Patients and treatment. J Pullen 1, JJ Shuster 2, M Link 3, M Borowitz 4, M Amylon 5, AJ Carroll 6, V Land 7, AT Look 8, B McIntyre 1 and B Camitta 9

Transcription

1 Leukemia (1999) 13, Stockton Press All rights reserved /99 $ Significance of commonly used prognostic factors differs for children with T cell acute lymphocytic leukemia (ALL), as compared to those with B-precursor ALL. A Pediatric Oncology Group (POG) study J Pullen 1, JJ Shuster 2, M Link 3, M Borowitz 4, M Amylon 5, AJ Carroll 6, V Land 7, AT Look 8, B McIntyre 1 and B Camitta 9 1 University of Mississippi Medical Center, Jackson, MS; 2 Pediatric Oncology Group Statistical Office, Gainesville, FL; 3 Stanford University, Packard Children s Hospital, Palo Alto, CA; 4 Johns Hopkins University, Baltimore, MD; 5 Stanford University, Stanford, CA; 6 University of Alabama Medical Center, Birmingham, AL; 7 Pediatric Oncology Group Operations Office, Chicago, IL; 8 St Jude Children s Research Hospital, Memphis, TN; and 9 Midwest Children s Cancer Center, Milwaukee, WI, USA T cell acute lymphocytic leukemia (T-ALL) and B-precursor ALL differ significantly in the clinical characteristics of the patients at presentation and in laboratory-defined characteristics of the leukemic cells. We assessed for pediatric patients with T-ALL the relative importance of prognostic factors previously demonstrated to predict outcome in B-precursor ALL. Presenting clinical and laboratory features were correlated with outcome for 441 children 12 months to 21 years of age with previously untreated T-ALL, registered on the Pediatric Oncology Group (POG) T3 protocol between 1986 and These T-ALL prognostic factor analyses were then compared to similar analyses for 1993 patients with B-precursor ALL enrolled during the same time period on the POG ALinC 14 protocol. Quantitative interaction between phenotype and each prognostic factor was studied to determine the relative importance of the prognostic factor for each of the two major immunophenotypes. We also analyzed the importance of maturational stage as a T-ALL prognostic factor, using a modified Ludwig definition of maturational stage. We conclude that several of the clinical and laboratory prognostic factors, which are used reliably for B- precursor ALL, are much less predictive in T-ALL (ie age, WBC, consensus risk group, hyperdiploidy, presence of translocations and CALLA expression). There was no significant difference between the phenotypes in the prognostic importance of race or gender. Our data demonstrate a significant difference in outcome among the three maturational stages of T- cell ALL, with the intermediate group faring best. Using traditional risk group criteria to stratify patients with T-ALL for therapy may not be appropriate. Keywords: acute lymphocytic leukemia; T cell; B-precursor; childhood; prognostic factors Introduction Children with T cell acute lymphocytic leukemia (T-ALL) have been shown by many investigators to have an unfavorable prognosis, as compared to those with B-precursor ALL, 1 9 although significant improvement in outcome for patients with T-ALL has been reported from some studies employing intensive, rotating pulses of chemotherapy Some cooperative groups have placed any patient with the T-ALL immunophenotype in a higher risk group in treatment assignment, regardless of other commonly used prognostic determinants, such as age and WBC. 8,9,14 17 Other groups have used the same age and WBC-defined risk categories in designating intensity of treatment for both B-precursor and for T-ALL; 11,18,19 but some of these have assigned any patient with a mediastinal mass and the T immunophenotype to a higher risk category. 11,20,21 Furthermore, many groups have employed the identical risk- Correspondence: J Pullen, University of Mississippi Medical Center, Division of Pediatric Hematology/Oncology, 2500 North State Street, Jackson, MS 39216, USA; Fax: (601) Received 31 May 1999; accepted 12 July 1999 adjusted treatment regimens for T-ALL and for B-precursor ALL patients, 5,10,11,14,20 while the Pediatric Oncology Group (POG) has designed different therapies for the T-ALL and B-precursor ALL immunophenotypes, respectively. 