Neurotoxicity lymphoblastic leukaemia: comparison …comparison oforal andintramuscular methotrexate and two doses of radiation J MChessells, T C S Cox, B Kendall, NP C Cavanagh, L

Archives ofDisease in Childhood 1990; 65: 416-422

Neurotoxicity in lymphoblastic leukaemia:comparison of oral and intramuscular methotrexateand two doses of radiation

J M Chessells, T C S Cox, B Kendall, N P C Cavanagh, L Jannoun, S Richards

AbstractSerial cranial computed tomograms were car-ried out in 136 children with acute lympho-blastic leukaemia who were receiving 24 Gy or18 Gy of cranial irradiation and continuingtreatment with doses of methotrexate givenweekly orally or intramuscularly. The findingswere correlated with treatment variables, thedevelopment of fits, and the inteliigence quo-tient (IQ). Reversible brain shrinkage, attri-buted to treatment with steroids, was foundon 87 of 114 initial scans (76%); 14 showedchanges in white matter during treatment(10%), and calcification was found in 13 eitherduring or after treatment (10%). Eight child-ren (6%) had fits, and in six of the eight therewere changes in white matter or calcificationon the scans. Comparison of the two radio-therapy dosages showed no difference in theincidence of abnormalities seen on computedtomography, fits, or serial IQ measurements,but children receiving intramuscular metho-trexate had a higher incidence of calcificationand a lower mean IQ at one year than thosewho received the drug orally, although thisdifference was not apparent later. Youngerchildren were more likely to develop changeson computed tomograms and fits, and to havelow IQs on completion of treatment, withchanges most apparent in those less than 2years of age. There were highly significantcorrelations between abnormalities on com-puted tomography, fits, and IQ. These find-ings confirm the neurological vulnerability ofyounger children with acute lymphoblasticleukaemia, show an association betweenabnormalities on computed tomography andintellectual deficit, and suggest that metho-trexate is more toxic when given intramuscu-larly than orally. They provide no evidencethat 18 Gy of cranial irradiation is less toxicthan 24 Gy, and indicate the need for alterna-tive treatment regimens.

Hospital forSick Children,Great Ormond Street,London WC1N 3JHJ M ChessellsT C S CoxB KendallN P C CavanaghL JannounClinical Trials ServiceUnit, OxfordS RichardsCorrespondence to:Dr Chessells.Accepted 7 December 1989

Now that large numbers of children with acutelymphoblastic leukaemia are surviving for longperiods, and many are cured, it has becomeapparent that an appreciable minority developlate effects oftreatment.' These include impairedgrowth and endocrine problems,' and a range ofneuropsychological disorders varying fromsubtle learning difficulties2 and abnormalitieson computed tomography3 to overt neurologicaldeterioration, dementia, and convulsions.4Though such complications are, not surpris-ipgly, more common among children who havereceived more than one course of treatment,l

their occurrence has lead to reappraisal ofcertain aspects of treatment, most notably thatdirected to the central nervous system (so called'central nervous system prophylaxis'). The mostcommonly used central nervous system treat-ment has been a course of cranial irradiation in adose of 24 Gy, and intrathecal injections ofmethotrexate. At the same time a desire todecrease the rate of bone marrow relapse haslead to investigation of drug dosage and timing,particularly during continuing (maintenance)treatment. Our concern about the variability ofabsorption of methotrexate lead us to comparethe oral and intramuscular routes of administra-tion during continuing treatment of acutelymphoblastic leukaemia. We report here theresults of investigations in a group of childrenwith acute lymphoblastic leukaemiawho receivedeither 24 Gy or 18 Gy of central nervous systemirradiation, and methotrexate given either orallyor intramuscularly.

Patients and methodsBetween 1979 and 1982, all children with non-Tcell acute lymphoblastic leukaemia referred tothis hospital were treated with a standardinduction protocol comprising daunorubicin,vincristine, prednisolone and crisantaspase (L-asparaginase) and then, after treatment of thecentral nervous system was completed, continu-ing treatment with 6-mercaptopurine daily fortwo weeks in three, prednisolone and vincris-tine every six weeks, and methotrexate in asingle weekly dose of 20 mg/M2. Children wererandomly allocated to receive the methotrexateby mouth or as a single weekly intramuscularinjection. This phase of treatment was contin-ued for a total of 96 weeks. All children receivedtreatment of the central nervous system whichcomprised methotrexate given intrathecally forsix doses, and a course of cranial irradiation.Cranial irradiation was given for the first part ofthe trial in a dose of 24 Gy in 15 fractions overthree weeks and, in January 1981, was reducedto 18 Gy in 10 fractions given over two weeks.Thus the allocation of the dose of radiotherapy,unlike the route of administration of the metho-trexate, was not random.

