Comparative Study of Epidermal Growth Factor Receptor

Mutation Analysis on Cytology Smears and Surgical

Pathology Specimens From Primary and Metastatic Lung

Carcinomas

Renu Khode, MD, Douglas A. Larsen, MD, Brianne C. Culbreath, BS, Shane Parrish, BS, Kimberly L. Walker,

BS, Lubna Sayage-Rabie, MD, Robert S. Beissner, MD, PhD; and Arundhati Rao, MD, PhD

BACKGROUND: The detection of epidermal growth factor receptor (EGFR) mutations on small biopsy or fine-

needle aspiration samples is required to guide therapy in nonsmall cell lung cancer (NSCLC). In this study, the

authors compared results from EGFR mutation testing on both cytologic smears and surgical specimens and

also compared the performance of platforms using 2 different technologies (pyrosequencing and real-time poly-

merase chain reaction) for both specimen types. METHODS: Specimens from 114 patients were divided into 2 subsets.

The first subset had 60 paired cytology smears and surgical specimens, including 37 paired specimens from the same site

and 23 paired specimens from different sites. The second subset consisted of nonpaired cytology smears and formalin-

fixed, paraffin-embedded (FFPE) tissues (including 8 cell blocks), which were compared on the pyrosequencing and real-

time polymerase chain reaction platforms. Laser-capture microscopy was used to enrich tumor in the FFPE specimens

before DNA extraction. RESULTS: All cytology smears that were used in the study were adequate for analysis on both

platforms. Comparison between smears and concurrent FFPE tissues from the same anatomic site had a concordance

rate of 97%. The concordance rate between the pyrosequencing platform and the real-time polymerase chain reaction

platform was 84% and 85% for FFPE tissues and cytology smears, respectively. CONCLUSIONS: The current results indi-

cated that direct extraction and analysis of EGFR mutations from cytology smears can be performed successfully on both

a pyrosequencing platform and a real-time polymerase chain reaction platform with results comparable to those achieved

in matched surgical specimens. In fine-needle aspiration/endobronchial ultrasound samples with limited tissue, cytology

smears can be important for molecular analysis. Cancer (Cancer Cytopathol) 2013;121:361–9. VC 2013 American Cancer

Society.

KEY WORDS: epidermal growth factor receptor, cytology surgical correlation, PyroMark Q24, Rotor-Gene Q.

INTRODUCTION

Lung cancer is the leading cause of death from cancer worldwide and is responsible for more than 158,000deaths in the United States alone.1 Patients with nonsmall cell lung cancer (NSCLC), including adenocar-cinoma, routinely receive standard chemotherapy, which provides a 1-year survival rate of approximately30%, and undergo surgery, after which, approximately 50% develop disease recurrence and die within 5years.2 This limited response rate has encouraged the introduction of molecular-targeted therapies toimprove the survival rates of these patients.

Epidermal growth factor receptor (EGFR), which is a member of the ErbB (erythroblastic leukemia viral

oncogene homolog) family of transmembrane tyrosine kinase receptor proteins, is activated by ligand binding

Received: August 17, 2012; Revised: November 9, 2012; Accepted: November 29, 2012

Published online January 30, 2013 in Wiley Online Library (wileyonlinelibrary.com)

DOI: 10.1002/cncy.21273, wileyonlinelibrary.com

Corresponding author: Arundhati Rao, MD, PhD, Scott & White Healthcare, 2401 South 31st Street, Temple, TX, 76504; Fax: (254) 724-6329;

arao@sw.org

Department of Pathology, Scott & White Memorial Hospital and Texas A&M University Health Science Center, Temple, Texas

Cancer Cytopathology July 2013 361

Original Article

followed by receptor dimerization and phosphorylation;

and activating mutations can lead to uncontrolled cell

proliferation, tumor invasion, and resistance to chemo-

therapy.3-7 EGFR tyrosine kinase inhibitors (TKIs),

such as gefitinib and erlotinib, occupy and prevent acti-

vation of the tyrosine kinase binding site, leading to in-

hibition of farther downstream effects,8 and are

particularly effective in approximately 10% to 20% of

lung tumors that contain somatic mutations in the

EGFR tyrosine kinase domain.9 The 2 most common

mutations, which account for 90% of all somatic EGFR

mutations, consist of an in-frame deletion of exon 19

and a point mutation in exon 21 (L858R).9,10 Other

mutations with known sensitivity to EGFR TKIs

include the G719 mutations in exon 18 and the L861

mutations in exon 21.9 Recently modified National

Comprehensive Cancer Network (NCCN) guidelines

recommend EGFR mutation testing for some histologic

subtypes of lung cancer, including adenocarcinoma,

large cell carcinoma, and (NSCLC, not otherwise speci-

fied, before instituting targeted EGFR TKI therapy.11 A

provisional clinical opinion generated by the American

Society of Clinical Oncology also recommends that

EGFR mutation testing take place in patients with

NSCLC who mat receive a TKI as first-line therapy.12

We explored the possibility of using stained cytology

smears as a specimen of choice for detecting EGFR

mutations by using paired cytology smears and surgical

biopsies. We chose cytology smears, because cytopatho-

logists quite often have very limited samples, and

adequate tissue is not received to perform cell blocks. In

the current study, we compared DNA yield and results

using 60 cytology smears on single slides with formalin-

fixed, paraffin-embedded (FFPE) surgical specimens,

including biopsies and resections.

