MEDICAL IMAGING—RADIATION ONCOLOGY—ORIGINAL ARTICLE

MRI abnormalities in children following sequentialchemotherapy, hyperfractionated accelerated radiotherapy andhigh-dose thiotepa for high-risk primitive neuroectodermaltumours of the central nervous systemStefanie C Thust,1,2 Esther Blanco,3 Antony J Michalski,3 WK Chong,1 Mark N Gaze,2,3 Kim Phipps3 andKshitij Mankad1

Departments of 1Paediatric Neuroradiology and 3Paediatric Oncology, Great Ormond Street Hospital for Children NHS Foundation Trust and2Department of Oncology, University College London Hospitals NHS Foundation Trust, London, UK

SC Thust MD, FRCR; E Blanco MRCPCh;

AJ Michalski PhD, MB, MRCP, FRCPCH;

WK Chong BMedSci, MBChB, MD, MRCP,

FRCR; MN Gaze MD, FRCR, FRCP, K PhippsRGN, RSCN, BSc Honours Research; K MankadMBBS, MRCP, FRCR.

CorrespondenceDr Kshitij Mankad, Neuroradiology Department,

Great Ormond Street Hospital

for Children NHS Foundation Trust, Great

Ormond Street, London WC1N 3JH, UK.

Email: stefanie.thust@gosh.nhs.uk

Conflicts of interest: In part, this work was

undertaken at UCLH/UCL, who received a

proportion of funding from the Department of

Health’s NIHR Biomedical Research Centres

funding scheme.

Submitted 07 May 2014; accepted 26 July

2014.

doi:10.1111/1754-9485.12232

Abstract

Introduction: Intensive postsurgical therapies have improved survival in chil-dren with primitive neuroectodermal tumour, but there is concern that thecombination of chemotherapy and radiotherapy may result in a compoundinjury to normal brain. The purposes of this analysis were to characterise whattypes of imaging abnormalities occur, identify risk factors and explore howtreatment-related changes may be distinguished from tumour.Method: One hundred fifty-three MRI studies in 14 children treated withsequential chemotherapy, hyperfractionated accelerated radiotherapy andhigh-dose thiotepa were retrospectively analysed at a paediatric nationalreferral centre.Results: We observed 11 episodes of new focal enhancing lesions, 5 of whichwere transient and judged to be treatment related. In addition, 7/14 (50%)of children demonstrated moderate to severe brain volume loss featuring aleukodystrophy pattern.Conclusion: Treatment-related brain MRI abnormalities occurred frequently inthis series with a risk of misdiagnosis as tumour. A proportion of patientssuffer generalised white matter injury, which has not been appreciated as aside effect of this particular therapy.

Key words: cerebral primitive neuroectodermal tumour; chemotherapy;medulloblastoma; paediatrics; radiation oncology; thiotepa.

Introduction

Intensive postsurgical therapies have improved survi-val in children with primitive neuroectodermal tumour(PNET) including medulloblastoma, although the progno-sis remains poor for patients who relapse followingtherapy.1–3 Potential side effects of radiation treatmentsuch as tissue necrosis and white matter injury are welldescribed, although it is possible to reduce the volume ofparenchymal changes with conformal radiation.4 Nowa-days, radiation necrosis is uncommon after initial radio-therapy (RT) with doses not exceeding tissue tolerance.Increasingly, neurotoxic effects of chemotherapy drugsare recognised on brain imaging, for example, in the

form of posterior reversible encephalopathy syndrome.5

A variety of drugs including methotrexate, busulfan andthiotepa have been implicated as potentially neurotoxic,whereby chemotherapy administration within combina-tion schedules including high-dose therapy makes it dif-ficult to ascertain whether a single or multiple agents areresponsible.6–8 There is evidence that the combination ofchemotherapy and RT may result in a compound injuryto normal brain parenchyma.9 Recent reports in childrenundergoing high-dose chemotherapy (HDCT) and RTidentified transient focal enhancing lesions (FELs) onT1-weighted gadolinium-enhanced MRI.8,10 Hyper-fractionated accelerated radiotherapy (HART) is a rela-tively new form of treatment, in which the total radiation

