

Male 248 (780) 216 (774)Age (year) median (IQR) 65 (59-70) 65 (60-70)Histology

Adenocarcinoma 250 (786) 225 (806)Squamous cell carcinoma 68 (214) 54 (194)

Tumor location Middle esophagus 45 (142) 34 (122)Distal esophagus 221 (695) 198 (710)GEJ 52 (164) 47 (168)

Tumor length (cm) median (IQR) 50 (30-70) 50 (30-60)cT-stage

T1 11 (35) 10 (36)T2 41 (129) 39 (140)T3 243 (764) 218 (781)T4a 23 (72) 12 (43)

cN-stageN0 85 (267) 70 (251)N1 117 (368) 100 (358)N2 99 (311) 93 (333)N3 15 (47) 15 (54)Missing 2 (06) 1 (04)

TreatmentSurgery-alone 50 (157) 45 (161)nCRT 212 (667) 194 (695)dCRT 29 (91) 21 (75)dRT 27 (85) 19 (68)

Abbreviations IQR = interquartile range GEJ= Gastroesophageal junction cT-stage = clinical T-stage cN-stage = clinical N-stage nCRT = neoadjuvant chemoradiotherapy dCRT = definitive chemoradiotherapy and dRT = definitive radiotherapy

29

2


Table 2 Endoscopic ultrasonography related dataall patients (n=318) n ()

EUS successful 200 (629)EUS incomplete 79 (248)

Stenosis 78 (245)Patient-related 1 (03)

EUS not performed 39 (123)Stenosis 30 (94)Patient-related 4 (13)On indication no EUS 1 (03)Unknown 4 (13)

In the nCRT group (n=194) EUS changed the proposed treatment based on upfront 18F-FDG PET-CT in 53 (273) In 52 (268) patients it influenced the radiation target volumes in 3 (15) the curability in 35 (180) the extent of lymph node dissection and in 17 (88) FNA added valuable information Of all patients in 31 (160) EUS affected the extent of both the radiation target volumes and the lymph node dissection usually at AJCC lymph node 2-5 implicating EUS to be valuable in identifying mediastinal lymph node metastases especially these upperhigh lymph nodesEUS influenced the treatment in 13 (289) surgery-alone treated patients (n=45) all by extending the lymph node dissection EUS identified mediastinalhigh lymph node metastasis in 9 (200) patients with 18F-FDG PETCT positive lymph nodes while in 4 (89) other patients the 18F-FDG PETCT positive lymph nodes turned out to be negative on EUS In 3 (67) of these 13 patients FNA was performed and conclusive EUS influenced the treatment in 14 (350) patients treated with dCRT or dRT in 11 (275) it changed the radiation target volumes in 2 (50) the curability and in 1 (25) patient FNA was additional In most of these patients EUS enlarged the radiation target volumes 7 (175) while it was decreased in 4 (100) patients

Additional value of EUSAdditional lymph nodes were detected in 150 (538) patients while less suspicious lymph nodes were found in 32 (115) EUS caused N upstaging in 107 (384) and N downstaging in 31 (111) patients In 77 (276) patients EUS found suspicious lymph nodes at other nodal stations than with 18F-FDG PETCT In total 7 (25) patients received a re-EUS examination after primary EUS because of suspect lymph nodes on the 18F-FDG PET or CT (n=3) an unclear outcome of the FNA (n=3) or an unclear outcome of the EUS itself (n=1)

30


Table 3 Influence of EUS compared to 18F-FDG PETCT all patients (n=279) n ()

nCRT (n=194) n ()

Surgery-alone (n=45) n ()

dCRT amp dRT (n=40) n ()

EUS influenced the treatment 80 (287) 53 (273) 13 (289) 14 (350)- Radiotherapy fields 63 (226) 52 (268) na 11 (275)

Larger RT field 45 (161) 38 (196) na 7 (175)Smaller RT field 17 (61) 13 (67) na 4 (100)Smaller and Larger 1 (04) 1 (05) na 0 (00)

- Incurable curable 5 (18) 3 (15) 0 (00) 2 (50)

- Influence on LN resection 48 (172) 35 (180) 13 (289) naPositive on EUS 29 (104) 20 (103) 9 (200) naNegative on EUS 19 (68) 15 (77) 4 (89) na

- FNA additional 21 (75) 17 (88) 3 (67) 1 (25)Fine-needle aspiration

Positive 22 (79) 14 (72) 1 (22) 7 (175)Negative 31 (111) 27 (139) 4 (89) 0 (00)Inconclusive 5 (18) 2 (10) 2 (44) 1 (25)

N updown staging up staging 107 (384) 87 (448) 12 (267) 8 (200)down staging 31 (111) 15 (77) 9 (200) 7 (175)

More LN found with EUS 150 (538) 113 (582) 18 (400) 19 (475)Less LN found with EUS 32 (115) 16 (82) 9 (200) 7 (175)Other LN localization 77 (276) 52 (268) 11 (244) 14 (350)Additional staging techniques

Bronchoscopy 8 (29) 2 (10) 0 (00) 6 (150)EBUS 6 (22) 4 (21) 1 (22) 1 (25)Ultrasound liver 11 (39) 6 (31) 4 (89) 1 (25)Ultrasound neck 21 (75) 16 (82) 2 (44) 3 (75)MRI liver 1 (04) 1 (05) 0 (00) 0 (00)CT liver 1 (04) 0 (00) 0 (00) 1 (25)CT thorax 2 (07) 2 (10) 0 (00) 0 (00)CT brain 1 (04) 1 (05) 0 (00) 0 (00)Mediastinoscopy 1 (04) 0 (00) 1 (22) 0 (00)

Number of EUS1 272 (975) 191 (985) 45 (100) 36 (900)2 7 (25) 3 (15) 0 (00) 4 (100)

Abbreviations nCRT = neoadjuvant chemoradiotherapy dCRT = definitive chemoradiotherapy dRT = definitive radiotherapy EUS = endoscopic ultrasonography RT = radiotherapy LN = lymph node FNA = fine-needle aspiration EBUS = endo bronchial ultra sound and na = not applicable

31

2


DISCUSSION

Accurate staging is of vital importance in the treatment decision-making process and radiotherapy target volume delineation and hence the adjusted correct treatment The value of EUS after an upfront 18F-FDG PETCT scan remains matter of debate Staging with EUS seems required since it has the highest sensitivity and accuracy in detecting lymph nodes suspicious for metastatic disease whereas high regional recurrence rates also suggested the need for a more accurate radiotherapy delineation of involved lymph node stations In clinical practice not all patients benefit from EUS because of severe stenosis in about 20-36 [11] The present study is the first to determine the influence of EUS on the given treatment with curative intent We found that EUS influenced treatment in approximately 29 of the patients Moreover EUS was especially important for adequate delineation of radiotherapy target volumes implicating its importance in locoregional control and treatment of esophageal cancer patients with nCRTSeveral studies previously determined the influence of EUS on treatment decision-making but not on the given treatment itself Two studies determined the influence of EUS after CT-alone with a questionnaire and found that EUS changed the treatment decision in 24-34 of the patients [4 6] A study by Van Zoonen determined that although EUS increased the specificity after CT and ultrasound of the neck the additional value on determining the surgical resectability was limited [16] Moreover Findlay et al found that the risk of EUS in T2-4a patients on CT outweighed the additional information of EUS on T- and N-stage [5] The only study that compared EUS with 18F-FDG PETCT was the study by Schreurs et al which found that 18F-FDG PETCT was the best predictor of curability of the resection [8] However as mentioned the studies above only determined the influence of EUS on treatment decision making and not on the given treatment itself whereas detection of metastatic lymph nodes located at some distance of the primary tumor is of fast importance for radiotherapy delineation in the era of nCRT Currently the role of EUS in treatment decision making might be limited In the current treatment paradigm treatment decision making in T2-4aN0-3M0 esophageal cancer patients is most commonly based on the presence of comorbidities that might not permit nCRT or dCRT As mentioned EUS is especially important in patients eligible for nCRT dCRT or dRT Refraining from EUS would have caused inadequate radiotherapy target volumes in approximately 196 of the nCRT and 175 of the dCRT and dRT patients In patients eligible for treatment with dCRT and dRT group this might lead to impaired locoregional control of disease On the other hand refraining from EUS would have caused an unnecessary large radiotherapy target volumes in approximately 7 of the patients treated with nCRT and 10 of the dCRT and dRT which could increase the risk of radiotherapy induced toxicities

32


In the future diffusion-weighted magnetic resonance imaging (DW-MRI) with contrast agents that are taken up by lymph nodes ie Gadofosveset or ultra-small iron particles might improve the detection of suspicious lymph nodes [17] Moreover in the coming years proton therapy will become more clinically available which will decrease the radiation associated (cardiopulmonary) toxicities [18] An important limitation of this study is its retrospective character which sometimes impedes the exact anatomical information about suspected lymph node metastases although all lymph node metastases were scored according to the AJCC system by the researchers The rate of FNA procedures was rather low in present study (n=58 208) which may have resulted in false positive EUS results and probably unnecessary upstaging because the accuracy of EUS for predicting lymph nodes on echo features in different cancers is approximately 80 [19] A large prospective study with well-defined FNA of suspected lymph nodes might determine the exact role of EUS in up- and downstaging although accurate EUS with FNA is time-consuming and should be performed in centralized institute because of its relative complexity In improving the detection of involved regions with potential harmless imaging modalities the recent use of DW-MRI seems to be of great importanceIn conclusion our study determined the additional value of EUS on the given treatment and found that EUS was of added value in approximately 29 EUS seems especially important for radiotherapy target volume delineation of mediastinal and high mediastinal lymph nodes metastases implicating its importance in staging esophageal cancer patients planned to be treated with curative intended nCRT Future prospective studies with current sophisticated imaging would determine the exact place of EUS-FNA in EC staging

33

2


REFERENCES

[1] Bruzzi JF Munden RF Truong MT et al PETCT of esophageal cancer its role in clinical management Radiographics 2007 271635-1652[2] van Vliet EP Heijenbrok-Kal MH Hunink MG Kuipers EJ Siersema PD Staging investigations for oesophageal cancer a meta-analysis Br J Cancer 2008 98547-557[3] Kelly S Harris KM Berry E et al A systematic review of the staging performance of endoscopic ultrasound in gastro-oesophageal carcinoma Gut 2001 49534-539[4] Mortensen MB Edwin B Hunerbein M Liedman B Nielsen HO Hovendal C Impact of endoscopic ultrasonography (EUS) on surgical decision-making in upper gastrointestinal tract cancer an international multicenter study Surg Endosc 2007 21431-438[5] Findlay JM Bradley KM Maile EJ et al Pragmatic staging of oesophageal cancer using decision theory involving selective endoscopic ultrasonography PET and laparoscopy Br J Surg 2015 1021488-1499[6] Gines A Fernandez-Esparrach G Pellise M Llach-Osendino J Mata A Bordas JM Impact of endoscopic ultrasonography (EUS) and EUS-guided fine-needle aspiration (EUS-FNA) in the management of patients with esophageal cancer A critical review of the literature Gastroenterol Hepatol 2006 29314-319[7] Wallace MB Nietert PJ Earle C Krasna MJ Hawes RH Hoffman BJ Reed CE An analysis of multiple staging management strategies for carcinoma of the esophagus computed tomography endoscopic ultrasound positron emission tomography and thoracoscopylaparoscopy Ann Thorac Surg 2002 741026-1032[8] Schreurs LM Janssens AC Groen H et al Value of EUS in Determining Curative Resectability in Reference to CT and FDG PET The Optimal Sequence in Preoperative Staging of Esophageal Cancer Ann Surg Oncol 2016 231021-1028[9] Westerterp M van Westreenen HL Deutekom M et al Patients perception of diagnostic tests in the preoperative assessment of esophageal cancer Patient Prefer Adherence 2008 2157-162[10] Heeren PA van Westreenen HL Geersing GJ van Dullemen HM Plukker JT Influence of tumor characteristics on the accuracy of endoscopic ultrasonography in staging cancer of the esophagus and esophagogastric junction Endoscopy 2004 36966-971[11] Catalano MF Van Dam J Sivak MV Jr M alignant esophageal strictures staging accuracy of endoscopic ultrasonography Gastrointest Endosc 1995 41535-539

34


[12] Edge SB AJCC Cancer Staging Handbook 2010[13] American Joint Committee on Cancer AJCC cancer staging manual 7th ed New York Springer 2009[14] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[15] Honing J Smit JK Muijs CT et al A comparison of carboplatin and paclitaxel with cisplatinum and 5-fluorouracil in definitive chemoradiation in esophageal cancer patients Ann Oncol 2014 25638-643[16] van Zoonen M van Oijen MG van Leeuwen MS van Hillegersberg R Siersema PD Vleggaar FP Low impact of staging EUS for determining surgical resectability in esophageal cancer Surg Endosc 2012 262828-2834[17] Heijnen LA Lambregts DM Martens MH et al Performance of gadofosveset-enhanced MRI for staging rectal cancer nodes can the initial promising results be reproduced Eur Radiol 2014 24371-379[18] Verma V Lin SH Simone CB2nd Mehta MP Clinical outcomes and toxicities of proton radiotherapy for gastrointestinal neoplasms a systematic review J Gastrointest Oncol 2016 7644-664[19] Chen VK Eloubeidi MA Endoscopic ultrasound-guided fine needle aspiration is superior to lymph node echofeatures a prospective evaluation of mediastinal and peri-intestinal lymphadenopathy Am J Gastroenterol 2004 99628-633

35

2



Impact of pretreatment nodal staging on response evaluation to neoadjuvant

chemoradiotherapy and prognosis in esophageal cancer

Willemieke PM Dijksterhuis Jan Binne Hulshoff

Hendrik M van Dullemen Gursah Kats-Ugurlu

Johannes GM Burgerhof Tijmen Korteweg

Veronique EM Mul Geke AP Hospers John ThM Plukker

Submitted to the Journal of Surgical Oncology

CHAPTER 3


ABSTRACT

Background and objectivesClinical lymph node (cN) staging in esophageal cancer (EC) remains difficult We evaluated the reliability of pretreatment nodal staging and the effect on disease-free survival (DFS) after neoadjuvant chemoradiotherapy (nCRT) and esophagectomy related to primary surgery

MethodsThree-hundred-ninety-five EC patients who underwent curative intended surgery with or without nCRT between 2000 and 2015 were included A surgery-alone and nCRT group (each n = 135) were formed by propensity score matching on clinical tumor and nodal stage and histopathology Pretreatment staging consisted of PET and CT or PET-CT and EUS Clinical and pathological N-stage (pN) was scored as correct (cN=pN) downstaged (cNgtpN) or upstaged (cNltpN) Prognostic impact on 5-year DFS was assessed with multivariate Cox regression analysis (factors with P value lt 01 on univariate analysis)

ResultsThe surgery-alone and nCRT group differed in correct (319 vs 281) nodal up (430 vs 163) and downstaging (252 vs 556) respectively (P lt 0001) Nodal up-staging was common in cT3T4a tumors and adenocarcinomas Prognostic factors for DFS were pN (P = 0002) and lymph-angioinvasion (P = 0016) in the surgery-alone vs upper abdominal cN metastases (P = 0012) and lymph-node ratio (P = 0034) in the nCRT group

ConclusionsIncorrect lymph node staging is common and might impede determination of response and prognosis after nCRT

38


INTRODUCTION

Accurate pretreatment staging of the primary tumor and lymph nodes (LNs) is crucial for proper treatment decision-making prediction of response to neoadjuvant therapy and for prognostication in patients with esophageal cancer (EC) [1-4] Clinical staging commonly consists of endoscopic ultrasonography (EUS) with fine needle aspiration (FNA) multi-row detector computed tomography (CT) and 18-F-fluorodeoxyglucose positron emission tomography (18F-FDG-PET) or integrated PET-CT [5 6] In patients treated with surgery-alone ldquoinadequaterdquo clinical nodal staging is high which may pursue after neoadjuvant chemoradiotherapy (nCRT) [6-11] The main effect of nCRT is downstaging of the primary tumor and metastatic LNs with nodal downstaging (clinical (c)Nstagegt pathological (p)Nstage) being present in 45-69 of the patients treated with nCRT [12 13] However due to underestimation of nodal involvement (30- 63) in patients treated with surgery-alone a clinical ldquonodal downstaging effectrdquo is commonly reported as well [7 9 10] This implicates that the amount of patients with lymph node response after nCRT is overestimated as pathologically negative nodes after nCRT include both patients initially staged as node negative (cN0=pN0) and sterilized nodes (cN+ to pN0) On the other hand patients falsely assessed as clinically node negative (cN0=pN+) might not be treated with nCRT or might receive an inadequate radiotherapy dosage on metastatic lymph nodes Furthermore the impact of site-specific localized LN metastases on survival remains unclear In surgery-alone patients with proven LN metastases on both diaphragmatic sides Talsma et al found a detrimental survival [3] However others failed to prove prognostic importance of cN-positive localization in EC patients after nCRT [13] Aim of this study was to evaluate the rate of nodal up- and downstaging in EC patients treated with primary surgery and with nCRT followed by surgery and its presuming effect on pathologic response to nCRT In addition we determined the prognostic impact of LN localization and nodal up- and downstaging on the 5-year disease-free survival

MATERIALS AND METHODS

Study populationData of patients with locally advanced adenocarcinoma (AC) and squamous cell carcinoma (SCC) of the esophagus (cT2-4aN0-3M0cT1N1-3M0) who underwent esophagectomy with curative intent between 2000 and 2015 (n = 419) were collected from a prospectively managed database From 2004-2009 nCRT was given in the randomized controlled CROSS trial and as standard in the same patientsrsquo categories from 2009 onwards Excluded were patients with concurrent

39

3


malignancies (n = 5) missing data (n = 12) salvage surgery or preceding endoscopic mucosal resection (n = 7) Of the remaining 395 patients 222 were treated with surgery-alone and 173 with nCRT and surgery After propensity score matching on clinical T-stage (cT) N-stage (cN) and histopathologic tumor type both groups consisted of 135 patients (Table 1) This study was performed according to the National Health Care guidelines with approval of our Institutional Ethical Board

Clinical stagingAll patients were clinically staged before treatment by 16-64 CT thorax-abdomen (2mm slices) and PET or PET-CT with EUS FNA by two experienced GI-endoscopists in our high-volume institute EUS was performed with (n = 48178) or without FNA (n = 222822) and endobronchoscopic ultrasonography (EBUS n = 5 19) with FNA on indication Before 2009 patients received a PET scan (n = 145 537) thereafter patients received an integrated PET-CT scan (n = 125 463) Patients were staged according to the 7th TNM system of the American Joint Committee on Cancer (AJCC) [14] LNs were considered clinically positive (cN+) if highly suspected (ge1cm on short axis) on CT EUSEBUS if FNA proven or with increased FDG-uptake on PET-CT

Treatment and pathologyPatients underwent a transthoracic open or minimally invasive esophagectomy with standard 2-field dissection of mediastinal para-esophageal and upper abdominal (along splenic common hepatic celiac artery and perigastric) lymph nodes by 3 experienced surgeons High paratracheal nodes (AJCC LN station 2) were usually dissected when indicated [14] Neoadjuvant chemoradiotherapy consisted of carboplatin (area under curve of 2 mgmin) and paclitaxel (50 mgm2) and 414 Gy 23 fractions of 18 Gy (the CROSS regimen) for five weeks followed by surgical resection within 6-10 weeks [15] All patients received full radiotherapy doses and over 75 completed nCRT while 23 had 4 of the 5 cycles of chemotherapy Pathological examination was performed by two experienced upper-GI pathologists using our standard pathologic protocol [16] Tumor response was scored according to Mandard tumor regression grade (TRG) ranging from pathologic complete response (TRG 1) to non-response (TRG 5) [17] None of the patients received adjuvant therapy

Clinical versus pathological nodal stagingPretreatment ie clinical nodal staging was correlated with pathological staging Nodal staging was scored as either correct in node negative (cN0=pN0) and node positive (cN1-3=pN1-3) patients as nodal downstaging (cNgtpN) or as upstaging

40


(cNltpN) The outcome of clinical nodal staging in the surgery-alone group was compared with the outcome of clinical nodal staging in the nCRT group In the surgery-alone group the accuracy sensitivity specificity positive and negative predictive value of clinical staging were determined These analyses could not be performed after nCRT because of genuine downstaging effect by sterilizing of nCRT We assumed the hypothetic impact of nodal response to nCRT by comparing the clinical and pathologic nodal classification after nCRT to the findings after primary surgery Nodal response cannot be assumed adequately if cN0-status altered to pN+ while cN0 to pN0 evaluation may be based on incorrect assessment as cN0 (false treatment response) or on potential sterilized effect of nCRT (true treatment response) In both groups nodal misstaging was based on up- and downstaging and assessed by histologic type and clinical T-stage

Site-specific LN metastases and prognostic value on DFSLN regions were marked prospectively according to a standard pathologic protocol including the AJCC node map [14] We determined the effect of nodal up-downstaging in patients (surgery-alone n = 124 nCRT n = 122) with adequately recorded suspected clinical lymph node locations Lymph nodes were scored in 3 regions upper mediastinal (paratrachealpara-aortic station 2-6) low mediastinal (para-esophagealsubcarinalpulmonary ligament station 7-9 and 15) and upper abdominal (station 16-20) Thereafter we determined the prognostic value of involvement of these nodal locations of DFS Potential prognostic factors in the univariate analyses included clinical and pathological supra- (mediastinal and para-esophageal) and sub-diaphragmatic LNs and the variable nodal up- and downstaging

Follow-upPatients received follow-up every 3 4 and 6 months during the first second and third year respectively and yearly thereafter Recurrences were determined with radiological imaging endoscopy andor histocytologic examination DFS was measured from the date of treatment until recurrence or end of follow-up

Statistical analysisChi-square test and likelihood ratio were used to determine differences in LN involvement and location DFS was displayed with Kaplan-Meier curves Factors with P value lt 010 on univariate regression analysis were included in a multivariate Cox regression analysis for DFS with P value lt 005 as significant Statistical analyses were performed with IBM SPSS Statistics for Windows Version 230 (Armonk NY IBM Corp)

41

3


RESULTS

Patients and tumor characteristicsAfter propensity score matching (n = 270) the groups differed in age (P = 0013) pT-stage and pN-stage (P lt 0001) lymph-angioinvasion (P = 0003) and LN ratio gt02 (P lt 0001) (Table 1) The median (IQR) number of resected LNs was higher in the nCRT group (15 (IQR 120-220) compared to 14 (IQR 120-188) after surgery-alone Pathologic ResponseComplete pathologic response of the primary tumor (ypT0) after nCRT was 207 (n = 28) while pathologic complete response (pCR ypT0N0) including nodal response was seen in 156 (n = 21) A significant higher pCR rate was seen in SCC (822 (364) vs 13113 (12) with AC P = 0007)

Table 1 Patient and tumor related characteristics in the surgery-alone vs nCRT group Surgery-alone(n = 135) n ()

nCRT(n = 135) n ()

P value

Male 114 (844) 106 (785) P = 0210a

Age in years median (IQR) 65 (57-71) 63 (57-68) P = 0013b

Histology P = 0737a


Tumor location P = 0988a

Middle esophagus 13 (96) 13 (96)Distal esophagus 96 (711) 97 (719)GEJ 26 (193) 25 (185)

Tumor length (cm) median (IQR) 5 (3-7) 5 (3-7) P = 0786b

cT-stage P = 0527c

T1 3 (22) 1 (07)T2 18 (133) 21 (156)T3ampT4a 114 (844) 113 (837)

cN-stage P = 0984c

N0 36 (267) 36 (267)N1 67 (496) 67 (496)N2 28 (207) 29 (215)N3 4 (30) 3 (22)

42


pT-stage P lt 0001a

pCR (ypT0N0) - 21 (156)T0 0 (00) 28 (207)T1 11 (81) 20 (148)T2 27 (200) 24 (178)T3ampT4 97 (718) 63 (467)

pN-stage P lt 0001a

N0 37 (274) 92 (681)N1 41 (304) 29 (215)N2 31 (230) 11 (81)N3 26 (193) 3 (22)

Perineural growth 40 (299) 28 (211) P = 0099a

Lymph-angioinvasion 51 (383) 29 (218) P = 0003a

Number of resected LN median (IQR)

140 (90-188) 150 (120-220) P = 0009b

Lymph node ratio gt02 60 (448) 12 (89) P lt 0001a

Follow-up in months median (IQR) 226 (116-516) 223 (112-426) P = 0551b

R1-resection 16 (119) 7 (52) P = 0050a

Abbreviations nCRT = neoadjuvant chemoradiotherapy IQR = interquartile range GEJ = gastroesophageal junction pretreatment staging cT = clinical tumor stage

cN = clinical nodal stage pT = pathological tumor stage pCR = pathologic complete response pN = pathological nodal stage LN = lymph node R1 = microscopic positive resection margin a = chi-Square test b = Mann-Whitney U test and c = likelihood ratio

43

3


Table 2 N-stage pattern in surgery-alone and nCRT group and regarding location of positive lymph nodes

Equal nodal staging(cNstage=pNstage)

Nodal up-staging (cNstagelt pNstage)

Nodal down-staging (cNstagegt pNstage)

P value

Node negative (cN0=pN0)

Node pos-itive (cN1-3=pN1-3)

Surgery alone (n=135) n 15 (111) 28 (207) 58 (430) 34 (252) P lt 0001a

nCRT (n=135) n 22 (163) 16 (119) 22 (163) 75 (556)Surgery-alone group

Adenocarcinoma (n=115) 12 (104) 25 (217) 54 (470) 24 (209) P = 0031b

Squamous cell carcinoma (n=20)

3 (150) 3 (150) 4 (200) 10 (500)

cT12 tumors (n=21) 7 (333) 4 (190) 3 (143) 7 (333) P = 0002b

cT34a tumors (n=114) 8 (70) 24 (211) 55 (482) 27 (237)nCRT group



5 (227) 4 (182) 0 (00) 13 (591)

cT12 tumors (n=22) 8 (364) 2 (91) 1 (45) 11 (500) P = 0037b

cT34a tumors (n=113) 14 (124) 14 (124) 21 (186) 64 (566)Mediastinal lymph nodes

Surgery alone (n=124) 102 (823) 2 (16) 5 (40) 15 (121) P = 0022b

nCRT (n=122) 101 (828) 0 (00) 0 (00) 21 (172)Para-esophageal nodes

Surgery alone (n=124) 31 (250) 41 (331) 24 (194) 28 (226) P lt 0001a

nCRT (n=122) 46 (377) 9 (74) 11 (90) 56 (459)Upper abdominal nodes P lt 0001a

Surgery alone (n=124) 52 (419) 27 (218) 35 (282) 10 (81)nCRT (n=122) 68 (557) 6 (49) 14 (115) 34 (279)

Abbreviations nCRT = neoadjuvant chemoradiotherapy cN = clinical nodal stage pN = pathological nodal stage cT = clinical tumor stage a = Pearson chi-square b = likelihood ratio

Nodal staging clinical versus pathological N-stageAfter nCRT the rate of positive nodes differed considerably compared to surgery alone (319 vs 726) with a remarkable reduction in N2N3 stages (103 vs 423 Table 1) Table 2 depicts the nodal up- and downstaging in both groups and in regard to lymph node locations

44


Overall clinical and pathological N-stage were equal (cN=pN) in 81 (300) patients In surgery-alone patients 43 (319) were staged adequately (15 cN0=pN0 and 28 cN1-3=pN1-3) with an overall accuracy of 319 Nodal upstaging occurred in 58 (430) and downstaging in 34 (252) patients after primary surgery In comparison the accuracy of clinical N+ detection (cN0=pN0cN+=pN+) was 681In the nCRT group 38 (281) patients had equal cNpN-stages (22 cN0=pN0 and 16 cN1-3=pN1-3) with nodal up- and downstaging in 22 (163) and 75 (556) patients respectively which differed significantly in surgery-alone patients (Plt 0001) A significant difference in the surgery-alone (P = 0031) and nCRT (0026)) groups was seen in regard to histologic type with relatively more upstaging in AC and more downstaging in SCC (Table 2) Nodal up- and downstaging also differed between cT1T2 and cT3T4a tumors in the surgery-alone (P = 0002) and nCRT group (P lt 0037) with a higher rate of nodal upstaging in locally advanced (cT3T4a) tumors (Table 2) Nodal downstaging was common in cT1T2 (333 n = 7) after surgery-alone but was more or less equal after nCRT in both cT1T2 (50 n = 11) and cT3T4a (566 n = 64) Equal nodal cNpN staging was often seen in cT1T2 tumors in the surgery-alone (524 n = 11) and nCRT group (455 n = 10)

Clinical and pathological location of LN metastasesNodal up-downstaging differed significantly between the locational subgroups upper mediastinal low mediastinal and upper abdominal LNrsquos (Table 2) Figure 1 displays the distribution and number of cN+ and pN+ per LN location In the surgery-alone group the sensitivity in detecting low mediastinal LN metastases (631) was higher than in the upper abdominal (436) and upper mediastinal (286) stations The specificity was high in mediastinal and upper abdominal (872 and 834) lymph nodes but lower in low-mediastinal nodes (525 Figure 1)After nCRT no upper mediastinal LN metastases were detected Upper abdominal LN metastases were commonly understaged in the surgery-along group particularly in the distalGEJ tumors while downstaging frequently occurred

45

3


in low mediastinal nodes (459) followed by upper abdominal nodes (279) compared to 226 and 81 respectively

Figure 1 Distribution of clinical and pathological lymph node involvement in the surgery-alone and nCRT groupAbbreviations nCRT = neoadjuvant chemoradiotherapy cN+ = clinical node positive pN+ = pathologic node positive N= = equal nodal staging Nuarr = nodal upstaging Ndarr = nodal downstaging cN0 = clinical node negative pN0 = pathologic node negative PPV = positive predictive value NPV = negative predictive value

Prognostic value of LN metastases location related to nodal up-downstaging Figure 2 displays the 5-year DFS of nodal up- and downstaging which differed significantly in both the surgery-alone (P lt 0001) and nCRT group (P = 0014) Independent prognostic factors for 5-year DFS were pathological N-stage (P lt 0002) and lymph-angioinvasion (P = 0016) in the surgery-alone group In the nCRT group clinical upper abdominal LN metastases (P = 0012) and lymph node ratio gt02 (P = 0034) were most prognostic (Table 3)

46


Table 3 Univariate and multivariate Cox regression analysis for 5-year disease-free survival in the surgery-alone and nCRT group

Univariate Cox regression analyses

Surgery-alone nCRTHR 95 CI P value HR 95 CI P value

Squamous cell carcinoma NS 043 018 ndash 102 0054cN+ upper abdominal NS 238 136 ndash 416 0002pT0 - - 100 0095

pT1-2 100 0001 285 107 ndash 760 0036pT3-4 282 151 ndash 527 271 104 ndash 708 0042pN0 100 0000 100 0022

pN1 346 162 ndash 737 0001 128 062 ndash 267 0506pN2 480 223 ndash 1030 0000 338 119 ndash 964 0023pN3 678 311 ndash 1479 0000 546 129 ndash 2319 0021pN+ above diaphragm 271 165 ndash 444 0000 210 107 ndash 411 0030pN+ abdominal 314 189 ndash 521 0000 200 100 ndash 402 0051Lymph-angioinvasion 264 163 ndash 429 0000 NSLymph node ratio gt02 328 202 ndash 535 0000 0002R1-resection 354 178 ndash 704 0000 481 168 ndash 1382 0004Equal nodal staging (cN=pN) 100 0000 NSNodal downstaging (cNgtpN) 062 030 ndash 130 0204Nodal upstaging (cNltpN) 277 161 ndash 478 0000

Multivariate Cox regression analysisSurgery-alone nCRT

HR 95 CI P value HR 95 CI P valuepT0 - - pT0 100 0143

pT1-2 100 0066 pT1-2 269 101 ndash 718 0049pT3-4 183 096 ndash 348 pT3-4 222 083 ndash 593 0111pN0 100 0002 cN+ upper

abdominal 191 105 ndash 347 0012

pN1 304 141 ndash 655 0004

pN2 364 165 ndash 804 0001

pN3 454 203 ndash 1019 0000

Lymph-angioinvasion

189 113 ndash 318 0016 Lymph node ratio gt02

409 139 ndash 1204

0034

Abbreviations nCRT = neoadjuvant chemoradiotherapy HR = hazard ratioCI = confidence interval cN+ = positive clinical nodal stage pT = pathological tumor stage pCR = pathologic complete response pN = pathological nodal stage pN+ = positive pathological nodal stage R1 = microscopic positive resection margin cN = clinical nodal stage NS = not significant = overall P-value of the categorical variables = significant (P lt 005)

47

3


Figure 2 Kaplan Meier curves for 5-year disease-free survival in nodal up- and downstaging in the surgery-alone and nCRT group

DISCUSSION

Adequate detection of LN metastases in EC is a strong prognostic parameter and essential in accurate delineation of radiotherapy tumor volumes and extent of radical nodal dissection Moreover it can make a difference in treatment decision making in cT1T2 tumors between surgery alone or surgery preceded by nCRT Unfortunately inappropriate preoperative LN staging is still common in EC even with current sophisticated methods [6 7] Not surprisingly nCRT

48


leads to nodal downstaging of 556 in this study However the low accuracy of nodal staging with gt25 downstaging in patients treated with surgery alone indicates that a considerable part of post-nCRT downstaging is in fact caused by inadequate staging rather than genuine response Inaccurate staging therefore impedes determination of complete response and prognosis after nCRT Even though we included all patients (T1-T4a) the 43 clinical nodal upstaging in the surgery-alone group was comparable with the 444 in Crabtree study with only T2 tumors [7] However the 163 upstaging in our nCRT group is considerably lower than the 369 in Crabtree group probably because approximately 60 of their patients was treated with only preoperative chemotherapy Even with our extensive staging (CT and PET or PET-CT and EUS if possible) the accuracy (cN0=pN0cN+=pN+) of 681 was slightly lower than the 744 in Crabtree group [7] The inaccuracy seems to be T-stage based on EUS as shown in the primary surgery group (Table 2) which implies a higher rate of pN upstaging in the more advanced T3T4 stages Others also reported a disputable reliance of EUS in assessing cN with even higher overstaging rates of pN0 tumors [6 8] Combined with overestimation of nodal involvement as expressed by 252 nodal downstaging in the surgery-alone and 556 in the nCRT group (Table 2) a substantial inaccuracy rate might be expected after neoadjuvant treatment If accurate staged it would have led to a change in the radiation fields with fewer regions to be targeted Moreover the downstaging effect was even higher in the cT1T2 tumors (50) but comparable to the nodal up and downstaging of cT3T4 in the neoadjuvant group which suggests a generally high nodal overstaging with probably less nodal CRT sensitivity This potential overestimation of response to nCRT might contribute to the large variety of 25 to 42 complete response in the literature [15 18 19] It is important to predict prognosis in complete responders since ypT0N0 patients with nodal downstaging (ypN0) appeared to have a worse survival compared with true equal staged cN0=pN0 patients [20] This was also expressed in a significant better prognosis of patients with nodal downstaging (cNgtpN) than those with nodal up-staging in both groups (P lt 0001 surgery-alone vs P = 0014 nCRT) probably caused by true N negative tumors (Figure 2)Nieman et al found survival to be less in patients with pN0 after neoadjuvant therapy vs patients with pN0 after primary surgery and suggested a negative prognostic impact in case of sterilized involved nodes [21] Currently the standard of pathologic LN evaluation in the resected specimen is based on presence or absence of viable tumor cells Others suggests that nodal collagen fibrosis or necrosis may be useful in determining evidence of prior nodal involvement but they are still not considered in most pathologic reports as it may be difficult to differentiate some of these changes from radiation induced effects [21]

49

3


Interestingly downstaging and pCR were more frequently in SCC patients treated with nCRT compared to patients with AC which supports earlier findings [15 22] However 50 nodal downstaging in SCC patients in our surgery-alone group is considerably higher than 154 reported by Park et al and was also more frequently in surgically treated cT1T2 patients (n = 7 333 Table 2) [23] Possible explanations for difficult nodal staging might be the complexity of longitudinal esophageal lymphatic drainage with skip metastases and a large number (gt50) of small LN metastases (lt5 mm) in EC impeding clinical detection [23 24] Another problem is the relatively high rate of LN metastases in the upper abdominal region which is prognostic in the nCRT group (P = 0012) [25] We identified nodal up-staging more commonly in the upper abdominal region in the surgery-alone group (28) which is probably related to incomplete EUS staging by severe stenosis in 20-36 [26] In line with tumor positive LN location Talsma et al found a worse prognosis in distal EC patients with suspected LN on EUS at both diaphragmatic sides and suggested a combined staging system including distribution of LN related to the diaphragm [3] Improvement of PET-CT and recently the application of diffusion-weighted magnetic resonance imaging eventually with nodal contrast agents (gadofosveset or iron nanoparticles) might increase the adequacy of detecting LN metastases in future studies [27 28] A limitation of this study might be the use of integrated PET-CT combined with a diagnostic CT after 2009 versus CT and PET between 2000-2009 Furthermore the accuracy of EUS might be higher if FNA could be applied to more distinct lymph nodes In our center FNA was performed in patients with clinically suspected LNs (n = 48177) considered relevant in determining the extent of radiation fields andor nodal resection In conclusion treatment decision-making in EC strongly depends on adequate clinical LN staging Nodal up- and downstaging was frequently found in patients treated with surgery alone (43 and 25 respectively) This inaccuracy rate in LN staging has an impact on prognosis to nCRT especially pertaining advanced tumors and upper abdominal nodes It might also influence radiotherapy target volume delineation with increased risk of locoregional recurrence or overtreatment with inappropriate administration or unnecessary chemoradiation which may induce (cardiopulmonary) toxicity Furthermore it is important to acknowledge imperfections of current preoperative nodal staging in assessing the genuine effect of nCRT to optimize future individualized treatment options including a possible wait-and-see strategy in future EC patients with a clinical complete response to nCRT

50


REFERENCES

[1] Peyre CG Hagen JA DeMeester SR et al The number of lymph nodes removed predicts survival in esophageal cancer an international study on the impact of extent of surgical resection Ann Surg 2008 248549-556[2] Kayani B Zacharakis E Ahmed K Hanna GB Lymph node metastases and prognosis in oesophageal carcinoma--a systematic review Eur J Surg Oncol 2011 37747-753[3] Talsma AK Ong CA Liu X et al Location of lymph node involvement in patients with esophageal adenocarcinoma predicts survival World J Surg 2014 38106-113[4] Rice TW Ishwaran H Hofstetter WL et al Esophageal Cancer Associations With (pN+) Lymph Node Metastases Ann Surg 2017 265122-129[5] van Vliet EP Heijenbrok-Kal MH Hunink MG Kuipers EJ Siersema PD Staging investigations for oesophageal cancer a meta-analysis Br J Cancer 2008 98547-557[6] Parry K Haverkamp L Bruijnen RC Siersema PD Offerhaus GJ Ruurda JP van Hillegersberg R Staging of adenocarcinoma of the gastroesophageal junction Eur J Surg Oncol 2016 42400-406[7] Crabtree TD Kosinski AS Puri V et al Evaluation of the reliability of clinical staging of T2 N0 esophageal cancer a review of the Society of Thoracic Surgeons database Ann Thorac Surg 2013 96382-390[8] Kutup A Link BC Schurr PG et al Quality control of endoscopic ultrasound in preoperative staging of esophageal cancer Endoscopy 2007 39715-719[9] Speicher PJ Ganapathi AM Englum BR Hartwig MG Onaitis MW DAmico TA Berry MF Induction therapy does not improve survival for clinical stage T2N0 esophageal cancer J Thorac Oncol 2014 91195-1201[10] Stiles BM Mirza F Coppolino A Port JL Lee PC Paul S Altorki NK Clinical T2-T3N0M0 esophageal cancer the risk of node positive disease Ann Thorac Surg 2011 92491-6 discussion 496-8[11] Dolan JP Kaur T Diggs BS et al Significant understaging is seen in clinically staged T2N0 esophageal cancer patients undergoing esophagectomy Dis Esophagus 2016 29320-325[12] Robb WB Dahan L Mornex F et al Impact of neoadjuvant chemoradiation on lymph node status in esophageal cancer post hoc analysis of a randomized controlled trial Ann Surg 2015 261902-908[13] Sepesi B Schmidt HE Lada M et al Survival in Patients With Esophageal Adenocarcinoma Undergoing Trimodality Therapy Is Independent of Regional Lymph Node Location Ann Thorac Surg 2016 1011075-1081[14] Edge SB Compton CC The American Joint Committee on Cancer the 7th edition of the AJCC cancer staging manual and the future of TNM Ann Surg Oncol 2010 171471-1474

