SEARCH

SEARCH BY CITATION

Keywords:

  • childhood lymphoblastic leukaemia;
  • prognosis

Abstract

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

The single most important prognostic determinant in childhood acute lymphoblastic leukaemia (ALL) is effective therapy and changes in therapy may influence the significance of other risk factors. The effect of intensified therapy on the importance of currently recognized phenotypic and genotypic determinants of outcome was assessed in 2090 children enrolled on the Medical Research Council United Kingdom acute lymphoblastic leukaemia XI (MRC UKALL XI) protocol. Treatment allocation was not determined by risk factors. Multivariate analysis confirmed the dominant influence on prognosis of age, sex and presenting white cell count (WCC). After allowing for these features, blast karyotype, d 8 marrow blast percentage and remission status at the end of induction therapy were the only remaining significant predictors of outcome. Organomegaly, haemoglobin concentration, French–American–British type, body mass index, presence of central nervous system disease at diagnosis, immunophenotype and presence of TEL/AML1 fusion gene (examined in a subset of 659 patients) either had no significant effect on outcome or were significant only in univariate analysis. Among karyotype abnormalities with an independent influence on prognosis, high hyperdiploidy (> 50 chromosomes) was shown to be favourable, whereas near haploidy (23–29 chromosomes), presence of the Philadelphia chromosome, t(4;11) and abnormalities affecting the short arm of chromosome 9 [abn (9p)] were adverse risk factors. Early responders to therapy, determined by residual marrow infiltration after 8 d of induction therapy, had a good outcome, while the small proportion of patients who did not achieve a complete remission by the end of induction therapy had a poor outcome. A third block of late intensification was shown to improve event-free survival by 8% at 5 years. The effect of these risk factors was not significantly different between those randomized to the third intensification block and those not randomized to a third block.

Treatment strategies for childhood acute lymphoblastic leukaemia (ALL) are increasingly based on complex risk stratification (Smith et al, 1996). These rely on methods to discriminate between risk groups in order to target high-risk patients for intensified treatment which does carry an increased risk of morbidity and mortality and, possibly, saving those at lower risk of relapse from such sequelae. Traditionally recognized risk groups defined by age, sex and presenting white cell count (WCC) (Chessells et al, 1995a, 1998; Pui et al, 1999) have been shown to contain subgroups of patients with different outcomes that are predicted by blast karyotype (Chessells et al, 1997; Forestier et al, 1997), early response to therapy (Riehm et al, 1987; Gaynon et al, 1997) and molecular genetic abnormalities (Borkhardt et al, 1997; Rubnitz et al, 1997). Characteristics such as the presence of Down's syndrome (Dordelmann et al, 1998), body mass index (BMI), liver and spleen size, presence of anterior mediastinal mass, presenting haemoglobin (Hb) level and platelet count, French–American–British morphological subtype (Lilleyman et al, 1992; Viana et al, 1994; Dordelmann et al, 1998) and immunophenotype (Donadieu et al, 1998; Hann et al, 1998) have been variably reported to confer prognostic significance.

Unlike most other study groups, the UK Medical Research Council (MRC) childhood ALL working party has only recently introduced risk-stratified targeting of intensified therapy. MRC UKALLX showed that all risk group patients benefited from two 5 d courses of intensification given at weeks 5 and 20 of therapy (Chessells et al, 1995b). Its successor trial, MRC UKALL XI, therefore accrued patients of all risk categories to a randomized comparison of even further intensification. This comparison was carried over into the next trial (MRC ALL 97). The overall result was an 8% improvement in event-free survival at 5 years for patients receiving a third course of intensification given at week 35 of therapy [68%, 95% confidence interval (CI) = 64–72% with additional intensification; 60%, 95% CI = 56–64% without additional intensification] (Hann et al, 2000).

Analysis of patients treated on MRC UKALL X revealed age, sex, presenting WCC (Chessells et al, 1998) and blast karyotype (Chessells et al, 1997) to be important prognostic factors. The present analyses of UKALL XI data were performed to assess the predictive value of these and other currently recognized prognostic variables, and to determine whether this was affected by further intensification of therapy. As an individual patient's risk group did not determine therapy, this study is not subject to the confounding effects of variable treatment present in most other recent trials reporting on prognostic factors.

Patients and methods

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Patients UKALL XI accrued 2090 children with ALL from August 1990 to March 1997. All children with ALL, except mature B-cell, aged 1 year (i.e. infants excluded) to 15 years inclusive were eligible for inclusion in the study. The reported analysis contains the follow-up of all patients to October 1999.

Treatment protocols Details of the randomization scheme and treatment schedule are shown in Fig 1 and Table . Until May 1991, when the full results from UKALLX became available (Chessells et al, 1997), all patients were randomized to receive a 4 week induction course, followed by either a 5 d intensification block at week 20 (regimen C) or two intensification blocks at weeks 5 and 20, identical to the schedule associated with the best outcome in UKALL X (regimen D). After May 1991, all patients received regimen D. Until March 1994, patients with central nervous system (CNS) disease received a modified intensification block at week 5 (regimen B); after this date, they all received modified regimen D. Of the 2090 patients entered, 70 were randomized to C and 72 were randomized to D before May 1991. Twelve patients entered before May 1991 and all 1930 patients who were entered after May 1991 were non-randomly assigned D. The other six patients entered all had CNS disease and were non-randomly assigned B as they were entered before March 1994.

image

Figure 1. UKALL XI treatment flow chart.

Download figure to PowerPoint

The 1688 patients accrued after March 1992 were eligible to be randomized to receive or not receive an 8 week third intensification block at week 35. Seven hundred and thirty seven of the 1688 patients were randomized to three intensification blocks (third block arm) and 733 were randomized to only two intensification blocks (non-third block arm). In 218 cases, the decision of whether or not to give the third block was made by the parent and clinician. All patients were scheduled to receive a total of 100 weeks of treatment.