bs_bs_banner

Journal of Medical Imaging and Radiation Oncology •• (2014) ••–••

© 2014 The Royal Australian and New Zealand College of Radiologists 1

dose is administered twice as daily fractions smaller thanstandard fraction size (hyperfractionation) over a shorterthan standard overall treatment time (acceleration). Theprinciple aims of HART are to overcome repopulationof malignant cells and to reduce the late effects oftreatment by using a smaller dose per fraction.Hyperfractionation has recently been associated withfavourable cognitive outcome compared with standardRT.11 The role of HART in the development of brainimaging abnormalities is not fully understood.

The authors’ experience at a paediatric neuro-oncology centre includes serial MRI assessment in chil-dren who have received sequential chemotherapy andHART followed by high-dose thiotepa with stem cellrescue. Some of our patients developed transient FELs,which were initially misinterpreted as tumour, suggest-ing there may be limited awareness of these radiologicalabnormalities. We also encountered generalised brainsignal changes, which have not previously beendescribed in context with this treatment. Here, we reportour formal analysis, which attempts to characterise whattypes of imaging abnormalities occur, identify riskfactors and explore how treatment-related changes maybe distinguished from tumour.

Method

Patients

All children aged 0–18 years with high-risk medullo-blastoma, supratentorial PNET (S-PNET) includingpineoblastoma, who had received treatment at GreatOrmond Street Hospital according to the Milan CancerCenter protocol published by Gandola et al.12 betweenDecember 2009 and April 2013, were included in thepresent study. Institutional approval was obtained for aretrospective review of demographic data, laboratorytest results, and RT and chemotherapy details usingthe patients’ clinical notes and hospital database. Serialimaging and relevant referral information was accessedvia the integrated radiology information system (GECentricity RIS-i version 5.0, GE Healthcare, Barrington,IL, USA).

Chemotherapy

Maximal surgical resection of the tumour was attemptedwherever possible. Postoperatively, children receivedfour cycles of chemotherapy at conventional doses. Thisinduction chemotherapy consisted of intravenous (IV)methotrexate 8 g/m2 plus vincristine 1.5 mg/m2 (week1), etoposide 2.4 g/m2 (week 2), cyclophosphamide4 g/m2 plus vincristine 1.5 mg/m2 (week 5) andcarboplatin 800 mg/m2 plus vincristine 1.5 mg/m2 (week8). Patients were then assessed for disease responsebefore proceeding to HART. The consolidation therapywas based on the pre-RT disease status. If in complete

remission (CR) pre-HART, patients received mainte-nance chemotherapy cycles over 54 weeks (orallomustine 80 mg/m2 as a single dose every 9 weeks fora total of 6 doses, plus vincristine 1.5 mg/m2 as IV bolusevery 3 weeks for a total of 18 doses). Those patientsnot in CR pre-HART, or those with S-PNET, went on toreceive two consecutive cycles of high-dose thiotepa(900 mg/m2/cycle) each with autologous stem cellsrescue. The autologous stem cells were collected duringinduction chemotherapy and a minimum dose of3.0 × 106/kg of CD34 + cells was infused after eachcourse of high-dose thiotepa.

RT

Treatment was delivered in two phases. In both, 6MVphotons from a linear accelerator were used. Two frac-tions were delivered daily with a minimum 6-h inter-fraction interval 5 days each week. In the first phase, thewhole brain and spine were treated with lateral opposedcranial fields with multi-leaf collimator shielding of thenon-target facial structures, and one or two direct pos-terior fields covering the spine down to include the theca.The prescribed dose to the craniospinal axis was 39 Gy in30 fractions of 1.3 Gy, except in children under the ageof 10 years who were judged on MRI following inductionchemotherapy to have achieved a complete or very goodpartial remission. Patients in this group received 31.2 Gyin 24 fractions.