To date, EGFR mutation testing has been known

primarily as a laboratory developed assay that works on

platforms using different technologies, including poly-

merase chain reaction (PCR) amplification and sequenc-

ing, amplification-refractory mutations system (ARMS),

peptide nucleic acid-locked PCR, and clamping or enrich-

ing of mutant alleles by restriction endonucleases before

PCR amplification.13-15 Conventional direct Sanger

sequencing requires tumor enrichment at or greater than

25%,14 whereas pyrosequencing is a real-time, quantita-

tive sequencing technology that does not depend on elec-

trophoresis; and studies have demonstrated that it is a

sensitive method for detecting mutations, including inser-

tions, deletions, and alterations in exons 19 and 21.16-19 A

recent article comparing EGFR detection by 3 methodol-

ogies has identified overall sensitivities of 67% for stand-

ard Sanger sequencing and 89% for pyrosequencing

compared with next-generation sequencing. The authors

also identified allele detection sensitivity of 11% for pyro-

sequencing compared with 21% for Sanger sequencing.20

In addition, sequencing technologies can identify novel

mutations not targeted by allele-specific PCR technolo-

gies. The recently US Food and Drug Administration

(FDA)-approved Qiagen Rotor-Gene Q system (Qiagen,

Valencia, Calif) uses a sensitive, real-time PCR based on

scorpion primers coupled with ARMS (the Qiagen EGFR

PCR Kit). It has been demonstrated that ARMS is very

sensitive and can detect as few as 1% of tumor cells in

lung cancer.21 It has also been used in some of clinical tri-

als with TKIs. In the current study, we chose to perform a

comparative study using sensitive sequencing (pyrose-

quencing) and PCR methodologies on a second subset of

available samples. Because the ARMS system has been

used in trials, we chose to compare the ARMS-based Qia-

gen EGFR PCR Kit on the Qiagen Rotor-Gene Q plat-

form with the Qiagen EGFR Pyro Kit on the PyroMark

Q24, in both fine-needle aspiration (FNA)-derived cytol-

ogy smears (including non-FFPE, direct smears) and

FFPE laser-capture microdissected surgical and cell block

specimens.

Unlike other targeted therapies, such as therapy with

v-raf-murine sarcoma viral oncogene homolog B1 (BRAF)

mutation-directed vemurafanib, EGFR mutation results

were not included in the data used for therapy approval,

and no FDA-approved companion diagnostic assay is cur-

rently available. Although NCCN guidelines recommend

surgical biopsy specimens as the specimen of choice, cyto-

logic samples (including FNA, pleural fluid, wash, and

brush samples) also frequently permit the initial, rapid,

and effective diagnosis of cancer. Several recent articles

have demonstrated the adequacy of FNA cytology samples

for EGFR mutation testing1,2,22 but have not compared

them with matched surgical specimens. We evaluated the

adequacy of cytology smears with matched surgical speci-

mens and further examined their concordance on differ-

ent molecular technologies, including pyrosequencing

and real-time PCR platforms.

Original Article

362 Cancer Cytopathology July 2013

MATERIALS AND METHODS

Study Design

Demographic data

The Institutional Review Board of Scott & White Me-

morial Hospital approved this study. All data were obtained

from 114 patients who received treatment within the Scott

& White Healthcare system between February 2008 and

March 2012. These included 58 men and 56 women, and

they ranged in age from 44 to 89 years (mean age, 66 years).

Sixty patients had matched cytology and FFPE samples. Of

these, 37 pairs were true matched samples from the same

site, and 23 pairs were samples from different sites in the

same patients (Figure 1). Informed consent was waived with

Institutional Review Board approval.

In total, 119 samples from 77 patients (54 unique

patients compared with Subset 1) were used to compare

the efficacy of detecting EGFR status on the PyroMark

Q24 and the Rotor-Gene Q. This subset was divided into

2 groups: 1) an FFPE biopsy group, which consisted of

lung resection specimens, small biopsy specimens of lung

and metastatic sites (adrenal gland, lymph node, liver,

kidney), and 8 cell blocks prepared from FNA material

and pleural fluids; and 2) a group of cytology direct

smears with FNA samples, as illustrated in Table 1.