12,22,23 Only 10 to 15% of children with ALL have leukemic cells demonstrating the T-ALL phenotype. 3,24,25 Therefore, when patients with T-ALL are assigned to the same treatment protocol as those with B-precursor ALL, comparatively fewer T-ALL patients are registered, even in the large cooperative groups. This makes it difficult to determine which prognostic factors are important in T-ALL, because results of two or more protocols and several different treatment regimens must be pooled to accumulate sufficient numbers of T-ALL patients. Since 1976, the POG has employed different treatment protocols for children with B-precursor ALL and T-ALL. Although this complicates the comparison of outcome between B- precursor and T-ALL patients, it creates an excellent opportunity for evaluating unique contributions of prognostic factors within each immunophenotypic group. The purpose of this paper is to assess for patients with T-ALL the relative importance of prognostic factors that are known to be predictive of outcome in patients with B-precursor ALL. We also address the relationship between outcome and maturational stage of the lymphoblasts in this group of uniformly treated patients with T-ALL. Materials and methods Patients and treatment Presenting clinical and laboratory features were correlated with outcome for the 441 children, 1.00 to years of age, with previously untreated T-ALL, who were entered and eligible for the pilot (POG 8691) and final (POG 8704) T- ALL No. 3 (T3) treatment protocols between January 1986 and January Outcome analyses for this paper were conducted as of April Patients with advanced stage T-NHL were also treated on the T3 protocol, but are not included in this paper. The POG 8691 T3 groupwide pilot protocol was open to new registrations from 2 January 1986 to 23 June 1987 and enrolled 100 patients, three of whom were ineligible. The POG 8704 T3 frontline protocol enrolled 359 patients (344 eligible) from 23 June 1987 to 9 January The T3 pilot and frontline protocols incorporated the same two treatment regimens (Table 1). Patients were randomized between regimens 1 and 2, except for the first 8 months of the pilot study, in which the two regimens were used sequentially. Regimen 1 employed intensive multi-drug induction and consolidation phases, followed by maintenance with rotating cycles of various agents, all previously demonstrated to be

2 Table 1 Induction: T3 treatment regimen Vincristine 1.5 mg/m 2 (max 2 mg) IVP weekly 5 begin day 1 Prednisone 40 mg/m 2 /day (max 60 mg/day) po t.i.d 28 days Cyclophosphamide 1000 mg/m 2 i.v. on day 1 Doxorubicin 50 mg/m 2 i.v. on day 1 Cytarabine 100 mg/m 2 /day IVCI for 5 days begin day 22 Cyclophosphamide 600 mg/m 2 i.v. on day 22 L-asparaginase IU/m 2 i.m. on days 27, 29 and 31 Consolidation: Teniposide 150 mg/m 2 i.v. twice weekly 4 doses begin day 43 Cytarabine 300 mg/m 2 IVP twice weekly 4 doses after teniposide Vincristine 1.5 mg/m 2 (max 2 mg) IVP weekly 4 begin day 71 Prednisone 40 mg/m 2 /day (max 60 mg/day) p.o. t.i.d 28 days begin day 71 Doxorubicin 50 mg/m 2 i.v. on day 71 CNS prophylaxis: Intrathecal triple drugs every other week 7 doses begin day 1, then q 9 weeks throughout maintenance Cranial XRT 2400 cgy begin day 71 for patients with WBC 50 K only Maintenance: 9 week cycle repeated 10 times: Cytarabine 150 mg/m 2 /day IVCI 3 days begin day 1 Cyclophosphamide 75 mg/m 2 IVP q 12 h 6 doses begin day 1 Vincristine 2 mg/m 2 (max 2 mg) i.v. on day 22 Doxorubicin 30 mg/m 2 i.v. on day 22 Prednisone 120 mg/m 2 /day po t.i.d. 5 days begin day 22 6MP 225 mg/m 2 /day p.o. t.i.d. 5 days begin day 22 Teniposide 150 mg/m 2 i.v. q 3 days 2 doses begin day 43 Cytarabine 300 mg/m 2 IVP q 3 days 2 doses following teniposide Randomization: L-asparaginase IU/m 2 i.m. weekly 20 doses begin day 99 Reprinted with permission from Amylon et al. 12 effective in T cell disease. Regimen 2 was the same as regimen 1, except that it incorporated weekly i.m. l-asparaginase for 20 doses, beginning immediately after the completion of consolidation treatment. These treatment regimens have been recently reported by Amylon et al. 12 The prognostic factor analyses for the patients with T-ALL were compared to similar analyses for POG patients with newly diagnosed B-precursor ALL enrolled during the same time period. The previously reported prognostic factor analyses for patients with B-precursor ALL 6,8,26,27 have been updated to April 1997 for this paper. The patients with B-precursor ALL (n = 1933) were registered on the POG 8602, ALinC 14 protocol from February 1986 to January 1991, using the risk group stratified treatments previously described. 