In the second phase, the whole posterior fossa (orthe tumour bed in the case of S-PNET) was treated withintensity-modulated arc therapy to a dose of 21 Gy in14 fractions of 1.5 Gy in those patients who hadreceived 39 Gy to the whole craniospinal axis, or28.5 Gy in 19 fractions of 1.5 Gy in those who hadreceived only 31.2 Gy. The maximum dose to the pos-terior fossa was therefore 60 Gy in 42 fractions overfour and a half weeks for those receiving the highercraniospinal dose, and 59.7 Gy in 43 fractions overfour and a half weeks for those receiving the lowercraniospinal dose. In addition, consideration was givento a boost of 9 Gy in six fractions over 3 days to clearlydefine small volume areas of macroscopic disease in theposterior fossa or spinal cord, which persisted afterinduction chemotherapy. No attempt was made to boostlarge areas of persistent widespread leptomeningealdisease.

Imaging

Serial MRI from presentation until the latest follow-upwas reanalysed by two reviewers in a consensus readingblinded to the original report. Imaging was performedimmediately post-surgery (within 72 h), every 3 monthswithin the first year and every 6 months thereafter if inremission with interval imaging in some cases, if therewas clinical concern between scheduled appointments.

SC Thust et al.

© 2014 The Royal Australian and New Zealand College of Radiologists2

Some of the interval scans were performed at otherinstitutions, but all MRI included T2-weighted, fluidattenuated inversion recovery, diffusion-weightedimaging (DWI)/apparent diffusion coefficient (ADC), andpre- and post-contrast T1-weighted sequences coveringthree planes. All brain abnormalities were documented ina descriptive manner through serial imaging from thepreoperative stage until the latest follow-up or untildeath. Changes were retrospectively categorised astumour, treatment-related focal abnormalities or gener-alised parenchymal changes. FELs were categorised astreatment-related (T-FEL) if these had newly appearedon serial imaging and then fully or partially resolved onfollow-up without interval treatment. Brain volume losswas assessed using a subjective visual grading systemclassing its degree as none, mild, moderate or severe. Toallow for potential differences, volume loss was sepa-rately recorded for the supratentorial and infratentorialcompartment. Postsurgical parenchymal changes wererecorded, but discounted from the analysis.

Results

In the study period, 14 children (aged 3–13 years, mean7.5 years) received serial brain imaging before and aftertreatment, totalling 153 MRI studies. Demographic andclinical characteristics of these patients are listed inTable 1 along with descriptions of the radiological abnor-malities and patient outcomes. Nine of 14 patients dem-onstrated a total of 11 incidences of new focal enhancingwhite matter lesions. Over serial scans, 3/11 FELsproved to represent disease recurrence, 2 at the resec-tion site and 1 distant. Five of 11 FELs were transientT-FELs, which resolved or regressed spontaneouslywithout interval treatment. In Patient 5, new onset FELcould not be characterised with certainty, the diseaseburden being too gross to make a distinction betweenT-FEL and recurrence. Patient 6 was thought to haveT-FEL, but died of disease before resolution could beobserved. Patient 9 was felt to have T-FEL, because thelesions showed similarity to T-FEL observed in otherpatients, but follow-up imaging is pending. One patient(Patient 4) with proven T-FEL was initially misdiagnosedas tumour recurrence. Examples of T-FEL are shown inFigure 1.