51

3


[15] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[16] Muijs C Smit J Karrenbeld A et al Residual tumor after neoadjuvant chemoradiation outside the radiation therapy target volume a new prognostic factor for survival in esophageal cancer Int J Radiat Oncol Biol Phys 2014 88845-852[17] Mandard AM Dalibard F Mandard JC et al Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma Clinicopathologic correlations Cancer 1994 732680-2686[18] Smit JK Guler S Beukema JC Mul VE Burgerhof JG Hospers GA Plukker JT Different recurrence pattern after neoadjuvant chemoradiotherapy compared to surgery alone in esophageal cancer patients Ann Surg Oncol 2013 204008-4015[19] Zanoni A Verlato G Giacopuzzi S et al Neoadjuvant concurrent chemoradiotherapy for locally advanced esophageal cancer in a single high-volume center Ann Surg Oncol 2013 201993-1999[20] Zanoni A Verlato G Giacopuzzi S et al ypN0 Does It Matter How You Get There Nodal Downstaging in Esophageal Cancer Ann Surg Oncol 2016 23998-1004[21] Nieman DR Peyre CG Watson TJ et al Neoadjuvant treatment response in negative nodes is an important prognosticator after esophagectomy Ann Thorac Surg 2015 99277-283[22] Burmeister BH Smithers BM Gebski V et al Surgery alone versus chemoradiotherapy followed by surgery for resectable cancer of the oesophagus a randomised controlled phase III trial Lancet Oncol 2005 6659-668[23] Park SY Kim DJ Jung HS Yun MJ Lee JW Park CK Relationship Between the Size of Metastatic Lymph Nodes and Positron Emission TomographicComputer Tomographic Findings in Patients with Esophageal Squamous Cell Carcinoma World J Surg 2015 392948-2954[24] Howard JM Johnston C Patterns of lymphatic drainage and lymph node involvement in esophageal cancer Abdom Imaging 2013 38233-243[25] Prenzel KL Holscher AH Drebber U Agavonova M Gutschow CA Bollschweiler E Prognostic impact of nodal micrometastasis in early esophageal cancer Eur J Surg Oncol 2012 38314-318[26] Catalano MF Van Dam J Sivak MVJr Malignant esophageal strictures staging accuracy of endoscopic ultrasonography Gastrointest Endosc 1995 41535-539

52


[27] Pultrum BB van der Jagt EJ van Westreenen HL et al Detection of lymph node metastases with ultrasmall superparamagnetic iron oxide (USPIO)-en-hanced magnetic resonance imaging in oesophageal cancer a feasibility study Cancer Imaging 2009 919-28[28] Lambregts DM Beets GL Maas M et al Accuracy of gadofosveset-en-hanced MRI for nodal staging and restaging in rectal cancer Ann Surg 2011 253539-545

53

3



Evaluation of progression prior to surgery after neoadjuvant chemoradiotherapy with computed

tomography in esophageal cancer patients

Jan Binne Hulshoff Justin K Smit

Eric J van der Jagt John Th M Plukker

American Journal of Surgery 2014 Jul208(1)73-9

CHAPTER 4


ABSTRACT

BackgroundThe risk of tumor progression during neoadjuvant chemoradiotherapy (CRT) in esophageal cancer (EC) is around 8 to 17 We assessed the efficacy of computed tomography (CT) to identify these patients before esophagectomy

MethodsNinety-seven patients with locally advanced EC treated with CarboplatinPaclitaxel and 414 Gy neoadjuvantly were restaged with CT Two radiologists reviewed pre- and post-CRT CT images The primary outcome was detection of clinically relevant progressive disease Missed metastases were defined as metastatic disease found during surgery or within 3 months after post-CRT CT

ResultsProgressive disease was detected in 9 patients (9) Both radiologists detected 5 patients with distant metastases (liver n = 4 lung metastasis n = 1) but missed progressive disease in 4 cases One radiologist falsely assessed 2 metastatic lesions but after agreement progressive disease was detected with sensitivity and specificity of 56 and 100 respectively

ConclusionCT is effective in detecting clinically relevant progressive disease in EC patients after neoadjuvant treatment

56


BACKGROUND

Esophageal cancer (EC) is now the 8th most commonly diagnosed malignancy worldwide [1] Currently the primary curative treatment for EC consists of neoadjuvant chemoradiotherapy (CRT) followed by surgery However the prognosis of patients with locally advanced EC remains poor even after subsequent curative-intended resection with an overall survival rate of 35 to 38 [2-3]A recently performed meta-analysis and outcome of the latest published randomized controlled Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study trial underline the significant positive effect of neoadjuvant CRT on survival and local recurrence rate [45] Even then only approximately 50 of the patients benefit from neoadjuvant CRT while 8 to 17 of the patients develop distant metastases in the time interval between CRT and surgery [6-9] Surgery in these patients is futile and should be avoided Instead maintenance of quality of life with palliative support on guidance of complaints should be the principal treatmentGuidelines regarding restaging after neoadjuvant CRT are lacking Conventional staging methods including endoscopic ultrasonography (EUS) and computed tomography (CT) are based on disturbance of anatomical structures Limitations of these techniques include difficulty with distinguishing vital tumor tissue from necrotic or fibrotic tissue and the delay between cell death and tumor shrinkage [10] On the other hand CT has a low false positivity with a high availability rate and is associated with relatively low costs Furthermore the spatial resolution of CT has improved over the last decenniumDetection of viable tumor tissue with 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) is based on increased metabolic activity of the tumor leading to a high 18F-FDG uptake [11] Integrated PET-CT scans combine the metabolic information from the PET with the anatomical location found with the CT scan Three studies showed a better detection of distant metastases with integrated PET-CT (3 out of 4 patients) compared with CT alone (2 out of 4 patients) [67] and 8 PET-CT however also has a high false positivity rate which has been reported to be as high as 8 [8]In our esophageal center all patients received a post-CRT CT to assess whether the patient has developed progressive disease during the time interval In this study we hypothesize that CT is effective in detecting distant metastasis and clinically relevant lymph node progression to restrict futile surgery after neoadjuvant CRT

57

4



PatientsBetween 2006 and 2011 116 patients with potentially resectable locally advanced EC who received a restaging CT within 3 weeks after neoadjuvant CRT were eligible for inclusion In total 19 patients were excluded because of missing post-CRT data (n=7) missing pre-CRT CT data (n=1) treatment in different medical centers (n=3) a premature post-CRT CT (n=6) made during the neoadjuvant treatment and clinically proven progressive disease during CRT treatment (n=2) Finally 97 of the 116 patients could be included in this study This study was performed in accordance with the guidelines of the national and our own Ethical Board We created a database containing patient and tumor characteristics treatment-related data and follow-up dataPatient characteristics are summarized in Table 1 The malefemale ratio is 331 with a median age of 65 years (interquartile range 60 to 70 years) and tumors staged clinically from stage IIa (T2N0M0) to stage IIIc (T4aN1-2M0T1-4aN3M0) at the time of diagnosis

Initial staging procedureInitial staging consisted of EUS with fine needle aspiration on indication and a CT of the chest and abdomen A full-body 18F-FDG PET or 18F-FDG PETCT was employed in all T2-T4a andor nodal-positive patients (N+) The outcomes of all staging methods were discussed in a multi-disciplinary meeting participated by at least one of the two experienced radiologist an oncologist a radiotherapeutic oncologist a nuclear medicine physician pathologist a gastroenterologist and a surgical oncologist involved in the treatment of EC patients All participants of this meeting agreed that the patients included in this study should be treated with neoadjuvant CRT

Restaging preoperative post-chemoradiotherapy computed tomographyAll patients received both pre- and post-CRT 64-multisliced CT scan (PhilipsMX 8000 Philips Medical Systems Best The Netherlands Siemens Somatom Sensation Siemens Erlanger Germany or GE medical systems Milwaukee WI) according to the EC protocol used in our center The CT includes the cervical thoracic and abdominal region and was performed with adequate oral and intravenous contrast Pre-CRT CT images with a maximum slice thickness of 5 mm from other medical centers were allowed All included patients had their preoperative post-CRT CT within at least 3 to 4 weeks after the end of the neoadjuvant treatment All post-CRT CT scans were performed in our center with a Somatom Sensation 64 CT (Siemens Erlanger Germany) with slices of 215 mm In case of suspicious progressive disease additional examinations (either PET magnetic resonance imaging ultrasonography cytology or a combination) were carried out to strengthen or prove progressive disease

58


Table 1 Patients tumor and treatment characteristicsnumber of patients

Male 75 (77)Age median (IQR) 65 (60-70) yrsBMI median (IQR) 256 (23-284)Histology

Adenocarcinoma 80 (82)Squamous cell carcinoma 16 (17)Adenosqaumous cell carcinoma 1 (1)

Tumor location Middle esophagus 5 (5)Distal esophagus 92 (95)

Initial stageIIa (T2N0M0) 4 (4)IIb (T3N0M0 T1-2N1M0) 20 (21)IIIa (T1-2N2M0 T3N1M0 T4aN0M0) 38 (39)IIIb (T3N2M0) 20 (21)

IIIc (T4aN1-2M0 TallN3M0) 15 (15)Complete CRT 85 (88)Time in weeks median (IQR)

Pathology and surgery 17 (15-19)Pre- and post-CRT CT 14 (12-15)End neoadjuvant CRT and surgery 6 (5-8)

Surgery (including diagnostic laparotomy) 92 (95)Complete resection 85 (88)Stage after neoadjuvant CRT

0 (pCR TisN0M0) 15 (15)Ia (T1N0M0) 11 (11) Ib (T1-2N0M0) 2 (2)IIa (T2N0M0) 9 (9)IIb (T3N0M0 T0-2N1M0) 31(32)IIIa (T1-2N2M0 T3N1M0 T4aN0M0) 11 (11)IIIb (T3N2M0) 5 (5)IIIc (T4aN1-2M0 T4bNallM0 TallN3M0) 4 (4)IV (TallNallM1) 9 (9)

Progressive disease between CRT and surgery 9 (9)Progression within 3 months after surgery 0 (0)Follow up in weeks (IQR) 36 (20-52)

CRT = chemoradiotherapy CT = computed tomography IQR = interquartile range pCR = pathologically complete response lowast Pathologically or clinically proven metastatic disease

59

4


Reassessment of progressive diseaseTwo experienced radiologists including a senior radiologist specialized in gastroesophageal cancer reviewed both the pre- and post-CRT CT images and restaged these images according to the 7th tumor node metastasis (TNM) edition [12] Both radiologists were blinded to the outcome of the CT and further treatments The primary outcome was treatment change because of progressive disease (yes or no) detected by comparing the pre- and post-CRT CT Progressive disease was defined as any radiological visible increase in tumor volume including both nodal progression and distant metastasis Locoregional progression without signs of invasion in surrounding organs was not included as clinically relevant progressive disease in this study because it would not alter the course of surgical treatment As a reference standard we used either the cytological or histological confirmation of progressive disease detected before and during surgery or based on pathological examination of the resected specimens In addition we considered all suspect lesions that occurred within 3 months after the post-CRT CT as missed with CT All patients received a PETCT scan 3 months after surgical treatment for other research purposes which was also used to detect progressive disease Patients suspected of progressive disease were discussed in a multidisciplinary board and received additional examinations such as magnetic resonance imaging ultrasonography or EUS-guided fine needle aspiration on indication In defining true progressive disease we included all cytologicallyhisthologically proven progressive disease detected before or during surgery and pathological proven or clinical evidence of recurrentpersistent disease found within 3 months after the post neoadjuvant CRT CTWe determined the sensitivity specificity negative predictive value and positive predictive value of both radiologists in detecting progressive disease by comparing the pre- and post-CRT CT Patients assessed differently by the 2 radiologists were re-evaluated until agreement was obtained

TreatmentNeoadjuvant treatment consisted of intravenous Paclitaxel (50 mgm2) and Carboplatin (area under the curve = 2) administered 5 times during a 4-week radiotherapy period with a total dose of 414 Gy given in 23 fractions of 18 Gy [5] As standard surgical procedure we performed a radical transthoracic esophagectomy through a left or right thoracolaparotomy combined with a 2-field lymphadenectomy of mediastinal and abdominal lymph nodes including nodes at the celiac trunk and those along the common hepatic and splenic artery at the upper border of the pancreas

60


PathologyPathological processing of all resected specimens and lymph nodes has been standardized according to a detailed protocol on radicality including the proximal the distal and the circumferential margin The extension of the primary tumor localization number of nodal involvement and LN ratio was accurately described

Statistical analysisSPSS Version 17 (SPSS Inc Chicago IL) was used to perform statistical analyses The distribution of all patient characteristics was assessed with a QQ- and PP-plot Normally distributed variables were reported as mean plusmn standard deviation non-normally distributed variables as median [interquartile range] and categorical variables were reported as amount [percentage]

RESULTS

In Table 1 the patient tumor and treatment-related characteristics are summarized Of the 97 patients 9 (9) patients had progressive disease after neoadjuvant CRT The initial post-CRT CT prevented futile surgery in 5 (597 5) patients 4 patients developed liver metastases and 1 patient developed lung metastases Three patients had progressive disease identified during surgery 1 patient had a previous undetected peritoneal metastasis 1 patient had a metastatic lesion on the top of the bladder and 1 patient had both liver and omental metastases In addition cervical lymph node metastasis was observed in 1 patient with PET-CT performed because of a suspect vertebral lesion which was assessed as benign on PET-CT More information on these lesions and the imaging techniques used to strengthen or prove metastatic disease is revealed in Table 2 Pathologic proof of progressive disease was obtained in 6 patients There were no patients with a metastasis found within 3 months after the post-CRT CT

61

4


Table 2 Characteristics of patients with progressive disease relevant for further treatment

cTNM Histological tumor type

Location tumor

Location metastasis

Initially Detected with CT

Additional techniques

Surgery Patho-logically proven

T4aN1 Adeno Distal Liver Yes PET No No T3N1 Adeno Distal Liver Yes Biopsy No Yes T3N0 Adeno Distal Liver Yes Biopsy No YesT3N1 Adeno Distal Liver Yes No No T3N1 Sqaumous Distal Lung Yes No NoT3N2 Adeno Distal nodal

Liver Omental

Yes NoNo

No Yes NoYesYes

T3N1 Adeno Distal Bladder top

No Yes Yes

T3N1 Adeno Distal Peritoneal No Yes YesT4aN1 Adeno Distal Cervical

lymph node

No PET + Biopsy

Yes

cTNM = clinical tumor node metastasis CT = computed tomography PET = positron emission tomography Resectable lymph node progression

Efficacy in detecting progressive diseaseBoth radiologists detected 4 patients with liver metastases and 1 patient with lung metastases CT images are depicted in Fig 1 Progressive disease was missed by the radiologists in 4 patients 1st patient with a bladder top metastasis the 2nd had a peritoneal metastasis the 3rd had a cervical lymph node metastasis and the 4th had both liver and omental metastases

62


Figure 1 CT images of the patients with progressive disease detected by both radiologists (A) Pre-CRT CT and (B) post-CRT CT Images 1 to 4 are of patients with liver metastases and image 5 is of a patient with lung metastasis

During reassessment of the CT images of patients with missed progressive disease the radiologists agreed that the missed cervical node metastasis which was detected on PETCT should have been found on the post-CRT CT However because both radiologists missed this cervical node involvement it was scored as missed progressive disease Fig 2 depicts both the CT and PET-CT images in this patient Resectable nodal progression was detected by both radiologists in 2 patients but this was not relevant for subsequent treatment However one of these patients also had a liver and omental metastasis identified during surgery

63

4


Figure 2 Images of the patient with cervical lymph node progression after neoadjuvant CRT that was missed by both radiologists (A) Pre-CRT CT (B) post-CRT CT and (C) PET-CT was made to assess the suspect vertebral lesions in this patient

Radiologist 1 falsely suggested 2 metastases ndash a lung and vertebral metastasis (Fig 3) ndash identifying progressive disease with a sensitivity and specificity of 56 and 98 and an negative predictive value (NPV) and Positive Predictive Value (PPV) of 96 and 71 respectively Radiologist 2 correctly assessed the vertebral lesion as nonspecific based on the osteoblastic character identifying progressive disease with a sensitivity and specificity of 58 and 100 and an NPV and PPV of 96 and 100 respectively

Figure 3 CT images of the patient falsely assessed by radiologist 1 as patient with progressive disease (A) Pre-CRT CT and (B) post-CRT CT Image 1 was falsely assessed as lung metastasis and image 2 was falsely assessed as vertebral metastasis

64


After the differently assessed CT images were discussed the radiologists agreed that the vertebral lesion was nonspecific for metastasis and the lung lesion was benign After agreement both radiologists detected clinically relevant progressive disease with a sensitivity and specificity of 56 and 100 and an NPV and PPV of 96 and 100 respectivelyRetrospectively with knowledge of the metastases and its location a suspicious hepatic lesion of approximately 1 cm could be seen in the patients with both liver and omental metastases Even after re-evaluation the CT-missed omental metastases could not be detected The CT-missed bladder top metastasis was located outside of the range of the abdominal CT

DISCUSSION

In this study CT after neoadjuvant CRT (post-CRT CT) showed to be effective in the detection of clinically relevant progressive disease which leads to a change in treatment in these EC patients Detection of progressive disease before surgery on post-CRT CT prevented futile surgery in 5 (5) of these patients but missed progressive disease in 4 (4) patients Detection of progressive disease that alters the initial treatment strategy is of great importance as the majority of these patients is beyond curative treatment and should be refrained from surgeryIn this study we used only small-slice (215 mm) post-CRT CT scans for adequate radiological assessment of suspect lymph nodes andor metastatic lesions by 2 radiologists Most previously published studies used thicker slice (55 mm) CT scans to detect tumor progression The study by Bruzzi et al had distant metastases as primary outcome comparing CT with PET-CT in the detection of distant metastasis after neoadjuvant CRT [6] With PET-CT it was possible to detect distant metastasis in 7 patients (8) after neoadjuvant CRT while CT alone detected distant metastasis in 5 (6) of these patients Both missed metastases were located outside the range of the routinely performed CT imaging of the chest and abdomen This was also observed in our study with a bladder top metastasis that was located outside of the range of routine CT imagingCerfolio et al compared CT with EUS and PET-CT in restaging patients with EC after neoadjuvant CRT [8] The additional value of a PET-CT was limited only one additional metastasis was found and both PET-CT and CT still missed an omental metastasis PET-CT also falsely suggested one more distant metastasisA more recent study Blom et al compared CT with PET-CT in detecting distant metastases after neoadjuvant CRT [7] With PET-CT 4 out of 5 patients with a distant interval metastasis were identified while CT alone detected only 2 of these lesions One of the metastases that were missed with CT was located in the iliac crest which is usually located outside the range of the CT The other metastasis missed with CT was located in the right scapula Both CT and PET-CT missed 1 patient with peritonitis carcinomatosa and PET-CT falsely suggested a pulmonary metastasis

65

4


One of the reasons most studies advise a PET-CT is that EC tends to metastasize to uncommon and unusual distant locations after CRT such as skeletal muscles brain peritoneum subcutaneous soft tissue pleura pancreas and thyroid gland [13] These distant metastases can be missed with CT because only thoracic and abdominal CT images are made PET scans in contrast are full-body scans However most metastases after neoadjuvant CRT were located in either the liver or the lungs [6714] In our study 6 out of 9 metastases were located in either the liver or the lungs of which CT detected 5 metastases Most of the metastases located in the liver and the lungs can be found with CT [15] In one of the 4 patients with missed tumor progression on CT in this study re-assessment of a PET-CT for suspected vertebral metastases showed progression of cervical lymph node which was initially not been detected on post-CRT CT In addition after postoperative re-examination of the CT a suspect lesion could be detected in the liver of the patient with both liver and omental metastases Omental metastases are generally hardly detected with CT and the missed peritoneal metastases in this study could not be detected even after revision It is known that CT is inaccurate in detecting peritoneal metastases of lt1 cm [1617] However the additional value of PETCT is also limited in depicting small-volume metastatic lesions of 5 to 7 mm [18] Currently accurate detection of peritoneal metastasis is only possible by surgical intervention including laparoscopy or laparotomy [1920] In our center we therefore start with a laparotomy at the first phase of the esophagectomy followed by subsequent curative resection through a thoracotomyOne of the disadvantages of this retrospective study is that we did not compare CT with PET-CT Future research should compare a thin-slice thickness CT with PET-CT in the detection of patients with progressive disease after neoadjuvant CRT According to the retrospective design of our study we were unable to obtain pathological prove of metastatic disease in 3 patients with distant metastasis Because of small numbers it was not possible to accurately determine the interobserver variability between both a skilled and a specialized radiologist on EC in reviewing CT images in this study The different findings between them 1 radiologist had 2 false-positive results might be caused by a difference in experiencesThe potentially extended interval caused by the neoadjuvant treatment and the rate of progression during neoadjuvant treatment emphasizes the importance of imaging after neoadjuvant CRT Different studies assessed PET-CT to be more effective in the detection of distant metastasis However compared with CT PET-CT has a lower availability is more expensive and has a high amount of false-positive outcomes CT is therefore a useful method to detect patients with clinically relevant progressive disease in EC patients treated with neoadjuvant CRT

66


DISCLOSURE

The authors have nothing to disclose

67

4


REFERENCES

[1] Jemal A Bray F Center MM Ferlay J Ward E Forman D Global cancer statistics CA Cancer J Clin 2011 Mar-Apr61(2)69-90 [2] Allum WH Stenning SP Bancewicz J Clark PI Langley RE Long-term results of a randomized trial of surgery with or without preoperative chemotherapy in esophageal cancer J Clin Oncol 2009 Oct 2027(30)5062-5067 [3] Omloo JM Lagarde SM Hulscher JB Reitsma JB Fockens P van Dekken H et al Extended transthoracic resection compared with limited transhiatal resection for adenocarcinoma of the middistal esophagus five-year survival of a randomized clinical trial Ann Surg 2007 Dec246(6)992-1000 discussion 1000-1 [4] Sjoquist KM Burmeister BH Smithers BM Zalcberg JR Simes RJ Barbour A et al Survival after neoadjuvant chemotherapy or chemoradiotherapy for resectable oesophageal carcinoma an updated meta-analysis Lancet Oncol 2011 Jul12(7)681-692 [5] van Hagen P Hulshof MC van Lanschot JJ Steyerberg EW van Berge Henegouwen MI Wijnhoven BP et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 May 31366(22)2074-2084 [6] Bruzzi JF Swisher SG Truong MT Munden RF Hofstetter WL Macapinlac HA et al Detection of interval distant metastases clinical utility of integrated CT-PET imaging in patients with esophageal carcinoma after neoadjuvant therapy Cancer 2007 Jan 1109(1)125-134 [7] Blom RL Schreurs WM Belgers HJ Oostenbrug LE Vliegen RF Sosef MN The value of post-neoadjuvant therapy PET-CT in the detection of interval metastases in esophageal carcinoma Eur J Surg Oncol 2011 Sep37(9)774-778 [8] Cerfolio RJ Bryant AS Ohja B Bartolucci AA Eloubeidi MA The accuracy of endoscopic ultrasonography with fine-needle aspiration integrated positron emission tomography with computed tomography and computed tomography in restaging patients with esophageal cancer after neoadjuvant chemoradiotherapy J Thorac Cardiovasc Surg 2005 Jun129(6)1232-1241 [9] Flamen P Van Cutsem E Lerut A Cambier JP Haustermans K Bormans G et al Positron emission tomography for assessment of the response to induction radiochemotherapy in locally advanced oesophageal cancer Ann Oncol 2002 Mar13(3)361-368 [10] Westerterp M van Westreenen HL Reitsma JB Hoekstra OS Stoker J Fockens P et al Esophageal cancer CT endoscopic US and FDG PET for assessment of response to neoadjuvant therapy--systematic review Radiology 2005 Sep236(3)841-851 [11] Bar-Shalom R Valdivia AY Blaufox MD PET imaging in oncology Semin Nucl Med 2000 Jul30(3)150-185

68


[12] Sobin LH Compton CC TNM seventh edition whatrsquos new whatrsquos changed communication from the International Union Against Cancer and the American Joint Committee on Cancer Cancer 2010 Nov 15116(22)5336-5339 [13] Bruzzi JF Truong MT Macapinlac H Munden RF Erasmus JJ Integrated CT-PET imaging of esophageal cancer unexpected and unusual distribution of distant organ metastases Curr Probl Diagn Radiol 2007 Jan-Feb36(1)21-29 [14] Watt I Stewart I Anderson D Bell G Anderson JR Laparoscopy ultrasound and computed tomography in cancer of the oesophagus and gastric cardia a prospective comparison for detecting intra-abdominal metastases Br J Surg 1989 Oct76(10)1036-1039 [15] van Vliet EP Steyerberg EW Eijkemans MJ Kuipers EJ Siersema PD Detection of distant metastases in patients with oesophageal or gastric cardia cancer a diagnostic decision analysis Br J Cancer 2007 Oct 897(7)868-876 [16] Metser U Jones C Jacks LM Bernardini MQ Ferguson S Identification and quantification of peritoneal metastases in patients with ovarian cancer with multidetector computed tomography correlation with surgery and surgical outcome Int J Gynecol Cancer 2011 Nov21(8)1391-1398 [17] Wang Z Chen JQ Imaging in assessing hepatic and peritoneal metastases of gastric cancer a systematic review BMC Gastroenterol 2011 Mar 91119-230X-11-19 [18] Sironi S Messa C Mangili G Zangheri B Aletti G Garavaglia E et al Integrated FDG PETCT in patients with persistent ovarian cancer correlation with histologic findings Radiology 2004 Nov233(2)433-440 [19] Bonavina L Incarbone R Lattuada E Segalin A Cesana B Peracchia A Preoperative laparoscopy in management of patients with carcinoma of the esophagus and of the esophagogastric junction J Surg Oncol 1997 Jul65(3)171-174 [20] de Graaf GW Ayantunde AA Parsons SL Duffy JP Welch NT The role of staging laparoscopy in oesophagogastric cancers Eur J Surg Oncol 2007 Oct33(8)988-992

69

4



Prognostic value of the circumferential resection margin in esophageal cancer patients after

neoadjuvant chemoradiotherapy

Jan Binne Hulshoff Zohra Faiz

Arend Karrenbeld Gursah Kats-Ugurlu

Johannes G M Burgerhof Justin K Smit

John Th M Plukker

Annals of Surgical Oncology 2015 22 1301-1309

CHAPTER 5


ABSTRACT

BackgroundCircumferential resection margins (CRM) for esophageal cancer (EC) defined by the College of American Pathologists (CAP gt0 mm) or the Royal College of Pathologists (RCP gt1 mm) as tumor-free (R0) are based on a surgery-alone approach We evaluated the usefulness of both definitions in current practice with neoadjuvant chemoradiotherapy (nCRT)

MethodsCRMs were measured in 209 patients (104 with nCRT) with locally advanced EC after transthoracic esophagectomy Local recurrence and cancer related death were scored as events Patients were followed for at least 2 years or until death Prognostic factors (Plt01 in univariate analyses) for 2-year disease-free survival (DFS) and local recurrence-free survival (LRFS) were incorporated in multivariate Cox regression analyses Both CRM measurements were analyzed separately and prognostic cutoff values (0ndash10 mm) were assessed in both groups

ResultsIndependent prognostic factors (Plt005) for 2-year DFS were tumor length lymph node ratio angioinvasion and CAP R0 in the surgery-alone group and pN stage (Plt001) in the nCRT group Prognostic factors (Plt005) for 2-year LRFS were CAP lymph node ratio and tumor length in the surgery-alone group and CAP and grade in the nCRT group Optimal CRM cutoff values between 00 and 02 mm were prognostic for 2-year DFS in the surgery-alone and at 03 mm for the nCRT group

ConclusionsnCRT affected the CRM cutoff values After nCRT the CRM R0 according to the CAP was only prognostic for 2-year LRFS However in the surgery-alone group it was prognostic for both the 2-year DFS and LRFS

72


BACKGROUND

Even with neoadjuvant chemoradiotherapy (nCRT) the overall 5-year survival rate after esophagectomy remains relatively low at 47 in patients with locally advanced esophageal cancer (EC) [1] A strong prognostic indicator after a curative intended esophagectomy is the circumferential resection margin (CRM) rendered as microscopic tumor-free (R0) or tumor-positive (R1) [2-8] Commonly used definitions of a circumferentially R0 resection are those of the College of American Pathologist (CAP CRM gt0 mm) and the Royal College of Pathologists (RCP CRM gt1 mm) [910] After nCRT the optimal CRM may be influenced by tumor downsizing which facilitates a R0 resection [1]The optimal CRM cutoff point after nCRT has not been defined yet Recently two meta-analyses showed a significant association of a positive CRM according to both definitions with poor outcome which was even worse in patients with stage T3 disease or after nCRT [1112] However these studies did not assess which CRM definition was more powerful after nCRT while contradictory results after nCRT were reported in three other studies without a surgery-alone control group [13-15] Two studies with only squamous cell carcinoma showed a significant better survival rate in patients with a CRM gt 1 mm whereas no survival benefit was observed in R0 resections according to the CAP and RCP in a study with only T3 stage adenocarcinomas [13-15]We assessed the optimal CRM cutoff point and the prognostic value of R0 resections according to the CAP and RCP criteria in EC patients treated either with nCRT or surgery alone


Data collection of this explorative retrospective study was provided from a prospective maintained database of EC patients according to the national guidelines and the rules approved by the local ethical commission (wwwccmonl) We included only patients with a locally advanced curatively resectable EC (stage II-III) treated between 1997 and 2013 in whom the CRM was adequately assessed by our expert pathologists Of the patients treated with nCRT (n = 127) between 2005 and 2013 23 were excluded because of the following criteria incomplete medical records (n = 0) postoperative mortality (death within 90 days or in-hospital n = 10) progressive disease within 3 months after surgery or microscopic irradical (R1 tumor cells lt1 mm) longitudinal margins (n = 0) or follow-up lt24 months (n=13) Based on these exclusion criteria a reference group of surgery-alone treated patients (n = 105) was constructed Patients and tumor-related factors were matched and were equally distributed between both groups (Table 1)

73

5


Tumors staged according to the 6th TNM edition were recoded into the 7th edition [1617] Before 2000 (n = 11) staging consisted of endoscopic ultrasonography (EUS) with fine-needle aspiration (FNA) computed tomography (CT) of the neck thorax and abdomen and occasionally 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET n = 8) After 2000 a standard 18F-FDG PET was added which was replaced by 18F-FDG PETCT after 2009 Two weeks after nCRT patients were restaged with a CT thorax and abdomen

TreatmentAll patients underwent a transthoracic esophagectomy with en bloc dissection of regional mediastinal and abdominal (including the celiac trunk region) lymph nodes Patients with nCRT were treated according to the Dutch Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study (CROSS) regimen consisting of intravenous paclitaxel (50 mgm2) and carboplatin (AUC 2 mlmin) administered five times during a 5-week concurrent radiation period (414 Gy23 fractions of 18 Gy) [1] Before 2009 patients received nCRT based on their participation in the CROSS trial from 2009 onwards nCRT became standard of care for locally advanced EC patients (T1-4aN1-3 T2-4aN0-3 n = 75)

Table 1 Patient characteristics in the surgery-alone and neoadjuvant chemoradiotherapy (nCRT) group

nCRT (n=104) Surgery-alone (n=105) P valueMale 79 (760) 82 (781) 0714Age median (IQR) 63 (56-67) yrs 64 (57-69) yrs 0228

Histology 0382Adenocarcinoma 88 (846) 84 (800)Squamous cell carcinoma 16 (154) 21 (200)

Tumor location 0654Middle esophagus 8 (77) 12 (114)Distal esophagus 50 (481) 49 (467)GEJ 46 (442) 44 (419)

Tumor length gt5 cm 59 (567) 60 (571) 0695

cT-stage 0221T2 16 (154) 9 (86)T3 83 (798) 93 (886)T4a 5 (48) 3 (29)

74


nCRT (n=104) Surgery-alone (n=105) P valuecN-stage 0176

N0 27 (260) 41 (390)N1 50 (481) 44 (419)N2 22 (212) 18 (171)N3 5 (48) 2 (19)

pT-stage lt 0001para

Tx 1 (10)T0 21 (202)T1 22 (212)T2 14 (135) 20 (190)T3 46 (442) 82 (781)T4a 0 (00) 3 (29)

pN-stage lt 0001N0 62 (596) 28 (267)N1 26 (250) 34 (324)N2 11 (106) 25 (238)N3 5 (48) 18 (171)

Perineural growth 22 (212) 33 (314) 0084Angioinvasion 22 (212) 51 (486) lt 0001Number of LN (gt4 LN+) 10 (96) 32 (305) lt 0001Lymph node ratio (gt02) 18 (173) 50 (476) lt 0001Follow-up in months median (IQR) 275 (150 ndash

420)29 (155 ndash 560) 0241

Tumor recurrence 63 (606) 75 (714) 0098Local recurrence 17 (163) 35 (333) 0005Death 60 (577) 83 (790) 0001Tumor related death 54 (519) 73 (695) 0009CRM in mm median (IQR) 33 (10-50 mm) 05 (0ndash14 mm) lt 0001

0 mm 9 (87) 27 (257) lt 00010 ndash 1 mm 13 (125) 40 (381) lt 0001gt1 mm 82 (788) 38 (362) lt 0001

Abbreviations nCRT= neoadjuvant chemoradiotherapy cT= clinical T-stage cN= clinical lymph node stage pT= pathological T-stage pN= pathologic lymph node stage LN= lymph node CRM= circumferential resection margin CAP= College of American Pathologists and RCP= Royal College of Pathologists and IQR= inter quartile range = using chi square test = using Mann-Whitney test and para = using Fisher exact test

75

5


PathologyResected specimens were examined according to a standardized protocol by two specialized gastrointestinal pathologists The resected specimen was pinned on a Styrofoam plate by the surgeon enabling accurate pathological assessment of the marked Clinical Tumor Volume and Gross Tumor Volume areas in patients treated with nCRT [18] CRM was measured according to the method of Quirke the specimens were inked with Indian ink and fixed in formalin during 24 h [6] The specimens were sliced into transverse cross-sections of 05 cm for macroscopic assessment and sampling of at least two sections with the smallest CRM2 The CRM was microscopically assessed on hematoxylin and eosin stained samples in tenths of millimetres Furthermore the pT-stage pN-stage the lymph-node ratio (gt02 metastatic lymph node ratio) number of positive lymph nodes (gt4) histological tumor type tumor grade angioinvasion and perineural tumor growth were assessed

Follow-upPatients were followed for at least 2 years or until death every 3 months during the first year after surgery every 6 months in the second year and every year thereafter for the next 10 years Tumor recurrence was defined as histocytologically proven suspected radiological imaging or clinically evident recurrence Local recurrence included recurrent disease at the anastomotic site or in the original tumormediastinal bed

Statistical AnalysisDistribution of continuous patient characteristics was reported as median [interquartile range] and categorical variables were reported in numbers and percentages The patients groups were compared with the MannndashWhitney test for continuous variables and χ2 or Fisher exact test for categorical response variables KaplanndashMeier curves and log-rank test determine the 5-year disease-free survival (DFS) and local recurrence-free survival (LRFS) of both CRM definitions Prognostic values of all variables for 2-year DFS were assessed with univariate Cox regression analysis Factors within the univariate analysis were age tumor type and grade (G1ndash2 vs G3ndash4) clinical T and N stage tumor length (gt5 cm measured endoscopic or with CT) treatment type (nCRT or surgery alone) and pathologic outcome T and N stage number of LN metastases (gt4) and metastatic lymph node ratio (gt02) perineural growth and angioinvasion Multivariate Cox regression was performed by incorporating all variables with a P-value lt01 on univariate analysis Both the CAP (CRM gt0 mm) and RCP (CRM gt1 mm) definition entered the multivariate analysis separately The prognostic value of R0 resections according to the RCP and CAP for the 2-year

76


DFS and 2-year LRFS was assessed with multivariate Cox regression analyses in both treatment groups To assess the optimal cutoff value of the CRM on 2-year DFS an explorative analysis was performed in both groups Univariate analyses were undertaken to assess the prognostic value of all cutoff values (from 00 to 10 mm) The observed interval is based on the assumption that the expected optimal CRM cutoff should be between 00 and 10 mm The Akaike Information Criterion (AIC) which quantifies the quality of a statistical model for a set of data was used to indirectly compare the prognostic value of the CAP and RCP model [19] It penalizes the number of explanatory variables by adding twice the number of variables in the model to the minus2 log likelihood in a formula AIC = minus2 log likelihood +2 k in which k is the number of explanatory variables in the model The model with the lowest AIC was considered to be most prognostic The backwards likelihood ratio method was used in the Cox regression analysis Analyses were performed with SPSS version 22

RESULTS

Patient characteristics are summarized in Table 1 All nCRT patients with CAP-R1 resections (n = 9 87 ) had stage pT3 Of the 27 (257 ) R1 resections in patients treated with surgery alone 24 had stage pT3 1 had pT2 and 2 had stage pT4a disease The median CRM differed significantly with 33 [interquartile range (IQR) 10ndash50] mm versus 05 (IQR 0ndash14) mm for the nCRT and surgery-alone group respectively The median follow up was 290 (IQR 155ndash560) months and 275 (IQR 150ndash420) in the surgery-alone and nCRT groups respectively

Prognostic Value of the CAP and RCP CriteriaFigure 1 displays the DFS of both treatment groups with a R0 resection or involved CRM (R1 resection) according to CAP (Fig 1a) and RCP (Fig 1b) With the log-rank test the CAP definition was prognostic for 5-year DFS in both the surgery (P = 0008) and nCRT group (Plt0001) and the RCP definition was prognostic in the nCRT group (Plt0001) but not in the surgery group (P = 0071) The 5-year DFS was not different (P = 0131) between CAP R1 patients treated with or without nCRT but differed (P = 0031) between patients with an RCP R1 resection in both groups

Table 2 displays all prognostic factors with a Plt01 on univariate analysis and Table 3 shows the multivariate Cox regression models containing either the CAP or RCP for 2-year DFS and LRFS in both groups Independent prognostic factors for 2-year DFS in the surgery-alone group were tumor length [P = 0006 hazard ratio (HR) 268 CI 133ndash543] lymph node ratio (P = 0047 HR 257 CI 101ndash

77

5


651) and CAP (P = 0012 HR 041 CI 021ndash083) Independent prognostic factors for 2-year LRFS were lymph node ratio (P = 0020 HR 311 CI 120ndash809) tumor length (P = 0002 HR 1099 CI 249ndash4843) and CAP (P = 0004 HR 027 CI 011ndash0658) Both for 2-year DFS and LRFS the model containing the CAP had a lower AIC than the RCP model and therefore was more prognostic

Figure 1 Disease-free survival in patients treated with neoadjuvant chemoradiotherapy and surgery-alone with circumferential microscopic tumor-free (R0) or involved resection margins (R1) according to a CAP (0 mm) and b RCP (1 mm)

78


Table 2 Prognostic factors with a Plt01 on univariate analysis for disease-free and local recurrence free survival in the surgery-alone and neoadjuvant chemoradiotherapy group