Cytology smear preparation

Cytology smears consisted of both alcohol-fixed and

air-dried specimens.23 Alcohol-fixed specimens were

stained with the standard Papanicolaou method, similar

to that described in the literature.22 Air-dried smears were

allowed to air dry completely; then, Diff-Quik staining

(Stat-Lab Medical Products, Lewisville, Tex) was per-

formed according to the supplier’s instructions. Air-dried

smears were dipped 10 times in each of 3 proprietary solu-

tions, then rinsed with 10 dips in water, and coverslipped.

Diagnoses

Of the 114 patient samples, 81 were diagnosed as

adenocarcinoma, and 33 were diagnosed as NSCLC,

because immunostains were either noncontributory or

were not performed. Immunohistochemistry stains

FIGURE 1. This is a workflow chart of the current study. EGFR indicates epidermal growth factor receptor; Q24, the PyroMark

Q24 platform (Qiagen, Valencia, Calif).

TABLE 1. Specimens Used in the Current Study

FFPE Specimens No. Cytology Smears No.

Total 56 Total 63

Surgical specimens: Lung and lymph

nodes and 3 metastatic sites

48 FNA: Lung and lymph nodes

(9 US FNA, 11 endobronchial US FNA,

and 37 CT-guided FNA)

57

Cell blocks: Lung, lymph node, and pleural fluid 8 Pleural fluids: Bronchial wash and brush material 6

Abbreviations; CT, computed tomography; FFPE, formalin-fixed, paraffin-embedded; FNA, fine-needle aspirate; US, ultrasound.

Comparison of EGFR Test Platforms/Khode et al

Cancer Cytopathology July 2013 363

using a basic panel of thyroid transcription factor-1

(TTF-1), napsin A, p63, and cytokeratin 5/6 (CK5/6)

were performed on the Ventana Benchmark Ultra (Ven-

tana Medical Systems, Oro Valley, Ariz) and/or the

Dako AutostainerPlusLink (Dako North America, Car-

pinteria, Calif) to classify 58 cases. If the quantity of tis-

sue in the block was limited, then immunostains were

not performed, and the tumor was classified as NSCLC.

Cell groups from cytology smears and tumor foci on

FFPE specimens were identified, circled by a team of 5

pathologists (including 3 cytopathologists), and submit-

ted to the molecular laboratory for DNA extraction and

EGFR mutation status testing. In FFPE specimens,

laser-capture microscopy was performed to ensure

enrichment >75%.

DNA Extraction and Molecular Testing

Pinpoint extraction

Tumor-containing slides were selected, and the tu-

mor cells were extracted from a single slide per case using

the Zymo Research Pinpoint Slide DNA Isolation System

(Zymo Research Corporation, Irvine, Calif) according to

manufacturer’s instructions, which entailed placing the

pinpoint solution over the area of interest and allowing it

to completely dry at room temperature.24 The cells were

then scraped off the slide using a scalpel (Fig. 2). In most

cases, only isolated tumor cells (approximately 50 per

field) were isolated; and, whenever possible, at least 300

cells were used. In rare cases, intermingled lymphocytes

were present, and enrichment beyond 20% of tumor cells

was not possible. The tubes were centrifuged briefly; then,

50 lL extraction buffer and 5 lL proteinase K were added

to the tube, and the tube was incubated at 55�C for 4

hours followed by heating at 95�C to 98�C for 10

minutes. Extraction was completed by adding 100 lL of

pinpoint binding buffer and column purification.

Laser-capture microscopy

FFPE tissue sections were selected by the patholo-

gists, and the tumor cells were isolated using the Arcturus

Pixcell II laser-capture microscope (Arcturus Bioscience,

Inc., Mountain View, Calif) to ensure maximum tumor

enrichment. The sections were cut and air dried over-

night, then the paraffin was removed from the slides using

xylene. Next, the slides were stained using Mayer hema-

toxylin, and laser-capture microscopy was used to isolate

the cells of interest from a single slide (Fig. 3). After laser

capture, the captured cells were incubated at 65�C for

more than 16 hours in the presence of proteinase K.

When the incubation was complete, the samples were

ready for PCR analysis.23

EGFR analysis by pyrosequencing: Qiagen

Q24 Pyrosequencer

Samples were tested on the Q24 Pyrosequencer

using the Qiagen EGFR Pyro Kit.25 The package insert

FIGURE 2. Cytology smears are shown (Left) before and (Right) after pinpoint extraction with the Zymo Research Pinpoint Slide

DNA Isolation System (Zymo Research Corporation, Irvine, Calif).

Original Article

364 Cancer Cytopathology July 2013

was followed for the testing using 4 master mixes to

amplify the DNA targets of interest followed by 5 pyrose-

quencing reactions. Recommended DNA input for the

pyrosequencing reactions was 50 ng of DNA per reaction.