Before March 1994, patients who were not in complete remission at the end of induction were recommended to receive 2 weeks of vincristine and prednisolone if they had less than or equal to 25% blasts, or they were off protocol if they had more than 25% blasts. After this date, patients were recommended to receive early intensification followed by the third block and then the second block, and were to be considered for allogeneic bone marrow transplantation.

The study included additional randomizations. Patients with presenting WCC < 50 × 109/l were eligible to be randomized to receive intrathecal methotrexate (IT-MTX) with (754 patients) or without (759 patients) high-dose intravenous methotrexate (HD-MTX), and those with presenting WCC geqslant R: gt-or-equal, slanted 50 × 109/l were eligible to be randomized to receive IT-MTX plus HD-MTX (158 patients) vs. fractionated (24 Gy in 15 fractions of 1·6 Gy each) cranial radiotherapy (155 patients) as CNS-directed treatment. The 31 children with CNS disease and the 46 infants aged under 2 years with WCC geqslant R: gt-or-equal, slanted 50 × 109/l were recommended to receive IT-MTX and cranial radiotherapy and IT-MTX with HD-MTX respectively. In 187 cases, the decision on which CNS-directed therapy was used was made by the parents and/or clinician. The dose of HD-MTX was 8 g/m2 for children aged 1–4 years and 6 g/m2 in older children. Ten percent of the total dose was infused over the first hour and the remaining 90% over the next 23 h, along with a hydration regimen. Intrathecal methotrexate was given between 24 and 26 h after the start of the HD-MTX infusion and folinic acid rescue at doses of 15 mg/m2 intravenously at 36 h, 15 mg/m2 every 3 h until 48 h and then 15 mg/m2 intravenously every 6 h until the MTX level was less than 1 × 10−7 molar.

The first 402 patients in the study received two daily doses of daunorubicin (45 mg/m2) at the start of induction. This was withdrawn from the protocol for patients accrued after March 1992 owing to concerns about late cardiotoxicity. A preliminary report on early response did not reveal a difference in outcome for these two groups of patients (Lilleyman et al, 1997).

Individual centres in the UK obtained approval from their local research ethics committee, and took informed consent from parents and patients before entering patients into the study.

Morphology Three morphologists performed a central review of FAB typing and assessment of early response to therapy for all patients entered into the study. Early response was assessed by counting the percentage of residual blasts in the marrow after 8 d of induction chemotherapy.

Cytogenetics and molecular genetics Chromosomal analysis was performed by regional cytogenetic centres within the UK. Karyotypes were reviewed and karyograms prepared by the cytogeneticists of the Leukaemia Research Fund/UK Cancer Cytogenetics Group karyotype database in ALL. In some cases, fluorescence in situ hybridization (FISH) studies were carried out to confirm chromosomal abnormalities. Karyotypes, coded according to the main structural and/or numerical chromosomal changes, were provided to the Clinical Trial Service Unit in Oxford (CTSU) for analysis. The presence of the TEL/AML1 fusion gene was investigated using reverse transcription polymerase chain reaction (RT-PCR) in a subset of 659 patients as part of a UK childhood case–control study.

Completeness of data Data on the following prognostic variables was available for all patients: age, sex, WCC at presentation, presence of CNS disease at diagnosis and remission status at end of induction (EOI).

For the remaining variables of interest, data was available for > 75% of patients except TEL/AML1, as detailed above.

Randomization method and statistical analysis Randomization was carried out by phoning a central office in which the patient details were entered onto computer. This allocated patients to the third block or not, balancing within age-, sex-, WCC- and CNS therapy-allocated groups.

Event-free survival (EFS) is the time from diagnosis to an event (either relapse or death) and includes non-remitters. Patients surviving event free were censored at 31 October 1999, when data was complete for all but nine patients (these were censored at the date they were last known to be event free). The relationships between possible prognostic factors and age group, WCC group and sex were analysed using the Mantel–Haenszel test for trend or Fisher's exact test (in 2 × 2 tables) as appropriate (SAS Procedures Guide, 1990).

Each potential prognostic factor was grouped if necessary and Kaplan–Meier actuarial EFS curves were compared with the log-rank test (Peto et al, 1977). Variables were first analysed with no adjustment and then after allowance for age, WCC and sex. Variables found to be statistically significant after adjustment were analysed in Cox models that allowed the inclusion of continuous variables. Models were fitted using step-wise selection with variables added to the model if they had a P-value < 0·05 but removed if they had a P-value > 0·01. Odds ratios (OR) were calculated from the observed minus expected number of events (O − E) and its variance [var(O − E)] in log-rank analyses, while conditional risk ratios (RR) were obtained from Cox analysis and allowed for all other factors in the final model.

Whether or not patients received daunorubicin in induction was allowed for in analyses of d 8 blast percentage, as patients receiving daunorubicin had a much lower median d 8 blast percentage than those who did not (median 7% with daunorubicin, 33% without daunorubicin).

The data was investigated within the subset of 1470 patients who were randomized to receive or not receive a third block of treatment to find out whether risk factors had a different effect in those receiving more or less intensification.

All P-values are two-tailed. Owing to the large number of significance tests performed and the associated increased probability of obtaining conventionally significant (P < 0·05) results by chance, only P-values < 0·01 are quoted.

Results

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Follow-up and treatment outcome by randomization

At a median follow-up of 5 years 9 months (range 2 years 7 months to 9 years 1 month), 5 years EFS for the whole group, including those not randomized, was 63% [95% confidence interval (CI) = 61% to 65%]. Among those randomized for the third block, 5 year EFS was significantly better for UKALL XI patients allocated the third block arm (69%, 95% CI = 65% to 72%) compared with those allocated no third block (60%, 95% CI = 57% to 64%) (OR 0·76, log-rank 2p = 0·002).

Prognostic factors

As shown in Tables II, III and IV, age, sex, presenting WCC, spleen and liver size, anterior mediastinal mass, immunophenotype, ploidy, t(4;11), Philadelphia chromosome (Ph+), abnormality of the short arm of chromosome 9 [abn (9p)], d 8 blast percentage and remission status at the end of induction (EOI) were significant prognostic variables in unstratified log-rank analyses (at P < 0·01).