Surgery-alone group2-year DFS 2-year LRFS

HR 95 CI P-value HR 95 CI P-valuepN0pN1pN2pN3

100336736805

110 ndash 1032247 ndash 2192258 ndash 2505

0001

003400000000

1004083391154


0007

007601600002

Tumor length 223 114 ndash 434 0019 845 197 ndash 3621 0004Perineural growth 200 111 ndash 360 0021 NSAngioinvasion 290 149 ndash 565 0002 NSNumber of LN 401 221 ndash 726 lt 0001 341 146 ndash 797 0005Lymph node ratio 396 208 ndash 755

lt 0001351 144 ndash 857 0006

CAP R0 045 025 ndash 082 0010 042 018 ndash 098 0044RCP R0 046 023 ndash 091 0025 052 020 ndash 132 0168

nCRT group2-year DFS 2-year LRFS

HR 95 CI P-value HR 95 CI P-valuecT2cT3cT4a

1005131298

124 ndash 2122236 ndash 7145

0013

00240003

NS

pT0pT1pT2pT3

100113183274


0054

083702790025

NS

pN0pN1pN2pN3

100258469687


lt 0001

000700000000

100056537731


0047

059000220086

Perineural growth 187 097 ndash 338 0062 NSAngioinvasion 181 098 ndash 353 0055 NSNumber of LN 390 190 ndash 798 lt 0001 800 192 ndash 3325 0004Lymph node ratio 278 148 ndash 520 0001 431 125 ndash 1495 0021CAP R0 028 013 ndash 061 0001 042 018 ndash 098 lt 0001RCP R0 040 022 ndash 074 0003 030 009 ndash 106 0061

Abbreviations DFS= disease-free survival LRFS= local recurrence free survival CI= confidence interval cT= clinical T-stage cN= clinical lymph node stage pT= pathological T-stage pN= pathologic lymph node stage LN= lymph node CRM= circumferential resection margin R0= tumor free resection margin CAP= College of American Pathologists and RCP= Royal College of Pathologists = overall P-value of the categorical variables

79

5


The only independent prognostic factors for 2-year DFS in the nCRT group was the pN-stage (overall P = 0004) pN1 (P = 0007 HR 270 CI 131ndash559) and pN2ndash3 (P = 0005 HR 339 CI 143ndash803) Both CAP (P = 0001 HR 006 CI 001ndash031) and tumor grade (P = 0008 HR 1691 CI 212ndash13505) were prognostic for 2-year LRFS For both 2-year DFS and LRFS the multivariate regression model containing the CAP definition had a lower AIC and therefore was more prognostic

Optimal CRM after Surgery Alone and after nCRTCRM cutoff values of 00 (P = 0012 HR = 041 CI 021ndash083 AIC = 3170) 01 (P = 0045 HR = 050 CI 025ndash098 AIC = 3200) and 02 mm (P = 0028 HR = 048 CI 025ndash092 AIC = 3188) were independent prognostic factors for 2-year DFS in the surgery-alone group Based on the AIC the 00-mm cutoff value (CAP) was the most prognostic However in the nCRT group the optimal cutoff value for 2-year DFS was 03 mm (P = 0045 HR = 035 CI 013ndash098 AIC = 3481)

80


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p

Surg

ery-

alon

e gr

oup

2-ye

ar D

FS2-

year

LR

FSH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 31

70)

CA

P0

410

21 ndash

08

30

012

CA

P m

odel

(AIC

= 16

82)

CA

P0

270

11 ndash

06

580

004

LN ra

tio2

571

01 ndash

65

10

047

LN ra

tio3

111

20 ndash

80

90

020

Tum

or le

ngth

268

133

ndash 5

43

000

6Tu

mor

leng

th10

99

249

ndash 4

843

000

2A

ngio

inva

sion

190

094

ndash 3

85

007

5N

umbe

r of L

N+

213

092

ndash 4

95

007

8R

CP

mod

el

(AIC

= 32

09)

RC

P0

830

38 ndash

17

80

627

RC

P m

odel

(AIC

= 17

42)

RC

P0

57

022

ndash 1

51

025

8LN

ratio

268

105

ndash 6

81

003

9LN

ratio

313

112

ndash 8

24

002

1A

ngio

inva

sion

195

094

ndash 4

03

007

2Pe

rineu

ral g

row

th1

890

99 ndash

35

90

053

Tum

or le

ngth

252

125

ndash 5

09

001

0Tu

mor

leng

th8

591

99 ndash

37

080

004

Num

ber o

f LN

192

084

ndash 4

39

012

3

81

5


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p -

cont

inue

dnC

RT

grou

p2-

year

DFS

2-ye

ar L

RH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 34

99)

CA

P0

470

18 ndash

12

30

124

CA

P m

odel

(AIC

= 73

5)

CA

P0

060

01 ndash

03

10

001

cT3

200

76 ndash

13

490

114

pN0

100

000

4

Gra

de16

91

212

ndash 1

350

50

008

pN1

270

131

ndash 5

59

000

7pN

2-3

339

143

ndash 8

03

000

5

RC

P m

odel

(AIC

= 35

03)

RC

P0

690

31 ndash

15

20

359

RC

P m

odel

(AIC

= 80

0)

RC

P1

010

08 ndash

13

500

995

cT1

000

275

pN0-

11

000

203

232

051

ndash 1

051

pN1-

28

810

31 ndash

252

23

pN0

100

001

4

Gra

de30

07

279

ndash 3

246

00

005

pN1

251

122

ndash 5

210

014

Num

ber o

f LN

073

006

ndash 8

94

080

4pN

2-3

299

123

ndash 7

32

001

6LN

ratio

260

044

ndash 1

542

029

4pT

0-1

100

035

9

pT2

196

070

ndash 5

49

020

2pT

3-4a

184

071

ndash 4

73

020

9

Abb

revi

atio

ns D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

CI=

con

fiden

ce in

terv

al S

CC

= sq

uam

ous c

ell c

arci

nom

a c

T= c

linic

al

T-st

age

cN

= c

linic

al ly

mph

nod

e st

age

pT=

pat

holo

gica

l T-s

tage

pN

= p

atho

logi

c ly

mph

nod

e st

age

LN

= ly

mph

nod

e C

RM

= ci

rcum

fere

ntia

l mar

gin

R

0= tu

mor

free

rese

ctio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

ove

rall

P-va

lue

of th

e ca

tego

ri-ca

l var

iabl

es

= Si

gnifi

cant

(Plt0

05)

82


DISCUSSION

The prognostic value of the circumferential margin (CRM) has been proven in EC patients after surgery alone but its significance after neoadjuvant treatment is not well defined yet This study conducted in stage II-III EC patients showed that both definitions of a free CRM were not prognostic for 2-year DFS in patients treated with nCRT The CAP definition (gt0 mm) however was an independent prognostic factor for 2-year DFS in the surgery-alone group and for LRFS in the nCRT and surgery-alone group The optimal CRM cutoff value for 2-year DFS was 03 and between 00 and 02 mm in the nCRT and surgery-alone group respectivelyThis study is one of the first to assess the optimal cutoff value of the CRM after nCRT previously published studies used either the RCP or CAP criteria of a free CRM Although neoadjuvant treatment decreases the rate of R1 resection by transversal and sagittal tumor reduction the induced fibrosis may contain different amounts of undetectable viable tumor cells [1] Therefore the CRM assessment depends upon accurate histological examination of residual tumor which might be related to tumor heterogeneity CRM gt1 mm showed to be prognostic but several studies reported conflicting results in patients treated with nCRT (Table 4) Chao et al described a significantly better disease-free and disease-specific survival whereas Liu et al noted a significantly better overall survival (OS) [1314] However Harvin et al failed to prove a survival benefit after nCRT with respect to both CAP and RCPndashCRM resections [15] This difference might be explained by the inclusion of different pathologic tumor types Harvin et al only included ypT3 or higher adenocarcinomas whereas Chao et al and Liu et al included only patients with squamous cell carcinomas [13ndash15] In our study histologic tumor type did not to affect the prognostic value of the CRMs for DFS and LRFS although the number of squamous cell carcinomas in the nCRT group was rather small (n = 16) Inclusion of pathologic T3 tumors in determining the optimal CRM seems comprehensible as circumferential R1 resections in pT2 tumors are generally considered to be caused by inadequate surgery [72021] Moreover Rao et al stated that CRM involvement in the EC specimen is related to advanced disease rather than being an indicator of completeness of resection [4] In our study only one patient staged as ypT2 disease had a R1 resection due to extensive angioinvasive tumor growth within the CRM which depends more on biologic aggressiveness rather than poor surgery Another factor that might influence the CRM is the used surgical method Suttie et al noted that the transhiatal approach resulted in significantly more CRM involvement compared with the transthoracic approach [22] Because the transthoracic approach is our standard method we could disregard this potential confounding effect

83

5


Tabl

e 4

Stu

dies

reg

ardi

ng p

rogn

ostic

val

ue o

f the

cir

cum

fere

ntia

l res

ectio

n m

argi

n af

ter

neoa

djuv

ant c

hem

orad

ioth

erap

y

Stud

y (y

ear)

His

tolo

gySt

age

Patie

nts (

n)nC

RT

()

Out

com

eC

RM

defi

nitio

nP-

valu

e Th

omps

on e

t al

(200

8)A

C S

CC

cT1-

424

012

4 (5

2)

5-ye

ar su

rviv

al

RC

PN

SC

hao

et a

l (2

011)

SC

Cyp

T315

115

1 (1

00

)LR

FSR

CP

lt 0

05D

FSR

CP

lt 0

05D

SSR

CP

lt 0

05H

arvi

n et

al

(201

2)A

Cyp

T316

016

0 (1

00

)O

S D

FS L

RFS

C

AP

NS

OS

DFS

LR

FSR

CP

NS

Reid

et a

l (2

012)

AC

SC

CcT

1-4

269

42 (1

6)

DFS

RC

Plt

001

OS

RC

P0

05O

rsquoFar

rell

et a

l (2

013)

AC

SC

C o

ther

s cT

315

782

(52

)O

SR

CP

NS

OS

CA

P0

02Li

u et

al

(201

3)SC

CcT

1-4

9494

(100

)

OS

RC

Plt0

01

Abb

revi

atio

ns S

CC

= sq

uam

ous c

ell c

arci

nom

a A

C=

aden

ocar

cino

ma

nC

RT=

neoa

djuv

ant c

hem

orad

ioth

erap

y c

T= c

linic

al T

-sta

ge y

pT=

path

olog

ic

T-st

age

afte

r nC

RT D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

DSS

= di

seas

e sp

ecifi

c su

rviv

al C

RM

= ci

rcum

fere

ntia

l res

ectio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

on m

ultiv

aria

te a

naly

sis

84


Three other studies assessed the value of the CRM in which only a part of the included patients received nCRT again with conflicting results [23-25] Thompson et al (n = 240 52 nCRT) did not find a survival benefit whereas Reid et al (n = 269 156 nCRT) found a significantly better DFS and OS in patients with a RCP R0 resection [2325] Farrell et al (n = 157 52 nCRT) found the CAP definition (P = 002) more prognostic for the OS than the RCP definition [24]As in patients treated with nCRT the optimal CRM definition in surgically treated patients also is unclear Two recent meta-analyses showed that both CRM definitions were associated with a poor survival although the CAP criteria differentiated higher-risk groups [1112] Moreover Chan et al found that the CAP definition based on the hazard ratio and subgroup analysis had a prognostic advantage over the RCP criteria [12] Concordant to these results we found that the optimal CRM cutoff value in the surgery-alone group analyzed with the Akaike Information Criterion was the CAPBeside the CRM lymph node metastasis associated variables were important prognostic factors in this study lymph node ratio gt02 was independent prognostic for both 2-year DFS and LRFS in the surgery-alone group and pN-stage was the only prognostic factor for 2-year DFS in the nCRT group One meta-analysis which underlined the importance of lymph node metastasis indicated that nodal metastases appeared to negate the prognostic value of the CRM [12] Moreover the presence of lymph node metastases and an involved CRM indicated a more advanced-staged disease [26] Another prognostic factor in surgery-alone patients was the tumor length which is in correspondence with previously published data [27]Pultrum et al assessed the optimal CRM in surgically treated patients using the area under the curve (AUC) analysis on receiver operating curves (ROC which does not incorporate the time factor [2] A method that includes the time factor is the more complex time-dependent ROC method according to Heagerty et al [28] For our limited explorative study however we prefer to use multivariate Cox regression analysis and suggest validating the results in a larger cohort

CONCLUSION

This study showed that both definitions of a tumor-free CRM (CAP gt 0 mm RCP gt 1 mm) were not prognostic for DFS in patients treated with nCRT A free CRM according the CAP definition was prognostic for 2-year DFS in the surgery-alone group and an optimal CRM cutoff between 00 and 02 and at 03 mm in the surgery-alone and nCRT groups respectively These findings should be validated in a large prospective study

DISCLOSURE


85

5


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[2] Pultrum BB Honing J Smit JK et al A critical appraisal of circumferential resection margins in esophageal carcinoma Ann Surg Oncol 2010 17812-820[3] Mirnezami R Rohatgi A Sutcliffe RP Hamouda A Chandrakumaran K Botha A Mason RC Multivariate analysis of clinicopathological factors influencing survival following esophagectomy for cancer Int J Surg 2010 858-63[4] Rao VS Yeung MM Cooke J Salim E Jain PK Comparison of circumferential resection margin clearance criteria with survival after surgery for cancer of esophagus J Surg Oncol 2012 105745-749[5] Deeter M Dorer R Kuppusamy MK Koehler RP Low DE Assessment of criteria and clinical significance of circumferential resection margins in esophageal cancer Arch Surg 2009 144618-624[6] Dexter SP Sue-Ling H McMahon MJ Quirke P Mapstone N Martin IG Circumferential resection margin involvement an independent predictor of survival following surgery for oesophageal cancer Gut 2001 48667-670[7] Griffiths EA Brummell Z Gorthi G Pritchard SA Welch IM The prognostic value of circumferential resection margin involvement in oesophageal malignancy Eur J Surg Oncol 2006 32413-419[8] Khan OA Fitzgerald JJ Soomro I Beggs FD Morgan WE Duffy JP Prognostic significance of circumferential resection margin involvement following oesophagectomy for cancer Br J Cancer 2003 881549-1552[9] College of American Pathologists Surgical Pathology Cancer Case Summary (Checklist) Esophagus College of American Pathologists Northfield 2005[10] Mapstone N Minimum Dataset for Oesophageal Carcinoma Histopathology Reports (1st edn) Royal College of Pathologists London 1998[11] Wu J Chen QX Teng LS Krasna MJ Prognostic significance of positive circumferential resection margin in esophageal cancer a systematic review and meta-analysis Ann Thorac Surg 2014 97446-453[12] Chan DS Reid TD Howell I Lewis WG Systematic review and meta-analysis of the influence of circumferential resection margin involvement on survival in patients with operable oesophageal cancer Br J Surg 2013 100456-464[13] Chao YK Yeh CJ Chang HK et al Impact of circumferential resection margin distance on locoregional recurrence and survival after chemoradiotherapy in esophageal squamous cell carcinoma Ann Surg Oncol 2011 18529-534[14] Liu CY Wang BY Lee MY Tsai YC Liu CC Shih CH The prognostic value of circumferential resection margin in esophageal squamous cell carcinoma after concurrent chemoradiation therapy and surgery J Chin Med Assoc 2013 76570-575

86


[15] Harvin JA Lahat G Correa AM et al Neoadjuvant chemoradiotherapy followed by surgery for esophageal adenocarcinoma significance of microscopically positive circumferential radial margins J Thorac Cardiovasc Surg 2012 143412-420[16] American Joint Committee on Cancer AJCC cancer staging manual 6th ed New York Springer 2002[17] American Joint Committee on Cancer AJCC cancer staging manual 7th ed New York Springer 2009[18] Muijs C Smit J Karrenbeld A et al Residual tumor after neoadjuvant chemoradiation outside the radiation therapy target volume a new prognostic factor for survival in esophageal cancer Int J Radiat Oncol Biol Phys 2014 88845-852[19] Akaike H A New Look at the Statistical Model Identification Automatic Control IEEE Transactions 1974 19(6)716ndash723[20] Verhage RJ Zandvoort HJ ten Kate FJ van Hillegersberg R How to define a positive circumferential resection margin in T3 adenocarcinoma of the esophagus Am J Surg Pathol 2011 35919-926[21] Khan OA Cruttenden-Wood D Toh SK Is an involved circumferential resection margin following oesphagectomy for cancer an important prognostic indicator Interact Cardiovasc Thorac Surg 2010 11645-648[22] Suttie SA Nanthakumaran S Mofidi R Rapson T Gilbert FJ Thompson AM Park KG The impact of operative approach for oesophageal cancer on outcome the transhiatal approach may influence circumferential margin involvement Eur J Surg Oncol 2012 38157-165[23] Thompson SK Ruszkiewicz AR Jamieson GG et al Improving the accuracy of TNM staging in esophageal cancer a pathological review of resected specimens Ann Surg Oncol 2008 153447-3458[24] OrsquoFarrell NJ Donohoe CL Muldoon C Costelloe JM King S Ravi N Reynolds JV Lack of independent significance of a close (lt1 mm) circumferential resection margin involvement in esophageal and junctional cancer Ann Surg Oncol 2013 202727-2733[25] Reid TD Chan DS Roberts SA Crosby TD Williams GT Lewis WG Prognostic significance of circumferential resection margin involvement following oesophagectomy for cancer and the predictive role of endoluminal ultrasonography Br J Cancer 2012 1071925-1931[26] Griffiths EA Pritchard SA Mapstone NP Welch IM Emerging aspects of oesophageal and gastro-oesophageal junction cancer histopathology - an update for the surgical oncologist World J Surg Oncol 2006 482[27] Eloubeidi MA Desmond R Arguedas MR Reed CE Wilcox CM Prognostic factors for the survival of patients with esophageal carcinoma in the US the importance of tumor length and lymph node status Cancer 2002 951434-1443[28] Heagerty PJ Lumley T Pepe MS Time-dependent ROC curves for censored survival data and a diagnostic marker Biometrics 2000 56337-344

87

5



Effect of extending the original eligibility criteria for the ldquoCROSSrdquo neoadjuvant

chemoradiotherapy on toxicity and survival in esophageal cancer

Ellen C de Heer dagger Jan Binne Hulshoff dagger

Daphne Klerk Johannes GM Burgerhof

Derk Jan A de Groot John Th M Plukker Geke AP Hospers

dagger Both authors contributed equally to the manuscript

Annals of Surgical Oncology 2017 February 10

CHAPTER 6


ABSTRACT

BackgroundPatients with curable esophageal cancer (EC) who proceed beyond the original Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study (CROSS) eligibility criteria are also treated with neoadjuvant chemoradiotherapy (nCRT) This study assessed the effect that extending the CROSS eligibility criteria for nCRT has on treatment-related toxicity and overall survival (OS) in EC

MethodsThe study enrolled 161 patients with locally advanced EC (T1N1-3T2-4aN0-3M0) treated with the CROSS schedule followed by esophagectomy Group 1 consisted of 89 patients who met the CROSS criteria and group 2 consisted of 72 patients who met the extended eligibility criteria ie a tumor length greater than 8 cm (n = 24) more than 10 weight loss (n = 35) more than 2ndash4 cm extension in the stomach (n = 21) celiac lymph node metastasis (n = 13) andor age over 75 years (n = 2) The study assessed the differences in nCRT-associated toxicity [National Cancer Institutersquos Common Terminology Criteria for Adverse Events (CTCAE) grade ge 3] and 90-day postoperative mortality Moreover the prognostic value for OS was assessed with multivariate Cox regression analysis

ResultsNo difference was found in nCRT-associated toxicity (P = 0117) postoperative complications (P = 0783) and 90-day mortality (P = 0492) The OS differed significantly (P = 0004) with a median of 373 months [95 confidence interval (CI) 104ndash642 months] for group 1 and 172 months (95 CI 138ndash207 months) for group 2 Pathologic N stage (P = 0023) pathologic T stage (P = 0043) and group 2 (P = 0008) were independent prognostic factors for OS

ConclusionsExtension of the CROSS study eligibility criteria for nCRT did not affect nCRT-associated toxicity postoperative complications and postoperative mortality but was prognostic for OS

90


BACKGROUND

Neoadjuvant chemoradiotherapy (nCRT) according to the Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study (CROSS) (carboplatinpaclitaxel and 414 Gy radiotherapy) followed by a radical surgical resection is the gold standard for locally advanced esophageal cancer (EC) in the Netherlands [1] This nCRT scheme increased the 5-year overall survival (OS) by 10-13 while the postoperative complication rate did not increase [12] Patients with a potentially curative resectable EC who do not meet the original CROSS study inclusion criteria are currently also treated with nCRT ie including patients aged over 75 years and those with a tumor length gt8 cm a tumor that extends gt2ndash4 cm into the gastric cardia andor gt10 body weight loss Moreover the original CROSS study excluded patients with celiac lymph node metastases because these nodes were previously classified as distant metastases (M1a) in the American Joint Committee on Cancer (AJCC) TNM 6th edition [3] The currently used 7th edition of the AJCC TNM classifies celiac node involvement as regional metastasis (N1-3) and these patients are treated with nCRT [4]Besides a small Dutch study which found that the extended inclusion criteria tumor length gt8 cm and age over 75 years did not influence the complication rate no study has assessed the influence of extension of all CROSS eligibility criteria for nCRT on toxicity and survival [5] This study was designed to assess the effect of extended eligibility criteria for treatment with nCRT on the toxicity and mortality (lt90 days posttreatment) of EC patients Furthermore we assessed the difference in disease-free (DFS) and overall survival (OS) between patients that met the original CROSS study inclusion criteria and patients in the extended inclusion group


Patients Data for this retrospective study were obtained from a prospectively maintained database and the study was conducted according to the national guidelines and the rules approved by the local ethics board All patients with locally advanced EC (TNM7 T1N1-3T2-4aN0-3M0) who underwent nCRT according to the CROSS schedule followed by surgery between 2005 and 2015 at the University Medical Center Groningen were eligible for inclusion All patients included in the study had a histologically proven adenocarcinoma or squamous cell carcinoma of the esophagus or esophagogastric junction In addition the patients had an adequate hematologic renal hepatic and pulmonary function together with a World Health Organization (WHO) performance status of 2 or lower

91

6


Based on the aforementioned criteria 177 patients were eligible for inclusion A total of 16 patients were excluded because of concurrent malignancies (n = 3) previous malignancies within 5 years before treatment (n = 3) missing blood values (n = 7) progressive disease due to distant metastases present on the restaging PETCT (n = 2) or a prolonged interval (gt6 months) between nCRT and surgery (n = 1) Consequently 161 patients were included in the study

MethodsThe patients were divided in two groups Group 1 consisted of 89 patients who met the original CROSS study inclusion criteria and group 2 consisted of 72 patients with the extended nCRT criteria Group 2 included 24 patients with a tumor longer than 8 cm 35 patients with more than 10 weight loss 21 patients with more than 2ndash4 cm tumor extension in the gastric cardia 13 patients with celiac lymph node metastasis and 2 patients older than 75 yearsThe primary objective was to assess the difference in nCRT-related toxicity (grade ge 3) between group 1 and 2 All treatment complications and severity were measured according to the National Cancer Institutersquos Common Terminology Criteria for Adverse Events (CTCAE) version 40 grading scale [6] The secondary outcomes were the difference in postoperative complications postoperative mortality (30- and 90-day rates) DFS and OS DFS was defined as the time between the start of nCRT and the date of tumor recurrence and OS as the time between the start of nCRT and the date of death or last follow-upIn addition we compared OS of the extended CROSS group with a reference dCRT group using a multivariate Cox regression analysis containing all confounders (gender cTN-stage tumor location tumor length histology and age)

StagingAll patients were staged with endoscopic ultrasonography combined with a fine needle aspiration biopsy when indicated computed tomography (CT) of the chest and abdomen and 18F-fluorodeoxyglucose (FDG) positron emission tomography (18F-PET) or integrated 18F-FDG PETCT When indicated additional imaging was performed Patients were staged according to the 7th edition of the the tumor-node-metastasis (TNM) classification [4]

TreatmentAll patients received nCRT according to the CROSS schedule consisting of five weekly intravenous administrations of carboplatin [area under the curve (AUC) 2 mgmlmin] and paclitaxel (50 mgm2) as well as concurrent external beam radiotherapy (414 Gy23 fractions) 5 days per week [12] After nCRT either a radical transthoracic or minimally invasive esophagectomy was performed

92


with en-bloc dissection of mediastinal and abdominal lymph nodes Definitive chemoradiotherapy (dCRT) consisted of either carboplatinpaclitaxel (AUC 2 and 50 mgm2) or cisplatin and fluorouracil (Cis-5FU 75 mgm2 and 1 gm2) combined with radiotherapy (40-60 Gy in 30 fractions)

PathologyResected specimens were pathologically assessed according to a standard protocol on histologic subtype radicality of the resection margins (proximal distal and circumferential) pathologic T (ypT-) stage pathologic lymph node (ypN-) stage tumor location perineural growth and lymphangio-invasion

Follow-up EvaluationAccording to the standard protocol patients were seen every 3 months during the first year every 4 and 6 months during the second and third year and subsequently once every succeeding year until 10 years after treatment During the follow-up tumor recurrence andor cause of death was accurately described Tumor recurrence was proven either pathologically or radiologically

Statistical AnalysisDifferences in patient characteristics and complications were assessed using the Chi square test or the likelihood ratio test for categorical variables and the Mann Whitney U test for non-normally distributed variables Kaplan-Meier curves were used to display the DFS and OS Univariate Cox regression analysis was performed on all possible prognostic factors for both DFS and OS All factors with a P value lower than 010 in the univariate Cox regression analysis were included in the multivariate Cox regression analysis A P value lower than 005 was considered statistically significant All statistical analyses were performed with IBM SPSS Statistics for Windows version 220 (IBM Corp Armonk NY USA)

RESULTS

Patientsrsquo CharacteristicsThe characteristics of the patients are summarized in Table 1 The group 2 patients (n = 72) were more likely to have a tumor involving the gastroesophageal (GE) junction (P = 0005) a higher clinical T stage (cT P = 0000) and a higher clinical N stage (P = 0024) than the group 1 patients (n = 89) In addition significantly more patients in group II died (P = 0004) and the follow-up period was significantly shorter for group 2 with a median follow-up of 162 months [interquartile range (IQR) 92ndash403 months] compared with 232 months (IQR 118ndash529 months) for group 1 (P = 0037) In group 1 and 2 respectively 798

93

6


and 806 of the patients were able to complete the entire nCRT regimen (Table 2) Of the patients in group 2 12 (167) fulfilled two extended criteria 4 (56) fulfilled three criteria and 1 (14) fulfilled four criteria The presence of two or more extended eligibility criteria within a patient (n = 17) versus only one extended criterion (n = 55) did not influence the OS (P = 0642) or DFS (P = 0198)

Table 1 Patient and tumor characteristics of group 1 (CROSS inclusion criteria) and group 2 (not eligible for CROSS)

Group 1 (n = 89) n ()

Group 2 (n = 72)n ()

P value

Male 71 (798) 57 (792) P = 0924 a

Age in years median (IQR) 63 (58 - 67) 64 (57 - 69) P = 0299 b

WHOECOG Performance Status P = 0843 a

0-1 85 (955) 64 (889)2 0 (00) 0 (00)Missing 4 (45) 8 (111)

Comorbidities total 44 (494) 38 (528) P = 0673 a

Cardiovascular 34 (382) 28 (389) P = 0798 a

Pulmonary 3 (34) 1 (14)Cardiovascular amp Pulmonary 5 (56) 6 (83)Other 2 (22) 3 (42)No comorbidities 45 (506) 34 (472)

Histology P = 0095 a


Tumor location P = 0005 Middle esophagus 7 (79) 5 (69)Distal esophagus 76 (854) 49 (681)GEJ 6 (67) 18 (250)

Tumor length (cm) median (IQR) 50 (30 - 60) 65 (50 - 90) P = 0000 b

cT-stage P = 0000 a

T1 0 (00) 2 (28)T2 25 (281) 5 (69)T3 63 (708) 56 (778)T4a 1 (11) 9 (125)

94


cN-stage P = 0024 a

N0 22 (247) 7 (97)N1 38 (427) 30 (417)N2 27 (303) 29 (403)N3 2 (22) 6 (83)

ypT-stage P = 0525 a

CR 15 (169) 13 (181)T0 4 (45) 2 (28)T1 17 (191) 8 (111)T2 11 (124) 9 (125)T3 42 (472) 40 (556)ypN-stage P = 0706 a

N0 57 (640) 44 (611)N1 18 (202) 16 (222)

N2 11 (124) 7 (97)N3 3 (34) 5 (69)

Perineural growth 15 (169) 18 (250) P=0204 a

Lymph-angioinvasion 14 (157) 19 (264) P=0097 a

LN ratio (gt02 LN+) 12 (135) 11 (153) P=0747 a

Follow-up in months median (IQR) 232 (118 - 529) 162 (92 - 403) P=0037 b

Abbreviations IQR = interquartile range GEJ = gastroesophageal junction cT = clinical T-stage cN = clinical N-stage ypT = pathologic T-stage ypN = pathologic lymph node stage and LN = lymph node a = Likelihood Ratio b = MannndashWhitney U test

Toxicity and post-operative survivalTable 2 displays the distribution of nCRT toxicity postoperative complications and postoperative mortality (30- and 90-day rates) between the two groups A total of 48 patients (298) experienced severe toxicity (grade ge 3) or received a blood transfusion The total toxicity rates did not differ between the two groups (P = 0117) nor did the number of postoperative complications (data shown in Table 2)Although more patients in group 2 (n = 7 97) died within 90 days after surgery than in group 1 (n = 6 67) this difference was not significant (P = 0492) In addition the 30-day postoperative mortality did not differ between the two groups (P = 0486) with a 30-day mortality rate of 22 (n = 2) in group 1 and 42 (n = 3) in group 2

95

6


Overall survival Figure 1 displays the KaplanndashMeier curves with the OS and DFS for both group 1 and 2 The OS differed significantly between the two groups (P = 0004 Fig 1a) with a median of 373 months (95 confidence interval [CI] 104ndash642 months) in group 1 and 172 months (95 CI 138ndash207 months) in group 2 Table 3 displays the extended CROSS criteria and the factors with a P value lower than 010 in the univariate analysis Independent prognostic factors for OS in the multivariate Cox regression analysis were ypN (P = 0023) ypT (P = 0043) and group 2 (P = 0008) In a multivariate Cox regression analysis that assessed each eligibility criterion separately only celiac lymph node involvement [hazard ratio (HR) 3583 95 CI 1884ndash6814 P = 0000] was an independent prognostic factor for OS

Table 2 Treatment toxicity and complicationsGroup 1 (n=89) n ()

Group 2 (n=72) n ()

P-value

Completed nCRT 71 (798) 58 (806) P = 0902 a

Hematologic toxicityThrombocytopenia ndash overall P = 0068 a

Not applicable 26 (292) 28 (389)Grade 1 54 (607) 43 (597)Grade 2 8 (90) 1 (14)Grade 3 1 (11) 0 (00)

Leukopenia ndash overall P = 0338 a

Not applicable 15 (169) 20 (278)Grade 1 34 (382) 21 (292)

Grade 2 26 (292) 19 (264)Grade 3 13 (146) 12 (167)Grade 4 1 (11) 0 (00)

Blood transfusion P = 0417 a 0 87 (978) 67 (931)1 0 (00) 1 (14)2 1 (11) 1 (14)3 1 (11) 2 (28)4 0 (00) 1 (14)

Other nCRT complications (gt3 grade)Anemia 0 (00) 0 (00) NA a

Bleeding 0 (00) 1 (14) P = 0203 a

Nausea 3 (34) 4 (56) P = 0501 a

96


Fatigue 1 (11) 1 (14) P = 0880 a

Neurotoxic 0 (00) 2 (28) P = 0071 a

Diarrhea 0 (00) 1 (14) P = 0203 a

Esophagitis 2 (22) 5 (69) P = 0144 a

ge Grade 3 or blood transfusion 22 (247) 26 (361) P = 0117 a

Post-operative complications Pulmonary (all grades) 49 (551) 38 (528) P = 0773 a

Pneumonia 41 (461) 28 (389) P = 0360 a

Respiratory insufficiency 19 (213) 13 (181) P = 0602 a

Pulmonary embolism 2 (22) 0 (00) P = 0122 a

Cardiac (all grades) 26 (292) 22 (306) P = 0835 a

Arrhytmia 25 (281) 22 (306) P = 0732 a

Myocardial infarction 1 (14) 0 (00) P = 0273 a

Sepsis 8 (90) 6 (83) P = 0883 a

Post-operative bleeding 2 (22) 1 (14) P = 0678 a

Chylothorax 11 (124) 3 (42) P = 0057 a

Cardiac arrest 2 (22) 3 (42) P = 0486 a

Esophageal anastomotic leak 8 (90) 12 (167) P = 0143 a

Renal failure 2 (22) 4 (56) P = 0276 a

Ileus 6 (67) 2 (28) P = 0237 a

All patients with complications (all grades)

60 (674) 50 (694) P = 0783 a

Post-operative mortality30-day mortality 2 (22) 3 (42) P=0486 a

90-day mortality 6 (67) 7 (97) P=0492 a

Abbreviations NA = not applicable a= Likelihood Ratio = pneumonia atelectasis respiratory insufficiency acute respiratory distress syndrome pleural effusion pneumothorax andor pulmonary embolism = arrhythmia andor myocardial infarction

97

6


Table 3 Prognostic factors on univariate and multivariate Cox regression analysis for overall survival

Univariate analysisHR (95 CI) P-value

Group 2 1802 (1200-2707) 0005Celiac lymph node metastasis 3969 (2188-7198) 00002 -4 cm cardia growth 1329(0721-2452) 0362gt 8 cm length 1217 (0699-2118) 0488gt 10 weightloss 1407 (0892-2217) 0142Squamous cell carcinoma 0543 (0295-1000) 0050ypT0 1000 0008ypT1 0589 (0254-1367)ypT2 1945 (0974-3884)ypT3 1778 (1019-3100) ypN0 1000 0000ypN1 1518 (0914-2522)ypN2 2144 (1158-3968)ypN3 5024 (2215-11398)R1 resection 3266 (1543-6912) 0002LN-ratio (gt 02 LN+) 229 (1437-4105) 0001Perineural growth 2076 (1314-3279) 0002Lymph-angioinvasion 1829 (1125-2874) 0015

Multivariate analysisHR (95 CI) P-value

ypT0 1000 0043ypT1 0540 (0224-1301)ypT2 1798 (0854-3789)ypT3 1294 (0704-2378)Group 2 1762 (1157-2685) 0008ypN0 1000 0023ypN1 1349 (0805-2263)

ypN2 1896 (0989-3635)ypN3 3415 (1446-8064)

Variables with P lt 010 on univariate analysis were included in the multivariate analysis Abbreviations CI = confidence interval ypT = pathologic T-stage ypN = pathologic lymph node stage LN = lymph node

98


Disease-free survivalThe difference in DFS between group 1 and 2 approached significance (P = 0073 Fig 1b) with a median of 425 months (95 CI 157ndash694 months) in group 1 and 182 months (95 CI 74ndash289 months) in group 2 Table 4 displays the extended CROSS criteria and the factors with a P value lower than 010 in the univariate analyses as well as the independent prognostic factors in the multivariate Cox regression analysis for DFS Gender (P = 0024) LN ratio (P = 0001) squamous cell carcinoma (P = 0031) and group 2 (P = 0027) were independent prognostic factors for DFS A closer look at specific subgroups of group 2 with multivariate Cox regression analysis showed that only celiac lymph node metastasis was an independent prognostic factor for DFS (HR 3741 CI 1822ndash7680 P = 0000)

Figure 1 The overall and disease-free survival in patients that met the original CROSS criteria or the extended CROSS eligibility criteria (a b) in patients with or without celiac lymph node metastases (c d) and in patients that met the original CROSS criteria or the extended CROSS eligibility criteria without celiac lymph node metastases (e f) And the overall survival in patients that met the extended CROSS eligibility criteria or patients from a definitive chemoradiotherapy reference group (g)

99

6


Table 4 Prognostic factors on univariate and multivariate Cox regression analysis for disease-free survival


Group 2 1509 (0959-2375) 0075Celiac lymph node metastasis 3898 (1923-7904) 00002 -4 cm cardia growth 1454 (0742-2849) 0275gt 8 cm length 1103 (0580-2097) 0764gt 10 weightloss 1229 (0720-2096) 0450Female 0484 (0255-0920) 0027Squamous cell carcinoma 0366 (0167-0802) 0012cT1 amp T2 1000 0084cT3 1961 (1003-3833)cT4a 2894 0984-8510)ypT0 1000 0023ypT1 1535 (0636-3706)ypT2 3056 (1298-7194)ypT3 2632 (1275-5435)ypN0 1000 0001ypN1 1470 (0812-2659)ypN2 3060 (1618-5785)ypN3 4374 (1682-11375)R1 resection 4389 (2043-9431) 0000LN-ratio (gt 02 LN+) 3106 (1758-5489) 0000Perineural growth 1694 (0993-2890) 0053Lymph-angioinvasion 1940 (1131-3327) 0016


Female 0474 (0248-0907) 0024Squamous cell carcinoma 0413 (0185-0923) 0031Group 2 1685 (1061-2676) 0027LN-ratio (gt 02 LN+) 2712 (1524-4826) 0001

Variables with P lt 010 on univariate analysis were included in the multivariate analysis Abbreviations CI = confidence interval cT = clinical T-stage ypT = pathologic T-stage ypN = pathologic lymph node stage LN = lymph node

100


Comparison of Survival Between the Extended CROSS and dCRT Reference Group Supplementary Table 1 depicts the characteristics of the dCRT and extended CROSS group The dCRT group (n = 80) and the extended CROSS group (n = 72) differed in cT stage (P = 0001) cN stage (P = 0000) squamous cell carcinoma (P = 0006) tumor location (P = 0001) age (P = 0021) and WHO performance status (P = 0007) The patients in the extended CROSS group showed an increased OS (P = 0010 Fig 1g) with the log-rank test but not in the Cox-regression model (Supplementary Table 2) that contained possible confounders The number of complications grade ge 3 did not differ between the two groups (P = 0115)

DISCUSSION

Several randomized studies including the CROSS study have shown that nCRT increases both OS and DFS for EC patients with locoregional disease compared with surgery alone [17] Moreover pathologic complete response rates of approximately 30 are commonly observed after nCRT [1] Extending the original criteria for CROSS nCRT is a logical step to improve survival in locally advanced EC In this study we assessed the impact of extended eligibility criteria for nCRT on toxicities OS and DFS in these patients No difference was found in the toxicity rates between patients in group 1 (original CROSS criteria) and group 2 (extended CROSS criteria) However the OS and DFS in group 2 were significantly lower in the multivariate Cox regression analysis Schrauwen et al (n = 116) found that the extended inclusion criteria based on tumor length greater than 8 cm (n = 7) and age over 75 years (n = 9) had no influence on the complication rates but were prognostic for OS with the log-rank test [5] However interpreting these results is difficult due to the low number of patients the absence of multivariate analysis and absence of celiac lymph node metastases in the analysis [5] The overall rate of toxicity (grade ge 3) or blood transfusion was not significantly higher in group 2 (247) than in group 1 (361) (P = 0117) The incidences of severe leukopenia (grade ge 3) in group 1 (157) and group 2 (167) were somewhat higher than the 6 in the original CROSS trial but within the range of 3ndash24 in the literature [189] Furthermore the observed rates of thrombocytopenia grade 3 or higher of 11 in group 1 and 0 in group 2 correspond well with the 1 rate of thrombocytopenia in the CROSS trial The 30-day mortality rates in group 1 (22) and group 2 (42) are also comparable with the mortality rate of 2 in the original CROSS study [1] Thus the CROSS nCRT schedule in group 2 is not associated with significantly higher hematologic or non-hematologic toxicity and can be safely applied in the extended patient category