Children who were under 2 years of age at thetime of diagnosis received regular intrathecalinjections of methotrexate every six weeks untiltheir second birthday, when radiotherapy wasgiven in a dose of 18 Gy in 10 fractions over twoweeks. The details of this protocol have beenpublished,5 and the design is shown in thefigure. It was planned that children on thisprotocol should receive serial computed tomo-

416

on June 18, 2020 by guest. Protected by copyright.

http://adc.bmj.com

/A

rch Dis C

hild: first published as 10.1136/adc.65.4.416 on 1 April 1990. D

ownloaded from

http://adc.bmj.com/

Neurotoxicity in lymphoblastic leukaemia: comparison oforal and intramuscular methotrexate and two doses ofradiation

'PLOD' protocolfor non-T,non-B acute lymphoblasticleukaemia. Numbers refer todoses ofdrug inmg/m2lsurface area exceptfor crisantaspase(L-asparagainase), which isin unitsln Isurface area.

Daunorubicin 45 l 4 4Cranial

radIotherapyIradiotherapy

Vincristine 1 5 Vincristine,

Prednisolone 40mg \Prcednisolone PrdnisoloneCrisantaspase tt lt lit t t

6-mercaptopurine 50-75 75-100 75-100 6-mercaptopurine

Intrathecal methotrexate I 1 I,TT

t tBone marrow Computed tomogram

Induction and central nervous system|prophylaxis1 2 3 4 5 6 7 8 9 10 11 12Time (week)

Methotrexate 4weekly Methotrexate

20 mg Orallylor absorptionintramuscularly

-.*-- Repeat -

Continuing treatment

1 2 3 4 5 6

Table I Numbers of patients and investigations

Age at diagnosis Total Route of methotrexate Cranial irradiation IQ assessment(years) patients'

Oral Intramuscular 24 Gy 18 (Gy Serial Late only

<2 18 (8) 13 (6) 5 (2) Deferred 0 112-5 65 (39) 31 (22) 34 (17) 25 (16) 40 (23) 27 235-10 44 (23) 24 (11) 20 (12) 19 (9) 25 (14) 28 1010+ 9 (4) 3 (1) 6 (3) 4 (1) 5 (3) 5 1

Total 136 71 65 48 70 60 45

*Number of boys in parentheses.

graphy, psychological assessment, and neuro-

logical examination. The numbers of children inwhom each of these investigations was carriedout is shown in table 1.Whenever possible all patients had a com-

puted tomogram within a week of diagnosis andagain at one year, two years, and (whenpossible) between four and five years fromdiagnosis. Routine scans were performed withsedation where necessary using an EMI 1010head scanner or Toshiba-60 without intra-venous contrast. The scans were assessed bytwo observers (TCSC and BK) who had noknowledge of the clinical variables. A gradingsystem was used to determine the degree ofbrain shrinkage (ventricular dilatation or widesuperficial subarachnoid spaces) changes inwhite matter, or calcification; the presence ofany abnormality was noted.The numerical grading system was devised by

comparison with normal control scans. Anattempt was thus made to determine whetherany of the changes noted were progressive.Progression was defined as an increase in thenumerical grading score seen on two or moreconsecutive scans. The serial scans for eachpatient were assessed together.

Patients had psychological assessments beforeradiotherapy and at one and two years (andlater) follow up when possible. Assessmentincluded measures of general intelligence (IQ)using the revised Wechsler intelligence scale forchildren for those over the age of 6 years.6 TheMcCarthy scales of children's abilities,7 andthe British ability scales,8 were used for childrenunder the age of 6 years. The attainments ofchildren of school age were assessed with theBritish ability scales. Assessment of all children

was not possible for logistic reasons, and somechildren, together with those aged less than 2 atdiagnosis, had a single IQ test on completion oftreatment.The patients were seen by a paediatric neuro-

logist within six months of completion oftreatment, or sooner in the event of overtneurological problems. A history was taken ofthe pregnancy and delivery, birth weight,developmental milestones, and of previous ill-nesses. Details of the family history wererecorded. A history was also taken of anysymptoms during the treatment: weakness,unsteadiness, headaches, or visual disturbance.Information about any fits, their nature andfrequency, were also recorded.The child was given a neurological examina-

tion and a record was made of clumsiness(dyspraxia) and joint mobility. Head circum-ference and visual acuity were recorded and anote was made of whether the child was left orright handed.The overall design of the trial, and the

pharmacokinetics of methotrexate, have beenreported in detail elsewhere. At the time of thefirst report there was no significant difference inevent free survival between children receivingmethotrexate orally or intramuscularly. Thedata were analysed again in December 1988(minimum follow up for all patients of six years)and there remains no difference in survivalbetween the children in the two treatmentgroups. Children who relapsed were withdrawnfrom the study, but results of investigationscarried out before relapse are included.The associations between variables (such as

abnormalities on computed tomograms, fits,and treatment regimens) were tested by x2