Table II.   Significance of age, sex and WCC in unadjusted and adjusted log-rank analysis.
   Log-rank analysis  
   UnadjustedAdjusted*
 N (%)OO/EP-valueO/EP-value
  • *

    Analysis of age group allows for sex and WCC group, analysis of sex allows for age group and WCC group, and analysis of WCC group allows for age group and sex.

  •   †% of patients entered into study.

  •   P-values are log-rank P-values for trend in age and WCC and heterogeneity in sex.

Age (years)
 1–41184 (57)3870·85< 0·00010·86< 0·0001
 5–9590 (28)2281·07 1·06 
 10+316 (15)1551·50 1·45 
Sex
 Male1189 (57)4961·17< 0·00011·16< 0·0001
 Female901 (43)2740·79 0·80 
WCC (× 109/l)
 < 201294 (62)4100·80< 0·00010·81< 0·0001
 20–49334 (16)1220·98 1·02 
 50–99210 (10)961·35 1·34 
100+252 (12)1422·18 2·02 
Table III.   Significance of cytogenetics and ploidy in unadjusted and adjusted log-rank analysis.
   Log-rank analysis
   UnadjustedAdjusted*
 N (%)OO/EP-valueO/EP-value
  • *

    Adjusted for age group, WCC group and sex.

  •   †% of patients entered into study, except TEL/AML1 = % of the 659 patients tested, and cytogenetics and ploidy = % of the 1658 patients with a successful cytogenetic result.

  •   P-values are for log-rank P-values for trend except when only two groups are compared, when P-values are for heterogeneity.

Cytogenetic sample
 Yes2025 (97)7390·990·050·990·05
 No63 (3)301·44 1·43 
Cytogenetic result
 Success1658 (79)6111·020·21·020·02
 Failure367 (18)1280·90 0·90 
Ploidy
 23–29 (near haploid)4 (< 1)49·26< 0·00018·410·0009
 30–44 (low hypodiploid)6 (< 1)20·83 1·39 
 45 (hypodiploid)98 (6)391·17 1·07 
 46 (normal)293 (18)1211·14 1·10 
 46 (pseudodiploid)486 (29)2071·24 1·09 
 47–50 (low hyperdiploid)204 (12)771·02 1·01 
 51–65 (high hyperdiploid)529 (32)1490·71 0·81 
 66–80 (near triploid)16 (1)50·80 0·82 
 > 80 (near tetraploid)21 (1)70·83 0·92 
< 45 chromosomes
 Yes10 (< 1)62·110·093·140·003
 No1647 (79)6050·99 0·99 
Ploidy ≤ hypodiploid
 Yes109 (5)461·260·11·190·2
 No1548 (74)5650·98 0·99 
Ploidy ≥ high hyperdiploid
 Yes566 (27)1610·71< 0·00010·820·0008
 No1091 (52)4501·17 1·09 
Cytogenetic abnormalities
 t(1;19)Yes47 (3)130·750·30·630·07
No1611 (77)5981·01 1·01 
 t(4;11)Yes15 (1)135·12< 0·00013·69< 0·0001
No1643 (79)5980·98 0·98 
 11q23Yes43 (3)130·760·30·690·2
No1615 (77)5981·01 1·01 
 Philadelphia +veYes25 (2)183·84< 0·00012·96< 0·0001
No1633 (78)5930·98 0·98 
 abn(6q)Yes173 (10)671·080·51·050·7
No1485 (71)5440·99 0·99 
 abn(12p)Yes205 (12)730·970·81·040·7
No1453 (70)5381·00 0·99 
 abn(9p)Yes131 (8)691·69< 0·00011·350·01
No1527 (73)5420·95 0·97 
 TEL/AML1Positive128 (19)430·910·50·960·8
Negative531 (81)1951·02 1·01 
Table IV.  Significance of other prognostic variables in unadjusted and adjusted log-rank analysis.
    Log-rank analysis
    UnadjustedAdjusted*
  N (%†)OO/EP-valueO/EP-value
  • *

    Adjusted for age group, WCC group and sex (d 8 blast percentage also allows for (no) daunorubicin in induction).

  •   †% of patients entered into study.

  •   P-values are for log-rank P-values for trend except when only two groups are compared, when P-values are for heterogeneity.

Body mass index (kg/m2)< 15 15–19519 (25) 1396 (67)184 5120·95 1·000·081·02 0·991·0
20–24104 (5)441·24 0·97 
25–2912 (1)72·01 1·48 
Haemoglobin (g/dl)< 5·0466 (22)1580·890·020·970·5
5·0–6·9600 (29)2080·91 0·92 
7·0–8·9525 (25)2131·15 1·15 
9·0+465 (22)1771·07 0·96 
Platelets (× 109/l)< 25651 (31)2370·980·11·030·09
25–49494 (24)2021·15 1·09 
50–99454 (22)1721·05 1·00 
100+458 (22)1450·83 0·87 
Marrow blasts at diagnosis percentage< 90 90–94432 (21) 400 (19)152 1650·97 1·140·81·01 1·130·5
95–99888 (43)3120·94 0·93 
100231 (11)881·06 1·05 
CNS diseaseYes31 (1)121·180·60·810·5
No2059 (99)7581·00 1·00 
Down's syndromeYes39 (2)161·120·61·090·7
No2018 (97)7401·00 1·00 
Anterior mediastinal massYes No145 (7) 1848 (88)69 6631·56 0·960·00031·03 1·000·8
T-cell immunophenotypeYes No207 (10) 1730 (83)102 6041·73 0·93< 0·00011·03 0·990·7
Null immunophenotypeYes No60 (3) 1877 (90)27 6791·34 0·990·11·23 0·990·3
FABL11810 (87)6661·000·71·000·7
L2123 (6)461·06 0·95 
SplenomegalyNo883 (42)2970·880·0070·980·7
Easily palpable732 (35)2781·06 1·03 
Not below umbilicus352 (17)1461·20 1·04 
Below umbilicus77 (4)271·00 0·75 
HepatomegalyNo784 (38)2590·860·0010·930·1
Easily palpable870 (42)3241·03 1·01 
Not below umbilicus352 (17)1531·28 1·15 
Below umbilicus38 (2)120·82 0·76 
D 8 blast percentage≤ 25858 (41)2570·76< 0·00010·79< 0·0001
26+920 (44)4011·25 1·21 
D 28 remission statusCR No CR2019 (97) 71 (3)719 510·95 3·70< 0·00010·96 2·95< 0·0001