101

6


The 5-year OS of 47 (median 486 months) found in the Dutch randomized CROSS trial is comparable with the 43 (median 373 months) in our group 1 [12] Conversely the extended criteria group 2 had a remarkably lower 5-year OS of 23 (median 172 months) The median survival after noninvasive dCRT an alternative for patients with considerable comorbidity is 16ndash21 months raising the question whether dCRT is worth considering for the extended patient category [10-13] Nevertheless direct comparison of survival rates in the dCRT and extended CROSS group is not possible because dCRT studies also included irresectable tumors and inoperable patients In the included dCRT reference group we found a significantly lower OS (P = 0010) with the univariate log-rank test However this test does not correct for baseline differences (gender cTN stage tumor localization tumor length histology and age) between the extended nCRT group and the dCRT group Hence a multivariate Cox regression analysis containing these confounding variables was performed in which the OS did not differ (P = 0445) between the extended CROSS group and the dCRT group This suggests that the difference in survival curves might be caused by baseline differences between the groups rather than superiority of nCRT followed by surgery over dCRTSeveral studies found a comparable outcome in patients with celiac and regional lymph node metastasis Celiac lymph node metastases are therefore currently classified as regional lymph nodes (N+) whereas previous classification systems regarded them as distant (M1a) [14-16] In the current study the presence of tumor-positive celiac lymph nodes (n = 13) was the only extended eligibility criterion with an independent prognostic value We compared the survival of patients with celiac lymph node metastases in the extended CROSS group (n = 13) with M1a patients in the dCRT group (the latter involving both irresectable higher mediastinal and celiac nodes n = 15) and found no difference in survival (P = 0336) However the groups were too small for a solid conclusion Davies et al found that celiac lymph node metastasis (determined by endoscopic ultrasound) was not prognostic for OS after dCRT which was confirmed by Gwynne et al [1013] However further research seems necessary to elucidate the value of dCRT for patients with celiac lymph node metastasis probably in a randomized controlled trial or a large retrospective study The potential limitations of our study include the small sample size especially the subgroup of patients with celiac lymph node metastases (n = 13) Moreover two of these patients died within 90 days after surgery which may have influenced the OS Another potential weakness is that we included only patients who received surgery whereas approximately 8 experience interval metastases between nCRT and surgery [17]

102


In conclusion extension of the original CROSS inclusion criteria for nCRT followed by surgery in EC did not influence the toxicity rate indicating safe application of the CROSS nCRT regimen in the extended patient category However the OS in the extended CROSS group was significantly lower than in the standard CROSS group and did not differ significantly from the OS in the dCRT reference group in the multivariate Cox regression analysis This implies that the additional value of nCRT followed by surgery compared with dCRT in the extended CROSS group might be limited The findings of this study support further research regarding the strategy to extend the original CROSS criteria for nCRT in patients with locally advanced EC and should focus more on patients with celiac node metastases

103

6


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[2] Shapiro J van Lanschot JJ Hulshof MC et al Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS) long-term results of a randomised controlled trial Lancet Oncol 2015 161090-1098[3] American Joint Committee on Cancer AJCC cancer staging manual 6th ed New York Springer 2002[4] American Joint Committee on Cancer AJCC cancer staging manual 7th ed New York Springer 2009[5] Schrauwen RWM Bisseling TM Bonenkamp JJ Radema SA Ruumltten H Rosman C Braam PM Uitkomsten van neoadjuvante chemoradiotherapie gevolgd door slokdarmresectie bij patieumlnten met een hogere leeftijd of een grotere tumorlengte NTvO Mar 2015 201512(2)50-57[6] Anonymous National Cancer Institute Common Terminology Criteria for Adverse Events v30and v40 (CTCAE) Available athttpctepcancergovprotocolDevelopmentelectronic_applicationsctchtm Accessed 2015[7] Sjoquist KM Burmeister BH Smithers BM et al Survival after neoadjuvant chemotherapy or chemoradiotherapy for resectable oesophageal carcinoma an updated meta-analysis Lancet Oncol 2011 12681-692[8] Blom RL van Heijl M Klinkenbijl JH et al Neoadjuvant chemoradiotherapy followed by esophagectomy does not increase morbidity in patients over 70 Dis Esophagus 2013 26510-516[9] van Meerten E Muller K Tilanus HW et al Neoadjuvant concurrent chemoradiation with weekly paclitaxel and carboplatin for patients with oesophageal cancer a phase II study Br J Cancer 2006 941389-1394[10] Davies L Lewis WG Arnold DT et al Prognostic significance of age in the radical treatment of oesophageal cancer with surgery or chemoradiotherapy a prospective observational cohort study Clin Oncol (R Coll Radiol) 2010 22578-585[11] Honing J Smit JK Muijs CT et al A comparison of carboplatin and paclitaxel with cisplatinum and 5-fluorouracil in definitive chemoradiation in esophageal cancer patients Ann Oncol 2014 25638-643[12] Shao MS Wong AT Schwartz D Weiner JP Schreiber D Definitive or Preoperative Chemoradiation Therapy for Esophageal Cancer Patterns of Care and Survival Outcomes Ann Thorac Surg 2016 1012148-2154

104


[13] Gwynne S Hurt C Evans M Holden C Vout L Crosby T Definitive chemoradiation for oesophageal cancer--a standard of care in patients with non-metastatic oesophageal cancer Clin Oncol (R Coll Radiol) 2011 23182-188[14] Hofstetter W Correa AM Bekele N et al Proposed modification of nodal status in AJCC esophageal cancer staging system Ann Thorac Surg 2007 84365-73 discussion 374-5[15] Talsma K van Hagen P Grotenhuis BA Steyerberg EW Tilanus HW van Lanschot JJ Wijnhoven BP Comparison of the 6th and 7th Editions of the UICC-AJCC TNM Classification for Esophageal Cancer Ann Surg Oncol 2012 192142-2148[16] Sepesi B Schmidt HE Lada M et al Survival in Patients With Esophageal Adenocarcinoma Undergoing Trimodality Therapy Is Independent of Regional Lymph Node Location Ann Thorac Surg 2016 1011075-1081[17] Stiekema J Vermeulen D Vegt E et al Detecting interval metastases and response assessment using 18F-FDG PETCT after neoadjuvant chemoradiotherapy for esophageal cancer Clin Nucl Med 2014 39862-867

105

6


Supplementary table 1 Patient and tumor characteristics of the extended CROSS group and the definitive chemoradiotherapy group

Group II (n=72) n ()

dCRT group (n=80) n ()

P value

Male 57 (792) 61 (763) P = 0666 a



0-1 64 (889) 72 (900)2 0 (00) 6 (75)Missing 8 (111) 2 (25)






Tumor location P = 0001 a



cT-stage P = 0001 a

T1 2 (28) 1 (13)T2 5 (69) 4 (50)T3 56 (778) 39 (488)T4a 9 (125) 32 (400)missing 0 (00) 4 (50)

cN-stage P = 0000 a

N0 7 (97) 14 (175)N1 30 (417) 61 (763)N2 29 (403) 4 (50)N3 6 (83) 1 (13)


Follow-up in months median (IQR) 162 (92 - 403) 138 (80 ndash 221) P = 0060 b

Abbreviations dCRT = definitive chemoradiotherapy IQR = interquartile range GEJ = gastroesophageal junction cT = clinical T-stage cN = clinical N-stage a = Likelihood Ratio b = MannndashWhitney U test

106


Supplementary table 2 Multivariate Cox regression analysis for overall survival in the extended CROSS and definitive chemoradiotherapy group

Overall survivalHR (95 CI) P value

Age gt75 0669 (0321-1393) 0283Female 0497 (0281-0881) 0017Tumor length 0967 (0892-1048) 0410Squamous cell carcinoma 0551 (0307-0988) 0046cT1 amp T2 1000 0062cT3 0584 (0257-1330)cT4a 1019 (0394-2636)cN0 1000 0197cN1 0581 (0314-1074)cN2 amp cN3 0726 (0346-1522)Mid 1000 0031Distal 0460 (0238-0888)GEJ 0337 (0145-0783)nCRT 100 0445dCRT 1233 (0720-2113)

All possible cofounding variables for OS were included in the multivariate models Abbreviations HR= hazard ratio CI = confidence interval GEJ = gastroesophageal junction cT = clinical T-stage cN = clinical N-stage nCRT = neoadjuvant chemoradiotherapy and dCRT = definitive chemoradiotherapy

107

6



Improving prediction of response to neoadjuvant chemoradiotherapy

PART II



Predicting response to neoadjuvant chemoradiotherapy in esophageal cancer with textural features derived from pre-treatment

18F-FDG PETCT imaging

Roelof J Beukinga Jan Binne Hulshoff Lisanne V van Dijk Christina T Muijs

Johannes GM Burgerhof Gursah Kats-Ugurlu Riemer HJA Slart Cornelis H Slump

Veacuteronique EM Mul John ThM Plukker

Journal of Nuclear Medicine 2016 October 13

CHAPTER 7


ABSTRACT

BackgroundAdequate prediction of tumor response to neoadjuvant chemoradiotherapy (nCRT) in esophageal cancer (EC) patients is important in a more personalized treatment The current best clinical method to predict pathologic complete response is the maximal standardized uptake value (SUVmax) in 18F-fluorodeoxyglucose positron emission tomographycomputed tomography (18F-FDG PETCT) imaging To improve the prediction of response we constructed a model to predict complete response to nCRT in EC based on pre-treatment clinical parameters and 18F-FDG PETCT derived textural features

MethodsFrom a prospectively maintained single institution database we reviewed 97 consecutive patients with locally advanced EC and a pre-treatment 18F-FDG PETCT between 2009 and 2015 All patients were treated with nCRT (CarboplatinPaclitaxel414Gy) followed by esophagectomy We analyzed clinical geometrical and pre-treatment textural features extracted from both 18F-FDG PET and CT Current most accurate prediction model with SUVmax as predictor variable was compared with five different response prediction models constructed using Least Absolute Shrinkage and Selection Operator regularized logistic regression Internal validation was performed to estimate the modelrsquos performances Pathologic response was defined as complete versus incomplete response (Mandard tumor regression grade system 1 vs 2-5)

ResultsPathologic examination revealed 19 (196) complete and 78 (804) incomplete responders Least Absolute Shrinkage and Selection Operator regularization selected the clinical parameters histologic type and clinical T-stage the 18F-FDG PET derived textural feature lsquolong run low gray level emphasisrsquo and the CT derived textural feature lsquorun percentagersquo Introducing these variables to a logistic regression analysis showed areas under the receiver operating characteristic curve (AUCs) of 078 compared to 058 in the SUVmax model The discrimination slopes were 017 compared to 001 respectively After internal validation the AUCs decreased to 074 and 054 respectively

ConclusionThe predictive values of the constructed models were superior to the standard method (SUVmax) These results can be considered as an initial step in predicting tumor response to nCRT in locally advanced EC Further research in refining the predictive value of these models is needed to justify omission of surgery

112


INTRODUCTION

EC is one of the most aggressive tumors with early recurrences even after radical surgery The standard treatment in locally advanced (T1N1-3M0 and T2-4aN0-3M0) resectable EC is nCRT followed by a radical esophagectomy In the Dutch CROSS (ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study) trial nCRT improved the 5-year overall survival rate from 34 to 47 [1] Not all patients benefit from nCRT 29 of the patients in the CROSS trial had a complete response 52 had a partial response and even 18 had no tumor response [1] For complete responders surgical intervention might not be beneficial and a lsquowait-and-see policyrsquo might suffice Hence adequate response prediction is important in developing personalized treatment in EC Moreover accurate response prediction may be relevant in patient counseling in future clinical trial strategies based on personalized treatment So far response prediction showed only promising results with functional imaging of tumor viability with 18F-fluorodexoglucose positron emission tomography (18F-FDG PET) and recently with diffusion weighted magnetic resonance imaging [2-4] Traditional image-derived indices used in PET rely on quantification of lesion SUVs and overall tumor volume which have been shown to be important factors for patient outcome and treatment response [56] Although useful these parameters omit available information related to the spatial distribution and specific features regarding intra-tumor radiotracer accumulation This may limit the possibility to further characterize the biological behavior of the tumor based on hypoxia induced heterogeneity and genomic instability Intratumoral heterogeneity is correlated with aggressive tumor behavior and a decreased response due to expression of specific receptors with high cellular proliferation and angiogenesis [7-9] Hence even small tumor biopsies lack complete molecular characterization due to spatially heterogeneity A novel approach is to quantify spatial heterogeneity of metabolism and tissue density characterized by 18F-FDG uptake and Hounsfield Units with textural features The concept of textural analysis is based on spatial arrangement of voxels in a predefined volume of interest (VOI) This spatial intratumoral heterogeneity can be depicted from different spatial interrelationships on 18F-FDG PETCT scans Therefore 18F-FDG PETCT textural features have been proposed to be valuable in response prediction [10-15] The aim of this study was to develop a model to predict complete response to nCRT in locally advanced EC based on pre-treatment clinical predictors and 18F-FDG PETCT derived textural features

113

7



PatientsIn this retrospective study potentially curatively resectable EC patients were consecutively selected who underwent nCRT followed by esophagectomy between December 2009 and March 2016 Patients with less than 4 courses of chemotherapy with missing 18F-FDG PETCT or with incomplete medical records were excluded yielding a total of 97 patients In line with the rules of the Dutch National Health Sciences our Institutional Review Board approved this retrospective study and the requirement to obtain informed consent was waivedData were obtained from a prospectively maintained single institution database including patient characteristics tumor and treatment related data and follow-up data All patients were clinically staged with esophagoscopy and biopsy endoscopic ultrasonography with fine needle aspiration if indicated and whole-body integrated 18F-FDG PETCT Patients were staged according to the 7th edition of the TNM system maintained by the American Joint Committee on Cancer [16] and discussed in the hospitalrsquos multidisciplinary esophageal tumor board

ImagingPETCT imaging was performed with an integrated 18F-FDG PETCT system (Biograph mCT 4-64 PETCT Siemens Knoxville TN USA) Patients fasted for at least 6 hours prior to PETCT with no restrictions on drinking water Serum glucose levels were measured before 18F-FDG administration with a weight-based dose of 3 MBqkg Sixty minutes after tracer injection patients were scanned in treatment position An inspiration breath-hold low-dose CT for attenuation correction was performed and PET acquisitions were obtained in caudalndashcranial direction with field of view 500times500times500 mm 3D setting 2-3 min per bed position matrices of 512 times 512 (098 times 098 mm pixel size) and 2-mm slice thickness Image data were reconstructed according to European Association of Nuclear Medicine guidelines [17]Radiotherapy treatment planning including target volume delineation and CT texture analysis was performed on a 16- or 64-multidetector row spiral CT machine (Somatom Sensation 16 or 64 Siemens Medical Systems Erlangen Germany) CT thoraxabdomen was obtained in cranialndashcaudal direction with matrices of 512 times 512 (098 times 098 mm pixel size) and 3-mm slice thickness

Treatment and PathologyBased on the experiences and the good results of the CROSS study in which our institute had participated our multidisciplinary tumor board decided to continue nCRT according to the CROSS schedule This treatment consisted of

114


weekly intravenous administered Paclitaxel (50 mgm2) and Carboplatin (AUC 2 mgminmLminus1) during 5 weeks with concurrent external radiotherapy (414 Gy in 23 fractions 5 days per week) [1] Transthoracic esophagectomy with two-field lymphadenectomy was performed within 6 to 8 weeks after completion of nCRT The resected specimens were examined according to a standard protocol [18] Resection margins were defined according to the definitions of the College of American Pathologist as microscopic tumor-free (R0 gt0mm) or tumor-positive (R1) Pathologic response was assessed by two expert gastrointestinal pathologists according to the Mandard tumor regression grade (TRG) ranging from complete response (TRG 1) without viable tumor cells left to partial response (TRG 2-4) with viable tumor cells left to no response at all (TRG 5) [19]

Volume of interestTextural analysis was performed on a VOI incorporating the gross tumor volume for radiation treatment planning Tumor delineation was performed manually with consensus between three experienced radiation oncologists on axial planes of the radiotherapy planning CT to enclose three-dimensional coverage of the entire tumor Involved lymph nodes were not included into the VOI because these lesions are too small (lt10 cm3) for reliable textural analysis [20] The gross tumor volume was rigidly registered to the 18F-FDG PETCT data series (RTx Workstation 10 Mirada Medical Oxford UK) Erroneous registrations were manually adjusted after consensus of the collaborating investigators (RJB CTM VEM and JThP)

Tonal discretization18F-FDG PETCT imaging data and VOI delineations were loaded into Matlab 2014b (Mathworks Natick MA an interactive image processing environment) for processing and analyses The SUV for semi-quantitative analysis of metabolism was corrected for individual variations in serum glucose level and was discretized to reduce the continuous scale to a finite set of values and to reduce noise throughout the entire study in 05 gmL increments according to Doanersquos optimal bin width [21] Similarly the Hounsfield unit scale for quantitative analysis of tumor density was discretized in 30 HU increments for textural analysis

Candidate predictorsFor each patient a total of 88 parameters were evaluated including 7 clinical parameters 16 geometry features the glycolytic volume based on tumor volume and SUVmean and 19 first order 24 second order and 22 higher order textural features extracted from 18F-FDG PET and CT (supplemental materials) First order textural features are statistics based on the gray level distribution of the image but do not consider relative positions of gray levels Second and higher

115

7


order textural features do consider relative positions of gray levels and therefore allow quantification of heterogeneity For various spatial interrelationships frequency distributions (Figure 1) were obtained ie the gray level co-occurrence ( spatial dependence) matrix for pairwise arrangement of voxels (extracted with a pixel-to-pixel distance equal to 1) [22] the gray level run-length matrix for alignment of voxels with the same intensity [23] and the gray level size-zone matrix for characteristics of homogenous zones [24] Directional voxel analysis was performed in 3 dimensions with a connectivity of 26 voxels and analysis in 13 angular directions All second and higher order textural features are weighted averages of these matrices to express the relative importance of their properties All extracted textural features were normalized to the range (01)

Figure 1 A) Manual delineation of the gross tumor volume on planning CT B) Co-registration of LD-CT (lava colormap) to the radiotherapy planning CT (grayscale colormap) C) Overlay of PET image (lava colormap) onto radiotherapy planning CT (grayscale colormap) D) Cropping of PET VOI E) Feature extraction I) Assessment of tumor shape by means of geometry features II) Global assessment of tonal distribution by means of first order textural features III) Assessment of pairwise arrangement of voxels by means of grey level co-occurrence matrix IV) Assessment of alignment of voxels with the same intensity by means of grey level run-length matrix V) Assessment of characteristics of homogenous zones by means of grey level size-zone matrix

Statistical analysisStatistical analysis was performed with R 322 open-source software using the glmnet package (version 20-2) and the rms package (version 44-0) available from the Comprehensive R Archive Network (httpwwwr-projectorg)Since textural feature values may be subject to inter-observer variability in the delineation of the tumor the original delineations were uniformly eroded by ball-shaped structuring elements with radii of 1 and 2 voxels For each delineation

116


textural features were extracted and the stability of each feature was evaluated with the intra-class correlation (ICC) Only stable features (ICCgt07) were considered for further analysis Predictors were then selected by a univariable logistic model with a response variable labeling complete (Mandard TRG 1) and incomplete response (Mandard TRG 2-5) All potential predictors that met the Akaike Information Criterion (AIC) were considered significant To discourage overfitting the AIC is based on rewarding goodness-of-fit and penalizing the complexity of the model The AIC requires χ2 gt 2∙df ie when considering a predictor with one degree of freedom df this implies an α = P(χ2ge2) = 0157 [25] Significant predictors were used to construct six multivariable logistic regression models for comparison with current most accurate prediction model with SUVmax as predictor variable (model 1) These models were constructed by introducing clinical parameters (model 2) clinical parameters and geometry features (model 3) clinical parameters geometry features and PET textural features (model 4) clinical parameters geometry features and CT textural features (model 5) and clinical parameters geometry features and PETCT textural features (model 6) to a Least Absolute Shrinkage and Selection Operator a technique for L1-norm regularization By increasing the shrinkage parameter λ the regularization shrinks the estimated coefficients and excludes variables when they become zero The λ-value that minimized the 10-fold cross validated mean squared error was repeatedly determined with 100 repetitions The optimal λ-value was robustly determined by averaging over these obtained λ-values The selected variables were fitted to the data with a logistic regression The modelrsquos calibration was evaluated using visual inspection of calibration plots and the HosmerndashLemeshow test The modelrsquos performance was quantified in terms of discrimination with the AUC and the discrimination slope The goodness-of-fit was evaluated with the -2 log-likelihood and the Nagelkerke R2 The model was internally validated by a bootstrap approach with 2000 repetitions Bootstrapping allowed for obtaining the optimism corrected measures for model performance and for shrinkage of the estimated regression coefficients using the optimism-corrected slope

RESULTS

Patients and treatmentPatientrsquos characteristics are shown in Table 1 Seventy-nine patients (814) received the complete nCRT regimen (all patients received the full radiotherapy dose) Resection with curative intent was performed within a mean time of 56 (standard deviation 14) days after completion of nCRT R0 resection was achieved in 90 (928) patients and R1 resection in seven (72) patients all with positive circumferential resection margins and one with a positive proximal resection margin Pathological findings revealed complete response in 19 patients (196) and incomplete response in 78 patients (804)

117

7


Table 1 Patientrsquos characteristicsCharacteristic N Characteristic N Gender Male 82 845 cN stage cN0 21 217

Female 15 155 cN1 45 464Age lt70 y 78 804 cN2 28 289

ge70 y 19 196 cN3 3 31Histology AC 88 907 ypT stage ypT0 19 196

SCC 9 93 ypT1 13 134Tumor grade Missing 6 62 ypT2 13 134

G1 49 505 ypT3 52 536G3 42 433 ypN stage ypN0 63 650

EUS tumor length lt5 cm 37 402 ypN1 18 186ge5 cm 58 598 ypN2 10 103

Localization Mid 4 41 ypN3 6 62Distal 62 639 CRM R0 90 928GEJ 31 320 R1 7 72

nCRT cycles 4 18 186 Proximal resection margin R0 96 9905 79 814 R1 1 10

cT stage cT1 2 21 Mandard TRG 1 19 196cT2 16 165 2 23 237cT3 74 763 3 37 381cT4a 5 52 4 15 155

5 3 31

Abbreviations AC adenocarcinoma SCC squamous cell carcinoma EUS endoscopic ultrasonography GEJ gastresophageal junction nCRT neoadjuvant chemoradiotherapy TRG tumor regression grade cTN clinical tumornodal stage ypTN pathologic tumornodal stage after nCRT and CAP College of American Pathologists Tumor-free (R0) resection margin defined according to criteria of the CAP as gt0mm

Model developmentFor the preselection 144 of the 147 (973) variables were found to be robust for contour variations These variables were introduced to univariable logistic regression analysis resulting in 24 significant variables predictive for response including 4 clinical parameters 0 geometry features 1 first order 8 second and 5 higher order PET textural features and 1 first 1 second and 4 higher order CT textural features All constructed prediction models performed significantly better than model 1 (based on SUVmax) Introduction of only significant clinical parameters to the Least Absolute Shrinkage and Selection Operator regularization process resulted in the selection of histologic type and clinical T-stage (model 2) These variables were selected in each subsequently constructed model Compared

118


to model 1 the AUC improved from 058 to 071 the discrimination slope improved from 001 to 014 and the AIC decreased (ΔAIC=1066) For model 3 no additional variables were selected compared to model 2 since no geometry features were significant at the univariable logistic regression analysis For model 4 the PET textural feature lsquolong run low gray level emphasis (LRLGLe-PET)rsquo was selected Adding this variable did slightly improve the discrimination and the likelihood compared to model 2 and 3 but resulted in a higher AIC (ΔAIC=-079) After internal validation the AUC was equal to 069 For model 5 the CT textural feature lsquorun percentage (RP-CT)rsquo was selected Although the AIC was almost equal compared to model 2 and 3 (ΔAIC=-002) adding this variable improved the discrimination slope to 016 and the AUC remarkably to 079 This also persisted after internal validation (AUC=076) Finally entering all variables to the modeling process resulted in the selection of all above mentioned variables (model 6) Model 6 had the best goodness-of-fit but not the lowest AIC (ΔAIC=-023 056 and -021 compared to models 2 and 3 4 and 5 respectively) The AUC slightly decreased to 078 while the discrimination slope was increased to 017 After internal validation the AUC decreased to 074 The model regression coefficients and the corresponding model performance measures are shown in Table 2 and Table 3 respectively Figure 2 gives the values of the selected textural features and their corresponding frequency distributions for a complete and a non-responder For the selected textural features the range of values to reproduce the normalization process and the found ICCs for quantifying contouring robustness are given in (supplemental materials)

119

7


Figure 2 Example of the values of the selected textural features and their corresponding frequency distributions for a complete and a non-responder Abbreviations GLRLM gray level run-length matrix LRLGLe-PET long run low gray level emphasis measured on PET and RP-CT run percentage measured on CT

120


Tabl

e 2

Est

imat

ed r

egre

ssio

n co

effici

ents

of t

he p

redi

ctio

n m

odel

s for

pat

holo

gic

com

plet

e re

spon

se w

ithou

t opt

imis

m c

orre

ctio

n

Mod

el 1

Mod

el 2

ndash 3

Mod

el 4

Mod

el 5

Mod

el 6

Varia

ble

Coe

fS

Ep

Coe

fS

EP

Coe

fS

Ep

Coe

fS

Ep

Coe

fS

Ep

Inte

rcep

t-0

88

050

008

-04

20

500

39-0

86

065

019

-11

50

720

11-1

83

091

004

SUV

max

-17

21

450

24

His

tolo

gy

AC

100

100

100

100

SC

C-1

70

061

001

-14

70

650

02-1

89

064

000

-16

30

670

02cT

stag

e

cT1

amp c

T21

001

001

001

00

cT3

amp c

T4a

203

078

001

198

080

001

227

081

000

223

083

001

LRLG

Le-P

ET0

560

550

310

710

590

23

RP-

CT

0

010

010

150

020

010

10

Abb

revi

atio

ns L

RLG

Le-P

ET l

ong

run

low

gra

y le

vel e

mph

asis

mea

sure

d on

PET

and

RP-

CT

run

perc

enta

ge m

easu

red

on C

T

121

7


Table 3 Estimates of model performance for the four prediction models

Goodness-of-fitDiscrimina-tion

Calibration Validation

-2LLH AIC R2 AUC DS Intercept Slope HLpR2 boot

AUC boot

Model 1 9446 9846 002 058 001 209 246 075 000 054Model 2 ndash 3 8180 8780 022 071 014 -004 094 100 017 070Model 4 8059 8859 023 071 015 -015 087 045 017 069Model 5 7982 8782 024 079 016 -014 086 042 018 076Model 6 7803 8803 027 078 017 -022 081 046 018 074

Abbreviations -2LLH -2 log-likelihood AIC Akaike information criterion R2 Nagelkerke R2 AUC area under the receiver operating characteristic DS discrimination slope HLp HosmerndashLemeshow p-value and boot internal validated with bootstrapping

DISCUSSION

An adequate method to predict pathologic complete response after nCRT has not yet been defined in EC patients In personalized treatment accurate response prediction will lead to a paradigm shift with omitting surgical treatment in complete responders or preventing unnecessary nCRT in non-responders Response evaluation of nCRT is commonly based on tumor metabolic response measured by SUVmax with 18F-FDG PET but with a low sensitivity and specificity of 67 and 68 respectively [26] Current study is the first in predicting complete response with 18F-FDG PETCT derived textural features in a homogeneous group of EC patients treated according to the CROSS regimen We demonstrated that all constructed prediction models showed significant improvement compared to predictions based on SUVmax alone and may therefore be considered as an initial step in predicting responseIn this study the most predictive textural features were LRLGLe-PET and RP-CT LRLGLe-PET depends on long runs (coarse texture) with low gray levels and was higher (ie low and homogeneous 18F-FDG uptake) for complete responders and lower (ie high and heterogeneous 18F-FDG uptake) for incomplete responders possibly due to tumor hypoxia and necrosis RP-CT measures the homogeneity of runs (fine texture) and was higher in complete responders In univariable logistic regression high LRLGLe-PET and RP-CT values were associated with squamous cell carcinoma (P=012 and P=013 respectively) confirming the higher complete response rates in squamous cell carcinoma [1]The clinical value of SUVmax was limited possibly because it is extracted from a single voxel and does not characterize the total 18F-FDG uptake This causes a high dependency on the quality of the PET images (including noise) and the voxel size which induces a low reproducibility

122


Several studies focused on response prediction in EC using 18F-FDG PETCT derived textural features (Table 4) Van Rossum et al concluded that 18F-FDG PET derived textural features provide statistical value [14] but this does not translate into a clinically relevant benefit which is in line with our findings They only performed 18F-FDG PET textural analysis while this study demonstrates the additional value of CT textural analysis Other studies demonstrated promising findings but are hampered by several limitations including small patient cohorts with heterogeneous treatment schedules lack of multivariable analyses and a substantial chance of model overfitting due to the lack of optimism correction [11-1315] A limitation of this study is the absence of external validation which is essential for implementation into clinical practice Moreover several factors should be considered which affect textural analysis (a) Changing the bin width influences the quantization noise and has a crucial effect on textural features [27] Although only an indication we used Doanersquos optimal bin width to discretize the SUVs and Hounsfield Units [21] (b) Respiratory gated PETCT acquisitions could be considered to reduce respiration-induced smearing and contrast degradation [28] The constructed prediction model may serve as a basic model which can be extended with new features for usage for other applications Current constructed model might be helpful towards a safe decision in postponing a burdensome surgical procedure in patients with a doubtful adequate physical condition after nCRT In patients treated with definitive chemoradiotherapy detection of non-responders might allow additional treatments when available while in complete responders an adjusted follow-up might be justified to identify candidates for salvage surgery Up to now the authors do not consider the predictive value of the constructed model high enough to justify the omission of surgery after nCRT in EC A potential approach to improve the constructed basic prediction models could be (a) Adding interim- or post-treatment textural analysis Studies investigating both pre- and post-treatment textural analysis mainly reported post-treatment textural features to be associated with response [121415] We performed a post-treatment textural analysis in patients with a post-treatment PETCT scan (n=20) and found 21 significant textural features for response in univariable regression analysis However post-treatment textural analysis suffers from radiation induced esophagitis which complicates delineation of the primary tumor and difficulties in ldquotumorrdquo delineation in complete responders (b) Texture could be characterized with more specific PET tracers like 18F-fluoroerythronitroimidazole (18F-FETNIM quantifying hypoxia) [29] or 18F-fluorothymidine (18F-FLT targeting cellular proliferation) [30] or by other functional imaging modalities including the apparent diffusion coefficient (ADC) in diffusion weighted magnetic resonance imaging [4] (c) Biological markers have shown to be potential molecular markers in individualizing EC treatment and may be incorporated to improve prediction models [31]

123

7


Table 4 Current literature describing 18F-FDG PET texture analysis in response prediction in EC

Study n Type nCRTTiming PETCT Outcome

Reported entered variables

Tixier et al [13]

41 AC SCC

60 Gy + Cisplatin or CarboplatinFluorouracil

Pre-nCRT

CR vs PR vs NR [32]

Pre Local homogeneityPre Local entropyPre CoarsenessPre Intensity variabilityPre Size-zone features

Hatt et al [11]

50 AC SCC

60 Gy + Cisplatin Fluorouracil

Pre-nCRT

CR + PR vs NR [32]

Pre MATV Pre EntropyPre Homogeneity Pre DissimilarityPre Intensity variability Pre Zone percentage

Tan et al [12]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

ΔSUVmax

SUVmax ratioΔSUVmean

Pre SkewnessPost InertiaPost CorrelationPost Cluster prominence

Zhang et al [15]

20 AC SCC

504 Gy + Cisplatin or Carboplatin

Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

Post OrientationTumor involves GEJΔInertiaPost EnergyPost EntropyΔSkewness

Van Rossum et al [14]

217 AC 45 or 504 Gy + Fluo-ropyrimidine with either a platinum compound or a taxane

Pre- and post-nCRT

TRG 1 vs 2-4 [33]

EUS tumor lengthcT stageInduction chemotherapyPost nCRT endoscopic biopsySubjective PET assess-ment Post nCRT TLGPre Cluster shadeΔRun percentageΔGLCM EntropyPost nCRT Roundness

Current study

97 AC SCC

414 Gy + Carboplatin Paclitaxel

Pre-nCRT

TRG 1 vs 2-5 [19]

HistologycT stagePre LRLGLe-PET Pre RP-CT

124


Abbreviations AC adenocarcinoma SCC squamous cell carcinoma TRG tumor response grade nCRT neoadjuvant chemoradiotherapy MATV metabolically active tumor volume SUV standardized uptake value GEJ gastroesophageal junction EUS endoscopic ultrasonography cT clinical T-stage TLG total lesion glycolysis GLCM grey level co-occurrence matrix LRLGLe-PET long run low gray level emphasis measured on PET and RP-CT run percentage measured on CT

CONCLUSION

The constructed models are a valuable initial step in predicting response to nCRT in locally advanced EC Adding the 18F-FDG PET derived textural feature lsquolong run low gray level emphasisrsquo and the CT derived textural feature lsquorun percentagersquo to a model with the clinical parameters histologic type and clinical T-stage is potentially predictive and was more accurate than response prediction based on SUVmax These models may serve as basic models in determining clinical complete responders and can be extended with new features for usage for other applications

DISCLOSURENone of the authors have any financial or other relationship that could be constructed as a conflict of interest

125

7


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 20123662074-2084[2] Chen YM Pan XF Tong LJ Shi YP Chen T Can 18F-fluorodeoxyglucose positron emission tomography predict responses to neoadjuvant therapy in oesophageal cancer patients A meta-analysis Nucl Med Commun 2011321005-1010[3] Ott K Weber WA Lordick F et al Metabolic imaging predicts response survival and recurrence in adenocarcinomas of the esophagogastric junction J Clin Oncol 2006244692-4698[4] van Rossum PS van Lier AL van Vulpen M et al Diffusion-weighted magnetic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer Radiother Oncol 2015115163-170[5] Boggs DH Hanna A Burrows W Horiba N Suntharalingam M Primary gross tumor volume is an important prognostic factor in locally advanced esophageal cancer patients treated with trimodality therapy J Gastrointest Cancer 201546131-137[6] Pan L Gu P Huang G Xue H Wu S Prognostic significance of SUV on PETCT in patients with esophageal cancer a systematic review and meta-analysis Eur J Gastroenterol Hepatol 2009211008-1015[7] Diaz-Cano SJ Tumor heterogeneity mechanisms and bases for a reliable application of molecular marker design Int J Mol Sci 2012131951-2011[8] Harris AL Hypoxia-a key regulatory factor in tumour growth Nat Rev Cancer 2002238-47[9] Rajendran JG Schwartz DL OrsquoSullivan J et al Tumor hypoxia imaging with [F-18] fluoromisonidazole positron emission tomography in head and neck cancer Clin Cancer Res 2006125435-5441[10] Chicklore S Goh V Siddique M Roy A Marsden PK Cook GJ Quantifying tumour heterogeneity in 18F-FDG PETCT imaging by texture analysis Eur J Nucl Med Mol Imaging 201340133-140[11] Hatt M Tixier F Cheze Le Rest C Pradier O Visvikis D Robustness of intratumour 18F-FDG PET uptake heterogeneity quantification for therapy response prediction in oesophageal carcinoma Eur J Nucl Med Mol Imaging 2013401662-1671[12] Tan S Kligerman S Chen W et al Spatial-temporal [18F]FDG PET features for predicting pathologic response of esophageal cancer to neoadjuvant chemoradiation therapy Int J Radiat Oncol Biol Phys 2013851375-1382

126


[13] Tixier F Le Rest CC Hatt M et al Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitant radiochemotherapy in esophageal cancer J Nucl Med 201152369-378[14] van Rossum PS Fried DV Zhang L et al The incremental value of subjective and quantitative assessment of 18F-FDG PET for the prediction of pathologic complete response to preoperative chemoradiotherapy in esophageal cancer J Nucl Med 201657691-700[15] Zhang H Tan S Chen W et al Modeling pathologic response of esophageal cancer to chemoradiation therapy using spatial-temporal 18F-FDG PET features clinical parameters and demographics Int J Radiat Oncol Biol Phys 201488195-203[16] Edge SB Compton CC The American Joint Committee on Cancer the 7th edition of the AJCC cancer staging manual and the future of TNM Ann Surg Oncol 2010171471-1474[17] Boellaard R OrsquoDoherty MJ Weber WA et al FDG PET and PETCT EANM procedure guidelines for tumour PET imaging version 10 Eur J Nucl Med Mol Imaging 201037181-200[18] Muijs C Smit J Karrenbeld A et al Residual tumor after neoadjuvant chemoradiation outside the radiation therapy target volume a new prognostic factor for survival in esophageal cancer Int J Radiat Oncol Biol Phys 201488845-852[19] Mandard AM Dalibard F Mandard JC et al Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma Clinicopathologic correlations Cancer 1994732680-2686[20] Hatt M Majdoub M Vallieres M et al 18F-FDG PET uptake characterization through texture analysis investigating the complementary nature of heterogeneity and functional tumor volume in a multi-cancer site patient cohort J Nucl Med 20155638-44[21] Doane DP Aesthetic frequency classifications The American Statistician 197630181-183[22] Haralick RM Shanmugam K Dinstein I Textural features for image classification IEEE Transactions on Systems Man and Cybernetics 19733610-621[23] Galloway MM Texture analysis using gray level run lengths Computer Graphics and Image Processing 19754172-179[24] Thibault G Angulo J Meyer F Advanced Statistical Matrices for Texture Characterization Application to Cell Classification IEEE Transactions on Biomedical Engineering 201461630-637

127

7


[25] Akaike H A new look at the statistical model identification IEEE Transactions on Automatic Control 197419716-723[26] Kwee RM Prediction of tumor response to neoadjuvant therapy in patients with esophageal cancer with use of 18F FDG PET a systematic review Radiology 2010254707-717[27] Leijenaar RT Nalbantov G Carvalho S et al The effect of SUV discretization in quantitative FDG PET Radiomics the need for standardized methodology in tumor texture analysis Sci Rep 2015511075[28] Sindoni A Minutoli F Pontoriero A Iati G Baldari S Pergolizzi S Usefulness of four dimensional (4D) PETCT imaging in the evaluation of thoracic lesions and in radiotherapy planning Review of the literature Lung Cancer 20169678-86[29] Yue J Yang Y Cabrera AR et al Measuring tumor hypoxia with 18F-FETNIM PET in esophageal squamous cell carcinoma a pilot clinical study Dis Esophagus 20122554-61[30] Yue J Chen L Cabrera AR et al Measuring tumor cell proliferation with 18F-FLT PET during radiotherapy of esophageal squamous cell carcinoma a pilot clinical study J Nucl Med 201051528-534[31] Honing J Pavlov KV Mul VE et al CD44 SHH and SOX2 as novel biomarkers in esophageal cancer patients treated with neoadjuvant chemoradiotherapy Radiother Oncol 2015117152-158[32] Therasse P Arbuck SG Eisenhauer EA et al New guidelines to evaluate the response to treatment in solid tumors European Organization for Research and Treatment of Cancer National Cancer Institute of the United States National Cancer Institute of Canada J Natl Cancer Inst 200092205-216[33] Chirieac LR Swisher SG Ajani JA et al Posttherapy pathologic stage predicts survival in patients with esophageal carcinoma receiving preoperative chemoradiation Cancer 20051031347-1355