The pyrosequencing kit tests for mutations in the exon 18

codon 719 region, deletions in exon 19, mutations in the

exon 20 codon 768 and codon 790 regions, and muta-

tions in exon 21 codon regions 858 through 861. A 1.2%

DNA flash gel was used to verify the presence and quality

of PCR products before pyrosequencing for each of the

exons. Testing on the Q24 Pyrosequencer can be com-

pleted in approximately 4 hours.

EGFR mutation analysis by real-time polymerase

chain reaction: Qiagen Rotor-Gene Q

Samples were tested on the Rotor-Gene Q real-time

instrument using the Qiagen EGFR PCR Kit according

to instructions in the package insert.26 Eight master mixes

were prepared for each specimen and tested on the

Rotor-Gene Q instrument. The real-time kit tests for 19

deletions in exon 19 (it detects the presence of any of 19

deletions but does not distinguish between them),

L858R, L861Q, G719X (it detects the presence of

G719S, G719A, or G719C but does not distinguish

between them), S768I, and 3 insertions in exon 20 (it

detects the presence of any of 3 insertions but does not

distinguish between them). In addition to the mutation

reactions, a control gene reaction is included in the kit to

assess specimen integrity and quantity. Testing using real-

time PCR can be completed in approximately 4 hours.

Both extraction and EGFR analysis methods were

previously validated in the laboratory. Negative and posi-

tive tissue controls were analyzed routinely for all reactions.

The mutation-positive cell lines HCC 2935, NCI-H1975,

H1650, and HCC827 were used as positive controls.

RESULTS

Sample Adequacy

Cytology smears with 1 to 15 cell groups that had approx-

imately 10 to 50 cells per group provided 0.11 to 244.18

ng DNA per microliter. A maximum of 5 lL was used per

amplification reaction. It is noteworthy that a sample with

the extremely low quantity of 0.55 ng per reaction from

cytology smears amplified successfully. All cytology sam-

ples used in the current study were adequate for analysis

on both the Rotor-Gene Q and PyroMark Q24 platforms.

The quantity of DNA extracted from the FFPE samples

was 5.22 to 127 ng, and 6 of 56 FFPE samples produced

invalid results on the Rotor-Gene Q because of an insuffi-

cient quantity of DNA (less than the recommended 50 ng

per reaction).

Comparison of EGFR Results on Cytology

and Surgical Specimens Using Both

Pyrosequencing and Real-Time Polymerase

Chain Reaction Platforms

Overall, 55 of 60 pairs of specimens had identical EGFR

results, for a combined concordance rate of 91% on both

platforms. Only 1 of 37 pairs of specimens from the same

site produced discrepant mutation results between the cy-

tology smear and the biopsy specimen, which were

obtained 1 month apart. In that patient, the cytology smear

FIGURE 3. Formalin-fixed, paraffin-embedded (FFPE) tissue sections are shown before and after laser-capture microscopy

(LCM).

Comparison of EGFR Test Platforms/Khode et al

Cancer Cytopathology July 2013 365

was wild type for EGFR, whereas the resection surgical

specimen had an exon 18 mutation (Table 2, Patient 1).

In total, 4 of 23 pairs of specimens from different

sites were discrepant, for a concordance rate of 82% (Ta-

ble 2, Patients 2-5). Two patients (2 and 5) had cytology

smears only from a primary lung site and had a surgical bi-

opsy specimen from a metastatic site. The other 2 patients

with discrepant results had cytology smears and biopsy

specimens from different metastatic sites and had no pri-

mary tumor available for analysis. In addition, the cytol-

ogy smear from Patient 2 contained very few tumor cells

in a predominantly lymphocytic background.

Comparison of EGFR Results on the PyroMark

Q24 and Rotor-Gene Q Platforms

Formalin-fixed, paraffin-embedded samples

Six cases had invalid results on the Rotor-Gene Q

platform because of a paucity of DNA. No invalid results

were obtained on the PyroMark Q24 platform. Sixteen of

56 cases were positive for EGFR mutations on the Pyro-

Mark Q24 platform, whereas only 9 cases were positive

on the Rotor-Gene Q platform. Thus, the calculated spec-

ificity of the Rotor-Gene Q was 97% compared with the

PyroMark Q24. The concordance rate noted for FFPE

specimens was 84% (Table 3).

Cytology samples

The PyroMark Q24 detected 9 cases with EGFR

mutations, whereas the Rotor-Gene Q detected 4 cases

with EGFR mutations. The concordance rate for cytology

smears was 85%.

The overall sensitivity of the EGFR PCR Kit on the

Rotor-Gene Q was lower than that of the EGFR Pyro Kit

on the PyroMark Q24 for both FFPE specimens and cy-

tology smears, but the specificity was 97% and 92.5% for

FFPE specimens and cytology smears, respectively. Con-

cordance between the 2 platforms was almost identical at

84% and 85%, respectively.

In total, 26 mutations were detected in 26 patients.