417


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/

Chessells, Cox, Kendall, Cavanagh, Jannoun, Richards

analysis. Because of the association betweenabnormalities and age, comparisons of treat-ments were adjusted for age by calculating thedifference between observed and expected find-ings and its variance, and summing over the age

group.9 Any differences in mean IQ betweentreatment groups at two years and betweenthose with and without abnormalities on com-

puted tomography were assessed by Student's ttest.A X2 test for trend was used to see if there

were any trends towards more abnormalities atyounger ages, and one way analysis of variancewas used to see if there was any associationbetween IQ two years after diagnosis and age.

The number of patients who had IQ testsboth at diagnosis and later was quite small, so tolook for any effect of treatment on serial IQ, theMann-Whitney test was used to compare theapparent percentage increases in IQ from diag-nosis to one year, and from diagnosis to twoyears, between treatment groups. This allowsfor the non-normal distribution of the percent-age increases. Most analyses were performedusing the statistical software package BMDP.10

Permission for the study and accompanyinginvestigations was given by the joint ethicscommittee of the Hospital for Sick Children andInstitute of Child Health.

ResultsOne hundred and sixty four children were

entered into the study, of whom 158 achievedremission. Three children with Down's syn-drome and one whose mother had Down'ssyndrome were excluded from the study,together with one infant (the only patient in theseries) in whom central nervous system diseasewas present at the time of first lumbar puncture.Seventeen patients were not available for com-puted tomography either because the parentsdid not consent to the investigation or they livedtoo far away, and thus this report concerns theremaining 136 (table 1). The patients aredivided by age: under 2 years, 2 to 5, 5 to 10,and over 10 years. The numbers ofinvestigationsperformed in each group are given.

CHILDREN UNDER 2 YEARSEighteen children were under 2 years old at thetime of diagnosis, five of whom received intra-muscular and 13 oral methotrexate. Fifteen of17 scans carried out on children within the firstthree weeks of presentation showed brainshrinkage, but this was not progressive in any ofthe 14 who had follow up scans. Change in

white matter was found in four of the 14children scanned again at one year or subse-quently, and calcification in three of 14. Changesin white matter detected at two years were notdetected on later follow up in three children, butcalcification was equally prominent in two ofthe three children with previous changes whowere scanned again at five years. The full scaleIQ obtained after completion of treatment in 11

children ranged from 61 to 124 (mean 90), butonly one child in the series had an IQ of morethan 100, and only three had IQs between 90and 100. One child, with a coexistent left sidedarachnoid cyst on computed tomography, hadhad fits before the onset of treatment, and threeothers had fits during or after treatment. Of the11 children who had neurological examinationson completion of treatment, one showed someclumsiness, three had exaggerated knee re-

flexes, and five had hypermobility of theirjoints.

CHILDREN AGED 2 TO 5 (table 2)There were 65 children in this group; 51 hadsome brain shrinkage on at least one scan, 43 on

the first. Two had progressive brain shrinkage.Change in white matter during the first twoyears was detected in seven children, but didnot progress on subsequent scans. At longerfollow up, change in white matter was onlydetectable in one child. Calcification, however,was noted in five children during the first twoyears, and a further four at four to five years'follow up. Full scale IQ gave a mean value of109 at one year, 107-6 at two years, and 118 atfour to five years. Four of the children in thisage group had had one or more fits associatedwith fever before the start of treatment; none ofthese developed fits during or after treatment.Five additional children had fits during or aftertreatment, four of whom were receiving metho-trexate intramuscularly. Neurological examina-tion of 46 children showed that 14 were clumsy,11 had exaggerated knee reflexes, and 26 hadhyperextensible joints.

CHILDREN OVER 5 YEARS OF AGE (table 3)As there were only nine children in the studywho were over 10 years of age at the time ofdiagnosis, the findings of those who were over 5

and those who were over 10 are shown in table3. There were 44 children below 10, and nineover 10. Brain shrinkage was noted in 29 of 43at the beginning of treatment but did notprogress in any of them. Change in white matterwas detected in two during the first two years

Table 2 Results of computed tomography and IQ testing in children aged 2-5 years

Before 1 2 4-5 Abnormal Progressiveradiotherapy year years years at any abnormality

time

Computed tomography:Total No of scans 54 59 55 40No that showed brain shrinkage 43 17 9 1 51 2No that showed changes in white matter 0 5 5 1 8 0No that showed calcification 0 1 5 7 9 0