128


129

7



Prediction of response to neoadjuvant chemoradiotherapy with spatio-temporal

18F-FDG PET image biomarkers in esophageal cancer patients

Roelof J Beukinga dagger Jan Binne Hulshoff dagger Veacuteronique EM Mul

Walter Noordzij Gursah Kats-Ugurlu Riemer HJA Slart John Th M Plukker


Submitted to the European Journal of Nuclear Medicine and Molecular Imaging

CHAPTER 8


ABSTRACT

PurposeAssessment of changes of spatio-temporal image biomarkers (IBMs) derived from baseline and restaging 18F-Fluorodeoxyglucose positron emission tomography (18F-FDG PET) scans in predicting pathologic complete response (pCR) to neoadjuvant chemoradiotherapy (nCRT) in patients with locally advanced esophageal cancer (EC)

MethodsFifty patients with locally advanced EC treated with nCRT followed by surgery were analyzed We analyzed 32 pre-selected IBMs (including textural features and histogram-distances) on baseline and restaging 18F-FDG PET scans and analyzed their relative difference (ΔIBMs) Pathologic complete response was defined as a Mandard tumor regression grade (TRG) of 1 The MannndashWhitney test was used to test the statistical significance (α = 005) The discriminatory value of each IBM was assessed with the area under the receiver operating characteristic curve (AUC)

ResultsOf all patients 7 (140) had a pCR while 45 (860) patients had a pIR In total 14 post-nCRT and 2 textural ΔIBMs differed significantly between the pCR and pIR groups None of the baseline IBMs and histogram distances were associated with pCR The best traditional IBM was post-nCRT SUVmax (AUC = 083) The most promising post-nCRT IBMs were the local heterogeneity features derived from gray level co-occurrence matrices joint entropy angular second moment contrast dissimilarity correlation and cluster prominence all exhibiting an AUC gt 077

ConclusionsSeveral post-nCRT IBMs were associated with pCR and were considered stable enough for further analysis whereas baseline IBMs relative IBM differences and histogram differences were generally considered of less predictive value

132


INTRODUCTION

Neoadjuvant chemoradiotherapy (nCRT) followed by esophagectomy is the common standard curative treatment of resectable locally advanced esophageal cancer (EC) [1 2] Pathologic complete response (pCR) after nCRT is achieved in 25-42 of the patients and is accompanied with less recurrences and a significantly longer survival [1 3-6] In fact patients with an imaging based prediction of pCR may benefit from a personalized treatment approach including a lsquowait-and-see policyrsquo following nCRT and salvage surgery in case of isolated locoregional recurrence Temporal change in maximum standardized uptake value (SUVmax) is considered as the current standard method to predict response to nCRT However this value still yields an insufficient sensitivity and specificity to differentiate complete responders from partial and non-responders of 67 and 68 respectively [7] One of the reasons for this inadequate performance may be that SUVmax is a single voxel representation which is susceptible to noise artifacts Moreover SUVmax ignores the intratumoral 18F-Fluorodexyglucose (18F-FDG) spatial distribution and does not represent the overall tumor burden It has been hypothesized that high tumor heterogeneity prior to nCRT usually due to hypoxia is associated with an impaired tumor response to nCRT Image biomarkers (IBMs) extracted from 18F-FDG positron emission tomography (PET) quantifying geometric intensity and textural characteristics of tumors can potentially depict intratumoral 18F-FDG spatial distribution As intratumoral heterogeneity is present in nearly all esophageal tumors quantitative IBMs are potent tools to improve the image interpretation process acquiring in-vivo tumor information In earlier studies several IBMs including textural features and histogram-distances were found to be potentially useful parameters exhibiting even higher diagnostic accuracies than the SUVmax [7-14] Besides the common usage of baseline imaging in predicting response in EC patients restaging imaging just prior to esophagectomy may be of greater value as it precisely reflects the tumor status Furthermore acquiring imaging data at different time points during treatment enables analysis of spatio-temporal change in IBMs with close follow-up of tumor response during treatment Recently relative changes of textural features derived from baseline (pre-nCRT) and restaging (post-nCRT) 18F-FDG PET to depict changes in intratumoral 18F-FDG heterogeneity have shown promising results in predicting response in EC patients [7-10 13]The aim of this study was to assess the value of spatio-temporal IBM changes extracted from baseline and restaging 18F-FDG PET scans in predicting pCR to nCRT in patients with locally advanced EC

133

8


METHODS

PatientsThe present study was conducted according to the national Dutch guidelines for retrospective studies and rules of the local Institutional Ethical Board A total of 56 patients were eligible for inclusion as they had locally advanced (T1N1-3M0 or T2-

4aN0-3M0) EC and were treated with the same nCRT regimen (ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study [CROSS] regimen) followed by esophagectomy with a two-field lymph node dissection in the University Medical Center Groningen between October 2014 and October 2016 [1] Excluded from the analyses were patients with 18F-FDG PETCT scans made in other medical centers (n = 2) those with non-18F-FDG-avid tumors (n = 1) and those with distant metastases found prior to or during surgery (n = 3) In total 50 patients were included and all clinical data were obtained from a prospectively maintained database

Staging treatment and pathologyStaging consisted of CT thoraxabdomen 18F-FDG PETCT and endoscopic ultrasonography (EUS) with fine needle aspiration if indicated After staging all patients were discussed in our multidisciplinary upper gastrointestinal tumor board All patients were treated with nCRT according to the CROSS regimen consisting of carboplatin (2 mg∙min∙mL-1) and paclitaxel (50 mgm2) combined with 414 Gy (in 23 fractions) All patients were restaged with 18F-FDG PETCT approximately 6-8 weeks after nCRT Surgery consisted of either open or minimally invasive transthoracic esophagectomy combined with a two-field lymph node dissection Two experienced gastrointestinal pathologists determined tumor response to nCRT according to the Mandard tumor regression grade (TRG) which was considered the gold standard [15] This five-point scoring system classifies the percentage of residual vital tumor cells and the degree of nCRT-induced fibrosis The studied group was divided into two subgroups by categorizing response into pCR (TRG 1) versus pathologic incomplete response (pIR TRG 2-5)

18F-FDG PETCT imagingIntegrated baseline and restaging 18F-FDG PETCT (Biograph mCT 64 PETCT Siemens Knoxville TN USA) scans were performed An inspiration breath-hold low-dose CT was performed before PET acquisitions were obtained with a field of view (FOV) of 500 times 500 times 500 mm in 3D setting Images were reconstructed (voxel-size of 31819 times 31819 times 2 mm resulting in 512 times 512 times 250 voxels matrices) according to the European Association of Nuclear Medicine guidelines [16]

134


Volume of interest delineationBased on the radiotherapeutic gross tumor volume (GTV) which was manually delineated by an expert radiation oncologist in gastrointestinal malignancies the volume of interest (VOI) was defined The GTV was rigidly registered to the CT component of the baseline and restaging PETCT images (RTx Workstation 10 Mirada Medical Oxford UK) Because registration errors could occur the VOI was manually corrected by the researchers The post-nCRT delineation included the pre-treatment localization of the primary tumor and was adjusted manually to compensate for regression of the tumor size

IBM pre-selection and extractionThe large number of IBMs and multiple testing can increase the risk of false positive identification of predictive IBMs and of multicollinearity among IBMs In this study we therefore reduced the number of IBMs to a pre-selected group that have shown to be of predictive value in previous reported research [7-13] For each image and VOI 4 traditional features 1 morphological feature 1 statistical feature 17 textural features 5 bin-to-bin histogram-distances and 4 cross-bin histogram-distances (Table 1) [17] Matlab 2014b (Mathworks Natick MA USA) was used for image processing and analysis 18F-FDG uptake was converted to SUV and was corrected for the serum glucose level [16] Original voxel dimensions were up-sampled to isotropic voxel-dimensions of 2 times 2 times 2 mm using trilinear spline interpolation Voxels enclosed for ge 50 coverage were included to the up-sampled VOI For each IBM the relative IBM difference (ΔIBM) was calculated according to

Textural features were extracted from discretized image stacks to reduce the continuous-scaled SUV to a limited number of gray levels and to reduce image noise Voxels were discretized in 05 gmL increments starting at a minimum of 0 gmL Images were analyzed in 3 dimensions with a connectivity of 26 voxels (13 angular directions and a pixel-to-pixel distance of 1) The 13 different gray level co-occurrence and gray level run-length matrices along each angular direction were merged into combined matrices before feature extraction

∆IBM = restaging IBM - baseline IBM

x 100baseline IBM

135

8


Table 1 Pre-selected IBMs that have shown to be of predictive value in previous reported researchTraditional features

volumeSUVmax

SUVpeak

total lesion glycolysis (TLG)Morphological feature

roundnessStatistical feature

skewnessTextural features - Grey level co-occurrence based features

joint entropyangular second moment (energy)contrast (inertia)dissimilarityinverse difference (homogeneity 1)inverse difference moment (homogeneity 2)correlationcluster shadecluster prominence

Textural features - Grey level run length based featureshigh grey level run emphasisshort run high grey level emphasislong run low grey level emphasisrun percentage

Textural features - Grey level size zone based featureslarge zone emphasissmall zone high grey level emphasiszone percentage

Textural features - Neighboring grey level dependence based featurecoarseness

Histogram distances - Bin-to-bin based featuresEuclidean distancecityblock distancesquared Euclidean distancecosine distanceχ2 pearson distance

Histogram distances - Cross-bin based featuresquadratic form distancematch distanceKolmogorov-Smirnov distancequadratic chi distance

136


Statistical analysisStatistical analysis was performed with Matlab 2014b (Mathworks Natick MA USA) and R 322 open-source software available from the Comprehensive R Archive Network (httpwwwr-projectorg) IBMs were tested for their robustness to delineation variations Therefore two additional segmentations were created by morphological erosion with two ball-shaped structuring elements with radii of 1 and 2 voxels The reliability of ratings was measured by the intraclass correlation coefficient (ICC) Only IBMs having an excellent reliability (ICC gt 075) were considered robust The statistical difference between the studied IBMs for the pCR versus pIR group was assessed with the MannndashWhitney U test Only those IBMs with a P lt 005 were considered predictive Each studied IBM was quantified in terms of discrimination with the area under the receiver operating characteristic curve (AUC)

RESULTS

Table 2 displays the patient and tumor characteristics Complete or nearly complete nCRT administration was achieved in 48 patients (960) 2 patients received le 3 cycles of chemotherapy and all received the full dose of radiotherapy The median tumor volume before treatment was 527 cm3 (interquartile range 376 cm3) Overall tumor volume decreased to 382 cm3 (interquartile range 211 cm3) on the restaging PET scan All tumors had a minimum volume of 10 cm3 which was required for textural analysis to provide valuable complementary information [14] Of all patients 7 (140) had a pCR while 45 (860) patients had a pIR

137

8


Table 2 Patient and tumor characteristicsCharacteristics (n = 50)Sex

Male 40 (800)Female 10 (200)

Age (years)a 615 (120)Histology

Adenocarcinoma 47 (940)Squamous cell carcinoma 3 (60)

Tumor locationMiddle esophagus 5 (100)Distal esophagus 36 (720)GEJ 9 (180)

Tumor length (cm)a 60 (40)cT-stage

T2 4 (80)T3 42 (840)T4a 4 (80)

cN-stageN0 7 (140)N1 20 (400)N2 19 (380)N3 4 (80)

Number of chemotherapy cycles2 1 (20)3 1 (20)4 7 (140)5 41 (820)

RadicalityR0 49 (980)R1 1 (20)

Mandard TRG1 7 (140)2 15 (300)3 18 (360)4 9 (180)5 1 (20)

a Expressed as median with interquartile range in parentheses Data are numbers with percentages in parentheses Abbreviations GEJ = gastroesophageal junction cT = clinical T-stage cN = clinical N-stage R0 = microscopically radical resection R1 = microscopically irradical resection and TRG = tumor regression grade

138


Prediction of pCRTable 3 displays the P values on the MannndashWhitney U test AUCs and ICCs of all studied IBMs The discriminatory value (AUC) of all pre-selected IBMs have been visualized in Fig 1 and Fig 2 At baseline neither the 4 studied traditional IBMs or the textural IBMs showed significant differences between the pCR and pIR groups However at post-nCRT analysis the traditional IBMs volume SUVmax SUVpeak and TLG were all considered robust for delineation variations as was indicated by excellent reliabilities with ICCs of 091 100 100 and 096 respectively All these traditional IBMs were found to be significantly different with P values of 0044 0006 0009 and 0015 respectively and exhibited fairly or even good performances with AUCs of 074 083 081 and 079 respectively Moreover 15 textural IBMs were found to be robust for delineation variations whereas the IBMs roundness cluster shade large zone emphasis and coarseness were considered not stable enough for further consideration After selecting only those IBMs which were stable enough 10 textural IBMs were found to have a P lt 005 on the MannndashWhitney U test including the IBMs joint entropy (P = 0014) angular second moment (P = 0015) contrast (P = 0020) dissimilarity (P = 0024) inverse difference (P = 0029) inverse difference moment (P = 0027) correlation (P = 0022) cluster prominence (P = 0008) all derived from the same underlying gray level co-occurrence matrix and high grey level run emphasis (P = 0031) and short run high grey level emphasis (P = 0027) both derived from the same underlying gray level run-length matrix All these IBMs exhibited fairly or even good performances with AUCs of 079 079 078 077 076 076 077 082 076 and 076 respectively Noteworthy the IBM cluster shade exhibited a significant P = 0007 but was not considered stable enough with respect to delineation variations (ICC = 043)

Considering relative IBM differences (ΔIBMs) only Δjoint entropy and Δcontrast were found to be both robust for delineation variations (ICC of 097 and 099 respectively) and significantly different (P values of 0047 and 0044 respectively) between the pCR and pIR groups Both IBMs exhibited an AUC of 074 Also the relative difference in cluster shade was found to have a significant P = 0047 but was this IBM was again not considered stable enough with respect to delineation variations (ICC = 000) None of the traditional ΔIBMs or histogram distances were found to be associated with pCR

139

8


Fig 1 AUCs of the selected traditional and textural IBMs with an ICC gt 08 used for distinguishing pCR from pIR measured at baseline PET (A) and restaging PET (B) with the relative difference (C)

Fig 2 AUCs of the selected histogram distances used for distinguishing pCR from pIR

140


Table 3 P values on the MannndashWhitney U test AUCs and ICCs of all studied IBMsBaseline IBMs Post-nCRT IBMs ΔIBMs

IBM P AUC ICC P AUC ICC P AUC ICCvolume 0057 073 093 0044 074 091 0955 051 093SUVmax 0198 065 100 0006 083 100 0054 073 100SUVpeak 0198 065 100 0009 081 100 0061 072 100TLG 0371 061 097 0015 079 096 0520 058 099roundness 0189 066 087 0615 056 067 0093 070 068skewness 0758 054 094 0117 069 085 0467 059 -006joint entropy 0314 062 097 0014 079 097 0047 074 097angular second moment 0737 054 094 0015 079 091 0105 069 089contrast 0138 068 098 0020 078 099 0044 074 099dissimilarity 0146 067 097 0024 077 098 0054 073 098inverse difference 0386 060 097 0029 076 097 0088 070 096inverse difference moment 0417 060 097 0027 076 097 0093 070 096correlation 0371 061 061 0022 077 085 0065 072 086cluster shade 0675 055 093 0007 082 043 0047 074 -007cluster prominence 0162 067 099 0008 082 089 0057 073 097high grey level run emphasis 0219 065 093 0031 076 096 0083 071 097short run high grey level emphasis

0208 065 093 0027 076 096 0065 072 097

long run low grey level emphasis

0502 058 096 0240 064 078 0117 069 081

run percentage 0434 059 092 0065 072 095 0083 071 096large zone emphasis 0124 068 057 0867 052 061 0074 071 079small zone high grey level emphasis

0146 067 096 0171 066 095 0074 071 083

zone percentage 0078 071 095 0823 053 080 0050 073 093coarseness 0117 069 040 0434 059 024 0675 055 064Euclidean distance 0314 062 091cityblock distance 0276 063 094squared Euclidean distance 0314 062 084cosine distance 0737 054 089χ2 pearson distance 0288 063 093quadratic form distance 0314 062 093match distance 0208 065 096Kolmogorov-Smirnov distance 0219 065 090quadratic chi distance 0162 067 089

Significant difference between the pCR and pIR groups (P lt 005)Abbreviations SUV = standardized uptake value TLG = total lesion glycolysis nCRT = neoadjuvant chemoradiotherapy AUC = area under the receiver operating characteristic curve and ICC = intraclass correlation

141

8


Fig 3 Comprehensive overview of whether the pre-selected IBMs were found to be significant (green) non-significant (red) or not evaluated (black) in present and current available studies dagger indicates the present study ∆ indicates the relative IBM difference between baseline and post-nCRT 18F-FDG PET scans

142


DISCUSSION

Clinical evaluation of response to nCRT remains difficult but highly important in patients with potentially curable locally advanced EC both for exploring future personalized treatment and in predicting prognostic outcome Although objective response defined as a 30 or greater decrease in summed tumor diameters according to the RECIST [18] is frequently used for trial analysis it originally was not intended for clinical use Hence investigators examined the value of alternative definitions such as biology-derived response [19] Besides an evolutionary association of gene-expression patterns (tumor genotype) with response phenotypes comprehensive image quantification with radiomics is an emerging noninvasive approach in predicting survival outcome and may be used in response prediction in cancer treatment In the present study we assessed the value of spatio-temporal changes of several pre-selected IBMs which were reported as being predictive or significant based on baseline andor post-nCRT PET scans by current available studies We found that 14 post-nCRT IBMs (ie 4 traditional IBMs and 10 textural IBMs) and 2 textural ΔIBMs were associated with pCR None of the studied baseline IBMs and histogram distances were associated with pCRBesides the present study five other studies focused on response prediction to nCRT in EC based on relative IBM differences to depict changes in intratumoral 18F-FDG heterogeneity or based on 18F-FDG uptake distribution dissimilarities [7-10 13] whereas two others were only based on baseline PETCT scans [11 12] Inherent to performing multiple statistical tests with a large number of IBMs is that numerous different IBMs have been reported as promising over the years To validate all these results integration and pooling of those results is required Fig 3 gives a comprehensive overview of the potential significance of all pre-selected IBMs in predicting pCR of the present and previous reported studies Although interpretation of all reported IBMs is hampered by the lack of standardization and differences in methodology the results of the present study generally are consistent with previous research Remarkably more significant IBMs are found in the post-nCRT analysis compared to baseline and relative difference analyses suggesting a more prominent role for post-nCRT analyses when predicting pCR in EC patients In the post-nCRT analyses the most frequently reported significant IBMs were the local heterogeneity features derived from gray level co-occurrence matrices joint entropy angular second moment contrast dissimilarity correlation and cluster prominence which were all confirmed by the present study The study by Tan et al was the only study assessing histogram distances for prediction of response and found multiple histogram distances to provide useful information However in the present study none of these histogram distances were able to discriminate between pCR and pIR [9] A potential reason for this discrepancy may be the small patient cohort (n = 20) in the study by Tan et al

143

8


Though promising the use of radiomics in selecting patients to different treatment strategies based on identification of response is still not practicable due to many unsolved problems and the complexity of data There is a significant lack of standardization including the usage of different scanners and image acquisition protocols among hospitals and different feature extraction methods which could lead to a large variability of IBMs complicating the validation of results [20] Moreover one of current major concerns in textural analysis is that not all textural features are deemed stable enough for clinical application especially with respect to the impact of different devices and inter-observer variability on tumor delineation The study by Hatt et al found that robust IBMs for delineation variations included entropy inverse difference dissimilarity and zone percentage [21] however they only assessed a limited number of IBMs pre-selected in the current study Therefore multiple artificial tumor delineations were created in the current study which were used to test the robustness of all pre-selected IBMs Our results were consistent with the study by Hatt et al Post-nCRT tumor delineation is in any case complicated in complete responders due to the absence of viable tumor tissue In those patients the baseline tumor delineation was registered with the restaging PET and the original tumor dimension was retained Moreover localization of the tumor on the restaging PET can be difficult as the metabolically active area is often contaminated with nCRT-induced esophagitisWith reproducibility being another major concern of textural analysis Tixier et al compared textural features of double baseline 18F-FDG PET scans and found that local and regional 18F-FDG uptake textural features had similar or even better reproducibility than traditional SUV measurements [22] To be more specific the textural features joint entropy inverse difference and dissimilarity exhibited a high reproducibility whereas the textural features angular second moment contrast and correlation were characterized by lower reproducibility Other features that were pre-selected by the present study were not evaluatedThe present study had some shortcomings including the low number of patients (n = 50) with the result that no multivariable prediction models could be developed due to the chance of overfitting Verification of the proposed IBMs and model development requires studies with large cohorts of patients however the majority of recent studies only included a limited number of patients Multivariable analyses are also required to determine whether incorporation of the proposed textural IBMs provides significant model improvement compared to models based on traditional IBMs alone and whether redundancy exists among IBMs With the availability of large patient cohorts robust machine-learning approaches could potentially further enhance the clinical applicability of radiomic-based models in cancer in general [23] Furthermore incorporation of other (image) biomarkers have great potential to contribute to reach this specific goal Especially the extraction of potentially robust IBMs from other

144


sophisticated imaging modalities such as diffusion weighted imaging which is increasingly used in EC or by incorporation of useful genomics such as cancer stem cell markersIn conclusion in response phenotyping IBMs extracted from post-nCRT PET generally were observed to be better related to pCR than baseline IBMs Among the significant post-nCRT IBMs were the traditional IBMs volume SUVmax SUVpeak and TLG and 10 different textural IBMs of which the local heterogeneity features joint entropy angular second moment contrast dissimilarity correlation and cluster prominence all derived from the gray level co-occurrence matrix were considered stable enough and suitable for further analysis Relative IBM differences were found significantly different between patients with pCR and pIR in only 2 IBMs while histogram differences were not once The focus of response phenotyping with regard to the prediction of pCR should be concentrated more on post-nCRT analysis rather than on baseline analysis

145

8


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ Steyerberg EW van Berge Henegouwen MI Wijnhoven BP Richel DJ Nieuwenhuijzen GA Hospers GA Bonenkamp JJ Cuesta MA Blaisse RJ Busch OR ten Kate FJ Creemers GJ Punt CJ Plukker JT Verheul HM Spillenaar Bilgen EJ van Dekken H van der Sangen MJ Rozema T Biermann K Beukema JC Piet AH van Rij CM Reinders JG Tilanus HW van der Gaast A CROSS Group Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 20123662074-84[2] Shapiro J van Lanschot JJ Hulshof MC van Hagen P van Berge Henegouwen MI Wijnhoven BP van Laarhoven HW Nieuwenhuijzen GA Hospers GA Bonenkamp JJ Cuesta MA Blaisse RJ Busch OR Ten Kate FJ Creemers GJ Punt CJ Plukker JT Verheul HM Bilgen EJ van Dekken H van der Sangen MJ Rozema T Biermann K Beukema JC Piet AH van Rij CM Reinders JG Tilanus HW Steyerberg EW van der Gaast A CROSS study group Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS) long-term results of a randomised controlled trial Lancet Oncol 2015161090-8[3] van Hagen P Wijnhoven BP Nafteux P Moons J Haustermans K De Hertogh G van Lanschot JJ Lerut T Recurrence pattern in patients with a pathologically complete response after neoadjuvant chemoradiotherapy and surgery for oesophageal cancer Br J Surg 2013100267-73[4] Meguid RA Hooker CM Taylor JT Kleinberg LR Cattaneo SM2nd Sussman MS Yang SC Heitmiller RF Forastiere AA Brock MV Recurrence after neoadjuvant chemoradiation and surgery for esophageal cancer does the pattern of recurrence differ for patients with complete response and those with partial or no response J Thorac Cardiovasc Surg 20091381309-17[5] Smit JK Guler S Beukema JC Mul VE Burgerhof JG Hospers GA Plukker JT Different recurrence pattern after neoadjuvant chemoradiotherapy compared to surgery alone in esophageal cancer patients Ann Surg Oncol 2013204008-15[6] Zanoni A Verlato G Giacopuzzi S Weindelmayer J Casella F Pasini F Zhao E de Manzoni G Neoadjuvant concurrent chemoradiotherapy for locally advanced esophageal cancer in a single high-volume center Ann Surg Oncol 2013201993-9[7] van Rossum PS Fried DV Zhang L Hofstetter WL van Vulpen M Meijer GJ Court LE Lin SH The Incremental Value of Subjective and Quantitative Assessment of 18F-FDG PET for the Prediction of Pathologic Complete Response to Preoperative Chemoradiotherapy in Esophageal Cancer J Nucl Med 201657691-700

146


[8] Tan S Kligerman S Chen W Lu M Kim G Feigenberg S DSouza WD Suntharalingam M Lu W Spatial-temporal [18F]FDG-PET features for predicting pathologic response of esophageal cancer to neoadjuvant chemoradiation therapy Int J Radiat Oncol Biol Phys 2013851375-82[9] Tan S Zhang H Zhang Y Chen W DSouza WD Lu W Predicting pathologic tumor response to chemoradiotherapy with histogram distances characterizing longitudinal changes in 18F-FDG uptake patterns Med Phys 201340101707[10] Zhang H Tan S Chen W Kligerman S Kim G DSouza WD Suntharalingam M Lu W Modeling pathologic response of esophageal cancer to chemoradiation therapy using spatial-temporal 18F-FDG PET features clinical parameters and demographics Int J Radiat Oncol Biol Phys 201488195-203[11] Tixier F Le Rest CC Hatt M Albarghach N Pradier O Metges JP Corcos L Visvikis D Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitant radiochemotherapy in esophageal cancer J Nucl Med 201152369-78[12] Beukinga RJ Hulshoff JB van Dijk LV Muijs CT Burgerhof JGM Kats-Ugurlu G Slart RHJA Slump CH Mul VEM Plukker JTM Predicting Response to Neoadjuvant Chemoradiotherapy in Esophageal Cancer with Textural Features Derived from Pretreatment 18F-FDG PETCT Imaging J Nucl Med 201758723-9[13] Yip SS Coroller TP Sanford NN Mamon H Aerts HJ Berbeco RI Relationship between the Temporal Changes in Positron-Emission-Tomography-Imaging-Based Textural Features and Pathologic Response and Survival in Esophageal Cancer Patients Front Oncol 2016672[14] Hatt M Majdoub M Vallieres M Tixier F Le Rest CC Groheux D Hindie E Martineau A Pradier O Hustinx R Perdrisot R Guillevin R El Naqa I Visvikis D 18F-FDG PET uptake characterization through texture analysis investigating the complementary nature of heterogeneity and functional tumor volume in a multi-cancer site patient cohort J Nucl Med 20155638-44[15] Mandard AM Dalibard F Mandard JC Marnay J Henry-Amar M Petiot JF Roussel A Jacob JH Segol P Samama G Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma Clinicopathologic correlations Cancer 1994732680-6[16] Boellaard R ODoherty MJ Weber WA Mottaghy FM Lonsdale MN Stroobants SG Oyen WJ Kotzerke J Hoekstra OS Pruim J Marsden PK Tatsch K Hoekstra CJ Visser EP Arends B Verzijlbergen FJ Zijlstra JM Comans EF Lammertsma AA Paans AM Willemsen AT Beyer T Bockisch A Schaefer-Prokop C Delbeke D Baum RP Chiti A Krause BJ FDG PET and PETCT EANM procedure guidelines for tumour PET imaging version 10 Eur J Nucl Med Mol Imaging 201037181-200

147

8


[17] Zwanenburg A Leger S Valliegraveres M Loumlck S Image biomarker standardisation initiative - feature definitions CoRR 2016abs161207003[18] Eisenhauer EA Therasse P Bogaerts J Schwartz LH Sargent D Ford R Dancey J Arbuck S Gwyther S Mooney M Rubinstein L Shankar L Dodd L Kaplan R Lacombe D Verweij J New response evaluation criteria in solid tumours revised RECIST guideline (version 11) Eur J Cancer 200945228-47[19] Oxnard GR Schwartz LH Response phenotype as a predictive biomarker to guide treatment with targeted therapies J Clin Oncol 2013313739-41[20] Galavis PE Hollensen C Jallow N Paliwal B Jeraj R Variability of textural features in FDG PET images due to different acquisition modes and reconstruction parameters Acta Oncol 2010491012-6[21] Hatt M Tixier F Cheze Le Rest C Pradier O Visvikis D Robustness of intratumour 18F-FDG PET uptake heterogeneity quantification for therapy response prediction in oesophageal carcinoma Eur J Nucl Med Mol Imaging 2013401662-71[22] Tixier F Hatt M Le Rest CC Le Pogam A Corcos L Visvikis D Reproducibility of tumor uptake heterogeneity characterization through textural feature analysis in 18F-FDG PET J Nucl Med 201253693-700[23] Parmar C Grossmann P Bussink J Lambin P Aerts HJ Machine Learning methods for Quantitative Radiomic Biomarkers Sci Rep 2015513087

148


149

8



Prediction of tumor response to neoadjuvant chemoradiotherapy with DW-MRI and 18F-FDG

PETCT in esophageal cancer

Jan Binne Hulshoff dagger Veronique EM Mul dagger

Roelof J Beukinga Peter Kappert

Jan Hendrik Potze Gursah Kats-Ugurlu


Annemarieke Bartels-Rutten Regina Beets-Tan

John ThM Plukker


In preparation

CHAPTER 9


ABSTRACT

IntroductionEsophageal cancer patients with a pathologic complete response (pCR) after neoadjuvant chemoradiotherapy (nCRT) with esophagectomy might benefit from omitting surgery The current feasibility study assessed the value of combined diffusion-weighted magnetic resonance imaging (DW-MRI) and 18F-fluoro-deoxyglucose positron emission tomographycomputed tomography (18F-FDG PETCT) in the detection of complete response

Patients and MethodA prospective feasibility study was conducted in patients with locally advanced esophageal cancer (T2-4aN0-3M0T1N1-3M0) selected for neoadjuvant chemoradiotherapy followed by surgery All included patients (n = 18) received both a 18F-FDG PETCT and DW-MRI scan before during (ge2 weeks of nCRT) and approximately 8 weeks after nCRT Clinical response was assessed on PET and MRI by two experienced radiologists and nuclear physicians Thereafter early and late response was assessed by measuring changes in the maximal standardized uptake value (SUVmax) and the apparent diffusion coefficient (ADC) values with logistic regression analyses

ResultsClinical detection of pCR after nCRT with 18F-FDG PET DW-MRI and combined 18F-FDG PET and DW-MRI resulted in a sensitivity of 67 100 and 67 and a specificity of 67 93 and 100 respectively The inter-agreement Kappa between both radiologists who assessed the T2 DW-MRI scans was low at 0357 (P = 0130) There was no association between the ΔSUVmax and ΔDW-MRI and pCR

Conclusion18F-FDG PETCT and DW-MRI were both valuable in the post-nCRT clinical response evaluation Both the ΔSUVmax and ΔDW-MRI did not predict pCR although the findings of this feasibility study should be confirmed in a large trial

152


BACKGROUND

Pathologic complete response (pCR) after neoadjuvant chemoradiotherapy (nCRT) is identified in 25 to 42 of the patients with locally advanced esophageal cancer (EC) [1-3] In the near future some of these patients might benefit from a lsquowait and seersquo policy with imaging-based follow-up after nCRT instead of a radical esophagectomy On the other hand patients with a limited or no response may not benefit from an intensive treatment with nCRT Therefore accurate prediction of good or even complete response during or after nCRT could allow a more personalized treatment approach reducing the risk for treatment related morbidities By measuring changes in the maximal standardized uptake value (SUVmax) derived from 18-F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) scans response can be assessed with a sensitivity and specificity of 67 and 68 respectively [4] A recently introduced quantitative method for evaluation of response is diffusion-weighted magnetic resonance imaging (DW-MRI) [5-11] DW-MRI is sensitive to changes in proton mobility in tissues (diffusion) The apparent diffusion coefficient (ADC) derived from DW-MRI scan provides absolute values of diffusion Changes of the ADC after nCRT might represent changes in tissue (micro)structure which might be useful for the assessment of response A change in ADC after nCRT was found to be predictive for tumor response in EC [8-11] Moreover the 18F-FDG PET and DW-MRI might have a complementary value since DW-MRI depicts changes in cellular (micro)structure while 18F-FDG PET scans depict metabolic activity The present feasibility study assessed the additional value of DW-MRI and 18F-FDG PETCT compared to PETCT alone in the clinical assessment of response to nCRT in patients with esophageal cancer Thereafter we determined the additional value of ADC and the SUVmax in the prediction of complete and major response


Patients This feasibility study was performed with approval of the local Ethical Board All patients with a 18F-FDG avide and resectable locally advanced esophageal carcinoma (T2-4aNallM0T1N1-3M0) who were selected for treatment with nCRT followed by surgery were eligible for inclusion All patients had a histologically proven adeno- or squamous cell carcinoma of the esophagus or gastro esophageal junction had given written informed consent and were aged over 18 years Patient with mental disorders prior malignancies lt5 years poorly controlled diabetes contra-indications for MRI scanning major obesity (body mass index gt40) and severe esophagitis prior to treatment were not included All

153

9


selected patients (n = 18) received a 18F-FDG PETCT and DW-MRI scan before during (2 weeks after the start of nCRT) and after nCRT (approximately 8 weeks after nCRT) Data were collected prospectively

Staging treatment and response assessmentAccording to our standard staging workup all patients received a 18F-FDG PETCT scan with diagnostic 64 multi slice CT chestabdomen and endoscopic ultrasonography (EUS) All patients were treated with nCRT according to the CROSS regimen with carboplatin and paclitaxel combined with external radiotherapy (414 Gy) Surgery consisted of either a minimally invasive or open transthoracic esophagectomy en-bloc with a two-field lymphadenectomy of mediastinal and upper abdominal lymph nodes The Mandard tumor regression grade (TRG) was used to determine pathologic response in which pathologic complete response was defined as TRG score 1 and good responders as TRG 1 amp 2 [12]

Imaging and delineation methodIntegrated 18F-FDG PETCT (Biograph mCT 64 PETCT Siemens Knoxville TN USA) scans were made before during (at 2 weeks after the onset of nCRT) and after nCRT A deep-inspiration breath-hold low-dose CT was performed PET images (field of view of 500times500times500 mm in 3D setting) were reconstructed (voxel-size of 098 times 098 times 2 mm resulting in 512times512times250 matrices) according to the guidelines of the European Association of Nuclear Medicine [13] Delineation of the radiation target volume was performed on a Somatom Sensation 64-multidetector row spiral CT (Siemens Medical Systems Erlangen Germany)MRI scans were performed with 15 T MRI scans (Magnetom Aera Siemens Medical Solutions Erlangen Germany) The body coil served as a transmitter and a 18-element body matrix receive-coil as a receiver A localizer scan was performed followed by a transversal T1-weighted VIBE DIXON and T2-weighted BLADE a sagittal and coronal breath-hold and non-breath-hold TRUFI and a transversal and sagittal DW DW acquisitions at several b-values (b = 0 50 100 250 500 750 1000 smm2) were triggered with free-breathing respiratory gating and were performed with these parameters TR 2900 msec TE 90 msec voxel-size 236 times 236 times 6 mm The gross tumor volume (GTV) delineated by a radiation oncologist specialized in upper gastrointestinal malignancies was used for SUVmax and ADC extraction The volume of interest was registered from the delineationplanning CT scan to the 18F-FDG PETCT and the T2-weighted BLADE and manually corrected by the researchers The volume of interest on the T2-weighted BLADE was then registered to all b-values for determination of the ADC The volume of interest

154


was also registered to during and after nCRT scans and manually corrected Mirada medical system (RTx Workstation 10 Mirada Medical Oxford UK) and Matlab 2014b (Mathworks Natick MA USA) were used for imaging processing All extracted SUVmax and ADC values were normalized to the range (01)

Clinical response predictionAn expert radiologist and a nuclear physician both blinded for the outcome visually assessed and scored the presence of clinical complete response on DW-MRI and 18F-FDG PETCT Clinical visual response assessment with MRI was based on the T2 BLADE MRI and higher b-values (500 750 1000 smm2) Patients were scored as complete responders if no diffusion restriction was observed On 18F-FDG PETCT patients were scored as complete responders if no suspect 18F-FDG avid lesion was observed The sensitivity specificity positive predictive value (PPV) and negative predictive value (NPV) of visual assessment of suspect lesions on all imaging modalities was determined with pathologic complete response as the gold standard The information obtained from the clinical assessment of the 18F-FDG PETCT and DW-MRI scan were combined to an overall conclusion If one of these imaging techniques displayed visible tumor the outcome was scored as non-complete response Thereafter a double read of the DW-MRIs was performed by an expert radiologist blinded to the previous reading results of his colleague for the assessment of tumor response on the DW-MRI The Cohenrsquos interoberserver agreement between the two radiologists were assessed

Assessment of response with the ΔADC and ΔSUVmaxADC and SUVmax were extracted from scans at three different time points baseline (T0) during nCRT (T1 at 2 weeks after the onset of nCRT) and after nCRT (T2 approximately 8 weeks after the end of nCRT) The ΔSUVmax was calculated with the following formula ΔSUVmax(t)=( ΔSUVmax(t) - ΔSUVmax(pre)) ΔSUVmax(pre) x 100 in which t stands for the time point The ΔADC was calculated in the same manner ΔADC(t)=(ΔADC(t) - ΔADC(pre)) ΔADC(pre) x 100 The association between the ΔSUVmax and ΔDW-MRI and pCR (Mandard TRG 1) or major response (Mandard TRG 1 and 2) was assessed

StatisticsStatistical analyses were performed with SPSS version 23 and R open-source software (R322) using the rms package (version 44-0 httpwwwr-projectorg) Univariate logistic regression analysis was performed to determine pCR and major response with the ΔSUVmax and ΔADC Area Under the Receiver Operating Characteristic curve (AUC) was used to determine the performance of each variable

155

9


RESULTS

Table 1 displays the patient characteristics Most included patients had an adenocarcinoma (n = 16 889) The most common tumor location was the distal esophagus and gastroesophageal junction (n = 15 833) nCRT was completed in 14 (778) patients Pathologic complete response (pCR Mandard TRG1) was observed in 3 (167) patients and non-complete or no-response in 15 (833) One of the patients with a pathologic complete response of the primary tumor however had residual lymph node metastases (ypT0N2) In one patient the DW-MRI during nCRT could not be performed because the patient had severe swallowing problems due to a nearly complete stenosis with high risk of aspiration

Clinical response assessment with 18F-FDG PETCT and DW-MRIVisual assessment of response with the 18F-FDG PET scan made during nCRT resulted in detection of 1 patient with complete response giving a sensitivity of 33 With the DW-MRI scan made during nCRT no patient was assessed as complete responder With visual assessment of post-nCRT 18F-FDG PET scans pCR was correctly detected in 2 patients missed in 1 patient and assessed incorrectly as complete responders in 5 patients resulting in a sensitivity and specificity of 67 and 67 respectively (Table 2) Assessment of the post-nCRT DW-MRI scan by the expert radiologist resulted in a correct detection of pCR in 3 patients and falsely assessed response in 1 patient resulting in a sensitivity and specificity of 100 and 93 respectively The combined information of the MRI and 18F-FDG PET resulted in a correct assessment of pCR in 2 patients and missed pCR in 1 patient giving a sensitivity and specificity of 67 and 100 respectively

156


Table 1 Patient characteristicsn ()

GenderMale 15 (833)Female 3 (167)Age (IQR) 605 (55-67)HistologyAdenocarcinoma 16 (889)Squamous cell carcinoma 2 (111)LocationMid 3 (167)Distal 15 (833)cTT2 3 (167)T3 15 (833)cNN0 3 (167)N1 6 (333)N2 8 (444)N3 1 (56)Amounts of chemo3 1 (65)4 3 (167)5 14 (778)Mandard1 (pCR) 3 (167)2 9 (500)3 6 (333)ypTT0 3 (167)T1 3 (167)T2 4 (222)T3 8 (444)ypNN0 10 (556)N1 6 (333)N2 2 (111)