T-FEL appeared at a mean of 7.6 months after startingchemotherapy treatment (range 1.2–17.6 months) andshowed regression or resolution at a mean of 5.0 months(range 2.5–8.4 months). In all cases, the T-FELs weresmall (a few millimetres), nodular and avidly enhancing.They occurred in the pons and cerebellum most fre-quently, but also in the cerebral white matter andoccasionally deep grey matter. There were no focal non-enhancing lesions. No cases of treatment-related menin-geal enhancement were observed; on the contrary,leptomeningeal metastases at diagnosis or as part ofrecurrence were present in 6/14 children. Metastases of

the floor of the third ventricle were present in 4/14patients at diagnosis, and no T-FELs were observed inthis location. All primary tumours and recurrences dem-onstrated marked restricted diffusion even in smalllesions, whereas none of the T-FELs were restricted(Fig. 2). Restricted diffusion was defined as increasedsignal on b1000 DWI with corresponding reducedsignal on the ADC map compared with normal-appearingbrain.

In 13 of 14 children, a degree of generalised neuro-parenchymal volume loss was noted with no alternativecause identifiable from the clinical data review.In most patients, the supratentorial and infratentorialcompartments were similarly affected. In 3/13 patientswith brain volume loss, the tumour compartment wasmost affected. Using a subjective visual grading system,total volume loss ranged from none to severe in bothcompartments. A single patient with metastatic pine-oblastoma suffered a severe intercurrent illness follow-ing his first course of high-dose thiotepa, which mayhave contributed to his global brain atrophy. No infec-tive organism or other cause could be identified. Overall,we observed moderate or severe brain volume loss in7/14 children. In all patients, brain volume loss wasassociated with a leukodystrophy pattern of cerebellarand/or cerebral white matter T2 signal hyperintensity.Examples of neuroparenchymal volume loss are shownin Figure 3.

Eleven of 14 patients received thiotepa (n = 8 hadtwo cycles and n = 3 had one cycle). Of these 11 thi-otepa recipients, 7 suffered moderate to severe brainvolume loss. The other four patients suffered mildvolume loss; however, two of these children (Patients 7and 13) died early in the course of their disease and it ispossible they could have subsequently developed furthervolume loss or T-FEL. In the 3/14 children without thi-otepa treatment, brain volume loss was mild or absent.All children with proven or suspected T-FEL had receivedthiotepa.

Discussion

Previous reports

Initial experience using sequential chemotherapy, HARTand thiotepa according to the ‘Milan’ protocol in 33children found no severe neurological complications ofthe treatment, although in 2/33 patients, a self-resolvingnon-specific encephalopathy was reported and a secondmyeloablative course omitted.12 In 2004, Fouladi et al.reported transient enhancing lesions in 22/134 childrenreceiving HDCT, craniospinal radiation (CSI) andtopotecan for medulloblastoma or S-PNET. Theseoccurred at a median of 7.8 months after therapy startand resolved at a median of 6.2 months with localisednecrosis in two patients and localised atrophy in one.10 In2008, Spreafico et al. noted transient enhancing lesions

MRI abnormalities in children

© 2014 The Royal Australian and New Zealand College of Radiologists 3

Tab

le1.

Patie

ntcl

inic

ald

ata

and

MR

Ifind

ings

Patie

nt

ID

Age

(yea

rs)

Tum

our

hist

olog

yM

etas

tase

sat

dia

gnos

is

Tota

l

rese

ctio

n

FEL

typ

eFE

Llo

catio

nO

nset

(mon

ths)

Diff

usio

nR

esol

utio

n

(mon

ths)

Infr

aten

tori

al

volu

me

↓Su

pra

tent

oria

l

volu

me

↓Th

iote

pa

(cou

rses

)

Out

com

e

110

Med

ullo

bla

stom

a–

NR

ecur

renc

eSu

rgic

alm

argi

n12

.9R

estr

icte

dN

oM

oder

ate

Mild

–A

live

with

dis

ease

210

Med

ullo

bla

stom

a3r

dY

––

––

–Se

vere

Mod

erat

e2

Aliv

e,no

evid

ence

ofd

isea

se

35

Med

ullo

bla

stom

a–

YR

ecur

renc

eD

ista

nt10

.6R

estr

icte

dN

oM

ildM

ild–

Die

dof

dis

ease

47

Med

ullo

bla

stom

a3r

dY

1)R

ecur

renc

e

2)T-

FEL

1)M

argi

n

2)C

ereb

ellu

m,

pon

s

1)2.