IQ testag:Total No tested 28 26 48 12Mean score (range) 108 7 (82-150) 109-0 (84-142) 107 6 (75-147) 1188 (96-147)

418


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/


Table 3 Results of computed tomography and IQ testing in children aged over 5 years

Before 1 2 4-5 Abnormnal Progressiveradiotherapy year years years at any abnormality

time

Computed tomography:Total No of scansNo that showed brain shrinkageNo that showed changes in white matterNo that showed calcification

IQ testing:Total No testedMean score (range)

43290

0

471920

411721

2420

1

3621

0

0

0

38 33 41 12113-9 (84-144) 114-8 (81-138) 112-9 (61-142) 115-8 (95-136)

but not later in any patient. Calcification wasfound in one child, but no child had progres-sive calcification or change in white matter.Calcification and change in white matter wereonly seen in children who received methotrexateintramuscularly. The mean IQ at diagnosis was113 9, at two years 112 9, and at long termfollow up 115-8. Thirteen of 51 childrenexamined on completion of treatment wereclumsy, 16 had exaggerated knee reflexes, and10 had hyperextensible joints. Three children inthis age group had had one or more fits beforetreatment in association with fever, but no childdeveloped convulsions during or after treatment,and no other child in this age group subsequentlydeveloped convulsions.

OVERT NEUROLOGICAL ABNORMALITIESEight children had histories of one or more fitsbefore treatment. In seven (all aged under 6 atthe time of diagnosis) the fits seemed to haveoccurred in association with fever. The eighthchild is discussed below. None of these childrenseemed susceptible to fits during treatment.Eight additional children developed fits duringor after treatment of whom two had no abnor-malities on computed tomography. The othersix developed convulsions during or after treat-ment, and the scan showed change in whitematter or calcification. In three of these therewas obvious simultaneous neurological deter-ioration. Further details on these children aregiven in table 4. All were aged under 5 at thetime of diagnosis but only one was under 2; fivewere receiving methotrexate intramuscularly.One child (case 6) developed an acute encepha-lopathy that was presumed to be viral, and atnecropsy measles encephalitis was found. In noother child was a viral cause found for neuro-logical problems.

Two children (cases 4 and 5) developedprogressive convulsions with calcification andchanges in white matter. Both were receivingmethotrexate intramuscularly, and both subse-quently relapsed, one receiving further metho-trexate. At subsequent follow up there wasneurological deterioration and persisting fits.Both were thought clinically to have metho-trexate leucoencephalopathy but unfortunatelypermission for necropsy was not obtained foreither. Three children survived, one (case 2),the only one in the series who received metho-trexate orally, is well with prominent calcificationon scan; his fits have stopped but he has a fullscale IQ of 79 and requires remedial help atschool. Case 1, who was 14 months old at thetime of diagnosis and received methotrexateintramuscularly and long term methotrexateintrathecally, is also well but has an IQ of 61and severe learning problems. Case 2 has hadsevere grand mal convulsions that started aftercompletion of treatment, and calcification wasfirst noted on her scan five years after diagnosis.She has appreciable learning problems, an IQ of79, and became pubertal at the age of 8-2 years.One additional infant (case 7 who was describedpreviously) was found to have an arachnoidcyst; he subsequently relapsed and died. Thetwo year scan of one further child (not men-tioned in the table) is suggestive of an infarct,but the child is well and free of symptoms, hasno learning problems, and an IQ of 124.

ANALYSIS OF IQAt the time of diagnosis the mean IQ of thewhole sample was within the normal highaverage range of intelligence and the distributionof values was normal. Table 5 shows the mean(SD) IQs of the two radiotherapy and metho-trexate groups for those patients who had tests

Table 4 Clinical details of children with fits and abnormal computed tomograms

Case Age at Route by Radiotherapy Findings on ComplicationsNo diagnosis which computed tomography

(years) methotrexategiven

1 1.1 Intramuscular Deferred Atrophy, white matter Fits; IQ 6120 Gy change, calcification

2 2-0 Oral 18 Gy Persistent calcification; Fits; IQ 79change in white matter afterone year of treatment

3 2-0 Intramuscular 24 Gy Late (4 years) calcification Grand mal fits; an episode ofhemiplegia; early puberty; IQ 75

4 4-0 Intramuscular 24 Gy Atrophy, white matter Persistent fits; progressivechange abnormality on computed

tomography; died in relapse5 3-3 Intramuscular 24 Gy Atrophy, white matter Relapse; further methotrexate

change, calcification given; died6 3-0 Intramuscular 24 Gy Calcification, white matter Progressive central nervous system

change deterioration and fits; measles7 10 Oral Deferred Subarachnoid cyst left Fits before treatment; relapse at

temporal region two years

419


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/

Chessells, Cox, Kendall, Cavanagh,Jlannoun, Richards

Table S Average IQ at diagnosis and during follow up according to dose of radiotherapy and route by which methotrexategiven