Abbreviations IQR = inter quartile range cT = clinical T-stage cN = clinical N-stage pCR = pathologic complete response ypT = pathologic T-stage ypN = pathologic N-stage

157

9


Table 2 Clinical response evaluation with post-nCRT imagingSensitivity Specificity PPV NPV

CT 33 80 25 8618F-FDG PET 67 67 29 91DW-MRI 100 93 75 10018F-FDG PET amp DW-MRI 67 100 100 94

Abbreviations PPV = positivie predictive value NPV = negative predictive value CT= computed tomography 18F-FDG PET = Fluorine-18 labelled fluorodeoxyglucose positron emission tomography combined with computed tomography DW-MRI = diffusion-weighted magnetic resonance imaging

Agreement in post-nCRT DW-MRI assessment Both radiologists agreed in assessing pCR on the post-nCRT DW-MRI images in 2 patients and on non-response in 12 patients The radiologists disagreed on 4 patients resulting in a low Cohenrsquos Kappa of 0357 (P = 0130)

Correlation between ΔADC and ΔSUVmax for pCR and major responseTable 3 displays the logistic regression coefficient of the ΔADC and ΔSUVmax for both pCR and major response Both the ΔADC and ΔSUVmax during and after nCRT did not result in a significant identification of pCRDetection of major response with ΔADC during and after nCRT resulted in a P value of 0111 and 0123 and an AUC of 083 and 076 respectively The ΔSUVmax during and after nCRT did not result in a significant identification of major response

158


Tabl

e 3

Out

com

e of

the Δ

AD

C a

nd Δ

SUV

max

in a

logi

stic

reg

ress

ion

for

pCR

and

res

pons

ePa

thol

ogic

com

plet

e re

spon

seR

espo

nder

sO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

CO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

C

SUV

max

T0

011

72

2e-0

560

71

062

20

471

846

00

222

30

802

047

SUV

max

T1

776

e-05

16e

-12

3742

30

294

071

033

30

016

30

580

056

SUV

max

T2

244

e-06

27e

-14

218

20

166

084

343

10

027

81

058

20

56ΔS

UV

max

dur

ing

170

80

018

89

082

30

380

684

00

310

084

50

61ΔS

UV

max

afte

r2

174

00

181

10

731

042

360

10

120

34

053

40

56A

DC

T0

005

20

08

10

251

073

063

00

017

70

787

054

AD

C T

1

004

56

4e-0

532

30

356

060

177

219

02

1562

376

013

50

80A

DC

T2

001

94

4e-0

58

20

201

076

001

40

00

840

041

083

ΔAD

C d

urin

g0

152

00

572

053

30

5433

173

70

341

5467

80

111

083

ΔAD

C a

fter

019

10

017

79

063

50

60

001

22E

-07

62

012

30

76

Abb

revi

atio

ns p

CR

= p

atho

logi

c co

mpl

ete

resp

onse

OR

= o

dds r

atio

CI =

con

fiden

ce in

terv

al A

UC

= a

rea

unde

r the

cur

ve A

DC

= a

ppar

ent d

iffus

ion

coeffi

cien

t SU

V =

stan

dard

ized

upt

ake

valu

e

159

9


DISCUSSION

The relatively high number of patients with a pCR after neoadjuvant chemoradiotherapy has prompted us to consider the option of a lsquowait and seersquo policy with intensive follow-up and salvage surgery in case of locoregional tumor regrowth However detection of complete responders remains difficult with the current 18F-FDG PET imaging technique DW-MRI scans might have a complementary value to the 18F-FDG PET because 18F-FDG PET scans depict metabolic tumor activity while DW-MRI scans displays tissue diffusion In the present study an expert radiologist predicted pCR on the post-nCRT DW-MRI scan with a 100 sensitivity and 93 specificity Although combining the 18F-FDG PET and DW-MRI decreased the sensitivity to 67 the specificity increased to 100 Therefore no patients were falsely assessed as complete responders with the combination of the 18F-FDG PET and DW-MRI Decreasing the number of patients falsely predicted as complete pathologic responders is essential in the clinical decision making So far most studies determined the effectiveness of measuring changes in SUVmax after nCRT in the prediction of response as only moderately effective A major drawback of the SUVmax is that it is based on a single voxel and is therefore susceptible to artefacts Especially in patients with a radiation induced esophagitis the usefulness of the SUVmax might be limited In the present study we did not find an association between the ΔADC or ΔSUVmax and pCR A possible explanation is the low proportion of patients with pCR (n = 3) Although the association of the early and late ΔADC was slightly better with good responders (Mandard 1 amp 2) it did not translate into a clinical important difference So far several studies determined the effectiveness of DW-MRI in detecting pathologic complete response during or after neoadjuvant chemoradiotherapy De Cobelli et al (n = 32) found that the ΔADC measured after nCRT was inversely correlated with the tumor regression grade [11] In the present study we were unable to reproduce these findings A possible explanation might be that our primary endpoint is pCR (TRG 1) and major response (TRG 1-2) while in the study by De Cobelli the differences between patients with TRG 1-3 and 4-5 was determined [11] In our opinion differentiating between these patients after nCRT does not have clinical implications However differentiating between these two response groups might have clinical implications on the DW-MRI scan made during nCRT as these non-responders may benefit from an earlier resection or alternative treatment Van Rossum et al (n = 20) found that the treatment-induced change in ADC during the first 2-3 weeks of nCRT was highly predictive of pCR [9] In the present study we were unable to reproduce these findings (P = 0533) Although the timing of scanning was equal determination of the volume of interest

160


differed as van Rossum et al had delineated on DW-MR images with a b-value of 800 smm2 while the volume of interest in the present study was based on radiotherapeutic target volume delineation extracted from the CT scan to the T2 MRI at different b-values Weber et al (n = 15) found that an increased ADC was concordant to PET-response in most cases but was not correlated to prognosis [10] However these patients were treated with neoadjuvant chemotherapy aloneAs in most studies the present study was hampered by the low number of patients (n = 18) Therefore a large prospective multicenter study should be initiated to clarify the ability of 18F-FDG PET and DW-MRI in predicting pCR Moreover application of radiomics derived from 18F-FDG PET and DW-MRI scans might improve detection of pCR The prediction may improve even further by incorporating other factors such as biomarkers and genetic factorsIn conclusion combining the information of the 18F-FDG PET and DW-MRI seems to improve the clinical detection of pCR However differences exist between radiologists in the assessment of DW-MRI images Although the present study did not find a predictive value of objective outcomes as the ΔADC and ΔSUVmax quantifying response with objective outcomes will become increasingly important for future clinical decision making

ACKNOWLEDGEMENTS

This pilot study was performed in a multi-disciplinary collaboration with dr Wouter B Nagengast dr Derk Jan de Groot dr Boudewijn van Etten dr Jan Willem Haveman and dr Shekar VK Mahesh

161

9


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[2] Smit JK Guler S Beukema JC Mul VE Burgerhof JG Hospers GA Plukker JT Different recurrence pattern after neoadjuvant chemoradiotherapy compared to surgery alone in esophageal cancer patients Ann Surg Oncol 2013 204008-4015[3] Zanoni A Verlato G Giacopuzzi S et al Neoadjuvant concurrent chemoradiotherapy for locally advanced esophageal cancer in a single high-volume center Ann Surg Oncol 2013 201993-1999[4] Kwee RM Prediction of tumor response to neoadjuvant therapy in patients with esophageal cancer with use of 18F FDG PET a systematic review Radiology 2010 254707-717[5] van Rossum PS van Hillegersberg R Lever FM et al Imaging strategies in the management of oesophageal cancer whats the role of MRI Eur Radiol 2013 231753-1765[6] van Rossum PS van Lier AL Lips IM et al Imaging of oesophageal cancer with FDG PETCT and MRI Clin Radiol 2015 7081-95[7] Sakurada A Takahara T Kwee TC et al Diagnostic performance of diffusion-weighted magnetic resonance imaging in esophageal cancer Eur Radiol 2009 191461-1469[8] Imanishi S Shuto K Aoyagi T Kono T Saito H Matsubara H Diffusion-weighted magnetic resonance imaging for predicting and detecting the early response to chemoradiotherapy of advanced esophageal squamous cell carcinoma Dig Surg 2013 30240-248[9] van Rossum PS van Lier AL van Vulpen M et al Diffusion-weighted magnetic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer Radiother Oncol 2015 115163-170[10] Weber MA Bender K von Gall CC et al Assessment of diffusion-weighted MRI and 18F-fluoro-deoxyglucose PETCT in monitoring early response to neoadjuvant chemotherapy in adenocarcinoma of the esophagogastric junction J Gastrointestin Liver Dis 2013 2245-52

162


[11] De Cobelli F Giganti F Orsenigo E et al Apparent diffusion coefficient modifications in assessing gastro-oesophageal cancer response to neoadjuvant treatment comparison with tumour regression grade at histology Eur Radiol 2013 232165-2174[12] Mandard AM Dalibard F Mandard JC et al Pathologic assessment of tu-mor regression after preoperative chemoradiotherapy of esophageal carcinoma Clinicopathologic correlations Cancer 1994 732680-2686[13] Boellaard R ODoherty MJ Weber WA et al FDG PET and PETCT EANM procedure guidelines for tumour PET imaging version 10 Eur J Nucl Med Mol Imaging 2010 37181-200

163

9



SUMMARY



Summary

Jan Binne Hulshoff

CHAPTER 10


SUMMARY AND FUTURE PERSPECTIVES

The use of neoadjuvant chemoradiotherapy prior to an esophagectomy with curative intent has increased the 5-year overall survival of patients with locally advanced esophageal cancer with 13 compared to surgery only [12] Especially patients with a pathologic complete response benefit with a significantly longer overall and disease free survival [34] In the future pathologic complete responders might even benefit from refrainment of surgery using an intensive imaging follow-up and eventually salvage surgery in case of isolated loco-regional recurrence On the other hand some of the patients do not benefit from neoadjuvant chemoradiotherapy (nCRT) Therefore these patients have a prolonged time until surgery and may suffer from useless toxic side-effects Early identification of these non-responders may give us the opportunity to switch to alternative treatments preferably on patientrsquos tumor related biomarkers The aim of this thesis was to improve patients selection with the focus on refining staging strategies and treatment outcome and on assessing the clinical prediction of response to neoadjuvant chemoradiotherapy This will lead towards a more personalized treatment by increasing staging accuracy and treatment related outcome (part I) with a better prediction of pathologic complete response to nCRT (part II)

Part I Improving staging and treatment outcome of neoadjuvant chemoradiotherapy in esophageal cancer patientsIn Chapter 2 we determined the additional value of endoscopic ultrasound (EUS) after 18F-FDG PETCT upfront We found that EUS had an impact on the given treatment in approximately 29 of the patients In many cases EUS changed the radiotherapy target volume delineation by identifying undetected lymph node metastases or by a more precise assessment of suspect lymph nodes on 18F-FDG PETCT Based on this impact on the nodal tumor delineation EUS is a requisite in patients with a disputable upfront 18F-FDG PET staging who are eligible for neoadjuvant chemoradiotherapy Even after extensive staging with current sophisticated methods the accuracy of detecting lymph node metastases remained low as shown in Chapter 3 in patients treated with surgery-alone The high number of patients with clinical under-staging (25) in the surgery-alone group emphasized the difficulty in determining the true lymph node response clinically in patients treated with neoadjuvant chemoradiotherapy Approximately 8-16 of the patients treated with nCRT had proven interval metastases detected by CT or 18F-FDG PETCT imaging between neoadjuvant treatment and surgery [5-8] In a personalized treatment approach these patients should be detected prior to the surgical treatment in order to receive a more

168


suitable case-related palliative treatment Although 18F-FDG PETCT scans detect more distant metastases CT scanning has a higher specificity and widely available with associated lower costs In Chapter 4 we found that CT was moderate effective in the detection of interval metastases in 5 of the 9 distant metastases in a group of 97 patients after nCRT (sensitivity of 56) However based on a higher sensitivity of 18F-FDG PETCT in recent studies PETCT is currently the preferred detection method [910] To improve the treatment outcome of patients with locally advanced esophageal cancer it is also important to determine which patients will benefit less from neoadjuvant chemoradiotherapy In Chapter 5 we determined that a tumor free margin of gt03 mm was the optimal cut-off point for an adequate circumferential resection margin in esophageal cancer treated with neoadjuvant chemoradiotherapy which was also prognostic for a prolonged disease-free survival In patients treated with surgery-alone cut-off values between 00 and 02 mm were prognostic for disease-free survival Neoadjuvant chemoradiotherapy therefore seemed to alter the prognostic cut-off point of the circumferential marginHowever even in patients with a tumor-free resection margin the so-called R0 resections the incidence of recurrences are high This emphasized the need for a better selection of patients in the treatment with neoadjuvant chemoradiotherapy in locally advanced esophageal cancer Many of these advanced cancers were not considered eligible for nCRT according to the original CROSS inclusion criteria However these patients are currently also treated with nCRT In Chapter 6 we determined the influence of extending the original eligibility criteria for neoadjuvant chemoradiotherapy We found that the post-operative complications and comorbidities were equal in both groups However patients with the extended CROSS eligibility criteria had a significantly worse overall and disease-free survival compared to those who met the original CROSS eligibility criteria The survival in the group with extended criteria did not even differ from a cohort of patients treated with definitive chemoradiotherapy And the prognosis was even worse in patients with celiac lymph node metastases worse

Part II Improving response prediction in esophageal cancer patients treated with neoadjuvant chemoradiotherapy

As mentioned between 25 and 42 of the patients have a pathologic complete response and might benefit from an organ preserving approach with refraining from surgery followed by an intensive imaging follow-up in the near future [11112] However the commonly used imaging method to predict response the standardized uptake value (SUVmax) derived from 18F-FDG PET scans still yields a low sensitivity and specificity of 67 and 68 respectively [13] Textural

169

10


radiomic features derived from 18F-FDG PET scans depict the spatial correlation of 18F-FDG uptake between voxels and therefore display heterogeneity in uptake [14-19] In Chapter 7 we assessed the impact of textural features derived from the pre-treatment baseline 18F-FDG PETCT scan We found that a predictive model containing the histologic tumor type clinical T-stage the 18F-FDG PET derived textural feature long run low gray level emphasis and the CT derived textural feature run percentage was a superior predictive model compared to the traditional model containing the SUVmax In this study we only determined the predictive value of baseline textural features while we did not determine the influence of changes in textural features after neoadjuvant chemoradiotherapy Moreover we also did not determine the additional value of measuring changes in histogram distances before and after neoadjuvant chemoradiotherapy [20] In Chapter 8 we therefore determined the value of both changes in textural features and histogram distances after neoadjuvant chemoradiotherapy in detecting response Most features predictive for pathologic complete response were textural features derived from the post nCRT 18F-FDG PET Only two features that represented a change in textural feature value between the pre- and post-nCRT 18F-FDG PET were predictive for response Several conventional textural features (volume SUVmax and the SUVpeak) were also found to differ significantly between patients with a pathologic complete response and patients without a complete response Surprisingly the 18F-FDG PET feature long run low gray level emphasis was not predictive in this study while none of the histogram distances were predictive for complete response More future research is needed to refine this potential predictive method before it can be clinical useful Finally in Chapter 9 we reported the outcome of a pilot study about the additional role of diffusion-weighted magnetic resonance imaging (DW-MRI) to 18F-FDG PETCT scans in identifying tumor response to neoadjuvant chemoradiotherapy The clinical assessment of tumor response with the post-treatment DW-MRI resulted in a high sensitivity While the sensitivity did not increase after additional information from post-treatment 18F-FDG PETCT the specificity increased by using the combined 18F-FDG PETCT and DW-MRI Prediction of pathologic complete response based on the ΔADC and ΔSUVmax derived from both scans during and after neoadjuvant treatment was inadequate for clinical application However the positive complementary value of DW-MRI to the clinical response assessment with 18F-FDG PETCT should be validated in a larger study

170


FUTURE PERSPECTIVES

For a more personalized approach to be worthy in esophageal cancer patients future research should focus on a better detection of metastatic lymph nodes thereby improving the selection of patients Subsequently an improved staging will lead to a more precise radiotherapy delineation of metastatic lymph nodes with a higher number of patients with lymph node response A promising and potent method to increase the detection of lymph node metastases in esophageal cancer might be the staging of patients with high-resolution MRI scans [21] Since it is difficult to differentiate between malignant and benign enlarged lymph nodes the use of DWI-MRI contrast agents such as gadofosveset or iron oxide nanoparticles might be helpful on the T2 and diffuse contrast enhances (DCE) scans [22-24]

Because the survival of patients that did not fulfill the CROSS criteria for neoadjuvant chemoradiotherapy is significantly lower determining patients who will benefit less from neoadjuvant chemoradiotherapy is of vast importance in improving the outcome of esophageal cancer patients Especially the effectiveness of neoadjuvant compared to definitive chemoradiotherapy in patients with celiac lymph node metastases should be studied extensively

In the future treatment paradigm detection of non-responders before or during neoadjuvant chemoradiotherapy might prevent unnecessary treatment with possible complications or lead to an early switch to alternative treatment options Textural features and quantitative outcomes derived from 18F-FDG PET and DW-MRI scans or PET-MRI might be helpful in the differentiation between non-responders and responders in an early phase (during nCRT) However since the outcomes of different textural features studies are difficult to compare a meta-analysis should be performed to determine a set of reproducible predictive textural features for both early non-responders as late complete responders In the prediction of pathologic complete response a multicenter study is preferable to determine the additional value of 18F-FDG PET and DW-MRI scans Application of integrated 18F-FDG PET-MRI scans might increase the complementary value of these modalities even further as it might facilitate detection of small residual tumor lesions [2526] Combining current optimal imaging techniques with genetic factors biomarkers and plasma-derived tumor factors into a predictive model that differentiates non-responders average responders and complete responders might also facilitate a more personalized treatment approach

171

10


REFERENCES

[1] van Hagen P Hulshof MC van Lanschot JJ et al Preoperative chemoradiotherapy for esophageal or junctional cancer N Engl J Med 2012 3662074-2084[2] Shapiro J van Lanschot JJ Hulshof MC et al Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS) long-term results of a randomised controlled trial Lancet Oncol 2015 161090-1098[3] Scheer RV Fakiris AJ Johnstone PA Quantifying the benefit of a pathologic complete response after neoadjuvant chemoradiotherapy in the treatment of esophageal cancer Int J Radiat Oncol Biol Phys 2011 80996-1001[4] Donahue JM Nichols FC Li Z et al Complete pathologic response after neoadjuvant chemoradiotherapy for esophageal cancer is associated with enhanced survival Ann Thorac Surg 2009 87392-8 discussion 398-9[5] Bruzzi JF Swisher SG Truong MT et al Detection of interval distant metastases clinical utility of integrated CT-PET imaging in patients with esophageal carcinoma after neoadjuvant therapy Cancer 2007 109125-134[6] Cerfolio RJ Bryant AS Ohja B Bartolucci AA Eloubeidi MA The accuracy of endoscopic ultrasonography with fine-needle aspiration integrated positron emission tomography with computed tomography and computed tomography in restaging patients with esophageal cancer after neoadjuvant chemoradiotherapy J Thorac Cardiovasc Surg 2005 1291232-1241[7] Flamen P Van Cutsem E Lerut A et al Positron emission tomography for assessment of the response to induction radiochemotherapy in locally advanced oesophageal cancer Ann Oncol 2002 13361-368[8] Blom RL Schreurs WM Belgers HJ Oostenbrug LE Vliegen RF Sosef MN The value of post-neoadjuvant therapy PET-CT in the detection of interval metastases in esophageal carcinoma Eur J Surg Oncol 2011 37774-778[9] Stiekema J Vermeulen D Vegt E et al Detecting interval metastases and response assessment using 18F-FDG PETCT after neoadjuvant chemoradiotherapy for esophageal cancer Clin Nucl Med 2014 39862-867[10] Findlay JM Gillies RS Franklin JM et al Restaging oesophageal cancer after neoadjuvant therapy with (18)F-FDG PET-CT identifying interval metastases and predicting incurable disease at surgery Eur Radiol 2016 263519-3533[11] Smit JK Guler S Beukema JC Mul VE Burgerhof JG Hospers GA Plukker JT Different recurrence pattern after neoadjuvant chemoradiotherapy compared to surgery alone in esophageal cancer patients Ann Surg Oncol 2013 204008-4015

172


[12] Zanoni A Verlato G Giacopuzzi S et al Neoadjuvant concurrent chemoradiotherapy for locally advanced esophageal cancer in a single high-volume center Ann Surg Oncol 2013 201993-1999[13] Kwee RM Prediction of tumor response to neoadjuvant therapy in patients with esophageal cancer with use of 18F FDG PET a systematic review Radiology 2010 254707-717[14] van Rossum PS Fried DV Zhang L et al The Incremental Value of Subjective and Quantitative Assessment of 18F-FDG PET for the Prediction of Pathologic Complete Response to Preoperative Chemoradiotherapy in Esophageal Cancer J Nucl Med 2016 57691-700[15] Yip SS Coroller TP Sanford NN Mamon H Aerts HJ Berbeco RI Relationship between the Temporal Changes in Positron-Emission-Tomography-Imaging-Based Textural Features and Pathologic Response and Survival in Esophageal Cancer Patients Front Oncol 2016 672[16] Hatt M Majdoub M Vallieres M et al 18F-FDG PET uptake characterization through texture analysis investigating the complementary nature of heterogeneity and functional tumor volume in a multi-cancer site patient cohort J Nucl Med 2015 5638-44[17] Hatt M Tixier F Cheze Le Rest C Pradier O Visvikis D Robustness of intratumour (1)(8)F-FDG PET uptake heterogeneity quantification for therapy response prediction in oesophageal carcinoma Eur J Nucl Med Mol Imaging 2013 401662-1671[18] Tixier F Le Rest CC Hatt M et al Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitant radiochemotherapy in esophageal cancer J Nucl Med 2011 52369-378[19] Tixier F Hatt M Le Rest CC Le Pogam A Corcos L Visvikis D Reproducibility of tumor uptake heterogeneity characterization through textural feature analysis in 18F-FDG PET J Nucl Med 2012 53693-700[20] Tan S Zhang H Zhang Y Chen W DSouza WD Lu W Predicting pathologic tumor response to chemoradiotherapy with histogram distances characterizing longitudinal changes in 18F-FDG uptake patterns Med Phys 2013 40101707[21] Alper F Turkyilmaz A Kurtcan S Aydin Y Onbas O Acemoglu H Eroglu A Effectiveness of the STIR turbo spin-echo sequence MR imaging in evaluation of lymphadenopathy in esophageal cancer Eur J Radiol 2011 80625-628[22] Lambregts DM Beets GL Maas M et al Accuracy of gadofosveset-enhanced MRI for nodal staging and restaging in rectal cancer Ann Surg 2011 253539-545[23] Heijnen LA Lambregts DM Martens MH et al Performance of gadofosveset-enhanced MRI for staging rectal cancer nodes can the initial promising results be reproduced Eur Radiol 2014 24371-379

173

10


[24] Pultrum BB van der Jagt EJ van Westreenen HL et al Detection of lymph node metastases with ultrasmall superparamagnetic iron oxide (USPIO)-enhanced magnetic resonance imaging in oesophageal cancer a feasibility study Cancer Imaging 2009 919-28[25] Lee G I H Kim SJ et al Clinical implication of PETMR imaging in preoperative esophageal cancer staging comparison with PETCT endoscopic ultrasonography and CT J Nucl Med 2014 551242-1247[26] Matthews R Choi M Clinical Utility of Positron Emission Tomography Magnetic Resonance Imaging (PET-MRI) in Gastrointestinal Cancers Diagnostics (Basel) 2016 6103390diagnostics6030035

174


NEDERLANDSE SAMENVATTING

Slokdarmkanker staat op de 8ste plaats van de meest voorkomende maligniteiten De twee voornaamste histologische subtypes het plaveiselcelcarcinoom en het adenocarcinoom hebben een verschillende ontstaanswijze Het plaveiselcelcarcinoom ontstaat uit de epitheliale bekledingscellen van de slokdarm terwijl het adenocarcinoom uit het klierweefsel (vaak uit Barrett dysplastisch veranderd slijmvlies) ontstaat De laatste decennia steeg de incidentie van slokdarmkanker in Nederland van ongeveer 800 in 1990 naar 2500 in 2015 De voornaamste reden voor deze forse toename is een stijging van het aantal patieumlnten met een adenocarcinoom als gevolg van chronische reflux oesofagitis en de toegenomen adipositas in de meeste Westerse landen Door een continue reflux van maaginhoud en galzure zouten in de slokdarm neemt het aantal patieumlnten met een door oesofagitis ernstig dysplastisch veranderd slijmvlies Barrett slokdarm als voorloper van het adenocarcinoom ook toe Bij het stellen van de diagnose slokdarmkanker heeft ongeveer de helft van de patieumlnten metastasen op afstand (M+) of een lokaal doorgroeiende tumor (T4) waardoor een curatieve behandeling niet meer mogelijk is Indien curatie mogelijk is is chirurgie tot heden het belangrijkste onderdeel in de behandeling van deze patieumlnten Hierbij moet worden opgemerkt dat in geval van een beperkte oppervlakkige tumorlaesie (T1a) een curatieve endo(sub)mucosale resectie verricht kan worden Bij deze tumoren is de aanwezigheid van kliermetastasen verwaarloosbaar klein tot afwezig in tegenstelling tot de dieper infiltrerende tumoren (geT2) waarbij de aanwezigheid van kliermetastasen wel relatief groot is Daarnaast is definitieve chemoradiotherapie een reeumlle alternatieve behandelingsoptie indien curatieve chirurgie onmogelijk is door een lokaal te vergevorderde tumor (stadium T4b tumor) of de operatieve procedure te riskant is zoals bij hoge leeftijd enof ernstige nevenaandoeningen (comorbiditeiten) Waarbij opgemerkt moet worden dat de kans op genezing wel in meer of mindere mate beperkt is Voor langdurige lokale controle is radicale chirurgie (R0) nog steeds een van de belangrijkste factoren voor een curatieve behandeling De 5-jaars overleving van patieumlnten die alleen chirurgisch behandeld worden door een in opzet curatieve slokdarmresectie ligt rond de 20-35 waarbij de meeste patieumlnten al een lokaal vergevorderd carcinoom hebben met ook uitgebreide lymfekliermetastasen De overleving van deze patieumlnten verbeterde in het afgelopen decennium mede door de gunstige resultaten van de Nederlandse CROSS (ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study) trial In de CROSS-studie werden patieumlnten vooraf behandeld met een gecombineerde chemotherapie (carboplatin and paclitaxel) en radiotherapie (414 Gy in 23 fracties van 18 Gy) alvorens

175

10


geopereerd te worden Door deze zo geheten neoadjuvante chemoradiotherapie (nCRT) steeg de 5-jaarsoverleving van 34 met alleen een chirurgische resectie naar rond de 47 Naast een absoluut hoger aantal radicale resecties is de pathologische afwezigheid van vitaal tumorweefsel na chemoradiotherapie een belangrijk voordeel van de neoadjuvante CROSS-behandeling Patieumlnten met een pathologisch complete respons (25-42) na nCRT hebben niet alleen een significant betere overleving maar ook minder lokale tumorrecidieven Gezien deze goede respons met afwezigheid van tumorweefsel wordt steeds meer getwijfeld over de noodzaak van een slokdarmresectie indien een adequate complete respons voorspeld zou kunnen worden Zeker omdat een slokdarmresectie een grote interventie is met een relatief hoge morbiditeit en postoperatieve mortaliteit tussen de 2-4 Anderzijds heeft ongeveer 20-25 van de patieumlnten geen baat bij de voorbehandeling met chemoradiotherapie Deze patieumlnten hebben echter wel last van bijwerkingen en complicaties van de gegeven chemoradiotherapie In de nabije toekomst is een selectievere behandeling wenselijk waarbij een voor de patieumlnt afgemeten beste behandeling kan worden toegepast Het ontwikkelen van handvaten waarbij beter voorspeld kan worden welke patieumlnt baat heeft van de voorbehandeling zal bijdragen aan betere zorg voor deze patieumlntenpopulatie

Doel van dit proefschriftHet doel van dit proefschrift is om de patieumlnten-selectie te verbeteren en zo toe te werken naar een optimale behandeling van patieumlnten met slokdarmkanker De focus ligt hierbij op het verbeteren van de tumor stadieumlring in het algemeen en in het voorspellen van een pathologisch complete respons op de voorbehandeling met chemoradiotherapie Deel I van dit proefschrift heeft als doel de accuraatheid van de stadieumlring te verbeteren en zo meer inzicht te verkrijgen in welke patieumlnten wel en geen baat hebben van de voorbehandeling Deel II heeft als doel om de voorspelling van een pathologisch complete respons te verbeteren

Deel I Verbetering van de stadieumlring en de uitkomst van de neoadjuvante chemoradiotherapie bij patieumlnten met slokdarmkanker

Verbeteren van de stadieumlringDe standaard diagnostiek bij patieumlnten met slokdarmkanker bestaat uit endoscopie computertomografie (CT) 18F-fluor-desoxyglucose positronemissietomografie (18F-FDG PET) al dan niet gecombineerd met een CT (18F-FDG -CT) en endoechografie (EUS) De diagnose wordt histologisch gesteld op basis van biopten die afgenomen worden tijdens endoscopie De uitkomst van alle deze stadieumlrende technieken wordt vervolgens gerapporteerd volgens de huidige TNM-classificatie van de American Joint Committee on Cancer (AJCC) waarbij de

176


T(umor)-stagering staat voor de lokale invasiediepte van de tumor de N(odale)-stagering voor het aantal lymfekliermetastasen en de M(etastase)-stagering voor de aanwezigheid van afstandsmetastasen Tijdens een multidisciplinair overleg (MDO) waar artsen van alle betrokken specialismen aanwezig zijn wordt op basis van de TNM-stadieumlring de lichamelijke en mentale staat van de patieumlnt de voorgeschiedenis en leeftijd van patieumlnt met de aanwezige comorbiditeiten en de voorkeur van de patieumlnt de optimale behandeling besproken en geadviseerd Alle patieumlnten met een lokaal gevorderde slokdarmcarcinoom (T1N1-3 en T2-4aN0-3) zonder afstandsmetastasen (M0) met een adequate conditie zonder significante comorbiditeiten komen in principe in aanmerking voor een voorbehandeling met chemoradiotherapie gevolgd door een slokdarmresectie Zoals eerder vermeld hebben niet alle patieumlnten baat bij deze voorbehandeling Verbeteren van de selectie van patieumlnten is nodig om de uitkomsten van de behandeling in het algemeen te verbeteren en de neoadjuvante behandeling te reserveren voor patieumlnten die potentieel baat hebben en onnodige behandeling te vermijden bij patieumlnten die geen baat zullen ondervinden bij de voorbehandeling De beste methode om de lokale invasiediepte van de tumor (T-stage) en de aanwezigheid van lymfekliermetastasen vast te stellen is met endoechografie (EUS) waarbij EUS een sensitiviteit en specificiteit van respectievelijk 80 en 70 heeft in het vaststellen van lymfekliermetastasen Een belangrijk voordeel van de EUS is dat met behulp van een dunne naald biopsie cytologisch materiaal verkregen kan worden van lymfeklieren Op deze manier kan een vermoeden op lymfekliermetastase pathologisch bewezen worden Met de toevoeging van de echogeleide dunne naald biopsie neemt de sensitiviteit en specificiteit van EUS zelfs nog verder toe Er zijn echter ook nadelen verbonden aan de stadieumlring met EUS zo wordt het onderzoek door patieumlnten meestal als onprettig ervaren bestaat er een kleine kans op een slokdarmbloeding op aspiratie en op perforatie van de slokdarm Daarnaast is de EUS erg afhankelijk van de kundigheid van de maag- darm- leverarts Een ander nadeel is dat er bij 20-36 van de patieumlnten een niet te passeren stenose bestaat waardoor het onmogelijk is om de tumor en eventuele kliermetastasen adequaat te beoordelen De beste techniek om afstandsmetastasen (M-stage) vast te stellen is de 18F-FDG PET-scan met een sensitiviteit en specificiteit van respectievelijk 71 en 93 Een 18F-FDG PET-scan is een metabolische beeldvormende techniek dat de opname van radioactief gelabeld glucose (18F-fluordesoxyglucose) laat zien Omdat een tumor metabool erg actief is worden de primaire tumor en eventuele klieren afstandsmetastasen op de 18F-FDG PET zichtbaar als een 18F-FDG stapelende laesie Wanneer de 18F-FDG PET gecombineerd wordt met een CT is de detectie van klieren afstandsmetastasen zelfs nog beter In de huidige stadieumlring is er echter geen standaard volgorde van stadieumlring de tumoren worden gestadieumlerd met 18F-FDG PETCT en EUS in een willekeurige volgorde Mede door de goede

177

10


detectie van afstands-metastasen is de meest kosteneffectieve methode om een patieumlnt eerst te stadieumlren met een 18F-FDG PETCT De 18F-FDG PET up-front bleek ook de beste methode om een potentiele curatieve resectabiliteit van de tumor vast te stellen Een bijkomend voordeel van een upfront 18F-FDG PETCT is de betere stadieumlring van kliermetastasen indien deze gevolgd wordt door een EUS en er gericht een dunne naald biopt verricht kan worden van een op 18F-FDG PETCT suspecte of dubieuze lymfeklier Verdachte kliermetastasen rond de primaire tumor zijn echter niet adequaat met cytologie te beoordelen door een te grote kans op fout-positieve uitslag daar de FNA door primair tumorweefsel wordt uitgevoerd EUS-FNA op geleide van extra-regionale 18F-FDG-PET positieve klieren en dus op enige afstand van de primair tumor is wel goed en betrouwbaar mogelijk Echter met de huidige 18F-FDG PETCT-scans is de detectie van lymfekliermetastasen buiten het oorspronkelijk tumorgebied mogelijk zelfs zo goed dat een EUS eventueel achterwege gelaten kan worden indien de uitkomst geen klinische gevolgen heeft Dit zou ook het risico op EUS geassocieerde complicaties kunnen verkleinen In hoofdstuk 2 werd de additionele waarde van EUS na 18F-FDG PETCT onderzocht in 29 van de gevallen beiumlnvloedde EUS na 18F-FDG PETCT de behandeling Vooral bij patieumlnten die behandeld werden met neoadjuvante chemoradiotherapie gevolgd door chirurgie bleek de EUS van additionele waarde na een 18F-FDG PET upfront in het adequaat vaststellen van lymfekliermetastasen met de daaraan gerelateerde gevolgen voor de bestralingsvelden Echter zelfs na een complete stadieumlring met 18F-FDG PETCT en EUS blijft de accuraatheid van de detectie van lymfeklier metastasen laag Dit terwijl een onnauwkeurige detectie van lymfekliermetastasen klinische gevolgen kan hebben niet alleen kan het ertoe leiden dat lymfekliermetastasen buiten het te bestralen gebied (doelvolume) blijft maar ook kunnen sommige patieumlnten onder-behandeld worden en er ten onrechte geen neoadjuvante chemoradiotherapie wordt gegeven In hoofdstuk 3 werd de accuraatheid van alle stadieumlringstechnieken tezamen bepaald in het zo goed mogelijk vaststellen van lymfekliermetastasen Lymfeklierstadieumlring bleek inaccuraat in zowel patieumlnten die alleen chirurgisch behandeld werden als in patieumlnten die behandeld werden met neoadjuvante chemoradiotherapie In 25 van de chirurgisch behandelde patieumlnten was er sprake van onder-stadieumlring Deze inadequate lymfeklierstadieumlring bemoeilijkt het bepalen van de daadwerkelijke tumorrespons Dit verklaart mede de grote spreiding in de huidige respons percentages van 25-42

Daarnaast heeft tussen de 8 en 17 van de patieumlnten een intervalmetastase een metastase gevonden in de periode tussen de start van voorbehandeling en de operatie Voor patieumlnten met een intervalmetastase is curatie meestal niet meer mogelijk Het is daarom nuttig om deze metastasen voacuteoacuter de operatie vast te stellen

178


zodat de patieumlnt geen nutteloze ingreep hoeft te ondergaan en gestart kan worden met een meer geschikte palliatieve behandeling Ondanks het relatief hoge aantal patieumlnten met een intervalmetastase is de her-stadieumlring niet opgenomen in de huidige landelijke richtlijn De beste methodes om intervalmetastasen vast te stellen zijn de CT en 18F-FDG PETCT-scan Hoewel de 18F-FDG PETCT-scan beter is in het vaststellen van deze intervalmetastasen heeft het ook een hoger percentage van vals-positieve uitkomsten dan de CT Aan de andere kant zijn de kosten van CT-scans lager In hoofdstuk 4 werd de effectiviteit onderzocht van de CT in de identificatie van interval metastasen gevonden met beeldvorming tijdens de operatie of binnen 3 maanden na de operatie Met behulp van CT werden 5 van de 9 intervalmetastasen gevonden in een groep van 97 patieumlnten Met een sensitiviteit van 56 was de CT matig effectief in de detectie van afstandsmetastasen Recente studies vonden een hogere sensitiviteit van 18F-FDG PETCT-scans in de detectie van intervalmetastasen Thans lijkt het daarom beter 18F-FDG PETCT-scan te gebruiken voor detectie van intervalmetastasen

Verbeteren van de uitkomst na voorbehandeling met chemoradiotherapieOm de uitkomst te kunnen verbeteren is het belangrijk om meer inzicht te verkrijgen in welke patieumlnten wel en welke geen baat hebben bij de neoadjuvante chemoradiotherapie Een van de belangrijkste prognostische variabelen in patieumlnten die chirurgisch behandeld worden is de radicaliteit van de resectie zich uitend in de circumferentieumlle resectie marge (CRM) De CRM kan door de patholoog als microscopisch vrij (R0-resectie) of microscopisch niet tumor-vrij (R1-resectie) afgegeven worden Er zijn echter twee verschillende definities van een vrije resectie marge De College of American Pathologists (CAP) definieert de CRM als vrij als vitale tumorcellen zich meer dan gt00 mm van de geiumlnkte buitenrand bevinden terwijl de Royal College of Pathologists (RCP) de CRM afgeeft als vrij als vitale tumorcellen gt10 mm van de geiumlnkte buitenrand zitten Ondanks dat de prognostische waarde van de CRM duidelijk is in primair chirurgisch behandelde patieumlnten is de waarde na een voorbehandeling met chemoradiotherapie onduidelijk In hoofdstuk 5 werd de prognostische waarde van de CRM op de ziektevrije overleving onderzocht na een voorbehandeling met chemoradiotherapie Zowel de CAP als de RCP-definitie van een vrije resectie marge was in het onderzoek niet prognostisch voor de ziektevrije overleving Een afkappunt van gt03 mm was echter wel prognostisch voor een langere ziektevrije overleving In de primair chirurgische controlegroep was een afkappunt tussen de 00-02 mm prognostisch voor ziektevrije overleving Vanwege de goede uitkomsten van de CROSS-studie en het hoge aantal patieumlnten met een pathologisch complete response werd de patieumlntselectie voor behandeling met chemoradiotherapie uitgebreid Hierdoor werden ook patieumlnten met het CROSS-schema behandeld die niet voldeden aan de originele

179

10


inclusiecriteria van de CROSS-studie Het is echter onbekend of deze uitbreiding van de behandelingscriteria ook gevolgen heeft voor het aantal complicaties het post-operatief overlijden en de overleving Uit hoofdstuk 6 kwam naar voren dat uitbreiding van de criteria voor behandeling met neoadjuvante chemoradiotherapie geen effect had op het optreden van complicaties en postoperatief overlijden De overleving in de groep die niet voldeed aan de traditionele CROSS-inclusiecriteria was echter wel significant lager en niet significant verschillend van de definitieve chemoradiotherapie referentiegroep Verder bleek in de onderzochte groep de prognose bij de aanwezigheid van coeliacale lymfekliermetastasen infaust te zijn