Patients with an anterior mediastinal mass, those with no remission at EOI, a T-cell phenotype, Ph+ or abn(9p) were, on average, older, whereas high hyperdiploid (> 50 chromosomes) patients were younger (P < 0·001 in each case, except Ph+, P = 0·009). Patients with hepatomegaly or splenomegaly, an anterior mediastinal mass, > 25% d 8 marrow blast percentage, T-cell disease, t(4;11), Ph+ and abn(9p) tended to have a higher WCC, while those with high hyperdiploidy had a lower WCC (P < 0·001 for all variables). T-cell patients and those with a high d 8 marrow blast percentage were more likely to be boys (P = 0·01 for both), and patients with < 45 chromosomes were more likely to be girls (P = 0·002).

After allowing for age, sex and presenting WCC in a stratified log-rank analysis, only ploidy, t(4;11), Ph+, abn(9p), d 8 blast percentage and EOI remission status remained significant predictors of outcome.

Cox regression analysis showed that age, sex, WCC, d 8 blast percentage, late remission and blast karyotype were all of independent prognostic importance (Table V). There was no evidence that any of these factors had more or less effect in those who were randomized to a third block than in those who were randomized to no third block. It is not possible to look meaningfully at either t(4;11) or Ph+ by third block randomization as the number of patients with these abnormalities was too small and some also received first remission transplants, which confounds the analysis.

Table V.  Step-wise Cox multivariate analysis of variables found significant in adjusted log-rank analysis (N = 1422*).
ParameterRange of valuesRisk Ratio (95% CI)P-value
  • *

    Data is restricted to the 1658 patients who had a successful cytogenetic sample. Two hundred and thirty-five patients are excluded from the analysis as they had missing d 8 % blasts and one patient owing to missing values for ploidy group/number of chromosomes.

Log(WCC + 1)(0·405–7·245)1·26 (1·18–1·35)< 0·0001
D 8 blast percentage(1–99)1·01 (1·01–1·01)< 0·0001
Near haploid(0 = no, 1 = yes)13·87 (5·09–37·81)< 0·0001
t(4;11)(0 = no, 1 = yes)3·61 (2·00–6·48)< 0·0001
abn(9p)(0 = no, 1 = yes)1·69 (1·30–2·21)< 0·0001
D 28 remission status(0 = CR, 1 = no CR)2·20 (1·47–3·28)< 0·0001
Sex(1 = male, 2 = female)0·71 (0·60–0·86)0·0003
≥ High hyperdiploid(0 = no, 1 = yes)0·68 (0·55–0·84)0·0004
t(9;22)(0 = no, 1 = yes)2·45 (1·46–4·12)0·0007
Age(1–14)1·04 (1·02–1·07)0·0009

Age, sex and WCC Age, sex and WCC all had a highly significant effect on EFS. Girls had a better prognosis than boys (RR = 0·71, 95% CI 0·60–0·86), whereas a poorer prognosis was associated with increasing age (RR = 1·04, 95% CI 1·02–1·07 for each year increase) and WCC (RR = 1·26, 95% CI 1·18–1·35 for each unit increase of log (WCC + 1). The effects of age and WCC are illustrated in Fig 2, which shows a difference in EFS of about 30% at 5 years between young children with low WCC (5 years EFS 69%) and older children with high WCC (5 years EFS 39%).

image

Figure 2. Event-free survival by age at diagnosis and WCC.

Download figure to PowerPoint

A hazard score using these three variables [0·22 × log (WCC + 1) + 0·0043 × age2 − 0·39 × sex: boy = 1, girl = 2] derived from UKALL VIII and UKALL X data defined a very high-risk group of patients (score > 0·8) with 5 years EFS < 40% (Chessells et al, 1995a). A similar score [0·26 × log (WCC + 1) + 0·0040 × age2−0·36 × sex: boy = 1, girl = 2] derived from UKALL XI data remains valid for defining such a high-risk group in both the third block and no third block arms of the trial.

Immunophenotype Null immunophenotype had no influence on prognosis and, although highly significant in univariate analysis (P < 0·001), T-cell phenotype lost its significance when stratified by age, sex and WCC (P = 0·7).

Blast karyotype Cytogenetic analysis was carried out on 2025 patients and a successful result was obtained in 1658 cases, of which 293 had a normal karyotype. Numerical chromosomal changes were related to outcome. In particular, high hyperdiploidy was associated with a relatively good outcome (RR 0·69, 95% CI = 0·55–0·84), whereas the small number with near haploidy (23–29 chromosomes) had a bad prognosis (RR 13·94, 95% CI = 5·11–38·03). The structural chromosomal abnormalities, t(4;11) (RR 3·62, 95% CI = 2·01–6·52) and Ph+ (RR 2·46, 95% CI = 1·47–4·13) were associated with a poor outlook. Patients with abn(9p) also had a significantly higher risk of an event (RR 1·69, 95% CI 1·29–2·21). EFS by blast karyotype features is illustrated in Fig 3, which shows that EFS was best in the high hyperdiploid patients (71%) and worst in the t(4;11) patients (13% at 5 years). All four patients with near haploidy died within 21/2 years of diagnosis.

image

Figure 3. Event-free survival by blast karyotype. There was overlap between specific cytogenetic abnormalities and ploidy. When this occured the patient was classified by cytogenetic abnormality rather than by ploidy. Fourteen abn(9p) patients and one t(9;22) patient were high hyperdiploid (> 50 chromosomes), all other patients with t(4;11), t(9;22) and abn(9p) had 30–50 chromosomes. The only overlap between cytogenetic abnormalities was for two patients with both t(9;22) and abn(9p) who were included as t(9;22) patients.