4

2)3.

8

1)R

estr

icte

d

2)N

ore

stri

ctio

n

1)1.

3

2)8.

0

Mod

erat

eM

oder

ate

2A

live,

noev

iden

ce

ofd

isea

se

53

Med

ullo

bla

stom

aLe

pto

men

inge

alN

Unk

now

nB

rain

stem

10.5

No

rest

rict

ion

No

furt

her

scan

s

Seve

reSe

vere

2D

ied

ofd

isea

se

66

Med

ullo

bla

stom

a3r

dN

Susp

ecte

d

T-FE

L?

Cer

ebel

lum

,

pon

s,in

sula

10.8

No

rest

rict

ion

Die

dN

one

Mild

2D

ied

ofd

isea

se

710

Med

ullo

bla

stom

aLe

pto

men

inge

alN

––

––

–M

oder

ate

Mod

erat

e2

Die

dof

com

plic

atio

ns

89

Med

ullo

bla

stom

aLe

pto

men

inge

alN

1)T-

FEL

2)T-

FEL

+

recu

rren

ce

1)Po

ns

2)C

ereb

ellu

m,

pon

s,m

edul

la

1)5.

2

2)10

.3

1)N

ore

stri

ctio

n

2)N

ore

stri

ctio

n

1)2.

5

2)3.

1

Seve

reSe

vere

1D

ied

ofd

isea

se

913

Med

ullo

bla

stom

a3r

d,

lep

tom

enin

geal

NT-

FEL?

Cer

ebel

lum

,

bra

inst

em,

thal

amus

,

tem

por

al

No

rest

rict

ion

Pend

ing

Mod

erat

eM

oder

ate

1A

live

with

dis

ease

105

S-P

NET

–N

T-FE

LC

ereb

ral,

cere

bel

lum

1.2

No

rest

rict

ion

8.4

Mild

Mild

2A

live,

noev

iden

ce

ofd

isea

se

1111

S-P

NET

–Y

T-FE

LB

asal

gang

lia,

tem

por

allo

be

17.6

No

rest

rict

ion

3.0

Mild

Mild

2A

live,

noev

iden

ce

ofd

isea

se

123

S-P

NET

–Y

––

––

–N

one

Non

e–

Aliv

e,no

evid

ence

ofd

isea

se

138

S-P

NET

Lep

tom

enin

geal

N–

––

––

Non

eM

ild2

Die

dof

com

plic

atio

ns

147

Pin

eob

last

oma

Lep

tom

enin

geal

NT-

FEL

––

–?

Seve

reSe

vere

1A

live,

noev

iden

ce

ofd

isea

se

FEL,

foca

lenh

anci

ngle

sion

;N,n

o;S-

PN

ET,s

upra

tent

oria

lpri

miti

vene

uroe

ctod

erm

altu

mou

r;T-

FEL,

trea

tmen

t-re

late

dfo

cale

nhan

cing

lesi

on;Y

,yes

.

SC Thust et al.

© 2014 The Royal Australian and New Zealand College of Radiologists4

Fig. 1. Axial T2- and post-contrast T1-weighted

images demonstrating new onset transient focal

enhancing lesions in two patients (Patients 10 and

11). Subsequent follow-up imaging demonstrates

complete (Patient 10) and partial (Patient 11) reso-

lution of enhancement with residual T2 signal

abnormality. A persistent postsurgical subdural

collection is also seen in Patient 11.

Fig. 2. Axial T2-, post-contrast T1-, b1000

diffusion-weighted images and ADC maps

showing marked restricted diffusion within a pre-

dominantly solid medulloblastoma (Patient 1),

restricted diffusion in local recurrence of a poorly

enhancing medulloblastoma in a different patient

(Patient 4) and no restricted diffusion within T-FEL

in two patients (Patients 4 and 10). T-FEL,

treatment-related focal enhancing lesion.