Deel II Verbeteren van het voorspellen van pathologisch complete respons op de voorbehandeling met chemoradiotherapie

Van alle voor-behandelde patieumlnten heeft ongeveer 29 een pathologisch complete respons wat betekent dat er geen vitaal tumorweefsel meer te vinden is bij de pathologische beoordeling Deze patieumlnten hebben in de toekomst mogelijk baat bij het achterwege laten van de slokdarmresectie In plaats daarvan zou een intensieve follow-up met lsquosalvage surgeryrsquo in het geval van een lokaal of regionaal recidief een optie kunnen zijn Echter om dit beleid te kunnen voeren is het van belang dat de predictie van pathologisch complete respons goed genoeg is om een valide klinische beslissing te kunnen nemen Tot nog toe wordt de predictie van respons veelal verricht door het meten van de afname van de maximale waarde van de 18F-FDG opname (SUVmax) met de PETCT De sensitiviteit en specificiteit van respectievelijk 67 en 68 is echter te laag voor het nemen van klinische beslissingen In de praktijk zou dit namelijk betekenen dat sommige patieumlnten met een veronderstelde pathologisch complete respons nog vitale tumorcellen in situ hebben Een van de grootste nadelen van de SUVmax is dat het gebaseerd is op eacuteeacuten pixel Daardoor is de SUVmax gevoelig voor ruis Een nieuwe methode om respons te voorspellen is met behulp van texture features die verkregen worden van 18F-FDG PET-scans Met deze methode wordt gekeken naar kenmerken in het beeld en van patronen in de 18F-FDG-opname binnen een tumor en daarmee kan de heterogeniciteit in 18F-FDG opname worden gedetecteerd Verschillende onderzoeken vonden dat texture features verkregen van 18F-FDG PETCT-scans beter de respons voorspelden dan de SUVmax In hoofdstuk 7 bleek een predictie model met daarin het histologische tumor type de klinische T-stage de 18F-FDG PET feature long run low gray level emphasis en de CT feature run percentage een betere voorspeller voor pathologisch complete respons dan de veel gebruikte SUVmax In hoofdstuk 7 werd alleen gekeken naar de voorspellende waarde van lsquobaselinersquo features die geeumlxtraheerd werden uit de 18F-FDG PET-scans voor de behandeling bij stadieumlring terwijl er ook 18F-FDG PET-scans gemaakt werden na

180


de behandeling met chemoradiotherapie Daarnaast kan respons ook voorspeld worden door het verschil te meten in 18F-FDG uptake histogrammen voacuteoacuter en naacute de chemoradiatie In hoofdstuk 8 werd de waarde bepaald van veranderingen in texture features en histogram afstanden naacute een voorbehandeling met chemoradiotherapie dus afkomstig van de post-nCRT 18F-FDG PET-scan De meeste features die significant verschilden tussen patieumlnten met een pathologisch complete respons en patieumlnten met een incomplete respons waren post-nCRT 18F-FDG PET-scan features Dit benadrukt het belang van de post-nCRT 18F-FDG PET features Maar 2 features die het verschil meten tussen de baseline en post-nCRT 18F-FDG PET verschilden significant tussen patieumlnten met een complete respons en patieumlnten met een niet-complete respons De baseline 18F-FDG PET feature long run low gray level emphasis was in het huidige onderzoek echter niet geassocieerd met pathologisch complete respons terwijl deze feature in hoofdstuk 6 wel predictief bleek voor responsHet moet opgemerkt worden dat in er in dit onderzoek alleen maar een univariate analyse werd verricht Toekomstig onderzoek met een multivariate analyse is daarom nodig om een beter beeld te geven van de toegevoegde waarde van texture features in de klinische setting Een andere imaging methode waarmee tumor respons voorspeld kan worden is met behulp van diffusie gewogen magnetic resonance imaging (DW-MRI) DW-MRI is gevoelig voor water mobiliteit (diffusie) wat gebruikt kan gebruikt worden om de apparent diffusion coefficient (ADC) te meten In verschillende onderzoeken was een verandering in de ADC-waarde na chemoradiotherapie voorspellend voor pathologisch complete respons In hoofdstuk 9 werd in een pilotstudie de waarde van DW-MRI en 18F-FDG PETCT onderzocht in de detectie van patieumlnten met een pathologisch complete respons Door de uitkomst van de klinische beoordeling van zowel de DW-MRI als de 18F-FDG PETCT-scan te combineren was het mogelijk om een sensitiviteit van 67 en een specificiteit van 100 te halen Het gebruik van kwantitatieve uitkomsten zoals het verschil in de maximale standardized uptake value (SUVmax) bij de 18F-FDG PET en de apparent diffusion coefficient (ADC) bij de DW-MRI leidde niet tot een adequate voorspelling van pathologisch complete respons Hierbij moet echter opgemerkt worden dat de uitkomsten van deze pilotstudie uitgebreid onderzocht dienen te worden in een grotere studie

181

10


PUBLICATIONS IN PUBMED

1 Impact of Endoscopic Ultrasonography on 18F-FDG-PETCT Upfront Towards Patient Specific Esophageal Cancer Treatment JB Hulshoff VEM Mul HE de Boer W Noordzij T Korteweg HM van Dullemen WB Nagengast V Oppedijk JPEN Pierie JTM Plukker Accepted for publication Ann Surg Oncol 2017 Mar 16

2 Effect of extending the original eligibility criteria for the ldquoCROSSrdquo neoadjuvant chemoradiotherapy on toxicity and survival in esophageal cancer EC de Heer J B Hulshoff D Klerk JGM Burgerhof DJA de Groot J Th M Plukker GAP Hospers Accepted for publication Ann Surg Oncol 2017 Feb 10

3 Predicting response to neoadjuvant chemoradiotherapy in esophageal cancer with textural features derived from pre-treatment 18F-FDG PETCT imaging RJ Beukinga JB Hulshoff LV Dijk C Muijs JGM Burgerhof G Kats-Ugurlu R Slart C Slump VEM Mul JTM Plukker J Nucl Med 2016 Oct 13 pii jnumed116180299

4 Prognostic Value of the Circumferential Resection Margin in Esophageal Cancer Patients After Neoadjuvant Chemoradiotherapy JB Hulshoff Z Faiz A Karrenbeld G Kats-Ugurlu JGM Burgerhof JK Smit JTM Plukker Ann Surg Oncol 2015 Dec22 Suppl 3S1301-9

5 Evaluation of progression prior to surgery after neoadjuvant chemoradiotherapy with computed tomography in esophageal cancer patients JB Hulshoff JK Smit EJ van der Jagt JTM Plukker Am J Surg 2014 Jul208(1)73-9

182


DANKWOORD

Dit proefschrift was niet tot stand gekomen zonder de steun en hulp van veel mensen De volgende personen wil ik graag bij naam en toenaam bedanken voor hun hulp bij het tot stand komen van dit proefschrift

Mijn promotor Prof Dr JTM Plukker beste John tijdens een korte stage in mijn derde jaar werd ik direct enthousiast door uw verhalen Mede door dit enthousiasme besloot ik mijn wetenschappelijke stage bij u te lopen en gaandeweg werd voor mij duidelijk dat ik graag een MD PhD traject wou gaan doen Gedurende dit traject had u altijd tijd ondanks uw vele klinische werkzaamheden Uw gedrevenheid specialisme overschrijdende blik bekwaamheid en patieumlntgerichtheid zullen mij altijd bijblijven

Dr T Korteweg beste Tijmen bedankt voor alle tijd die u als co-promotor besteed heeft aan het beoordelen van de MRI-beelden Deze periode was voor mij een zeer leerzame ervaring en in de toekomst hoop ik deze kennis verder uit te breiden

Dr HM van Dullemen beste Hendrik graag wil ik u bedanken voor uw rol als co-promotor en als enthousiaste medebegeleider Ook al zat het in de opstartfase soms tegen (twee onderzoeken bleken helaas onhaalbaar) bleef u enthousiast Uw relativeringsvermogen inzet en hulp waren zeer belangrijk in de opstartfase

Drs V Mul beste Veronique bedankt voor de goede samenwerking en kritische noot bij de verschillende onderzoeken Uw kennis van en inzet in de delineatie van de radiotherapeutische doelvolumina heb ik zeer gewaardeerd Met het binnenhalen van de KWF-subsidie voor het PETMRI-onderzoek belooft het een mooie tijd te gaan worden

MSc J Beukinga beste Jorn jouw technische opleiding kennis en goede uitleg hebben een zeer grote bijdrage gehad in het voltooien van dit proefschrift Daarnaast ben je een goede collega en een goede kameraad Ik hoop dat we in de toekomst nog veel kunnen samenwerken met onderzoek

Drs A Karrenbeld en Dr G Kats beste Arend en Gursah graag wil ik jullie bedanken voor jullie bijzondere inzet bij de verschillende studies Jullie benaderbaarheid en kritische blik waren erg belangrijk bij de verschillende onderzoeken waaraan jullie participeerden Daarnaast wil ik de pathologie assistenten en medewerkers graag bedanken voor het verwerken van de resectie preparaten

183

10


Dr A Bartels-Rutten beste Annemarieke bedankt voor uw inzet bij het MRI-onderzoek

MSc JGM Burgerhof beste Hans bedankt voor uw hulp bij het oplossen van verschillende statistische vraagstukken

Jan Hendrik Potze en Peter Kappert bedankt voor jullie bereidwilligheid om na de werktijden en soms op vrije dagen patieumlnten te scannen Jullie inzet is van groot belang geweest voor het PETMRI-onderzoek Daarnaast was jullie enthousiasme en benaderbaarheid belangrijk in het opzetten van dit onderzoek

Beste voorgangers beste Dirk Bosch en Justin Smit bedankt voor jullie begeleiding tijdens mijn wetenschappelijke stage en in het begin van mijn MD-PhD In de toekomst gaan we ongetwijfeld weer samenwerken

Verder wil ik alle studenten bedanken die meegewerkt hebben aan de verschillende onderzoeken Enkele student-onderzoekers wil ik nadrukkelijk bedanken voor hun inzet Willemieke Dijksterhuis Hiske de Boer Kristien Jipping Ellen de Heer en Thijs van der Linden bedankt voor jullie inzet en enthousiasme

Graag wil ik de MD-PhDlsquoers in het Triade gebouw met wie ik veel omging bedanken Een goede balans tussen werk en plezier is erg belangrijk en jullie hadden hier een groot aandeel in In het bijzonder wil nog de volgende mensen bedanken Rik Mencke bedankt voor je onderzoeksadvies in het begin van mijn onderzoek carriegravere Beste Andries Groen het was altijd leuk met jou om over onderzoek te filosoferen en het was ook nog best gezellig Beste Leon van Dullemen je hebt veel adviezen gegeven niet alleen op onderzoeksgebied maar ook als goede vriend

Lieve familie en vrienden graag wil ik jullie bedanken voor jullie steun toeverlaat en hulp bij het tot stand komen van dit proefschrift

184


CURRICULUM VITEA

Jan Binne Hulshoff werd geboren op 11 mei 1988 in Drachten als zoon van Evert en Agatha Hulshoff In 2006 haalde hij zijn vwo-diploma op het Drachtster Lyceum En na 2 jaar uitgeloot te zijn geweest en Biomedische Wetenschappen te hebben gestudeerd aan de Rijksuniversiteit Groningen mocht hij in 2008 gelukkig beginnen met de studie Geneeskunde Tijdens zijn bachelor had hij al een grote interesse in zowel de chirurgie als de radiologie Dit leidde in 2011 tot een wetenschappelijke stage onder begeleiding van prof JTM Plukker bij de Oncologische Chirurgie wat de basis vormde voor dit proefschrift Vervolgens doorliep hij zijn M2 stages in het MCL te Leeuwarden waarna hij in 2015 zijn semiarts stage heeft gelopen bij de Oncologische Chirurgie en de Maag- Darm- en Leverziekten in het UMCG

Momenteel is hij werkzaam bij de chirurgie in het Nij Smellinghe ziekenhuis te Drachten en vanaf September 2017 zal hij beginnen met de opleiding tot radioloog in het UMCG

185

10


IMPROVING PATIENT SELECTION TOWARDS

PERSONALIZED TREATMENT DECISIONS IN

ESOPHAGEAL CANCER

Jan Binne Hulshoff


Hulshoff JB









esophageal cancer

Proefschrift






door


te Smallingerland






7

21

23

37

55

71

89

109

111

131

151

165

167175182183185

TABLE OF CONTENT






Summary





Jan Binne Hulshoff

CHAPTER 1


INTRODUCTION


8




9

1





10





11

1








12



13

1


REFERENCES


14



15

1



16



17

1



18



19

1




PART I









John ThM Plukker


CHAPTER 2


ABSTRACT





24


BACKGROUND


25

2





26






27

2


RESULTS



28








T1 11 (35) 10 (36)T2 41 (129) 39 (140)T3 243 (764) 218 (781)T4a 23 (72) 12 (43)




29

2








30



nCRT (n=194) n ()












Number of EUS1 272 (975) 191 (985) 45 (100) 36 (900)2 7 (25) 3 (15) 0 (00) 4 (100)


31

2


DISCUSSION


32



33

2


REFERENCES


34



35

2










CHAPTER 3


ABSTRACT





38


INTRODUCTION




39

3






40






41

3


RESULTS



nCRT(n = 135) n ()

P value

Male 114 (844) 106 (785) P = 0210a


Histology P = 0737a





cT-stage P = 0527c

T1 3 (22) 1 (07)T2 18 (133) 21 (156)T3ampT4a 114 (844) 113 (837)

cN-stage P = 0984c

N0 36 (267) 36 (267)N1 67 (496) 67 (496)N2 28 (207) 29 (215)N3 4 (30) 3 (22)

42


pT-stage P lt 0001a


pN-stage P lt 0001a

N0 37 (274) 92 (681)N1 41 (304) 29 (215)N2 31 (230) 11 (81)N3 26 (193) 3 (22)




140 (90-188) 150 (120-220) P = 0009b



R1-resection 16 (119) 7 (52) P = 0050a



43

3






P value







3 (150) 3 (150) 4 (200) 10 (500)

cT12 tumors (n=21) 7 (333) 4 (190) 3 (143) 7 (333) P = 0002b




5 (227) 4 (182) 0 (00) 13 (591)

cT12 tumors (n=22) 8 (364) 2 (91) 1 (45) 11 (500) P = 0037b









44




45

3





46












pN1 304 141 ndash 655 0004

pN2 364 165 ndash 804 0001

pN3 454 203 ndash 1019 0000

Lymph-angioinvasion


409 139 ndash 1204

0034


47

3



DISCUSSION


48



49

3



50


REFERENCES


51

3



52



53

3








CHAPTER 4


ABSTRACT





56


BACKGROUND


57

4






58













59

4




60




RESULTS


61

4




Location tumor

Location metastasis





Liver Omental

Yes NoNo

No Yes NoYesYes


No Yes Yes


lymph node

No PET + Biopsy

Yes



62




63

4





64



DISCUSSION


65

4



66


DISCLOSURE


67

4


REFERENCES


68



69

4








John Th M Plukker


CHAPTER 5


ABSTRACT





72


BACKGROUND




73

5









cT-stage 0221T2 16 (154) 9 (86)T3 83 (798) 93 (886)T4a 5 (48) 3 (29)

74



N0 27 (260) 41 (390)N1 50 (481) 44 (419)N2 22 (212) 18 (171)N3 5 (48) 2 (19)


Tx 1 (10)T0 21 (202)T1 22 (212)T2 14 (135) 20 (190)T3 46 (442) 82 (781)T4a 0 (00) 3 (29)

pN-stage lt 0001N0 62 (596) 28 (267)N1 26 (250) 34 (324)N2 11 (106) 25 (238)N3 5 (48) 18 (171)


420)29 (155 ndash 560) 0241




75

5





76



RESULTS




77

5




78





100336736805


0001

003400000000

1004083391154


0007

007601600002


lt 0001351 144 ndash 857 0006




1005131298

124 ndash 2122236 ndash 7145

0013

00240003

NS

pT0pT1pT2pT3

100113183274


0054

083702790025

NS

pN0pN1pN2pN3

100258469687


lt 0001

000700000000

100056537731


0047

059000220086



79

5




80


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p

Surg

ery-

alon

e gr

oup

2-ye

ar D

FS2-

year

LR

FSH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 31

70)

CA

P0

410

21 ndash

08

30

012

CA

P m

odel

(AIC

= 16

82)

CA

P0

270

11 ndash

06

580

004

LN ra

tio2

571

01 ndash

65

10

047

LN ra

tio3

111

20 ndash

80

90

020

Tum

or le

ngth

268

133

ndash 5

43

000

6Tu

mor

leng

th10

99

249

ndash 4

843

000

2A

ngio

inva

sion

190

094

ndash 3

85

007

5N

umbe

r of L

N+

213

092

ndash 4

95

007

8R

CP

mod

el

(AIC

= 32

09)

RC

P0

830

38 ndash

17

80

627

RC

P m

odel

(AIC

= 17

42)

RC

P0

57

022

ndash 1

51

025

8LN

ratio

268

105

ndash 6

81

003

9LN

ratio

313

112

ndash 8

24

002

1A

ngio

inva

sion

195

094

ndash 4

03

007

2Pe

rineu

ral g

row

th1

890

99 ndash

35

90

053

Tum

or le

ngth

252

125

ndash 5

09

001

0Tu

mor

leng

th8

591

99 ndash

37

080

004

Num

ber o

f LN

192

084

ndash 4

39

012

3

81

5


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p -

cont

inue

dnC

RT

grou

p2-

year

DFS

2-ye

ar L

RH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 34

99)

CA

P0

470

18 ndash

12

30

124

CA

P m

odel

(AIC

= 73

5)

CA

P0

060

01 ndash

03

10

001

cT3

200

76 ndash

13

490

114

pN0

100

000

4

Gra

de16

91

212

ndash 1

350

50

008

pN1

270

131

ndash 5

59

000

7pN

2-3

339

143

ndash 8

03

000

5

RC

P m

odel

(AIC

= 35

03)

RC

P0

690

31 ndash

15

20

359

RC

P m

odel

(AIC

= 80

0)

RC

P1

010

08 ndash

13

500

995

cT1

000

275

pN0-

11

000

203

232

051

ndash 1

051

pN1-

28

810

31 ndash

252

23

pN0

100

001

4

Gra

de30

07

279

ndash 3

246

00

005

pN1

251

122

ndash 5

210

014

Num

ber o

f LN

073

006

ndash 8

94

080

4pN

2-3

299

123

ndash 7

32

001

6LN

ratio

260

044

ndash 1

542

029

4pT

0-1

100

035

9

pT2

196

070

ndash 5

49

020

2pT

3-4a

184

071

ndash 4

73

020

9

Abb

revi

atio

ns D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

CI=

con

fiden

ce in

terv

al S

CC

= sq

uam

ous c

ell c

arci

nom

a c

T= c

linic

al

T-st

age

cN

= c

linic

al ly

mph

nod

e st

age

pT=

pat

holo

gica

l T-s

tage

pN

= p

atho

logi

c ly

mph

nod

e st

age

LN

= ly

mph

nod

e C

RM

= ci

rcum

fere

ntia

l mar

gin

R

0= tu

mor

free

rese

ctio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

ove

rall

P-va

lue

of th

e ca

tego

ri-ca

l var

iabl

es

= Si

gnifi

cant

(Plt0

05)

82


DISCUSSION


83

5


Tabl

e 4

Stu

dies

reg

ardi

ng p

rogn

ostic

val

ue o

f the

cir

cum

fere

ntia

l res

ectio

n m

argi

n af

ter

neoa

djuv

ant c

hem

orad

ioth

erap

y

Stud

y (y

ear)

His

tolo

gySt

age

Patie

nts (

n)nC

RT

()

Out

com

eC

RM

defi

nitio

nP-

valu

e Th

omps

on e

t al

(200

8)A

C S

CC

cT1-

424

012

4 (5

2)

5-ye

ar su

rviv

al

RC

PN

SC

hao

et a

l (2

011)

SC

Cyp

T315

115

1 (1

00

)LR

FSR

CP

lt 0

05D

FSR

CP

lt 0

05D

SSR

CP

lt 0

05H

arvi

n et

al

(201

2)A

Cyp

T316

016

0 (1

00

)O

S D

FS L

RFS

C

AP

NS

OS

DFS

LR

FSR

CP

NS

Reid

et a

l (2

012)

AC

SC

CcT

1-4

269

42 (1

6)

DFS

RC

Plt

001

OS

RC

P0

05O

rsquoFar

rell

et a

l (2

013)

AC

SC

C o

ther

s cT

315

782

(52

)O

SR

CP

NS

OS

CA

P0

02Li

u et

al

(201

3)SC

CcT

1-4

9494

(100

)

OS

RC

Plt0

01

Abb

revi

atio

ns S

CC

= sq

uam

ous c

ell c

arci

nom

a A

C=

aden

ocar

cino

ma

nC

RT=

neoa

djuv

ant c

hem

orad

ioth

erap

y c

T= c

linic

al T

-sta

ge y

pT=

path

olog

ic

T-st

age

afte

r nC

RT D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

DSS

= di

seas

e sp

ecifi

c su

rviv

al C

RM

= ci

rcum

fere

ntia

l res

ectio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

on m

ultiv

aria

te a

naly

sis

84



CONCLUSION


DISCLOSURE


85

5


REFERENCES


86



87

5










CHAPTER 6


ABSTRACT





90


BACKGROUND




91

6






92






RESULTS


93

6




Group 1 (n = 89) n ()

Group 2 (n = 72)n ()

P value

Male 71 (798) 57 (792) P = 0924 a



0-1 85 (955) 64 (889)2 0 (00) 0 (00)Missing 4 (45) 8 (111)








cT-stage P = 0000 a

T1 0 (00) 2 (28)T2 25 (281) 5 (69)T3 63 (708) 56 (778)T4a 1 (11) 9 (125)

94


cN-stage P = 0024 a

N0 22 (247) 7 (97)N1 38 (427) 30 (417)N2 27 (303) 29 (403)N3 2 (22) 6 (83)



N0 57 (640) 44 (611)N1 18 (202) 16 (222)

N2 11 (124) 7 (97)N3 3 (34) 5 (69)



LN ratio (gt02 LN+) 12 (135) 11 (153) P=0747 a




95

6




Group 2 (n=72) n ()

P-value









Bleeding 0 (00) 1 (14) P = 0203 a

Nausea 3 (34) 4 (56) P = 0501 a

96


Fatigue 1 (11) 1 (14) P = 0880 a

Neurotoxic 0 (00) 2 (28) P = 0071 a

Diarrhea 0 (00) 1 (14) P = 0203 a

Esophagitis 2 (22) 5 (69) P = 0144 a



Pneumonia 41 (461) 28 (389) P = 0360 a




Arrhytmia 25 (281) 22 (306) P = 0732 a


Sepsis 8 (90) 6 (83) P = 0883 a


Chylothorax 11 (124) 3 (42) P = 0057 a



Renal failure 2 (22) 4 (56) P = 0276 a

Ileus 6 (67) 2 (28) P = 0237 a


60 (674) 50 (694) P = 0783 a


90-day mortality 6 (67) 7 (97) P=0492 a


97

6







ypN2 1896 (0989-3635)ypN3 3415 (1446-8064)


98




99

6








100



DISCUSSION


101

6



102



103

6


REFERENCES


104



105

6





P value

Male 57 (792) 61 (763) P = 0666 a



0-1 64 (889) 72 (900)2 0 (00) 6 (75)Missing 8 (111) 2 (25)









cT-stage P = 0001 a

T1 2 (28) 1 (13)T2 5 (69) 4 (50)T3 56 (778) 39 (488)T4a 9 (125) 32 (400)missing 0 (00) 4 (50)

cN-stage P = 0000 a

N0 7 (97) 14 (175)N1 30 (417) 61 (763)N2 29 (403) 4 (50)N3 6 (83) 1 (13)




106






107

6




PART II









CHAPTER 7


ABSTRACT





112


INTRODUCTION


113

7






114






115

7





116



RESULTS


117

7











5 3 31



118



119

7



120


Tabl

e 2

Est

imat

ed r

egre

ssio

n co

effici

ents

of t

he p

redi

ctio

n m

odel

s for

pat

holo

gic

com

plet

e re

spon

se w

ithou

t opt

imis

m c

orre

ctio

n

Mod

el 1

Mod

el 2

ndash 3

Mod

el 4

Mod

el 5

Mod

el 6

Varia

ble

Coe

fS

Ep

Coe

fS

EP

Coe

fS

Ep

Coe

fS

Ep

Coe

fS

Ep

Inte

rcep

t-0

88

050

008

-04

20

500

39-0

86

065

019

-11

50

720

11-1

83

091

004

SUV

max

-17

21

450

24

His

tolo

gy

AC

100

100

100

100

SC

C-1

70

061

001

-14

70

650

02-1

89

064

000

-16

30

670

02cT

stag

e

cT1

amp c

T21

001

001

001

00

cT3

amp c

T4a

203

078

001

198

080

001

227

081

000

223

083

001

LRLG

Le-P

ET0

560

550

310

710

590

23

RP-

CT

0

010

010

150

020

010

10

Abb

revi

atio

ns L

RLG

Le-P

ET l

ong

run

low

gra

y le

vel e

mph

asis

mea

sure

d on

PET

and

RP-

CT

run

perc

enta

ge m

easu

red

on C

T

121

7






AUC boot



DISCUSSION


122



123

7





Tixier et al [13]

41 AC SCC


Pre-nCRT

CR vs PR vs NR [32]


Hatt et al [11]

50 AC SCC


Pre-nCRT

CR + PR vs NR [32]


Tan et al [12]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

ΔSUVmax



Zhang et al [15]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]




Pre- and post-nCRT

TRG 1 vs 2-4 [33]


Current study

97 AC SCC


Pre-nCRT

TRG 1 vs 2-5 [19]


124



CONCLUSION



125

7


REFERENCES


126



127

7



128


129

7









CHAPTER 8


ABSTRACT





132


INTRODUCTION


133

8


METHODS





134






x 100baseline IBM

135

8



volumeSUVmax

SUVpeak










136



RESULTS


137

8



Male 40 (800)Female 10 (200)





T2 4 (80)T3 42 (840)T4a 4 (80)

cN-stageN0 7 (140)N1 20 (400)N2 19 (380)N3 4 (80)



Mandard TRG1 7 (140)2 15 (300)3 18 (360)4 9 (180)5 1 (20)


138




139

8




140




0208 065 093 0027 076 096 0065 072 097


0502 058 096 0240 064 078 0117 069 081


0146 067 096 0171 066 095 0074 071 083



141

8



142


DISCUSSION


143

8



144



145

8


REFERENCES


146



147

8



148


149

8










John ThM Plukker


In preparation

CHAPTER 9


ABSTRACT





152


BACKGROUND




153

9





154






155

9


RESULTS



156





157

9







158


Tabl

e 3

Out

com

e of

the Δ

AD

C a

nd Δ

SUV

max

in a

logi

stic

reg

ress

ion

for

pCR

and

res

pons

ePa

thol

ogic

com

plet

e re

spon

seR

espo

nder

sO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

CO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

C

SUV

max

T0

011

72

2e-0

560

71

062

20

471

846

00

222

30

802

047

SUV

max

T1

776

e-05

16e

-12

3742

30

294

071

033

30

016

30

580

056

SUV

max

T2

244

e-06

27e

-14

218

20

166

084

343

10

027

81

058

20

56ΔS

UV

max

dur

ing

170

80

018

89

082

30

380

684

00

310

084

50

61ΔS

UV

max

afte

r2

174

00

181

10

731

042

360

10

120

34

053

40

56A

DC

T0

005

20

08

10

251

073

063

00

017

70

787

054

AD

C T

1

004

56

4e-0

532

30

356

060

177

219

02

1562

376

013

50

80A

DC

T2

001

94

4e-0

58

20

201

076

001

40

00

840

041

083

ΔAD

C d

urin

g0

152

00

572

053

30

5433

173

70

341

5467

80

111

083

ΔAD

C a

fter

019

10

017

79

063

50

60

001

22E

-07

62

012

30

76

Abb

revi

atio

ns p

CR

= p

atho

logi

c co

mpl

ete

resp

onse

OR

= o

dds r

atio

CI =

con

fiden

ce in

terv

al A

UC

= a

rea

unde

r the

cur

ve A

DC

= a

ppar

ent d

iffus

ion

coeffi

cien

t SU

V =

stan

dard

ized

upt

ake

valu

e

159

9


DISCUSSION


160



ACKNOWLEDGEMENTS


161

9


REFERENCES


162



163

9



SUMMARY



Summary

Jan Binne Hulshoff

CHAPTER 10





168





169

10



170


FUTURE PERSPECTIVES




171

10


REFERENCES


172



173

10



174




175

10






176



177

10




178




179

10





180



181

10








182


DANKWOORD








183

10









184


CURRICULUM VITEA



185

10


Hulshoff JB









esophageal cancer

Proefschrift






door


te Smallingerland






7

21

23

37

55

71

89

109

111

131

151

165

167175182183185

TABLE OF CONTENT






Summary





Jan Binne Hulshoff

CHAPTER 1


INTRODUCTION


8




9

1





10





11

1








12



13

1


REFERENCES


14



15

1



16



17

1



18



19

1




PART I









John ThM Plukker


CHAPTER 2


ABSTRACT





24


BACKGROUND


25

2





26






27

2


RESULTS



28








T1 11 (35) 10 (36)T2 41 (129) 39 (140)T3 243 (764) 218 (781)T4a 23 (72) 12 (43)




29

2








30



nCRT (n=194) n ()












Number of EUS1 272 (975) 191 (985) 45 (100) 36 (900)2 7 (25) 3 (15) 0 (00) 4 (100)


31

2


DISCUSSION


32



33

2


REFERENCES


34



35

2










CHAPTER 3


ABSTRACT





38


INTRODUCTION




39

3






40






41

3


RESULTS



nCRT(n = 135) n ()

P value

Male 114 (844) 106 (785) P = 0210a


Histology P = 0737a





cT-stage P = 0527c

T1 3 (22) 1 (07)T2 18 (133) 21 (156)T3ampT4a 114 (844) 113 (837)

cN-stage P = 0984c

N0 36 (267) 36 (267)N1 67 (496) 67 (496)N2 28 (207) 29 (215)N3 4 (30) 3 (22)

42


pT-stage P lt 0001a


pN-stage P lt 0001a

N0 37 (274) 92 (681)N1 41 (304) 29 (215)N2 31 (230) 11 (81)N3 26 (193) 3 (22)




140 (90-188) 150 (120-220) P = 0009b



R1-resection 16 (119) 7 (52) P = 0050a



43

3






P value







3 (150) 3 (150) 4 (200) 10 (500)

cT12 tumors (n=21) 7 (333) 4 (190) 3 (143) 7 (333) P = 0002b




5 (227) 4 (182) 0 (00) 13 (591)

cT12 tumors (n=22) 8 (364) 2 (91) 1 (45) 11 (500) P = 0037b









44




45

3





46












pN1 304 141 ndash 655 0004

pN2 364 165 ndash 804 0001

pN3 454 203 ndash 1019 0000

Lymph-angioinvasion


409 139 ndash 1204

0034


47

3



DISCUSSION


48



49

3



50


REFERENCES


51

3



52



53

3








CHAPTER 4


ABSTRACT





56


BACKGROUND


57

4






58













59

4




60




RESULTS


61

4




Location tumor

Location metastasis





Liver Omental

Yes NoNo

No Yes NoYesYes


No Yes Yes


lymph node

No PET + Biopsy

Yes



62




63

4





64



DISCUSSION


65

4



66


DISCLOSURE


67

4


REFERENCES


68



69

4








John Th M Plukker


CHAPTER 5


ABSTRACT





72


BACKGROUND




73

5









cT-stage 0221T2 16 (154) 9 (86)T3 83 (798) 93 (886)T4a 5 (48) 3 (29)

74



N0 27 (260) 41 (390)N1 50 (481) 44 (419)N2 22 (212) 18 (171)N3 5 (48) 2 (19)


Tx 1 (10)T0 21 (202)T1 22 (212)T2 14 (135) 20 (190)T3 46 (442) 82 (781)T4a 0 (00) 3 (29)

pN-stage lt 0001N0 62 (596) 28 (267)N1 26 (250) 34 (324)N2 11 (106) 25 (238)N3 5 (48) 18 (171)


420)29 (155 ndash 560) 0241




75

5





76



RESULTS




77

5




78





100336736805


0001

003400000000

1004083391154


0007

007601600002


lt 0001351 144 ndash 857 0006




1005131298

124 ndash 2122236 ndash 7145

0013

00240003

NS

pT0pT1pT2pT3

100113183274


0054

083702790025

NS

pN0pN1pN2pN3

100258469687


lt 0001

000700000000

100056537731


0047

059000220086



79

5




80


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p

Surg

ery-

alon

e gr

oup

2-ye

ar D

FS2-

year

LR

FSH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 31

70)

CA

P0

410

21 ndash

08

30

012

CA

P m

odel

(AIC

= 16

82)

CA

P0

270

11 ndash

06

580

004

LN ra

tio2

571

01 ndash

65

10

047

LN ra

tio3

111

20 ndash

80

90

020

Tum

or le

ngth

268

133

ndash 5

43

000

6Tu

mor

leng

th10

99

249

ndash 4

843

000

2A

ngio

inva

sion

190

094

ndash 3

85

007

5N

umbe

r of L

N+

213

092

ndash 4

95

007

8R

CP

mod

el

(AIC

= 32

09)

RC

P0

830

38 ndash

17

80

627

RC

P m

odel

(AIC

= 17

42)

RC

P0

57

022

ndash 1

51

025

8LN

ratio

268

105

ndash 6

81

003

9LN

ratio

313

112

ndash 8

24

002

1A

ngio

inva

sion

195

094

ndash 4

03

007

2Pe

rineu

ral g

row

th1

890

99 ndash

35

90

053

Tum

or le

ngth

252

125

ndash 5

09

001

0Tu

mor

leng

th8

591

99 ndash

37

080

004

Num

ber o

f LN

192

084

ndash 4

39

012

3

81

5


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p -

cont

inue

dnC

RT

grou

p2-

year

DFS

2-ye

ar L

RH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 34

99)

CA

P0

470

18 ndash

12

30

124

CA

P m

odel

(AIC

= 73

5)

CA

P0

060

01 ndash

03

10

001

cT3

200

76 ndash

13

490

114

pN0

100

000

4

Gra

de16

91

212

ndash 1

350

50

008

pN1

270

131

ndash 5

59

000

7pN

2-3

339

143

ndash 8

03

000

5

RC

P m

odel

(AIC

= 35

03)

RC

P0

690

31 ndash

15

20

359

RC

P m

odel

(AIC

= 80

0)

RC

P1

010

08 ndash

13

500

995

cT1

000

275

pN0-

11

000

203

232

051

ndash 1

051

pN1-

28

810

31 ndash

252

23

pN0

100

001

4

Gra

de30

07

279

ndash 3

246

00

005

pN1

251

122

ndash 5

210

014

Num

ber o

f LN

073

006

ndash 8

94

080

4pN

2-3

299

123

ndash 7

32

001

6LN

ratio

260

044

ndash 1

542

029

4pT

0-1

100

035

9

pT2

196

070

ndash 5

49

020

2pT

3-4a

184

071

ndash 4

73

020

9

Abb

revi

atio

ns D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

CI=

con

fiden

ce in

terv

al S

CC

= sq

uam

ous c

ell c

arci

nom

a c

T= c

linic

al

T-st

age

cN

= c

linic

al ly

mph

nod

e st

age

pT=

pat

holo

gica

l T-s

tage

pN

= p

atho

logi

c ly

mph

nod

e st

age

LN

= ly

mph

nod

e C

RM

= ci

rcum

fere

ntia

l mar

gin

R

0= tu

mor

free

rese

ctio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

ove

rall

P-va

lue

of th

e ca

tego

ri-ca

l var

iabl

es

= Si

gnifi

cant

(Plt0

05)

82


DISCUSSION


83

5


Tabl

e 4

Stu

dies

reg

ardi

ng p

rogn

ostic

val

ue o

f the

cir

cum

fere

ntia

l res

ectio

n m

argi

n af

ter

neoa

djuv

ant c

hem

orad

ioth

erap

y

Stud

y (y

ear)

His

tolo

gySt

age

Patie

nts (

n)nC

RT

()

Out

com

eC

RM

defi

nitio

nP-

valu

e Th

omps

on e

t al

(200

8)A

C S

CC

cT1-

424

012

4 (5

2)

5-ye

ar su

rviv

al

RC

PN

SC

hao

et a

l (2

011)

SC

Cyp

T315

115

1 (1

00

)LR

FSR

CP

lt 0

05D

FSR

CP

lt 0

05D

SSR

CP

lt 0

05H

arvi

n et

al

(201

2)A

Cyp

T316

016

0 (1

00

)O

S D

FS L

RFS

C

AP

NS

OS

DFS

LR

FSR

CP

NS

Reid

et a

l (2

012)

AC

SC

CcT

1-4

269

42 (1

6)

DFS

RC

Plt

001

OS

RC

P0

05O

rsquoFar

rell

et a

l (2

013)

AC

SC

C o

ther

s cT

315

782

(52

)O

SR

CP

NS

OS

CA

P0

02Li

u et

al

(201

3)SC

CcT

1-4

9494

(100

)

OS

RC

Plt0

01

Abb

revi

atio

ns S

CC

= sq

uam

ous c

ell c

arci

nom

a A

C=

aden

ocar

cino

ma

nC

RT=

neoa

djuv

ant c

hem

orad

ioth

erap

y c

T= c

linic

al T

-sta

ge y

pT=

path

olog

ic

T-st

age

afte

r nC

RT D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

DSS

= di

seas

e sp

ecifi

c su

rviv

al C

RM

= ci

rcum

fere

ntia

l res

ectio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

on m

ultiv

aria

te a

naly

sis

84



CONCLUSION


DISCLOSURE


85

5


REFERENCES


86



87

5










CHAPTER 6


ABSTRACT





90


BACKGROUND




91

6






92






RESULTS


93

6




Group 1 (n = 89) n ()

Group 2 (n = 72)n ()

P value

Male 71 (798) 57 (792) P = 0924 a



0-1 85 (955) 64 (889)2 0 (00) 0 (00)Missing 4 (45) 8 (111)








cT-stage P = 0000 a

T1 0 (00) 2 (28)T2 25 (281) 5 (69)T3 63 (708) 56 (778)T4a 1 (11) 9 (125)

94


cN-stage P = 0024 a

N0 22 (247) 7 (97)N1 38 (427) 30 (417)N2 27 (303) 29 (403)N3 2 (22) 6 (83)



N0 57 (640) 44 (611)N1 18 (202) 16 (222)

N2 11 (124) 7 (97)N3 3 (34) 5 (69)



LN ratio (gt02 LN+) 12 (135) 11 (153) P=0747 a




95

6




Group 2 (n=72) n ()

P-value









Bleeding 0 (00) 1 (14) P = 0203 a

Nausea 3 (34) 4 (56) P = 0501 a

96


Fatigue 1 (11) 1 (14) P = 0880 a

Neurotoxic 0 (00) 2 (28) P = 0071 a

Diarrhea 0 (00) 1 (14) P = 0203 a

Esophagitis 2 (22) 5 (69) P = 0144 a



Pneumonia 41 (461) 28 (389) P = 0360 a




Arrhytmia 25 (281) 22 (306) P = 0732 a


Sepsis 8 (90) 6 (83) P = 0883 a


Chylothorax 11 (124) 3 (42) P = 0057 a



Renal failure 2 (22) 4 (56) P = 0276 a

Ileus 6 (67) 2 (28) P = 0237 a


60 (674) 50 (694) P = 0783 a


90-day mortality 6 (67) 7 (97) P=0492 a


97

6







ypN2 1896 (0989-3635)ypN3 3415 (1446-8064)