Download figure to PowerPoint

11q23 abnormalities other than t(4;11) lacked prognostic significance. This may be accounted for by the fact that the majority of cases in this group had non-specific abnormalities involving the chromosome band 11q23 in which the involvement of MLL was mainly unknown. Several reports have indicated that the presence of the TEL/AML1 fusion gene is associated with an extremely good prognosis (Borkhardt et al, 1997; Rubnitz et al, 1997). Although RT-PCR results were only available for 659 patients, the EFS of tested patients was not significantly different from that of patients who were not tested. Analysis of the subset of 659 patients in this study tested specifically, found no significant difference in outcome between TEL/AML1-positive and -negative patients (OR 1·05 95% CI = 0·75–1·48).

Early response to therapy The percentage of residual blasts in the marrow after 8 d of induction therapy was highly predictive of outcome (RR 1·008 95% CI = 1·005–1·011 for an increase of 1% residual blasts in the marrow; RR 1·58, 95% CI = 1·31–1·90 for > 25% vs. leqslant R: less-than-or-eq, slant 25% blasts, Fig 4). Non-remitters at EOI also had a poor outlook (RR 2·20, 95% CI = 1·47–3·28).

image

Figure 4. Event-free survival by d 8 blast percentage. The data is divided according to whether the patient was entered before or after March 1992, when the protocol was altered to remove daunorubicin from the induction schedule and introduce the option of randomization to a third block of treatment.

Download figure to PowerPoint

Interactions between variables T-cell immunophenotype did not have a significant influence on outcome independent of age, sex and WCC, but there was a significant interaction between T-cell disease, WCC and outcome (P < 0·001). Within the low WCC group (< 20 × 109/l), T-ALL patients fared worse than those with other immunophenotypes (P < 0·001), while within the high WCC (> 100 × 109/l) group, patients with T-ALL fared non-significantly better than those with other immunophenotypes (P = 0·2).

Similar interactions exist between WCC, outcome and both t(4;11) and Ph+ (P < 0·001 in both cases). Overall, patients with t(4;11) and Ph+ had a poor prognosis whatever their WCC, but the effect of t(4;11) was worse in the low WCC group, whereas that of Ph+ was worse in the high WCC group. Although achieving statistical significance, there is uncertainty about the clinical significance of these last interactions as they are based on small numbers of non-uniformly treated patients, many of whom went on to receive an allogeneic marrow transplant.

Discussion

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

We have assessed the significance of currently recognized prognostic factors in patients treated on MRC UKALL XI, a randomized study of additional intensification, which accrued all patients with childhood ALL except those diagnosed in infancy and those with mature B-cell ALL. Even after such intensive therapy, age, sex and WCC at presentation retained their dominant influence on prognosis.

We can confirm the previously reported association of high hyperdiploidy (Raimondi et al, 1992) with a good prognosis, and near haploid and Ph+ (Uckun et al, 1998a, 1999) with an adverse prognosis on such standard therapy. Unlike previous reports (Pui et al, 1990; Uckun et al, 1998b), we could find no adverse prognostic significance for near tetraploidy or t(1:19)(q23;p13) (Uckun et al, 1998b). It would appear that the very small number (15 in this study) of t(4:11)-positive patients outside infancy have the same poor prognosis as infants. Other abnormalities involving chromosome 11q23 apparently did not confer an adverse risk, but this is a heterogeneous group, some of whom may have rearrangements of the MLL gene and a poor outlook, and others without this rearrangement who may have standard risk disease. Abnormalities of chromosome arm 9p conferred an adverse risk independent of other prognostic factors with a RR of 1·69, confirming the recent Children's Cancer Group (CCG) report (Heerema et al, 1999). We were unable to confirm the reported association of TEL/AML1 fusion abnormality with a good prognosis (Borkhardt et al, 1997; Rubnitz et al, 1997). Taken together with the failure to detect prognostic significance in a prospective study by the FRALLE group (Baruchel et al, 1999) and reports from the Berlin–Frankfurt–Münster (BFM) group indicating that its incidence is similar at diagnosis and relapse (Harbott et al, 1997; Seeger et al, 1998), our finding adds to the evidence questioning the prognostic significance of this abnormality. The contradictory evidence on this issue might be a reflection of differences in treatment approaches as these can emphasize or mask the significance of individual prognostic factors (Loh et al, 1998). Furthermore, our study again emphasizes the fact that effective therapy is the most important prognostic factor and can overcome adverse factors, as has been demonstrated over the last decade for ALL of mature B cells (Sandlund et al, 1996).

We can confirm the prognostic value, first reported by the CCG (Miller et al, 1989), of assessing early response to therapy by counting residual blasts in the marrow after 8 d of induction therapy. The RR for event risk (1·58) associated with the presence of > 25% marrow blasts at this time-point is similar to that reported by the CCG (Steinherz et al, 1996; Gaynon et al, 1997). The BFM method of assessing early response to therapy by residual peripheral blood blast count after 7 d of preinduction prednisolone (Riehm et al, 1987) is more specific, but less sensitive. It captures only around 11% of patients with adverse risk compared with the 50% captured by assessing the marrow blast count. The handful of patients failing to remit at the end of induction had a poor outcome with < 30% EFS at 5 years, whatever subsequent therapy they received.