At diagnosis At I year At 2 years

No Mean (SD) No Mean (SD) No Mean (SD)tested IQ tested IQ tested IQ

Radiotherapy: 24 Gy 21 116-0 (12-7) 21 115-9 (13-6) 21 115-7 (15-4)Radiotherapy: 18 Gy 33 111-5 (16-4) 33 111-2 (15-9) 33 113-6 (15-2)Methotrexate: oral 26 115-2 (15-5) 26 117-2 (14-3) 26 118-5 (16-2)Methotrexate: intramuscular 28 111-4 (14-8) 28 109-1 (14-9) 28 110-6 (13-4)

at the time of diagnosis, and at one and twoyears.Whereas there was an apparent slight increase

in the mean IQs of children receiving metho-trexate orally, those of children receivingmethotrexate intramuscularly showed anapparent decrease. Analysis of variance showedno significant difference between these results.Estimation of the percentage increase in IQfrom presentation to one year using the Mann-Whitney (non-parametric) test, showed a signi-ficant difference between the children receivingmethotrexate orally and intramuscularly, butthis difference was not sustained when thevalues at presentation and at two years werecompared. There were no significant differencesbetween the two radiotherapy groups.

RISK FACTORS FOR DEVELOPMENT OFABNORMALITIESThere was no association between age at thetime of diagnosis and the appearance of brainshrinkage on computed tomography, butyounger children were significantly more likelyto develop changes in white matter (p<0025),calcification, (p<0 025), fits (p<0.01), andhyperextensible joints (p<001) than olderchildren; they were not, however, more likely tobe clumsy. The IQ on completion of treatmentwas significantly lower (p=0 0006) in the under2 year olds, and a regression analysis of IQ andage showed a significant correlation (p=001).

There was no association between any of thefindings on computed tomography or any

neurological abnormality, and the sex of thechild.

Children who received 24 Gy of irradiationwere at no more risk of abnormalities visible onscans than those receiving 18 Gy (table 6). Table6 also shows the influence of route of adminis-tration of the methotrexate. Although morechildren who received methotrexate intramus-cularly developed change in white matter (9/65compared with 5/71), this difference was notsignificant, nor was the difference in the inci-dence of fits, but calcification was found moreoften in those receiving methotrexate intra-muscularly (p=003). This significance dis-appeared when allowance was made for age.

INTER-RELATIONSHIP OF ABNORMALITIES (table 7)There was no relationship between the presenceof brain shrinkage on computed tomographyand the development of fits or the assessment ofIQ. There seemed to be an association betweenclumsiness on neurological follow up examina-tion and the presence of atrophy on the finalfollow up scan. Children with change in whitematter or calcification seen on the scan weresignificantly more likely to have fits, but therewas no association between these findings andany other neurological symptoms. There was ahighly significant correlation between theoccurrence of fits during or after treatment andthe IQ, and between these findings and changeson computed tomography.

PROBLEMS OF GROWTH AND DEVELOPMENTProspective assessment of endocrine function

Table 6 Results according to dose of radiotherapy and route of administration of methotrexate

Dose of Route of administrationradiotherapy* of methotrexate

24 Gy 18 Gy Oral Intramuscular

Total No 48 70 71 65Abnormalities on computed tomography:No that showed brain shrinkage 33 54 53 49No that showed changes in white matter 3 7 5 9No that showed calcification 3 7 3 10

No that developed fits 4 1 2 6IQ at two years:No tested 31 59 52 48Mean IQ 110 109-4 109-1 105-5

*Children aged <2 years were excluded.

Table 7 Inter-relationship of abnormalities

Abnormalities on computed tomography Fits

Brain Changes in Calcificationshrinkage white matter

Fits NS 0-0001 <0 0005 Not applicableLow IQ NS 0-03 0-0001 <0 0005Headache NS NS NS NSClumsiness 0-05 NS NS NSHyperextensibility of joints NS NS NS NSExaggerated reflexes NS NS NS NS

420


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/


was not a part of this study. The youngestchildren have now reached the age of 9-10years, and precocious or early puberty has beennoted in nine girls and premature adrenarche inone boy. There is no association between thedose of cranial irradiation or the route ofadministration of methotrexate, and the onset ofearly puberty. Assessment of growth hormonewas not undertaken, but one additional boy hasdeclined further investigation and has clinicalgrowth hormone deficiency.