26,28 Each protocol was reviewed and approved by the National Cancer Institute, as well as by the local institutional review boards of each POG institution participating in these studies. Informed consent was obtained from the patients, their parents, or both, as deemed appropriate, according to the Department of Health and Human Services guidelines. Laboratory evaluation All POG patients with newly diagnosed ALL from February 1986 to January 1991 were registered on the POG 8600 biologic classification study. (Patients with T-ALL diagnosed from January 1991 to January 1992 had the initial classification studies performed as part of the subsequent POG 9000 ALL classification study, which included the same diagnostic workup for T-ALL at the same reference laboratories, as described here for the POG 8600 classification study.) Registration on the classification study was required prior to enrollment on either the T3 treatment protocol for T-ALL, or the ALinC 14 treatment protocol for B-precursor ALL. The classification protocol included a battery of laboratory tests, as outlined in the subsequent sections. Cytochemical studies: Bone marrow cells obtained at diagnosis were stained at each POG local institution according to standard techniques, including the use of Wright Giemsa, myeloperoxidase (MPO), Sudan black B, and naphthyl and butyrate esterase stains. The diagnosis of ALL was based on the morphology and cytochemistry criteria of the French American British (FAB) Cooperative Group. Only patients with leukemic cells negative for MPO and Sudan black B and negative for non-specific esterase stains were eligible for the POG 8600 New ALL classification protocol. Immunophenotyping: Specimens of pre-treatment heparinized bone marrow (approximately 90% of samples) or peripheral blood containing WBC/ l with more than 75% blasts, when marrow could not be aspirated (approximately 10% of samples) were placed in RPMI-1640 with 15% fetal calf serum and shipped at room temperature by overnight carrier to either the Stanford immunophenotyping reference laboratory or to the Duke immunophenotyping reference laboratory. The bone marrow samples were separated on Ficoll Hypaque gradients and immunophenotyping studies were performed in both reference laboratories. Cell surface antigens were detected by a standard indirect immunofluorescence assay using a panel of monoclonal antibodies directed against B-lineage, T-lineage, and myeloid-associated antigens and were assayed for fluorescence activity by flow cytometry. Antigens routinely tested (when sufficient cells were available for performing the entire panel) included CD1, CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, CD22, CD24, CD13, CD15, CD33, CD34, CD45 and HLADR. Samples were also sent for sig,,, and cig testing at the St Jude Children s Research Hospital immunophenotyping reference laboratory. These results in relationship to outcome have been previously described and will not be presented here. Depending on the pattern of reactivity, cases were classified as T cell ( 20% of the blasts expressing at least two of the antigens CD7, CD5 and CD2, or expressing only CD7 in the absence of myeloid antigens) or B-precursor (CD19 positive; 1697