Ideally, risk stratification should facilitate tailoring of drug combinations, schedules and doses to biologically distinct subgroups of childhood ALL. This approach proved successful for mature B-cell lymphoma and leukaemia in which fundamental changes in drug combinations and scheduling from standard lymphoma and leukaemia therapy resulted in dramatic improvements in cure rates in the late seventies (Sandlund et al, 1996). Unfortunately, to date, similar approaches have not met with the same success in other subtypes of ALL (Crist et al, 1992). Therefore, in practice, risk stratification has been principally employed to target high-risk groups for more intensive treatments that entail greater risks of death or long-term side-effects on the assumption that the benefits may outweigh the risks in such patient groups. However, it would appear that low-risk groups, sometimes excluded from intensive protocols, may gain as much (Hutchinson et al, 1996; Gustaffson et al, 1998), if not more, from dose intensification than high-risk patients. This was the reported reason for the paradoxical risk reversals observed by the US CCG group when putatively higher-risk patients achieved a better outcome with more intensive regimens than lower-risk patients who received less intensive risk-adjusted therapy (Gaynon et al, 1993; Tubergen et al, 1993; Nachman et al, 1998). When the low-risk patients eventually received more intensive therapy, the expected risk ordering re-emerged. Similarly, an attempt at withdrawal of intensive reinduction therapy from ‘low-risk’ patients by the German BFM group resulted in an increased incidence of treatment failures (Henze et al, 1990).

Although the risk criteria assessed here are important predictors of outcome, the majority of events arise in patients without obvious adverse risk factors. There is considerable heterogeneity of outcome within currently defined risk groups that needs clarification to allow more precise estimation of prognosis. Detection of submicroscopic levels of residual disease by molecular (Brisco et al, 1994; Van Dongen et al, 1998) and flow cytometric (Coustan-Smith et al, 1998) methods might provide the most sensitive and specific method for capturing adverse risk subsets of low-risk patients who may benefit from interventions currently reserved for high-risk patients. Until these methods are brought into clinical application, risk stratification should be applied judiciously to select very high-risk patients for experimental approaches rather than to withdraw therapy from apparently lower-risk patients.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

We would like to thank all of the paediatric haematologists/oncologists who entered patients into the trial, the Medical Research Council and Imperial Cancer Research Fund for Central Support, and the Leukaemia Research Fund of Great Britain for supporting the cytogenetic database.

Members of the MRC childhood leukaemia working party during the period of these studies were: C.C. Bailey, C.J. Barton, V.A. Broadbent, A. Caswell, J.M. Chessells, P.J. Darbyshire, S.I. Dempsey, I.J. Durrant, O.B. Eden (Chairman), K.M. Forman, B. Gibson, A. Goodman, I.M. Hann, C. Haworth, F.G. Hill, M.E.M. Jenney, D.J. King, S. Kinsey, J.S. Lilleyman, M. Madden, J.R. Mann, S.T. Meller, C.D. Mitchell, A. Oakhill, M. Radford, M.M. Reid, S. Richards, O.P. Smith, R.F. Stevens, A. Thomas, F. Vargha-Khadem, A.J. Vora, D. Webb, K. Wheatley, A. Will and K.P. Windebank.