DiscussionA wide range of neuropsychological problemsoccur in children who are treated for acutelymphoblastic leukaemia. These include learn-ing difficulties and low or declining IQscores,2 11 usually more pronounced in youngerchildren,'2 and even neurological deterioration,dementia, and necrotising leucoencephalo-pathy.4 Though leucoencephalopathy has mostfrequently been reported in children treated forone or more episodes of leukaemia of the centralnervous system, usually with radiation andprolonged intrathecal chemotherapy,'3 it hasalso been described in children in first remissionwho have received methotrexate intrathecallytogether with cranial irradiation, and subsequentmethotrexate given intravenously.'4 15 Com-puted tomography of the brain in methotrexateradiation encephalopathy has been well des-cribed, and abnormalities include ventricularand subarachnoid dilatation, low attenuation ofwhite matter, and intracerebral calcification.'6Similar changes may be found, however, inchildren without symptoms who are treated foracute lymphoblastic leukaemia, although theincidence of such change varies and may well beassociated with the type of central nervoussystem treatment. It has been reported thatcomputed tomographic changes such as calcifi-cation and cortical atrophy show strong correla-tion with the extent of intellectual impairment. 7Seizures, though a feature of methotrexateencephalopathy, also occur in up to 10% ofchildren with acute lymphoblastic leukaemiaand may be more common in those receiving thedrug intrathecally and intravenously than inthose receiving cranial irradiation and shortterm methotrexate intrathecally."'

Clearly there is no single cause of theseproblems, and it has even been argued thatmany, if not all of the learning problems may beattributed to the effects of leukaemia itself orthose of chronic illness, rather than to treatmentdirected at the central nervous system.'9 Pros-pective psychometric studies are hampered bythe young age of many of the patients and thehigh attrition rate of the disease. A recent meta-analysis ofmost papers published on the subject,however, has concluded that most, if not all, ofthe problems are the result of the treatmentrather than the disease.20The three main therapeutic components pre-

disposing to central nervous system damage inchildren treated for acute lymphoblasticleukaemia have been identified as cranial irradi-ation, methotrexate given intravenously, andmethotrexate given intrathecally.4 Analysis of

their individual contribution to damage is diffi-cult, because all three treatments are frequentlygiven in conjunction.These concerns have lead to reappraisal and

modification of treatment directed at the centralnervous system but such changes must betempered by the realisation that all childrenwith acute lymphoblastic leukaemia are at somerisk of central nervous system relapse, and thatthe outcome after overt central nervous systemleukaemia is poor."

Regular intrathecal chemotherapy withoutcranial irradiation may be adequate for somepatients,22 and there is some retrospectiveevidence that it is less damaging,23 24 butpatients receiving regular methotrexate intra-thecally and intravenously have been reportedas showing more abnormalities on computedtomography and similar IQ scores when com-pared with those receiving short term treatmentintrathecally together with cranial irradia-

2526tion.Our decision to reduce the dose of cranial

irradiation during the study was prompted bythese concerns and by the report from theAmerican Children's Cancer Study Group that adose of 18 Gy was as effective as 24 Gy intreatment of the central nervous system,27 andwe hoped that it would prove less toxic.Similarly, by deferring irradiation in childrenunder 2 years old, we hoped to reduce theproblems in young children. The randomisedcomparison of methotrexate given orally andintramuscularly was undertaken with knowledgeof the variable concentrations obtained afteroral treatment,28 confirmed by us during thiswork,29 and in the hope that methotrexate givenintramuscularly would decrease the bone marrowand testicular relapse rate. The work reportedhere was planned prospectively to monitor anypotential toxicity. We had calculated that thedose of methotrexate of 20 mg/M2 needed forintramuscular use was much lower than that of40-60 mg that was reported to cause neuro-logical dysfunction. 14 IFOur results, however, show that there was no

difference in the incidence of abnormalities oncomputed tomography, fits, or changes in IQs,in the two groups of children who receivedeither 18 or 24 Gy. Moreover, as we havepreviously reported, the incidence of precociouspuberty, and thus presumably pituitary-hypo-thalamic dysfunction, remains similar.30 Thuswe have no evidence to suggest that the lowerdose of irradiation is less toxic.By contrast, we found some evidence that

methotrexate given intramuscularly was moretoxic than when given orally; more childrenreceiving the drug intramuscularly developedattenuation of white matter and fits during orafter treatment in association with abnormalitieson scans. These results were not significant, butthey are suggestive, particularly in view of theadditional finding of increased calcification inthe group receiving the drug intramuscularly.There was also some evidence that IQ measure-ments declined in children receiving metho-trexate intramuscularly, although this findingwas not sustained on longer follow up. Four ofthe five children who developed repeated fits in