MRI abnormalities in children

© 2014 The Royal Australian and New Zealand College of Radiologists 5

in 18/49 children treated with dose-intensive sequentialchemotherapy, HART and thiotepa occurring with amedian of 8 months after RT and beginning to regressafter 13 months.8 Both studies judged the nature ofchanges to be predominantly transient with no reports ofany generalised brain injury. A recent study summaris-ing UK experience on HART without HDCT in 34 patientsreported no imaging abnormalities.13

T-FEL

Morphologically, the T-FELs identified in our patient groupare similar to the enhancing lesions in the aforementionedtwo studies, as is the timeframe of their occurrence andregression. Unlike pseudoprogression, which is defined astreatment-related progression of an existing lesion,14

T-FEL tended to be located remote from the primaryresection, but often in the same tentorial compartment.Knowledge of a typical timeframe for the development ofT-FEL is crucial to avoid misinterpretation as tumour, butunfortunately, the timing alone does not exclude recur-rence. For children with residual disease, at the onset ofthe T-FEL, the pre-existing neoplasia had decreased insize over serial scans, a feature which could possiblyaid interpretation in such patients. On the contrary, inpatients with recurrence, any pre-existing neoplasiawas progressive or static at their onset. A distribution ofT-FEL involving the cerebellum and brainstem appears

common, although lesions also occurred within cerebralwhite matter and deep grey matter. In our patient group,supratentorial T-FELs were more common in the S-PNETgroup, possibly reflecting the high-dose radiation field.One previous study by Negretti et al. evaluated childrenwith various tumours including high-grade gliomas andfound that pseudoprogression type changes afterbusulfan-thiotepa treatment and CSI occurred exclusivelywithin the 50–55 Gy isodoses.7 The current literatureon diffusion appearances of treatment-related brainabnormalities is conflicting: Asao et al. found significantlylower ADC values in adult brain tumours compared withradiation necrosis, suggesting this may help differentiatethe two.15 However, diffusion restriction has beendescribed within a proportion of T-FEL in children follow-ing proton therapy and HDCT, suggesting restrictiondoes not always equal tumour.16 Transient diffusionrestriction in white matter has been recognised in adultsas a feature of methotrexate neurotoxicity.6 In our study,all tumours and even small recurrences/metastases dem-onstrated marked restricted diffusion, whereas none ofthe T-FEL did. This may be a useful distinguishing feature,but any diagnosis based on diffusivity should be madewith caution until more conclusive evidence becomesavailable. To our knowledge, ours is the first report todescribe diffusion appearances of T-FEL in children fol-lowing a regimen of sequential chemotherapy, HART andthiotepa.

Fig. 3. Serial images in three patients demon-

strating parenchymal volume loss and white

matter changes developing over time. Patient 8:

axial T2 and coronal FLAIR showing infratentorial

and supratentorial brain volume loss. Patient 5:

axial T2 showing supratentorial volume loss and

white matter signal abnormality with leukodystro-

phy pattern. Patient 14: axial T2 and coronal FLAIR

showing white matter signal abnormality with leu-

kodystrophy pattern and global brain volume loss.

FLAIR, fluid attenuated inversion recovery.

SC Thust et al.