Exon 19 deletions were the most common (12 of 26

patients); and exon 21 mutations (9 of 26 patients), exon

20 mutations (4 of 26 patients), and an exon 18 mutation

(1 of 26 patients) were the next most frequent among the

remaining patients. There was 1 patient who had both an

exon 19 deletion and an exon 20 mutation.

DISCUSSION

The discovery of activating EGFR mutations that can be

targeted by TKIs is a major step in effective therapy for

lung cancer. Cytology smears prepared from FNA are

TABLE 3. Sensitivity and Specificity of the Rotor-Gene Q Platform for Formalin-Fixed, Paraffin-Embedded Specimens and Cytology SmearsCompared With the PyroMark Q24 Platform

PyroMark Q24

Rotor-Gene QMutationPresent Wild Type Concordance

No. of FFPE specimens 56 84%

Mutation present 9 1

Wild type 7 33

Invalid Rotor-Gene results 6

No. of cytology smears 63 85%

Mutation present 4 4

Wild type 5 50

Abbreviations: FFPE, formalin-fixed, paraffin-embedded.

TABLE 2. Characteristics of Patients With Discordance Between Epidermal Growth Factor Receptor Resultsin Surgical and Cytologic Specimens

Patient No. Age, y SexSurgicalSpecimen

EGFR Statuson SurgicalSpecimen (%)

CytologySpecimen

EGFR Status onCytologySpecimen (%)

Time IntervalBetween Collectionof Surgical andCytology Specimen

Same site

1 75 Woman Lung resection Exon 18 (5.4) FNA, lung Wild type 1 mo

Different site

2 64 Woman Metastatic site Wild type FNA of primary, lung Exon 19 del (18.9) 31 mo

3 73 Woman Metastatic site Exon 20 (10.9) FNA, different

metastatic site, lymph nodes

Wild type 12 d

4 74 Man Metastatic site Exon 21 (10.7) FNA, different

metastatic site, lymph nodes

Wild type 3 mo

5 68 Man Metastatic site Exon 19 del FNA of primary, lung Wild type 39 mo

Abbreviations, del, deletion; EGFR, epidermal growth factor receptor; FNA, fine-needle aspirate.

Original Article

366 Cancer Cytopathology July 2013

used routinely in the diagnosis of lung carcinoma. How-

ever, molecular testing for EGFR status has been recom-

mended based on findings from biopsy or resection

specimens rather than cytology smears. This has been

highlighted in several studies, including Smouse et al,8

who tested 12 of 239 samples, and Clark,27 who tested 13

of 59 samples that represented cytologic material. A larger

group of 209 cytology cases was analyzed by Billah et al,22

but those authors did not compared their results with con-

current surgical pathology biopsies. More recently,

Chowdhuri et al assessed the feasibility of using laser-cap-

ture microdissection for EGFR testing on 12 cytology

samples28; however, laser-capture microdissection is an

expensive technology that is not available to all

laboratories.

In the current study, we performed a comparison

between cytology smears and concurrent surgical speci-

mens from the same anatomic site in a group of 37

patients and observed a concordance rate of 97%. The

only patient in this group that had a discrepant result was

Patient 1, who had both a cytology smear and a surgical

resection specimen that revealed well differentiated ade-

nocarcinoma. This discrepancy may be attributed to the

sampling of different clones and innate tumor heterogene-

ity, as noted by Sakurada et al.29 The second group of 23

patients had matched pairs of cytology smears and surgical

specimens from different anatomic sites and produced 4

discordant results, for a concordance rate of 82%. Dis-

crepancies in all 4 of those patients (Table 2, Patients 2-5)

may be attributed to differences in metastatic tumor

EGFR status. In addition, we noted that the cytology

smears from Patient 2 had cell groups intimately admixed

with the background lymphoid cells. The pinpoint extrac-

tion from this patient had only 1.57 ng per microliter

isolated DNA, with less than 300 tumor cells. This case

highlights the importance of accurate estimation of the

proportion of inflammatory and stromal cells to tumor

cells, as noted by Ladanyi and Pao.9 The sensitivity of al-

lele detection depends on the methodology used, and

even sensitive pyrosequencing technologies have detection

limits of 10%, which easily may be masked by a dispro-

portionate representation of inflammatory cells. Pinpoint

extraction is a relatively simple macrodissection enrich-

ment method, but clean preparations of tumor cells can

be difficult in FNA cytology smears from lymph nodes.