421


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/

422 Chessells, Cox, Kendall, Cavanagh,J_announ, Richards

association with abnormalities on their scanswere receiving methotrexate intramuscularly.Though all three of the five who survived haveserious learning problems, none has the focalneurological abnormality or continued deter-ioration described in gross encephalopathyassociated with higher dose methotrexate givenintravenously.Our analysis of the risk factors for neuro-

psychological damage in this study confirmsonce again the vulnerability of young children,but not that of girls that has been reported byothers.3' In retrospect, our decision to deferirradiation in those under 2 years old possiblycontributed to these results, as these childrenreceived prolonged intrathecal chemotherapy aswell as irradiation. New regimens for this agegroup are clearly indicated.Our results also show that younger children

are also more likely to develop abnormalities oncomputed tomography. It is difficult to com-pare our incidence of abnormalities with thosereported by others because the details of treat-ment vary; for example, one of the highest ratesof abnormalities has been reported in childrenreceiving cranial irradiation and monthly intra-thecal chemotherapy.3 The high incidence ofbrain shrinkage we noted on the initial scansreflected treatment with steroids and was pre-sumably reversible damage. The incidence oflong term atrophy was much lower. We noted,like others,25 that the appearance of attenuationof white matter did not necessarily persist infollow up scans; this abnormality would havebeen more effectively assessed by nuclearmagnetic resonance.32

Calcification, in contrast to change in whitematter, may not develop for some years after thestart of treatment,33 and several of our patientsshowed this late change. There was a highlysignificant correlation between the IQ and thepresence of calcification or attentuation of whitematter on scan, a relationship also noted in agroup of 23 children in a previous study.'7 Wehave thus shown that in a group of childrenreceiving treatment for acute lymphoblasticleukaemia, abnormalities of computed tomo-graphy, fits, and a low IQ all tend to occur inassociation; that younger children are morevulnerable to all of these problems; and thatvulnerability may be exacerbated by the use ofmoderate intramuscular doses of methotrexate.Neuropsychological problems were not ame-liorated by the reduction of cranial irradiationfrom 24 to 18 Gy, or by deferring radiotherapyin the younger children. These results highlightthe need for effective alternative treatment ofthe central nervous system, particularly in theyounger child, but attempts to evaluate suchalternatives should be accompanied by prospec-tive neuropsychological assessment.We thank Miss J Stevens for her dedication and enthusiasm, andthe Medical Research Council and the Leukaemia Research Fundfor financial support.

1 Wheeler K, Leiper AD, Jannoun L, Chessells JM. Medicalcost of curing childhood acute lymphoblastic Ieukaemia. BrMed J 1988;296:162-6.

2 Jannoun L, Chessells JM. Long-term psychological effects ofchildhood Ieukaemia and its treatment. Pediatric Hematologyand Oncology 1987;4:293-308.

3 Peyland Ramu N, Poplack DG, Pizzo PA, Adornato BT, DiChiro G. Abnormal CT scans of the brain in asymptomatic

children with acute lymphocytic leukemia after prophylactictreatment of the central nervous system with radiation andintrathecal chemotherapy. N EnglJ Med 1978;298:815-8.

4 Bleyer WA. Neurologic sequelae and ionizing radiation: anew classification. Cancer Treat Rep 1981;65:89-98.

5 Chessells JM, Leiper AD, Tiedemann K, Hardisty RM,Richards S. Oral methotrexate is as effective as intra-muscular in maintenance therapy of acute lymphoblasticleukaemia. Arch Dis Child 1987;62:172-6.

6 Wechsler D. Wechsler intelligence scale for children. Revisedmanual. New York: Psychological Corporation, 1974.

7 McCarthy D. McCarthy scales of children's abilities manual.New York: Psychological Corporation, 1970.

8 Elliott C, Murray D, Pearson L. Manualfor the British abilityscales. Windsor: NFER Publishing Co, 1978.

9 Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockadeduring and after myocardial infarction: overview of therandomized trials. Prog Cardiovasc Dis 1985;27:335-71.

10 Dixon WJ. BDMP statistical software. Los Angeles: BDMPStatistical Software Inc, 1985.

11 Meadows TA, Massari DJ, Ferguson J, Gordon J, Littman P,Moss K. Declines in IQ scores and cognitive dysfunction inchildren with acute lymphocytic leukaemia treated withcranial irradiation. Lancet 1981i;i:1015-8.

12 Jannoun L. Are cognitive and educational developmentaffected by age at which prophylactic therapy is given inacute lymphoblastic leukaemia? Arch Dis Child 1983;58:953-8.

13 Rubinstein LJ, Herman MM, Long TF, Wilbur JR. Dis-seminated necrotizing leukoencephalopathy: a complicationof treated central nervous system leukemia and lymphoma.Cancer 1975;35:291-305.