98




99

6








100



DISCUSSION


101

6



102



103

6


REFERENCES


104



105

6





P value

Male 57 (792) 61 (763) P = 0666 a



0-1 64 (889) 72 (900)2 0 (00) 6 (75)Missing 8 (111) 2 (25)









cT-stage P = 0001 a

T1 2 (28) 1 (13)T2 5 (69) 4 (50)T3 56 (778) 39 (488)T4a 9 (125) 32 (400)missing 0 (00) 4 (50)

cN-stage P = 0000 a

N0 7 (97) 14 (175)N1 30 (417) 61 (763)N2 29 (403) 4 (50)N3 6 (83) 1 (13)




106






107

6




PART II









CHAPTER 7


ABSTRACT





112


INTRODUCTION


113

7






114






115

7





116



RESULTS


117

7











5 3 31



118



119

7



120


Tabl

e 2

Est

imat

ed r

egre

ssio

n co

effici

ents

of t

he p

redi

ctio

n m

odel

s for

pat

holo

gic

com

plet

e re

spon

se w

ithou

t opt

imis

m c

orre

ctio

n

Mod

el 1

Mod

el 2

ndash 3

Mod

el 4

Mod

el 5

Mod

el 6

Varia

ble

Coe

fS

Ep

Coe

fS

EP

Coe

fS

Ep

Coe

fS

Ep

Coe

fS

Ep

Inte

rcep

t-0

88

050

008

-04

20

500

39-0

86

065

019

-11

50

720

11-1

83

091

004

SUV

max

-17

21

450

24

His

tolo

gy

AC

100

100

100

100

SC

C-1

70

061

001

-14

70

650

02-1

89

064

000

-16

30

670

02cT

stag

e

cT1

amp c

T21

001

001

001

00

cT3

amp c

T4a

203

078

001

198

080

001

227

081

000

223

083

001

LRLG

Le-P

ET0

560

550

310

710

590

23

RP-

CT

0

010

010

150

020

010

10

Abb

revi

atio

ns L

RLG

Le-P

ET l

ong

run

low

gra

y le

vel e

mph

asis

mea

sure

d on

PET

and

RP-

CT

run

perc

enta

ge m

easu

red

on C

T

121

7






AUC boot



DISCUSSION


122



123

7





Tixier et al [13]

41 AC SCC


Pre-nCRT

CR vs PR vs NR [32]


Hatt et al [11]

50 AC SCC


Pre-nCRT

CR + PR vs NR [32]


Tan et al [12]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

ΔSUVmax



Zhang et al [15]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]




Pre- and post-nCRT

TRG 1 vs 2-4 [33]


Current study

97 AC SCC


Pre-nCRT

TRG 1 vs 2-5 [19]


124



CONCLUSION



125

7


REFERENCES


126



127

7



128


129

7









CHAPTER 8


ABSTRACT





132


INTRODUCTION


133

8


METHODS





134






x 100baseline IBM

135

8



volumeSUVmax

SUVpeak










136



RESULTS


137

8



Male 40 (800)Female 10 (200)





T2 4 (80)T3 42 (840)T4a 4 (80)

cN-stageN0 7 (140)N1 20 (400)N2 19 (380)N3 4 (80)



Mandard TRG1 7 (140)2 15 (300)3 18 (360)4 9 (180)5 1 (20)


138




139

8




140




0208 065 093 0027 076 096 0065 072 097


0502 058 096 0240 064 078 0117 069 081


0146 067 096 0171 066 095 0074 071 083



141

8



142


DISCUSSION


143

8



144



145

8


REFERENCES


146



147

8



148


149

8










John ThM Plukker


In preparation

CHAPTER 9


ABSTRACT





152


BACKGROUND




153

9





154






155

9


RESULTS



156





157

9







158


Tabl

e 3

Out

com

e of

the Δ

AD

C a

nd Δ

SUV

max

in a

logi

stic

reg

ress

ion

for

pCR

and

res

pons

ePa

thol

ogic

com

plet

e re

spon

seR

espo

nder

sO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

CO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

C

SUV

max

T0

011

72

2e-0

560

71

062

20

471

846

00

222

30

802

047

SUV

max

T1

776

e-05

16e

-12

3742

30

294

071

033

30

016

30

580

056

SUV

max

T2

244

e-06

27e

-14

218

20

166

084

343

10

027

81

058

20

56ΔS

UV

max

dur

ing

170

80

018

89

082

30

380

684

00

310

084

50

61ΔS

UV

max

afte

r2

174

00

181

10

731

042

360

10

120

34

053

40

56A

DC

T0

005

20

08

10

251

073

063

00

017

70

787

054

AD

C T

1

004

56

4e-0

532

30

356

060

177

219

02

1562

376

013

50

80A

DC

T2

001

94

4e-0

58

20

201

076

001

40

00

840

041

083

ΔAD

C d

urin

g0

152

00

572

053

30

5433

173

70

341

5467

80

111

083

ΔAD

C a

fter

019

10

017

79

063

50

60

001

22E

-07

62

012

30

76

Abb

revi

atio

ns p

CR

= p

atho

logi

c co

mpl

ete

resp

onse

OR

= o

dds r

atio

CI =

con

fiden

ce in

terv

al A

UC

= a

rea

unde

r the

cur

ve A

DC

= a

ppar

ent d

iffus

ion

coeffi

cien

t SU

V =

stan

dard

ized

upt

ake

valu

e

159

9


DISCUSSION


160



ACKNOWLEDGEMENTS


161

9


REFERENCES


162



163

9



SUMMARY



Summary

Jan Binne Hulshoff

CHAPTER 10





168





169

10



170


FUTURE PERSPECTIVES




171

10


REFERENCES


172



173

10



174




175

10






176



177

10




178




179

10





180



181

10








182


DANKWOORD








183

10









184


CURRICULUM VITEA



185

10



esophageal cancer

Proefschrift






door


te Smallingerland






7

21

23

37

55

71

89

109

111

131

151

165

167175182183185

TABLE OF CONTENT






Summary





Jan Binne Hulshoff

CHAPTER 1


INTRODUCTION


8




9

1





10





11

1








12



13

1


REFERENCES


14



15

1



16



17

1



18



19

1




PART I









John ThM Plukker


CHAPTER 2


ABSTRACT





24


BACKGROUND


25

2





26






27

2


RESULTS



28








T1 11 (35) 10 (36)T2 41 (129) 39 (140)T3 243 (764) 218 (781)T4a 23 (72) 12 (43)




29

2








30



nCRT (n=194) n ()












Number of EUS1 272 (975) 191 (985) 45 (100) 36 (900)2 7 (25) 3 (15) 0 (00) 4 (100)


31

2


DISCUSSION


32



33

2


REFERENCES


34



35

2










CHAPTER 3


ABSTRACT





38


INTRODUCTION




39

3






40






41

3


RESULTS



nCRT(n = 135) n ()

P value

Male 114 (844) 106 (785) P = 0210a


Histology P = 0737a





cT-stage P = 0527c

T1 3 (22) 1 (07)T2 18 (133) 21 (156)T3ampT4a 114 (844) 113 (837)

cN-stage P = 0984c

N0 36 (267) 36 (267)N1 67 (496) 67 (496)N2 28 (207) 29 (215)N3 4 (30) 3 (22)

42


pT-stage P lt 0001a


pN-stage P lt 0001a

N0 37 (274) 92 (681)N1 41 (304) 29 (215)N2 31 (230) 11 (81)N3 26 (193) 3 (22)




140 (90-188) 150 (120-220) P = 0009b



R1-resection 16 (119) 7 (52) P = 0050a



43

3






P value







3 (150) 3 (150) 4 (200) 10 (500)

cT12 tumors (n=21) 7 (333) 4 (190) 3 (143) 7 (333) P = 0002b




5 (227) 4 (182) 0 (00) 13 (591)

cT12 tumors (n=22) 8 (364) 2 (91) 1 (45) 11 (500) P = 0037b









44




45

3





46












pN1 304 141 ndash 655 0004

pN2 364 165 ndash 804 0001

pN3 454 203 ndash 1019 0000

Lymph-angioinvasion


409 139 ndash 1204

0034


47

3



DISCUSSION


48



49

3



50


REFERENCES


51

3



52



53

3








CHAPTER 4


ABSTRACT





56


BACKGROUND


57

4






58













59

4




60




RESULTS


61

4




Location tumor

Location metastasis





Liver Omental

Yes NoNo

No Yes NoYesYes


No Yes Yes


lymph node

No PET + Biopsy

Yes



62




63

4





64



DISCUSSION


65

4



66


DISCLOSURE


67

4


REFERENCES


68



69

4








John Th M Plukker


CHAPTER 5


ABSTRACT





72


BACKGROUND




73

5









cT-stage 0221T2 16 (154) 9 (86)T3 83 (798) 93 (886)T4a 5 (48) 3 (29)

74



N0 27 (260) 41 (390)N1 50 (481) 44 (419)N2 22 (212) 18 (171)N3 5 (48) 2 (19)


Tx 1 (10)T0 21 (202)T1 22 (212)T2 14 (135) 20 (190)T3 46 (442) 82 (781)T4a 0 (00) 3 (29)

pN-stage lt 0001N0 62 (596) 28 (267)N1 26 (250) 34 (324)N2 11 (106) 25 (238)N3 5 (48) 18 (171)


420)29 (155 ndash 560) 0241




75

5





76



RESULTS




77

5




78





100336736805


0001

003400000000

1004083391154


0007

007601600002


lt 0001351 144 ndash 857 0006




1005131298

124 ndash 2122236 ndash 7145

0013

00240003

NS

pT0pT1pT2pT3

100113183274


0054

083702790025

NS

pN0pN1pN2pN3

100258469687


lt 0001

000700000000

100056537731


0047

059000220086



79

5




80


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p

Surg

ery-

alon

e gr

oup

2-ye

ar D

FS2-

year

LR

FSH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 31

70)

CA

P0

410

21 ndash

08

30

012

CA

P m

odel

(AIC

= 16

82)

CA

P0

270

11 ndash

06

580

004

LN ra

tio2

571

01 ndash

65

10

047

LN ra

tio3

111

20 ndash

80

90

020

Tum

or le

ngth

268

133

ndash 5

43

000

6Tu

mor

leng

th10

99

249

ndash 4

843

000

2A

ngio

inva

sion

190

094

ndash 3

85

007

5N

umbe

r of L

N+

213

092

ndash 4

95

007

8R

CP

mod

el

(AIC

= 32

09)

RC

P0

830

38 ndash

17

80

627

RC

P m

odel

(AIC

= 17

42)

RC

P0

57

022

ndash 1

51

025

8LN

ratio

268

105

ndash 6

81

003

9LN

ratio

313

112

ndash 8

24

002

1A

ngio

inva

sion

195

094

ndash 4

03

007

2Pe

rineu

ral g

row

th1

890

99 ndash

35

90

053

Tum

or le

ngth

252

125

ndash 5

09

001

0Tu

mor

leng

th8

591

99 ndash

37

080

004

Num

ber o

f LN

192

084

ndash 4

39

012

3

81

5


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p -

cont

inue

dnC

RT

grou

p2-

year

DFS

2-ye

ar L

RH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 34

99)

CA

P0

470

18 ndash

12

30

124

CA

P m

odel

(AIC

= 73

5)

CA

P0

060

01 ndash

03

10

001

cT3

200

76 ndash

13

490

114

pN0

100

000

4

Gra

de16

91

212

ndash 1

350

50

008

pN1

270

131

ndash 5

59

000

7pN

2-3

339

143

ndash 8

03

000

5

RC

P m

odel

(AIC

= 35

03)

RC

P0

690

31 ndash

15

20

359

RC

P m

odel

(AIC

= 80

0)

RC

P1

010

08 ndash

13

500

995

cT1

000

275

pN0-

11

000

203

232

051

ndash 1

051

pN1-

28

810

31 ndash

252

23

pN0

100

001

4

Gra

de30

07

279

ndash 3

246

00

005

pN1

251

122

ndash 5

210

014

Num

ber o

f LN

073

006

ndash 8

94

080

4pN

2-3

299

123

ndash 7

32

001

6LN

ratio

260

044

ndash 1

542

029

4pT

0-1

100

035

9

pT2

196

070

ndash 5

49

020

2pT

3-4a

184

071

ndash 4

73

020

9

Abb

revi

atio

ns D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

CI=

con

fiden

ce in

terv

al S

CC

= sq

uam

ous c

ell c

arci

nom

a c

T= c

linic

al

T-st

age

cN

= c

linic

al ly

mph

nod

e st

age

pT=

pat

holo

gica

l T-s

tage

pN

= p

atho

logi

c ly

mph

nod

e st

age

LN

= ly

mph

nod

e C

RM

= ci

rcum

fere

ntia

l mar

gin

R

0= tu

mor

free

rese

ctio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

ove

rall

P-va

lue

of th

e ca

tego

ri-ca

l var

iabl

es

= Si

gnifi

cant

(Plt0

05)

82


DISCUSSION


83

5


Tabl

e 4

Stu

dies

reg

ardi

ng p

rogn

ostic

val

ue o

f the

cir

cum

fere

ntia

l res

ectio

n m

argi

n af

ter

neoa

djuv

ant c

hem

orad

ioth

erap

y

Stud

y (y

ear)

His

tolo

gySt

age

Patie

nts (

n)nC

RT

()

Out

com

eC

RM

defi

nitio

nP-

valu

e Th

omps

on e

t al

(200

8)A

C S

CC

cT1-

424

012

4 (5

2)

5-ye

ar su

rviv

al

RC

PN

SC

hao

et a

l (2

011)

SC

Cyp

T315

115

1 (1

00

)LR

FSR

CP

lt 0

05D

FSR

CP

lt 0

05D

SSR

CP

lt 0

05H

arvi

n et

al

(201

2)A

Cyp

T316

016

0 (1

00

)O

S D

FS L

RFS

C

AP

NS

OS

DFS

LR

FSR

CP

NS

Reid

et a

l (2

012)

AC

SC

CcT

1-4

269

42 (1

6)

DFS

RC

Plt

001

OS

RC

P0

05O

rsquoFar

rell

et a

l (2

013)

AC

SC

C o

ther

s cT

315

782

(52

)O

SR

CP

NS

OS

CA

P0

02Li

u et

al

(201

3)SC

CcT

1-4

9494

(100

)

OS

RC

Plt0

01

Abb

revi

atio

ns S

CC

= sq

uam

ous c

ell c

arci

nom

a A

C=

aden

ocar

cino

ma

nC

RT=

neoa

djuv

ant c

hem

orad

ioth

erap

y c

T= c

linic

al T

-sta

ge y

pT=

path

olog

ic

T-st

age

afte

r nC

RT D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

DSS

= di

seas

e sp

ecifi

c su

rviv

al C

RM

= ci

rcum

fere

ntia

l res

ectio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

on m

ultiv

aria

te a

naly

sis

84



CONCLUSION


DISCLOSURE


85

5


REFERENCES


86



87

5










CHAPTER 6


ABSTRACT





90


BACKGROUND




91

6






92






RESULTS


93

6




Group 1 (n = 89) n ()

Group 2 (n = 72)n ()

P value

Male 71 (798) 57 (792) P = 0924 a



0-1 85 (955) 64 (889)2 0 (00) 0 (00)Missing 4 (45) 8 (111)








cT-stage P = 0000 a

T1 0 (00) 2 (28)T2 25 (281) 5 (69)T3 63 (708) 56 (778)T4a 1 (11) 9 (125)

94


cN-stage P = 0024 a

N0 22 (247) 7 (97)N1 38 (427) 30 (417)N2 27 (303) 29 (403)N3 2 (22) 6 (83)



N0 57 (640) 44 (611)N1 18 (202) 16 (222)

N2 11 (124) 7 (97)N3 3 (34) 5 (69)



LN ratio (gt02 LN+) 12 (135) 11 (153) P=0747 a




95

6




Group 2 (n=72) n ()

P-value









Bleeding 0 (00) 1 (14) P = 0203 a

Nausea 3 (34) 4 (56) P = 0501 a

96


Fatigue 1 (11) 1 (14) P = 0880 a

Neurotoxic 0 (00) 2 (28) P = 0071 a

Diarrhea 0 (00) 1 (14) P = 0203 a

Esophagitis 2 (22) 5 (69) P = 0144 a



Pneumonia 41 (461) 28 (389) P = 0360 a




Arrhytmia 25 (281) 22 (306) P = 0732 a


Sepsis 8 (90) 6 (83) P = 0883 a


Chylothorax 11 (124) 3 (42) P = 0057 a



Renal failure 2 (22) 4 (56) P = 0276 a

Ileus 6 (67) 2 (28) P = 0237 a


60 (674) 50 (694) P = 0783 a


90-day mortality 6 (67) 7 (97) P=0492 a


97

6







ypN2 1896 (0989-3635)ypN3 3415 (1446-8064)


98




99

6








100



DISCUSSION


101

6



102



103

6


REFERENCES


104



105

6





P value

Male 57 (792) 61 (763) P = 0666 a



0-1 64 (889) 72 (900)2 0 (00) 6 (75)Missing 8 (111) 2 (25)









cT-stage P = 0001 a

T1 2 (28) 1 (13)T2 5 (69) 4 (50)T3 56 (778) 39 (488)T4a 9 (125) 32 (400)missing 0 (00) 4 (50)

cN-stage P = 0000 a

N0 7 (97) 14 (175)N1 30 (417) 61 (763)N2 29 (403) 4 (50)N3 6 (83) 1 (13)




106






107

6




PART II









CHAPTER 7


ABSTRACT





112


INTRODUCTION


113

7






114






115

7





116



RESULTS


117

7











5 3 31



118



119

7



120


Tabl

e 2

Est

imat

ed r

egre

ssio

n co

effici

ents

of t

he p

redi

ctio

n m

odel

s for

pat

holo

gic

com

plet

e re

spon

se w

ithou

t opt

imis

m c

orre

ctio

n

Mod

el 1

Mod

el 2

ndash 3

Mod

el 4

Mod

el 5

Mod

el 6

Varia

ble

Coe

fS

Ep

Coe

fS

EP

Coe

fS

Ep

Coe

fS

Ep

Coe

fS

Ep

Inte

rcep

t-0

88

050

008

-04

20

500

39-0

86

065

019

-11

50

720

11-1

83

091

004

SUV

max

-17

21

450

24

His

tolo

gy

AC

100

100

100

100

SC

C-1

70

061

001

-14

70

650

02-1

89

064

000

-16

30

670

02cT

stag

e

cT1

amp c

T21

001

001

001

00

cT3

amp c

T4a

203

078

001

198

080

001

227

081

000

223

083

001

LRLG

Le-P

ET0

560

550

310

710

590

23

RP-

CT

0

010

010

150

020

010

10

Abb

revi

atio

ns L

RLG

Le-P

ET l

ong

run

low

gra

y le

vel e

mph

asis

mea

sure

d on

PET

and

RP-

CT

run

perc

enta

ge m

easu

red

on C

T

121

7






AUC boot



DISCUSSION


122



123

7





Tixier et al [13]

41 AC SCC


Pre-nCRT

CR vs PR vs NR [32]


Hatt et al [11]

50 AC SCC


Pre-nCRT

CR + PR vs NR [32]


Tan et al [12]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

ΔSUVmax



Zhang et al [15]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]




Pre- and post-nCRT

TRG 1 vs 2-4 [33]


Current study

97 AC SCC


Pre-nCRT

TRG 1 vs 2-5 [19]


124



CONCLUSION



125

7


REFERENCES


126



127

7



128


129

7









CHAPTER 8


ABSTRACT





132


INTRODUCTION


133

8


METHODS





134






x 100baseline IBM

135

8



volumeSUVmax

SUVpeak










136



RESULTS


137

8



Male 40 (800)Female 10 (200)





T2 4 (80)T3 42 (840)T4a 4 (80)

cN-stageN0 7 (140)N1 20 (400)N2 19 (380)N3 4 (80)



Mandard TRG1 7 (140)2 15 (300)3 18 (360)4 9 (180)5 1 (20)


138




139

8




140




0208 065 093 0027 076 096 0065 072 097


0502 058 096 0240 064 078 0117 069 081


0146 067 096 0171 066 095 0074 071 083



141

8



142


DISCUSSION


143

8



144



145

8


REFERENCES


146



147

8



148


149

8










John ThM Plukker


In preparation

CHAPTER 9


ABSTRACT





152


BACKGROUND




153

9





154






155

9


RESULTS



156





157

9







158


Tabl

e 3

Out

com

e of

the Δ

AD

C a

nd Δ

SUV

max

in a

logi

stic

reg

ress

ion

for

pCR

and

res

pons

ePa

thol

ogic

com

plet

e re

spon

seR

espo

nder

sO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

CO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

C

SUV

max

T0

011

72

2e-0

560

71

062

20

471

846

00

222

30

802

047

SUV

max

T1

776

e-05

16e

-12

3742

30

294

071

033

30

016

30

580

056

SUV

max

T2

244

e-06

27e

-14

218

20

166

084

343

10

027

81

058

20

56ΔS

UV

max

dur

ing

170

80

018

89

082

30

380

684

00

310

084

50

61ΔS

UV

max

afte

r2

174

00

181

10

731

042

360

10

120

34

053

40

56A

DC

T0

005

20

08

10

251

073

063

00

017

70

787

054

AD

C T

1

004

56

4e-0

532

30

356

060

177

219

02

1562

376

013

50

80A

DC

T2

001

94

4e-0

58

20

201

076

001

40

00

840

041

083

ΔAD

C d

urin

g0

152

00

572

053

30

5433

173

70

341

5467

80

111

083

ΔAD

C a

fter

019

10

017

79

063

50

60

001

22E

-07

62

012

30

76

Abb

revi

atio

ns p

CR

= p

atho

logi

c co

mpl

ete

resp

onse

OR

= o

dds r

atio

CI =

con

fiden

ce in

terv

al A

UC

= a

rea

unde

r the

cur

ve A

DC

= a

ppar

ent d

iffus

ion

coeffi

cien

t SU

V =

stan

dard

ized

upt

ake

valu

e

159

9


DISCUSSION


160



ACKNOWLEDGEMENTS


161

9


REFERENCES


162



163

9



SUMMARY



Summary

Jan Binne Hulshoff

CHAPTER 10





168





169

10



170


FUTURE PERSPECTIVES




171

10


REFERENCES


172



173

10



174




175

10






176



177

10




178




179

10





180



181

10








182


DANKWOORD








183

10









184


CURRICULUM VITEA



185

10






7

21

23

37

55

71

89

109

111

131

151

165

167175182183185

TABLE OF CONTENT






Summary





Jan Binne Hulshoff

CHAPTER 1


INTRODUCTION


8




9

1





10





11

1








12



13

1


REFERENCES


14



15

1



16



17

1



18



19

1




PART I









John ThM Plukker


CHAPTER 2


ABSTRACT





24


BACKGROUND


25

2





26






27

2


RESULTS



28








T1 11 (35) 10 (36)T2 41 (129) 39 (140)T3 243 (764) 218 (781)T4a 23 (72) 12 (43)




29

2








30



nCRT (n=194) n ()












Number of EUS1 272 (975) 191 (985) 45 (100) 36 (900)2 7 (25) 3 (15) 0 (00) 4 (100)


31

2


DISCUSSION


32



33

2


REFERENCES


34



35

2










CHAPTER 3


ABSTRACT





38


INTRODUCTION




39

3






40






41

3


RESULTS



nCRT(n = 135) n ()

P value

Male 114 (844) 106 (785) P = 0210a


Histology P = 0737a





cT-stage P = 0527c

T1 3 (22) 1 (07)T2 18 (133) 21 (156)T3ampT4a 114 (844) 113 (837)

cN-stage P = 0984c

N0 36 (267) 36 (267)N1 67 (496) 67 (496)N2 28 (207) 29 (215)N3 4 (30) 3 (22)

42


pT-stage P lt 0001a


pN-stage P lt 0001a

N0 37 (274) 92 (681)N1 41 (304) 29 (215)N2 31 (230) 11 (81)N3 26 (193) 3 (22)




140 (90-188) 150 (120-220) P = 0009b



R1-resection 16 (119) 7 (52) P = 0050a



43

3






P value







3 (150) 3 (150) 4 (200) 10 (500)

cT12 tumors (n=21) 7 (333) 4 (190) 3 (143) 7 (333) P = 0002b




5 (227) 4 (182) 0 (00) 13 (591)

cT12 tumors (n=22) 8 (364) 2 (91) 1 (45) 11 (500) P = 0037b









44




45

3





46












pN1 304 141 ndash 655 0004

pN2 364 165 ndash 804 0001

pN3 454 203 ndash 1019 0000

Lymph-angioinvasion


409 139 ndash 1204

0034


47

3



DISCUSSION


48



49

3



50


REFERENCES


51

3



52



53

3








CHAPTER 4


ABSTRACT





56


BACKGROUND


57

4






58













59

4




60




RESULTS


61

4




Location tumor

Location metastasis





Liver Omental

Yes NoNo

No Yes NoYesYes


No Yes Yes


lymph node

No PET + Biopsy

Yes



62




63

4





64



DISCUSSION


65

4



66


DISCLOSURE


67

4


REFERENCES


68



69

4








John Th M Plukker


CHAPTER 5


ABSTRACT





72


BACKGROUND




73

5









cT-stage 0221T2 16 (154) 9 (86)T3 83 (798) 93 (886)T4a 5 (48) 3 (29)

74



N0 27 (260) 41 (390)N1 50 (481) 44 (419)N2 22 (212) 18 (171)N3 5 (48) 2 (19)


Tx 1 (10)T0 21 (202)T1 22 (212)T2 14 (135) 20 (190)T3 46 (442) 82 (781)T4a 0 (00) 3 (29)

pN-stage lt 0001N0 62 (596) 28 (267)N1 26 (250) 34 (324)N2 11 (106) 25 (238)N3 5 (48) 18 (171)


420)29 (155 ndash 560) 0241




75

5





76



RESULTS




77

5




78





100336736805


0001

003400000000

1004083391154


0007

007601600002


lt 0001351 144 ndash 857 0006




1005131298

124 ndash 2122236 ndash 7145

0013

00240003

NS

pT0pT1pT2pT3

100113183274


0054

083702790025

NS

pN0pN1pN2pN3

100258469687


lt 0001

000700000000

100056537731


0047

059000220086



79

5




80


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p

Surg

ery-

alon

e gr

oup

2-ye

ar D

FS2-

year

LR

FSH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 31

70)

CA

P0

410

21 ndash

08

30

012

CA

P m

odel

(AIC

= 16

82)

CA

P0

270

11 ndash

06

580

004

LN ra

tio2

571

01 ndash

65

10

047

LN ra

tio3

111

20 ndash

80

90

020

Tum

or le

ngth

268

133

ndash 5

43

000

6Tu

mor

leng

th10

99

249

ndash 4

843

000

2A

ngio

inva

sion

190

094

ndash 3

85

007

5N

umbe

r of L

N+

213

092

ndash 4

95

007

8R

CP

mod

el

(AIC

= 32

09)

RC

P0

830

38 ndash

17

80

627

RC

P m

odel

(AIC

= 17

42)

RC

P0

57

022

ndash 1

51

025

8LN

ratio

268

105

ndash 6

81

003

9LN

ratio

313

112

ndash 8

24

002

1A

ngio

inva

sion

195

094

ndash 4

03

007

2Pe

rineu

ral g

row

th1

890

99 ndash

35

90

053

Tum

or le

ngth

252

125

ndash 5

09

001

0Tu

mor

leng

th8

591

99 ndash

37

080

004

Num

ber o

f LN

192

084

ndash 4

39

012

3

81

5


Tabl

e 3

Mul

tivar

iate

ana

lysi

s of m

odel

s con

tain

ing

the

CR

M d

efini

tion

acco

rdin

g to

the

CA

P (C

RM

0 m

m) o

r th

e R

CP

(CR

M 1

mm

) in

the

surg

ery-

alon

e an

d ne

oadj

uvan

t che

mor

adio

ther

apy

grou

p -

cont

inue

dnC

RT

grou

p2-

year

DFS

2-ye

ar L

RH

R95

C

IP-

valu

eH

R95

C

IP-

valu

eC

AP

mod

el

(AIC

= 34

99)

CA

P0

470

18 ndash

12

30

124

CA

P m

odel

(AIC

= 73

5)

CA

P0

060

01 ndash

03

10

001

cT3

200

76 ndash

13

490

114

pN0

100

000

4

Gra

de16

91

212

ndash 1

350

50

008

pN1

270

131

ndash 5

59

000

7pN

2-3

339

143

ndash 8

03

000

5

RC

P m

odel

(AIC

= 35

03)

RC

P0

690

31 ndash

15

20

359

RC

P m

odel

(AIC

= 80

0)

RC

P1

010

08 ndash

13

500

995

cT1

000

275

pN0-

11

000

203

232

051

ndash 1

051

pN1-

28

810

31 ndash

252

23

pN0

100

001

4

Gra

de30

07

279

ndash 3

246

00

005

pN1

251

122

ndash 5

210

014

Num

ber o

f LN

073

006

ndash 8

94

080

4pN

2-3

299

123

ndash 7

32

001

6LN

ratio

260

044

ndash 1

542

029

4pT

0-1

100

035

9

pT2

196

070

ndash 5

49

020

2pT

3-4a

184

071

ndash 4

73

020

9

Abb

revi

atio

ns D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

CI=

con

fiden

ce in

terv

al S

CC

= sq

uam

ous c

ell c

arci

nom

a c

T= c

linic

al

T-st

age

cN

= c

linic

al ly

mph

nod

e st

age

pT=

pat

holo

gica

l T-s

tage

pN

= p

atho

logi

c ly

mph

nod

e st

age

LN

= ly

mph

nod

e C

RM

= ci

rcum

fere

ntia

l mar

gin

R

0= tu

mor

free

rese

ctio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

ove

rall

P-va

lue

of th

e ca

tego

ri-ca

l var

iabl

es

= Si

gnifi

cant

(Plt0

05)

82


DISCUSSION


83

5


Tabl

e 4

Stu

dies

reg

ardi

ng p

rogn

ostic

val

ue o

f the

cir

cum

fere

ntia

l res

ectio

n m

argi

n af

ter

neoa

djuv

ant c

hem

orad

ioth

erap

y

Stud

y (y

ear)

His

tolo

gySt

age

Patie

nts (

n)nC

RT

()

Out

com

eC

RM

defi

nitio

nP-

valu

e Th

omps

on e

t al

(200

8)A

C S

CC

cT1-

424

012

4 (5

2)

5-ye

ar su

rviv

al

RC

PN

SC

hao

et a

l (2

011)

SC

Cyp

T315

115

1 (1

00

)LR

FSR

CP

lt 0

05D

FSR

CP

lt 0

05D

SSR

CP

lt 0

05H

arvi

n et

al

(201

2)A

Cyp

T316

016

0 (1

00

)O

S D

FS L

RFS

C

AP

NS

OS

DFS

LR

FSR

CP

NS

Reid

et a

l (2

012)

AC

SC

CcT

1-4

269

42 (1

6)

DFS

RC

Plt

001

OS

RC

P0

05O

rsquoFar

rell

et a

l (2

013)

AC

SC

C o

ther

s cT

315

782

(52

)O

SR

CP

NS

OS

CA

P0

02Li

u et

al

(201

3)SC

CcT

1-4

9494

(100

)

OS

RC

Plt0

01

Abb

revi

atio

ns S

CC

= sq

uam

ous c

ell c

arci

nom

a A

C=

aden

ocar

cino

ma

nC

RT=

neoa

djuv

ant c

hem

orad

ioth

erap

y c

T= c

linic

al T

-sta

ge y

pT=

path

olog

ic

T-st

age

afte

r nC

RT D

FS=

dise

ase-

free

surv

ival

LR

FS=

loca

l rec

urre

nce

free

surv

ival

DSS

= di

seas

e sp

ecifi

c su

rviv

al C

RM

= ci

rcum

fere

ntia

l res

ectio

n m

argi

n C

AP=

Col

lege

of A

mer

ican

Pat

holo

gist

s an

d R

CP=

Roy

al C

olle

ge o

f Pat

holo

gist

s =

on m

ultiv

aria

te a

naly

sis

84



CONCLUSION


DISCLOSURE


85

5


REFERENCES


86



87

5










CHAPTER 6


ABSTRACT





90


BACKGROUND




91

6






92






RESULTS


93

6




Group 1 (n = 89) n ()

Group 2 (n = 72)n ()

P value

Male 71 (798) 57 (792) P = 0924 a



0-1 85 (955) 64 (889)2 0 (00) 0 (00)Missing 4 (45) 8 (111)








cT-stage P = 0000 a

T1 0 (00) 2 (28)T2 25 (281) 5 (69)T3 63 (708) 56 (778)T4a 1 (11) 9 (125)

94


cN-stage P = 0024 a

N0 22 (247) 7 (97)N1 38 (427) 30 (417)N2 27 (303) 29 (403)N3 2 (22) 6 (83)



N0 57 (640) 44 (611)N1 18 (202) 16 (222)

N2 11 (124) 7 (97)N3 3 (34) 5 (69)



LN ratio (gt02 LN+) 12 (135) 11 (153) P=0747 a




95

6




Group 2 (n=72) n ()

P-value









Bleeding 0 (00) 1 (14) P = 0203 a

Nausea 3 (34) 4 (56) P = 0501 a

96


Fatigue 1 (11) 1 (14) P = 0880 a

Neurotoxic 0 (00) 2 (28) P = 0071 a

Diarrhea 0 (00) 1 (14) P = 0203 a

Esophagitis 2 (22) 5 (69) P = 0144 a



Pneumonia 41 (461) 28 (389) P = 0360 a




Arrhytmia 25 (281) 22 (306) P = 0732 a


Sepsis 8 (90) 6 (83) P = 0883 a


Chylothorax 11 (124) 3 (42) P = 0057 a



Renal failure 2 (22) 4 (56) P = 0276 a

Ileus 6 (67) 2 (28) P = 0237 a


60 (674) 50 (694) P = 0783 a


90-day mortality 6 (67) 7 (97) P=0492 a


97

6







ypN2 1896 (0989-3635)ypN3 3415 (1446-8064)


98




99

6








100



DISCUSSION


101

6



102



103

6


REFERENCES


104



105

6





P value

Male 57 (792) 61 (763) P = 0666 a



0-1 64 (889) 72 (900)2 0 (00) 6 (75)Missing 8 (111) 2 (25)









cT-stage P = 0001 a

T1 2 (28) 1 (13)T2 5 (69) 4 (50)T3 56 (778) 39 (488)T4a 9 (125) 32 (400)missing 0 (00) 4 (50)

cN-stage P = 0000 a

N0 7 (97) 14 (175)N1 30 (417) 61 (763)N2 29 (403) 4 (50)N3 6 (83) 1 (13)




106






107

6




PART II









CHAPTER 7


ABSTRACT





112


INTRODUCTION


113

7






114






115

7





116



RESULTS


117

7











5 3 31



118



119

7



120


Tabl

e 2

Est

imat

ed r

egre

ssio

n co

effici

ents

of t

he p

redi

ctio

n m

odel

s for

pat

holo

gic

com

plet

e re

spon

se w

ithou

t opt

imis

m c

orre

ctio

n

Mod

el 1

Mod

el 2

ndash 3

Mod

el 4

Mod

el 5

Mod

el 6

Varia

ble

Coe

fS

Ep

Coe

fS

EP

Coe

fS

Ep

Coe

fS

Ep

Coe

fS

Ep

Inte

rcep

t-0

88

050

008

-04

20

500

39-0

86

065

019

-11

50

720

11-1

83

091

004

SUV

max

-17

21

450

24

His

tolo

gy

AC

100

100

100

100

SC

C-1

70

061

001

-14

70

650

02-1

89

064

000

-16

30

670

02cT

stag

e

cT1

amp c

T21

001

001

001

00

cT3

amp c

T4a

203

078

001

198

080

001

227

081

000

223

083

001

LRLG

Le-P

ET0

560

550

310

710

590

23

RP-

CT

0

010

010

150

020

010

10

Abb

revi

atio

ns L

RLG

Le-P

ET l

ong

run

low

gra

y le

vel e

mph

asis

mea

sure

d on

PET

and

RP-

CT

run

perc

enta

ge m

easu

red

on C

T

121

7






AUC boot



DISCUSSION


122



123

7





Tixier et al [13]

41 AC SCC


Pre-nCRT

CR vs PR vs NR [32]


Hatt et al [11]

50 AC SCC


Pre-nCRT

CR + PR vs NR [32]


Tan et al [12]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]

ΔSUVmax



Zhang et al [15]

20 AC SCC


Pre- and post-nCRT

TRG 1+2 vs 3-5 [19]




Pre- and post-nCRT

TRG 1 vs 2-4 [33]


Current study

97 AC SCC


Pre-nCRT

TRG 1 vs 2-5 [19]


124



CONCLUSION



125

7


REFERENCES


126



127

7



128


129

7









CHAPTER 8


ABSTRACT





132


INTRODUCTION


133

8


METHODS





134






x 100baseline IBM

135

8



volumeSUVmax

SUVpeak










136



RESULTS


137

8



Male 40 (800)Female 10 (200)





T2 4 (80)T3 42 (840)T4a 4 (80)

cN-stageN0 7 (140)N1 20 (400)N2 19 (380)N3 4 (80)



Mandard TRG1 7 (140)2 15 (300)3 18 (360)4 9 (180)5 1 (20)


138




139

8




140




0208 065 093 0027 076 096 0065 072 097


0502 058 096 0240 064 078 0117 069 081


0146 067 096 0171 066 095 0074 071 083



141

8



142


DISCUSSION


143

8



144



145

8


REFERENCES


146



147

8



148


149

8










John ThM Plukker


In preparation

CHAPTER 9


ABSTRACT





152


BACKGROUND




153

9





154






155

9


RESULTS



156





157

9







158


Tabl

e 3

Out

com

e of

the Δ

AD

C a

nd Δ

SUV

max

in a

logi

stic

reg

ress

ion

for

pCR

and

res

pons

ePa

thol

ogic

com

plet

e re

spon

seR

espo

nder

sO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

CO

RL

ower

95

CI

Upp

er

95

CI

P va

lue

AU

C

SUV

max

T0

011

72

2e-0

560

71

062

20

471

846

00

222

30

802

047

SUV

max

T1

776

e-05

16e

-12

3742

30

294

071

033

30

016

30

580

056

SUV

max

T2

244

e-06

27e

-14

218

20

166

084

343

10

027

81

058

20

56ΔS

UV

max

dur

ing

170

80

018

89

082

30

380

684

00

310

084

50

61ΔS

UV

max

afte

r2

174

00

181

10

731

042

360

10

120

34

053

40

56A

DC

T0

005

20

08

10

251

073

063

00

017

70

787

054

AD

C T

1

004

56

4e-0

532

30

356

060

177

219

02

1562

376

013

50

80A

DC

T2

001

94

4e-0

58

20

201

076

001

40

00

840

041

083

ΔAD

C d

urin

g0

152

00

572

053

30

5433

173

70

341

5467

80

111

083

ΔAD

C a

fter

019

10

017

79

063

50

60

001

22E

-07

62

012

30

76

Abb

revi

atio

ns p

CR

= p

atho

logi

c co

mpl

ete

resp

onse

OR

= o

dds r

atio

CI =

con

fiden

ce in

terv

al A

UC

= a

rea

unde

r the

cur

ve A

DC

= a

ppar

ent d

iffus

ion

coeffi

cien

t SU

V =

stan

dard

ized

upt

ake

valu

e

159

9


DISCUSSION


160



ACKNOWLEDGEMENTS


161

9


REFERENCES


162



163

9



SUMMARY



Summary

Jan Binne Hulshoff

CHAPTER 10





168





169

10



170


FUTURE PERSPECTIVES




171

10


REFERENCES


172



173

10



174




175

10






176



177

10




178




179

10





180



181

10








182


DANKWOORD








183

10









184


CURRICULUM VITEA



185

10

University of Groningen Improving patient selection ...

Documents

Transcript of University of Groningen Improving patient selection ...