Kalikaki et al30 demonstrated that EGFR mutation status

differed between primary tumors and corresponding me-

tastases from 7 of 25 patients (28%). Their hypothesis for

this included: 1) the inclusion of new mutations during

the evolution of the metastasis, 2) the administration of

TKIs, and 3) chemotherapy. In summary, the probable

reasons for the discrepancy in 5 of those patients included,

but were not limited to, 1) tumor heterogeneity, 2) differ-

ence in EGFR mutation status of primary and metastatic

sites, 3) the effect of radiation or chemotherapy on the

mutation status, and 4) an error in tumor selection with a

low level of tumor content. In our study, there was a very

high concordance rate (97%) for EGFR testing on speci-

mens obtained from the same site. There was a slightly

lower concordance rate of 82% for specimens obtained

from different sites, similar to a previous report by Schmid

et al (concordance rate, 14% using direct bidirectional

sequencing),31 who established the importance of repeat

testing on new, additional metastatic sites before therapy.

Exon 19 deletions (12 of 26 patients) were the most

frequently noted mutations in our study, similar to

previously reported mutation frequencies by Riely et al32

and Marchetti et al.33

We performed EGFR testing successfully on both

Papanicolaou-stained and Diff-Quik–stained smears

using small groups of cells on a single slide. Because

removing coverslips is time-consuming, it is more conven-

ient for pathologists to anticipate that additional cellular

material will be needed for molecular analysis and to

maintain at least 1 uncoverslipped, stained slide at the

time of the FNA procedure that can be immediately sent

for EGFR testing. A study by Killian et al34 suggested that

a higher quality of DNA was obtained from rapid Roma-

nowsky-stained direct smears, and the obtained DNA was

stable even after 10 years of storage. Thus, the preserva-

tion of stained, uncoverslipped slides can play a key role in

performing therapeutically important mutation analysis.

The clinical trials of EGFR-directed therapy pub-

lished to date have not been conducted with standardized

companion diagnostics, although ARMS, pyrosequenc-

ing, and Sanger sequencing all have been used. In the

IPASS trial (Iressa Pan-Asia Study), testing was performed

using ARMS, fluorescence in situ hybridization, and

immunohistochemistry methods35; in the INTEREST

trial (Iressa Nonsmall Cell Lung Cancer Trial Evaluating

Response and Survival Against Taxotere), direct gene

sequencing, fluorescence in situ hybridization, and

Comparison of EGFR Test Platforms/Khode et al

Cancer Cytopathology July 2013 367

immunohistochemistry were used.36 It is important to

note that no methodology for EGFR mutation detection

currently has received FDA approval, but many molecular

methods have been tested in an attempt to optimize

EGFR testing. In a study performed by Horiike et al,37

the results indicated that the EGFR Scorpions Kit (Qia-

gen) was superior to direct sequencing, especially for

detecting the major deletion mutations in exon 19 and

L858R. The Qiagen Rotor-Gene Q system uses a sensitive

real-time PCR based on scorpion primers coupled with

ARMS. It detects mutations within the EGFR gene but

cannot quantify or distinguish which specific mutation

has occurred.38 Dufort et al compared pyrosequencing

versus conventional BigDye Terminator sequencing (Invi-

trogen, Carlsbad, Calif) in 58 samples and noted that

pyrosequencing was a sensitive method.19 An analysis per-

formed by Ellison et al15 to compare ARMS and DNA

sequencing for various mutation analysis (RAS [rat sar-

coma], BRAF, and EGFR) concluded that ARMS was

more sensitive and robust for detecting somatic mutations

than standard DNA sequencing. We elected to compare

direct pyrosequencing (a sensitive sequencing method)

using the PyroMark Q24 platform versus the Qiagen

EGFR PCR Rotor-Gene Q system (a qualitative, targeted

ARMS, PCR-based test). The concordance between the 2

methods was 85% and 84% for cytology and FFPE,

respectively.

It has been reported that ARMS is affected less by

DNA artifacts from fragmented DNA in FFPE samples

during the macrodissection steps, and ARMS has been

proposed as ideal for samples in which the tumor content

is very low, for example, circulating free tumor DNA in

blood and in cytology samples.15 The DNA quality and

quantity is automatically estimated by the Qiagen Rotor-

Gene Q system, and no failed cytology samples were iden-

tified, indicating that this was not an issue in our study

with pinpoint extraction. However, in this study, the sen-

sitivity of the Qiagen EGFR PCR Kit was lower for all

samples compared with the sensitivity of pyrosequencing,

with 13 of 25 positive results classified as wild type using

the PCR method. A larger prospective clinical trial with

documented response to therapy may be needed to fully

establish the clinical validity of each platform for different

specimen types in therapeutic settings.

In conclusion, direct extraction and analysis of

EGFR mutations from cytology smears (which often are

the source of initial diagnostic tissues) is a convenient and

robust method for samples obtained from FNA and bron-

chial wash/brush samples. An overall 91% concordance

rate was observed between EGFR mutation analysis on cy-

tology and lung surgical specimens. The concordance rate

was 97% when both samples were collected from the same

anatomic site and 82% when they were collected from dif-

ferent anatomic sites. This high concordance rate supports

the use of direct cytology samples for EGFR testing, opti-

mizing diagnosis and rapid mutation testing. Platform

selection may be vital for the accurate detection of EGFR

mutation status. Based on our data, the PyroMark Q24

platform was more sensitive than the Rotor-Gene Q plat-

form, although specificity was high (>95%), and the

overall concordance between platforms was>80%.