References

  1. Top of page
  2. Abstract
  3. Patients and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  • Baruchel, A., Cayuela, J.M. & Landman, J. (1999) Prospective assessment of the prognostic value of TEL-AML1 expression in childhood acute lymphoblastic leukemia (ALL). Blood, 94, 502a.
  • Borkhardt, A., Cazzaniga, G., Viehmann, S., Valsecchi, M.G., Ludwig, W.D., Burci, L., Mangioni, S., Schrappe, M. & Riehm, H. (1997) Incidence and clinical relevance of TEL-AML1 fusion genes in children with acute lymphoblastic leukemia enrolled in the German and Italian multicenter therapy trials. Associazione Italiana Ematologia Oncologia Pediatrica and the Berlin-Frankfurt-Munster Study Group. Blood, 90, 571577.
  • Brisco, M.J., Condon, J., Hughes, E., Nech, S.H., Sykes, P.H., Seshadri, R., Toogood, I., Water, K., Tauro, G. & Eckhert, H. (1994) Outcome prediction in childhood acute lymphoblastic leukaemia by molecular quantification of residual disease at the end of induction. Lancet, 343, 196200.
  • Chessells, J.M., Richards, S.M., Bailey, C.C., Lilleyman, J.S. & Eden, O.B. (1995a) Sex and treatment outcome in childhood lymphoblastic leukaemia: report from the MRC UKALL trials. British Journal of Haematology, 89, 372380.
  • Chessells, J.M., Bailey, C.C. & Richards, S.M. (1995b) Intensification of treatment and survival in all children with lymphoblastic leukaemia: results of UK Medical Research Council trial UKALL X. Lancet, 345, 143148.
  • Chessells, J.M., Swansbury, G.J., Reeves, B., Bailey, C.C. & Richards, S.M. (1997) Cytogenetics and prognosis in childhood lymphoblastic leukaemia: results of MRC UKALL X. Medical Research Council Working Party in Childhood Leukaemia. British Journal of Haematology, 99, 93100.
  • Chessells, J.M., Hall, E., Prentice, H.G., Durrant, J., Bailey, C.C. & Richards, S.M. (1998) The impact of age on outcome in lymphoblastic leukaemia: MRC UKALL X and XA compared: a report from the MRC Paediatric and Adult Working Parties. Leukaemia, 12, 463473.
  • Coustan-Smith, E., Behm, F., Sanchez, , Boyett, J.M., Hancock, M.L., Raimondi, S.C., Rubnitz, J.E., Ribera, G.K., Sandlund, J.T., Pui, C.H. & Campoma, D. (1998) Immunological detection of minimal residual disease in children with acute lymphoblastic leukemia. Lancet, 351, 550554.DOI: 10.1016/s0140-6736(97)10295-1
  • Crist, W., Shuster, J., Look, T., Borowitz, M., Behm, F., Bowman, P., Frankel, L., Pullen, J., Krance, R. & Steuber, P. (1992) Current results of studies of immunophenotype-age and leukocyte-based therapy for children with acute lymphoblastic leukemia. The Pediatric Oncology Group. Leukemia, 6, 162166.
  • Donadieu, J., Auclerc, M.F., Baruchel, A., Leblanc, T., Landman-Powker, J., Perel, Y., Michel, G., Gornu, G. & Bordigoni, P. (1998) Critical study of prognostic factors in childhood acute lymphoblastic leukaemia: differences in outcome are poorly explained by the most significant prognostic variables. British Journal of Haematology, 102, 729739.DOI: 10.1046/j.1365-2141.1998.00818.x
  • Dordelmann, M., Schrappe, M., Reiter, A., Zimmerman, M., Graf, N., Schott, G., Lampert, F., Harbbott, J., Niemeyer, M. & Ludwig, W.D. (1998) Down's syndrome in childhood acute lymphoblastic leukemia: clinical characteristics and treatment outcome in four consecutive BFM trials. Berlin-Frankfurt-Munster Group. Leukemia, 12, 645651.DOI: 10.1038/sj.leu.2400989
  • Forestier, E., Gustafsson, G., Von Heideman, A., Heim, S., Hernnell, O., Mitelman, F., Nordenson, I. & Swolin, B. (1997) Prognostic impact of bone marrow karyotype in childhood acute lymphoblastic leukaemia: Swedish experiences 1986–91. Acta Paediatrics, 86, 819825.
  • Gaynon, P.S., Steinherz, P.J., Bleyer, W.A., Ablin, A.R., Albo, V.C., Finkelstein, J.Z., Grossman, N.J., Novak, L.J. & Pyesmany, A.F. (1993) Improved therapy for children with acute lymphoblastic leukemia and unfavourable presenting features: a follow up report of the Children's Cancer Study Group CCG-106. Journal of Clinical Oncology, 11, 22342242.
  • Gaynon, P.S., Desai, A.A., Bostrom, B.C., Hutchinson, P.J., Lange, B.J., Nachman, J.B., Reaman, G.H., Sathar, N.H., Steinherz, P.J., Trigg, M.E., Tubergen, D.G. & Uckun, F.M. (1997) Early response to therapy and outcome in childhood acute lymphoblastic leukemia: a review. Cancer, 80, 17171726.DOI: 10.1002/(sici)1097-0142(19971101)80:9<1717::aid-cncr4>3.0.co;2-b
  • Gustaffson, G., Kreuger, A., Clausen, N., Garwicz, S., Kristinsson, J., Lies, O., Moe, P.H., Perkkio, M., Yssing, M. & Saarinen-Pihkala, U.M. (1998) Intensified treatment of acute childhood lymphoblastic leukaemia has improved prognosis, especially in non-high risk patients: the Nordic experience of 2648 patients diagnosed between 1981 and 1996. Nordic Society of Paediatric Haematology and Oncology (NOPHO). Acta Paediatrics, 87, 11511161.DOI: 10.1080/080352598750031149
  • Hann, I., Vora, A., Richards, S., Hill, F., Gibson, B., Lilleyman, J.S., Kinsley, S., Mitchell, C. & Eden, O.B. (2000) Benefit of intensified treatment for all children with acute lymphoblastic leukaemia: results from MRC UKALL XI and MRC AL97 randomized trials. Leukemia, 14, 356363.DOI: 10.1038/sj.leu.2401704
  • Hann, I.M., Richards, S.M., Eden, O.B. & Hill, F.G. (1998) Analysis of the immunophenotype of children treated on the Medical Research Council United Kingdom Acute Lymphoblastic Leukaemia Trial XI (MRC UKALLXI). Medical Research Council Childhood Leukaemia Working Party. Leukemia, 12, 12491255.DOI: 10.1038/sj.leu.2401093
  • Harbott, J., Viehmann, S., Borkhardt, A., Henze, G. & Lampert, F. (1997) Incidence of TEL/AML1 fusion gene analysed consecutively in children with acute lymphoblastic leukemia in relapse. Blood, 90, 49334937.
  • Heerema, N.A., Sather, H., Sensel, M.G., Lui-Marcs, W., Lange, B.J., Bostrom, B.C., Nachman, J.B., Steinherz, P.G. & Hutchinson, R. (1999) Association of chromosome arm 9p abnormalities with adverse risk in childhood acute lymphoblastic leukemia: a report from the Children's Cancer Group. Blood, 94, 15371544.
  • Henze, G., Fengler, R., Reiter, A., Ritter, J. & Riehm, H. (1990) Impact of early intensive reinduction therapy on event-free survival in children with low-risk acute lymphoblastic leukemia. Hamatological Bluttransfusion, 33, 483488.