© 2014 The Royal Australian and New Zealand College of Radiologists6

Brain volume loss

Generalised white matter signal changes resembling aleukodystrophy pattern and a degree of parenchymalvolume loss were present in all but one of our patientsfollowing completion of chemoradiation treatment. Usinga subjective visual grading system, 7/14 patients werejudged to have developed moderate or severe volumeloss. Similar abnormalities have been described inthe late 1990s with use of chemoradiation includingHART,17–19 but not in any of the recent studies, indicatingthis complication may be under-recognised. The findingin our study of frequent global white matter injury,possibly due to compound chemoradiation toxicity, is acause for concern. It has previously been shown that insurvivors of CSI for medulloblastoma, decreased volumeeven in normal-appearing white matter may be associ-ated with debilitating cognitive decline.20 Both Fouladiet al. and Negretti et al. found a statistically significantcorrelation between T-FEL and reduced IQ, supportingthe hypothesis of neurotoxicity, which could also bepostulated as a possible mechanism for the generalisedwhite matter abnormality.7,10 One (Patient 14) of our 14patients suffered a severe illness following treatment,which was characterised by diffuse cerebral, cerebellar,brainstem and deep grey matter signal abnormalitywithout enhancement. Morphologically, this resembledencephalitis and was followed by gross diffuse brainatrophy. It is possible that some or all of these changesoccurred secondary to neurotoxicity, as no infectiousorganism or other alternative aetiology could beidentified.

Patient risk factors

From the study data, it was not possible to conclusivelydetermine risk factors for treatment-related brainimaging abnormalities, the main limitations being thesmall number of patients and administration of combi-nation therapy. However, our results suggest that thi-otepa may play a role in the development of both brainvolume loss and T-FEL, possibly by means of a com-pound injury. The relative lack of imaging abnormalitiesin patients receiving HART without high-dose thiotepain our patient cohort and another recent study supportthis.13 For some patients, T-FEL and volume lossslightly predominated within the tumour compartment,which would have received the greatest radiation dosehighlighting HART as a potential contributing factor.Aside from the patient with pineoblastoma, children inthe medulloblastoma group developed greater brainvolume loss than the PNET group, the cause of whichremains unclear. Further studies to accurately quantifybrain volume loss in relation to neurocognitive functionand to investigate a larger number of patients would beuseful.

Conclusion

Brain MRI abnormalities in children receiving dose-intensive sequential chemotherapy, HART and thiotepaoccurred frequently in this series. The risk of misdiagnosisof T-FEL as tumour is substantial with potentially seriousconsequences for management. T-FELs were small, multi-nodular, occurred at a mean of 7.6 months (range 1.2–17.6 months) following start of treatment and resolved ata mean of 5.0 months (range 2.5–8.4 months) thereafter.They occurred most frequently in the cerebellum andbrainstem, but occasionally in the cerebral hemispheres.T-FEL demonstrated no restricted diffusion, whereas alltumours and recurrences were restricted. All cases ofleptomeningeal enhancement and third ventricularlesions represented metastatic disease, not T-FEL. Aproportion of patients suffered generalised white matterinjury, which has not been appreciated as a side effect ofthis particular therapy and could have implications forlong-term neuropsychological outcome.

References

1. Gajjar A, Chintagumpala M, Ashley D et al. Risk-adapted craniospinal radiotherapy followed byhigh-dose chemotherapy and stem-cell rescue inchildren with newly diagnosed medulloblastoma (StJude Medulloblastoma-96): long-term results from aprospective, multicentre trial. Lancet Oncol 2006;10: 813–20.

2. Massimino M, Gandola L, Spreafico F et al. Nosalvage using high-dose chemotherapy plus/minusreirradiation for relapsing previously irradiatedmedulloblastoma. Int J Radiation Biol Phys 2009;73: 1358–63.

3. Pizer B, Donachie PHJ, Robinson K et al. Treatmentof recurrent central nervous system primitiveneuroectodermal tumours in children andadolescents: results of a Children’s Cancer andLeukaemia study. Eur J Cancer 2011; 47: 1389–97.

4. Steen RG, Koury BSM, Grania CI et al. Effect ofionizing radiation on the human brain: white matterand gray matter T1 in pediatric brain tumor patientstreated with conformal radiation therapy. Int J RadiatOncol Biol Phys 2001; 49: 79–91.

5. Lucchini G, Grioni D, Colombini A et al.Encephalopathy syndrome in children withhemato-oncological disorders is not always posteriorand reversible. Pediatr Blood Cancer 2008; 51:629–33.