FUNDING SOURCES

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

1. Centers for Disease Control and Prevention. Lung Cancer Statistics2008. Atlanta, GA: 2008. Available at: http://www.cdc.gov/cancer/lung/statistics/. Accessed November 5, 2012.

2. Fathi AT, Brahmer JR. Chemotherapy for advanced stage non-small cell lung cancer. Semin Thorac Cardiovasc Surg.2008;20:210-216.

3. D’Angelo SP, Park B, Azzoli CG, et al. Reflex testing of resectedstage I through III lung adenocarcinomas for EGFR and KRASmutation: report on initial experience and clinical utility at a singlecenter. J Thorac Cardiovasc Surg. 2011;141:476-480.

4. Gupta R, Dastane AM, Forozan F, et al. Evaluation of EGFRabnormalities in patients with pulmonary adenocarcinoma: theneed to test neoplasms with more than 1 method. Mod Pathol.2009;22:128-133.

5. Cohen S. Purification of the receptor for epidermal growth factorfrom A-431 cells: its function as a tyrosyl kinase. Methods Enzy-mol. 1983;99:379-387.

6. Jorissen RN, Walker F, Pouliot N, Garrett TP, Ward CW,Burgess AW. Epidermal growth factor receptor: mechanisms ofactivation and signalling. Exp Cell Res. 2003;284:31-53.

7. Wells A. Molecules in focus: EGF receptor. Int J Biochem CellBiol. 1999;31:637-643.

8. Smouse JH, Cibas ES, Janne PA, Joshi VA, Zou KH, Linde-man NI. EGFR mutations are detected comparably in cytologicand surgical pathology specimens of nonsmall cell lung cancer.Cancer (Cancer Cytopathol). 2009;117:67-72.

9. Ladanyi M, Pao W. Lung adenocarcinoma: guiding EGFR-tar-geted therapy and beyond. Mod Pathol. 2008;21:S16-S22.

10. Otani H, Toyooka S, Soh J, et al. Detection of EGFR genemutations using the wash fluid of CT-guided biopsy needle inNSCLC patients. J Thorac Oncol. 2008;3:472-476.

Original Article

368 Cancer Cytopathology July 2013

11. Ettinger DS, Akerley W, Bepler G, et al. Non small cell lungcancer. J Natl Compr Canc Netw. 2010;8:740-801.

12. Keedy VL, Temin S, Somerfield MR, et al. American Society ofClinical Oncology provisional clinical opinion: epidermal growthfactor receptor (EGFR) mutation testing for patients with advancednon-small-cell lung cancer considering first-line EGFR tyrosine ki-nase inhibitor therapy. J Clin Oncol. 2011;29:2121-2127.

13. Eberhard DA, Giaccone G, Johnson BE. Biomarkers of responseto epidermal growth factor receptor inhibitors in Non-Small-CellLung Cancer Working Group: standardization for use in the clini-cal trial setting. J Clin Oncol. 2008;26:983-994.

14. Pao W, Ladanyi M. Epidermal growth factor receptor mutationtesting in lung cancer: searching for the ideal method. Clin CancerRes. 2007;13:4954-4955.

15. Ellison G, Donald E, McWalter G, et al. A comparison ofARMS and DNA sequencing for mutation analysis in clinical bi-opsy samples [serial online]. J Exp Clin Cancer Res. 2010;29:132.

16. Ronaghi M, Uhlen M, Nyen P. A sequencing method based onreal time pyrophosphate. Science. 1998;281:363, 365.

17. Guo DC, Qi Y, He R, Gupta P, Milewicz DM. High through-put detection of small genomic insertions or deletions by pyrose-quencing. Biotechnol Lett. 2003;25:1703-1707.

18. Takano T, Ohe Y, Sakamoto H, et al. Epidermal growth factorreceptor gene mutations and increased copy numbers predict gefiti-nib sensitivity in patients with recurrent non-small-cell lung cancer.J Clin Oncol. 2005;23:6829-6837.

19. Dufort S, Richard MJ, Lantuejoul S, de Fraipont F. Pyrose-quencing, a method approved to detect the 2 major EGFR muta-tions for anti EGFR therapy in NSCLC [serial online]. J Exp ClinCancer Res. 2011;30:57.

20. Querings S, Altmuller J, Ansen S, et al. Benchmarking of muta-tion diagnostics in clinical lung cancer specimens [serial online].PLoS ONE. 2011;6:e19601.

21. Kimura H, Kasahara K, Kawaishi M, et al. Detection ofepidermal growth factor receptor mutations in serum as apredictor of the response to gefitinib in patients with non-small cell lung cancer. Clin Cancer Res. 2006;12:3915-3921.