  • Hutchinson, R., Neerhout, R. & Bertolone, S. (1996) Should therapy be intensified for patients with good risk ALL? Blood, 88, 668a.
  • Lilleyman, J.S., Hann, I.M., Stevens, R.F., Richards, S.M., Eden, O.B., Chessells, J.M. & Bailey, C.C. (1992) Cytomorphology of childhood lymphoblastic leukaemia: a prospective study of 2000 patients. British Journal of Haematology, 81, 5257.
  • Lilleyman, J.S., Gibson, B.E., Stevens, R.F., Will, A.M., Hann, I.M., Richards, S.M. & Hill, F.G. (1997) Clearance of marrow infiltration after 1 week of therapy for childhood lymphoblastic leukaemia: clinical importance and the effect of daunorubicin. British Journal of Haematology, 97, 603606.DOI: 10.1046/j.1365-2141.1997.1002914.x
  • Loh, M.L., Silverman, L.B., Young, M.L., Neuberg, D., Grolub, T.R., Sallan, S.E. & Gilliland, D.G. (1998) Incidence of TEL/AML1 infusion in children with relapsed acute lymphoblastic leukemia. Blood, 92, 47924797.
  • Miller, D.R., Coccia, P.F., Bleyer, W.A., Lukens, J.N., Siegel, S.E., Sather, H.N. & Hammond, G.D. (1989) Early response to induction therapy as a predictor of disease-free survival and late recurrence of childhood acute lymphoblastic leukemia: a report from the Children's Cancer Study Group. Journal of Clinical Oncology, 7, 18071815.
  • Nachman, J.B., Sather, H.N., Sensel, M.G., Trigg, M.E., Breslow, N.E., Lukens, J.N., Wolff, L., Uckun, F.M. & Gaynon, P.S. (1998) Augmented post-induction therapy for children with high risk acute lymphoblastic leukemia and a slow response to initial therapy. New England Journal of Medicine, 338, 16631671.DOI: 10.1056/NEJM199806043382304
  • Peto, R., Pike, M.C., Armitage, P., Breslow, N.E., Cox, D.R., Howard, S.V., Mantel, N., McPherson, K., Peto, J. & Smith, P.G. (1977) Design and analysis of randomized clinical trials requiring prolonged observation of each patients. II. Analysis and examples. The British Journal of Cancer, 35, 139.
  • Pui, C.H., Carroll, A.J., Head, D.R., Riamondi, S.C., Shuster, J.J., Crist, W.M., Link, M.P., Borowit, Z.M. & Behm, F.G. (1990) Near-Triploid and Near-Tetraploid acute lymphoblastic leukaemia of childhood. Blood, 76, 590596.
  • Pui, C.H., Boyett, J.M., Relling, M.V., Harrison, P.L., Rivera, G.K., Behm, F.G., Sandlund, J.T., Ribeiro, R.C., Rubnitz, J.E., Grajjar, A. & Evans, W.E. (1999) Sex differences in prognosis for children with acute lymphoblastic leukemia. Journal of Clinical Oncology, 17, 818824.
  • Raimondi, S.C., Roberson, P.K., Pui, C.H., Behm, F. & Rivera, G.K. (1992) Hyperdiploid (47–50) acute lymphoblastic leukaemia in children. Blood, 79, 32453252.
  • Riehm, H., Reiter, A., Schrappe, M., Berthol, F., Dopfer, R., Gerein, V., Ludwig, R., Ritter, J., Stollman, B. & Henze, G. (1987) Corticosteroid-dependent reduction of leucocyte count in blood as a prognostic factor in acute lymphoblastic leukemia in childhood (therapy study ALL-BFM 83). Klinische Pediaatrics, 199, 151160.
  • Rubnitz, J.E., Downing, J.R., Pui, C.H., Shurtleffs, A., Raimondi, S.C., Evans, W.E., Head, D.R., Crist, W.M. & Rivera, G.K. (1997) TEL gene rearrangement in acute lymphoblastic leukemia: a new genetic marker with prognostic significance. Journal of Clinical Oncology, 15, 11501157.
  • Sandlund, J.T., Downing, J.R. & Crist, W. (1996) Non-Hodgkin's Lymphoma in Childhood. New England Journal of Medicine, 334, 12381247.DOI: 10.1056/NEJM199605093341906
  • SAS Institute Inc. (1990) SAS Procedures Guide, Version 6 (3rd edn). SAS Institute Inc., Cary, NC.
  • Seeger, K., Adams, H.P., Buchwald, D., Beyermann, B., Bremens, B., Niemeyer, C., Ritter, J., Schwabe, D., Harms, D., Schrappe, M. & Henze, G. (1998) TEL-AML1 fusion transcript in relapsed childhood acute lymphoblastic leukemia. The Berlin-Frankfurt-Munster Study Group. Blood, 91, 17161722.
  • Smith, M., Arthur, D.C., Camitta, B., Carroll, A.J., Crist, W., Graynon, P., Gelber, R., Heerema, N., Korn, E.L. & Link, M. (1996) Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. Journal of Clinical Oncology, 14, 1824.
  • Steinherz, P.G., Gaynon, P.S., Breneman, J.C., Cherlow, J.M., Grossman, N.J., Kersssey, J.H., Johnstone, H.S., Sather, H.N. & Trigs, M.E. (1996) Cytoreduction and prognosis in acute lymphoblastic leukemia – the importance of early marrow response: report from the Children's Cancer Group. Journal of Clinical Oncology, 14, 389398.
  • Tubergen, D.G., Gilchrist, G.S., O'Brien, R.T., Coccia, P.F., Sather, H.N., Waskerwitz, M.J. & Hammond, G.D. (1993) Improved outcome with delayed intensification for children with acute lymphoblastic leukemia and intermediate presenting features: a Childrens Cancer Group phase III trial. Journal of Clinical Oncology, 11, 527537.
  • Uckun, F.M., Nachman, J.B., Sather, H., Sensel, M.G., Kraft, P., Steinherz, P.G., Lange, B., Hutchinson, R., Reaman, G.H., Gaynon, P.S. & Heerema, N.A. (1998a) Clinical significance of Philadelphia chromosome positive pediatric acute lymphoblastic leukemia in the context of contemporary intensive therapies: a report from the Children's Cancer Group. Cancer, 83, 20302039.DOI: 10.1002/(sici)1097-0142(19981101)83:9<2030::aid-cncr21>3.3.co;2-h
  • Uckun, F.M., Sensel, M.G., Sather, H.N., Gaynon, P.S., Arthur, D.C., Lange, B.J., Steinherz, P.G. & Kraft, P. (1998b) Clinical significance of translocation t(1;19) in childhood acute lymphoblastic leukemia in the context of contemporary therapies: a report from the Children's Cancer Group. Journal of Clinical Oncology, 16, 527535.
  • Uckun, F.M., Nachman, J.B., Ssather, H., Sensel, M.G., Kraft, P., Steinherz, P.G., Lange, B., Hutchinson, R., Reaman, G.H., Gaynon, P.S. & Heerema, N.A. (1999) Poor treatment outcome of Philadelphia chromosome positive pediatric acute lymphoblastic leukemia despite intensive chemotherapy. Leukemia and Lymphoma, 33, 101106.
  • Van Dongen, J.J., Seriu, T., Panzer-Grumayer, E.R., Biondi, A., Pongers-Willemse, M.J., Coral, L., Stolz, F., Schrappe, M., Masera, G., Kauups, W.A. & Gadner, H. (1998) Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet, 352, 17311738.DOI: 10.1016/s0140-6736(98)04058-6
  • Viana, M.B., Murao, M., Ramos, G., Oliveira, H.M., De Carvalho, R.I., De Bastos, M., Colosimo, E.A. & Silverstrini, W.S. (1994) Malnutrition as a prognostic factor in lymphoblastic leukaemia: a multivariate analysis. Archives of Diseases of Childhood, 71, 304310.