6. Fisher MJ, Khademian ZP, Simon EM et al. Diffusion-weighted imaging of early methotrexate neurotoxicityin children. AJNR Am J Neuroradiol 2005; 26: 1686–9.

7. Negretti L, Blanchard P, Couanet D et al.Pseudoprogression after high-dose busulfan-thiotepawith autologous stem cell transplantation andradiation therapy in children with brain tumors:impact on survival. Neuro-Oncol 2012; 14: 1413–21.

MRI abnormalities in children

© 2014 The Royal Australian and New Zealand College of Radiologists 7

8. Spreafico F, Gandola L, Marchianò A et al. Brainmagnetic resonance imaging after high-dosechemotherapy and radiotherapy for childhood braintumors. Int J Radiat Oncol Biol Phys 2008; 70:1011–19.

9. Ruben JD, Dally M, Bailey M et al. Cerebral radiationnecrosis with emphasis on radiation parameters andchemotherapy. Int J Radiat Oncol Biol Phys 2006;65: 499–508.

10. Fouladi M, Chintagumpala M, Laningham FH et al.White matter lesions detected by magnetic resonanceimaging after radiotherapy and high-dosechemotherapy in children with medulloblastoma orprimitive neuroectodermal tumor. J Clin Oncol 2004;15: 4551–60.

11. Kennedy C, Bull K, Chevignard M et al. Quality ofsurvival and growth in children and young adults inthe PNET4 European controlled trial ofhyperfractionated versus conventional radiationtherapy for standard-risk medulloblastoma. Int JRadiat Oncol Biol Phys 2014; 88: 292–300.

12. Gandola L, Massimino M, Cefalo G et al.Hyperfractionated accelerated radiotherapy in theMilan strategy for metastatic medulloblastoma.J Clin Oncol 2009; 27: 566–71.

13. Taylor RE, Howman AJ, Wheatley K et al.Hyperfractionated Accelerated Radiotherapy (HART)with maintenance chemotherapy for metastatic(M1-3) Medulloblastoma – a safety/feasibility study.Radiother Oncol 2014; 111: 41–6.

14. DaCruz LCH Jr, Rodriguez I, Dominguez RC,Gasparetto EL, Sorensen AG. Pseudoprogression and

pseudoresponse. AJNR Am J Neuroradiol 2011; 32:1978–85.

15. Asao C, Korogi Y, Kitajima M et al. Diffusion-weightedimaging of radiation-induced brain injury fordifferentiation from tumor recurrence. AJNR Am JNeuroradiol 2005; 26: 1455–60.

16. Sabin ND, Merchant TE, Harreld JH et al. Imagingchanges in very young children with brain tumorstreated with proton therapy and chemotherapy. AJNRAm J Neuroradiol 2013; 34: 446–50.

17. Hertzberg H, Huk WJ, Ueberall MA et al. CNS lateeffects after ALL therapy in childhood. Part I:neuroradiological findings in long-term survivorsof childhood ALL – an evaluation of the interferencesbetween morphology and neuropsychologicalperformance. The German Late EffectsWorking Group. Med Pediatr Oncol 1997; 28:387–400.

18. Russo C, Fischbein N, Grant E, Prados MD. Lateradiation injury following hyperfractionatedcraniospinal radiotherapy for primitiveneuroectodermal tumor. Int J Radiat Oncol Biol Phys1999; 44: 85–90.

19. Dietrich U, Wanke I, Mueller T et al. Whitematter disease in children treated for malignantbrain tumours. Child Nerv Syst 2001; 17:731–8.

20. Shan ZY, Liu JZ, Glass JO, Gajjar A, Li CS,Reddick WE. Quantitative morphologic evaluationof white matter in survivors of childhoodmedulloblastoma. Magn Reson Imaging 2006; 24:1015–22.

SC Thust et al.

© 2014 The Royal Australian and New Zealand College of Radiologists8