14 McIntosh S, Klatskin EH, O'Brien RT, et al. Chronicneurologic disturbance in childhood leukaemia. Cancer1976;37:853-7.

15 Ch'ien LT, Aur RHA, Verzosa MS, et al. Progression ofmethotrexate-induced leukoencephalopathy in childrenwith leukaemia. Med Pediatr Oncol 1981;9:133-41.

16 Peylan-Ramu N, Poplack DG, Blei CL, Herdt JR, VermessM, Di Chiro G. Computer assisted tomography in metho-trexate encephalopathy. J Comput Assist Tomogr 1977;1:216-21.

17 Brouwers P, Riccardi R, Fedio P, Poplack DG. Long-termneuropsychologic sequelae of childhood leukaemia: corre-lation with CT brain scan abnormalities. J Pediatr 1985;106:723-8.

18 Ochs J, Bowman WP, Pui C-H, Abromowitch MA, Mason C,Simone JV. Seizures in childhood lymphoblastic leukaemiapatients. Lancet 1984,i: 1422-4.

19 Williams JM, Davis KS. Central nervous system prophylactictreatment for childhood leukaemia: neuropsyhological out-come studies. Cancer Treat Rev 1986;13:113-27.

20 Cousens P, Waters B, Said J, Stevens M. Cognitive effects ofcranial irradiation in leukaemia: a survey and meta-analysis. Journal of Child Psychiatry 1988;29:839-52.

21 Pinkerton CR, Chessells JM. Failed CNS prophylaxis inchildhood ALL. BrJ Haematol 1984;57:553-61.

22 Littman P, Coccia P, Bleyer WA, et al. Central nervoussystem (CNS) prophylaxis in children with low risk acutelymphoblastic leukaemia (ALL). Int J Radiat Oncol BiolPhys 1987;13:1443-9.

23 Pfefferbaum-Levine B, Copeland DR, Fletcher JM, RiedHL, Jaffe N, McKinnon WR. Neuropsychologic assess-ment of long-term survivors of childhood leukemia. Am JPediatr Hematol Oncol 1984;6:123-7.

24 Rowland JH, Glidewell OJ, Sibley RF, et al. Effects ofdifferent forms of central nervous system prophylaxis onneuropsychologic function in childhood leukemia. J ClinOncol 1984;2:1327-35.

25 Ochs JJ, Parvey LS, Whitaker JN, et al. Serial cranialcomputed-tomography scans in children with leukemiagiven two different forms of central nervous systemtherapy. J Clin Oncol 1983;1:793-8.

26 Mulhern RK, Wasserman AL, Fairclough D, Ochs JJ.Memory function in disease-free survivors of childhoodacute lymphocytic leukaemia given CNS prophylaxis withor without 1800 cGy cranial irradiation. J Clin Oncol1988;6:315-20.

27 Nesbit ME, Robison LL, Littman PS, et al. Presymptomaticcentral nervous system therapy in previously untreatedchildhood acute lymphoblastic leukaemia: comparison of1800 rad and 2400 rad. Lancet 1981;ii:461-6.

28 Balis FM, Savitch JL, Bleyer WA. Pharmacokinetics of oralmethotrexate in children. Cancer Res 1983;43:2342-5.

29 Pearson ADJ, Mills S, Amineddine HA, Long DR, CraftAW, Chessells JM. Pharmacokinetics of oral and intra-muscular methotrexate in children with acute lympho-blastic leukaemia. Cancer Chemother Pharmacol 1987;20:243-7.

30 Leiper AD, Stanhope R, Kitching P, Chessells JM. Pre-cocious and premature puberty associated with the treat-ment of acute lymphoblastic leukaemia. Arch Dis Child1987;62:1 107-12.

31 Robison LL, Nesbit ME, Sather HN, Meadows AT, OrtegaJA, Hammond GD. Factors associated with IQ scores inlongterm survivors of childhood acute lymphoblasticleukaemia. Am3' Pediatr Hematol Oncol 1984;6:115-21.

32 Packer RJ, Zimmerman RA, Bilaniuk LT. Magnetic reson-ance imaging in the evaluation of treatment related centralnervous system damage. Cancer 1986;58:635-40.

33 Riccardi R, Brouwers .P, DiC Chiro G, Poplack DG.Abnormal computed tomography brain scans in childrenwith acute Iymphoblastic leukaemia: serial long-term followup.JZ Clin Oncol 1985;3:12-8.


http://adc.bmj.com

/A

rch Dis C


ownloaded from

http://adc.bmj.com/

Documents

Neurotoxicity lymphoblastic leukaemia: comparison …comparison oforal andintramuscular methotrexate and two doses of radiation J MChessells, T C S Cox, B Kendall, NP C Cavanagh, L