22. Billah S, Stewart J, Staerkel G, Chen S, Gong Y, Guo M.EGFR and KRAS mutations in lung carcinoma: molecular testingby using cytology specimens. Cancer (Cancer Cytopathol). 2011;119:111-117.

23. Arcturus Bioscience, Inc. Arcturus PicoPure DNA Extraction Kithandbook. Mountain View, CA: Arcturus Bioscience, Inc.; 2012.Available at: http://search.cosmobio.co.jp/cosmo_search_p/search_gate2/docs/MOD_/KIT0103.20090123.pdf. Accessed July 10,2012.

24. Zymo Research Corporation. Pinpoint Slide DNA Isolation SystemInstructions. Catalog no. D3001. Irvine, CA: Zymo Research Cor-poration; 2012. Available at: http://www.zymoresearch.com/media/downloads/16/D3001i.pdf. Accessed July 10, 2012.

25. Qiagen. EGFR Pyro Handbook. Valencia, CA: Qiagen; 2012.Available at: http://www.qiagen.com/selectlocation.aspx?redir-

ect¼%2fliterature%2frender.aspx%3fid%3d200796. Accessed July10, 2012.

26. Qiagen. EGFR Rotor-Gene Q PCR Kit Handbook. Valencia, CA:Qiagen; 2012. Available at: www.qiagen.com/literature/ren-der.aspx?id¼201622. Accessed July 10, 2012.

27. Clark DP. Seize the opportunity: underutilization of fine needleaspiration biopsy to inform targeted cancer therapy decisions. Can-cer (Cancer Cytopathol). 2009;117:289-297.

28. Chowdhuri SR, Xi L, Pham TH, et al. EGFR and KRASmutation analysis in cytologic samples of lung adenocarcinomaenabled by laser capture microdissection. Mod Pathol. 2012;25:548-555.

29. Sakurada A, Lara-Guerra H, Liu N, Shepherd FA, Tsao MS.Tissue heterogeneity of EGFR mutation in lung adenocarcinoma.J Thorac Oncol. 2008;3:527-529.

30. Kalikaki A, Koutsopoulos A, Trypaki M, et al. Comparison ofEGFR and K-RAS gene status between primary tumors and corre-sponding metastases in NSCLC. Br J Cancer. 2008;99:923-929.

31. Schmid K, Oehl N, Wrba F, Pirker R, Pirker C, Filipits M.EGFR/KRAS/BRAF mutations in primary lung adenocarcinomasand corresponding locoregional lymph node metastases. Clin Can-cer Res. 2009;15:4554-4560.

32. Riely GJ, Politi KA, Miller VA, Pao W. Update on epidermalgrowth factor receptor mutations in non-small cell lung cancer.Clin Cancer Res. 2006;12:7232-7241.

33. Marchetti A, Martella C, Felicioni L, et al. EGFR mutations innon-small-cell lung cancer: analysis of a large series of cases and de-velopment of a rapid and sensitive method for diagnostic screeningwith potential implications on pharmacologic treatment. J ClinOncol. 2005;23:857-865.

34. Killian JK, Walker RL, Suuriniemi M, et al. Archival fine-needleaspiration cytopathology (FNAC) samples: untapped resource forclinical molecular profiling. J Mol Diagn. 2010;12:739-745.

35. Fukuoka M, Wu YL, Thongprasert S, et al. Biomarker analysesand final overall survival results from a phase III, randomized,open-label, first-line study of gefitinib versus carboplatin/paclitaxelin clinically selected patients with advanced non-small-cell lungcancer in Asia (IPASS). J Clin Oncol. 2011;29:2866-2874.

36. Douillard JY, Shephard FA, Hirsh V, et al. Molecular predictorsof outcome with gefitinib and docetaxel in previously treated non-small-cell lung cancer: data from the randomized phase III INTER-EST trial. J Clin Oncol. 2010;28:744-752.

37. Horiike A, Kimura H, Nishio K, et al. Detection of epidermalgrowth factor receptor mutation in transbronchial needle aspiratesof non-small cell lung cancer. Chest. 2007;131:1628-1634.

38. Lopez-Rios F, Angulo B, Gomex B, et al. Comparison of molec-ular testing methods for the detection of EGFR mutations in for-malin-fixed paraffin-embedded tissue (FFPET) specimens of non-small cell lung cancer (NSCLC). Abstract 120. Paper presented at:European Society of Medical Oncology Third European LungCancer Conference; April 18-21, 2012; Geneva, Switzerland.2012. Available at: http://www.dianasterapeuticas.com/web/pdf/ESMO_Handout_PR.pdf. Accessed July 16, 2012.

Comparison of EGFR Test Platforms/Khode et al

Cancer Cytopathology July 2013 369