3 1698 CD22 positive or negative; CD24 positive or negative; CALLA positive or negative; CD7 and CD5 negative) or B cell (sig positive and or positive). Cytogenetics: Bone marrow samples were placed in RPMI-1640 medium supplemented with 15% fetal calf serum and shipped overnight to the POG cytogenetics reference laboratory at the University of Alabama at Birmingham. On arrival, samples were subjected to short-term (overnight) cultures. Routine methods were used for culture harvest, slide preparation and GTG banding. Karyotypic designations were assigned according to the ISCN (1985) 31 and ISCN (1991). 32 DNA index: Cellular DNA content determinations were performed at St Jude Children s Research Hospital (SJCRH) on bone marrow samples of leukemic cells stained with propidium iodide. Flow cytometry was used to derive the DNA index (ratio of DNA content in leukemic G 0 /G 1 cells vs that in normal diploid cells), which was then used to distinguish two prognostic categories: DNA index (DI) 1.16 or 1.16 (approximately equal to 52 or 52 chromosomes), as previously described. 26 If more than one leukemic line was demonstrated for a particular sample, the lowest value was used as the DI for the study. DI testing was performed on only 80% of samples received at the SJCRH reference laboratory, because, for the remainder of samples, no cells were left after the mandatory SJCRH reference laboratory cig, sig, and, testing. Reference laboratory statistics: In the Stanford ALL immunophenotyping reference laboratory 95.4% of the samples were processed satisfactorily, as were 91.4% in the Duke ALL immunophenotyping reference laboratory. The POG cytogenetics reference laboratory processed 80% of samples satisfactorily. Informative (abnormal) karyotypes were obtained in 54% of samples. Non-informative results were obtained in 46% of samples ( normal karyotypes in 26% and unsatisfactory results in 20%). The POG ploidy (DNA Index) reference laboratory processed 99% of samples tested with satisfactory results. Definitions: Complete remission (CR) was defined as 5% lymphoblasts in the bone marrow (with evidence of regeneration of normal marrow cells) and 5 mononuclear cells per l in the CSF. CNS leukemia was diagnosed by the presence of 5 mononuclear cells per microliter and identifiable leukemic blasts in Wright-stained cytocentrifuged samples of CSF, or by the presence of cranial nerve palsies, with or without leukemic blasts in the CSF. The diagnosis of testicular leukemia required confirmation by open biopsy. T-ALL maturational stage was defined according to a modification of the Ludwig T-ALL maturational staging definitions 33 as follows: early stage (CD1 negative, CD3 negative); intermediate stage (CD1 positive, CD3 negative or positive); and mature stage (CD1 negative, CD3 positive). These modifications were made because only a few of our patient samples had been tested for cytoplasmic CD3. All patients categorized as to maturational stage had blasts that were CD7 positive. Statistical considerations Event-free survival, the time from registration to the earliest of induction failure, relapse, second malignancy, death, or last clinical contact, was used as the dependent variable in all inferential analyses. The significance of binary (yes/no) prognostic factors within phenotypes was assessed by the twosided logrank test. 34 In addition, a formal test of quantitative interaction between phenotype and the prognostic factor was conducted by a Cox proportional hazards model 35 fitting three independent variables: phenotype (1 = B-precursor and 0 = T cell), prognostic factor (1 = unfavorable, 0 = favorable) and interaction (1 = T cell, unfavorable and 0 = not 1). The natural antilog of the interaction coefficient represents the ratio of the hazard ratios. For example, if the instantaneous ratio of failure probabilities (hazard ratio) for B-precursor (favorable to unfavorable) is 0.60 and the corresponding ratio for T cell is 0.90, the quotient is 0.60/0.90 = The values less than 1.00 imply the prognostic factor is stronger in B-precursor ALL; values greater than unity imply more strength in T-ALL, while the value 1.00 suggests equal strength. There are two major advantages of studying interaction as opposed to conclusions based on one being significant while another is not. First, with many more B-precursor patients, there is greater power in B-precursor ALL to detect real differences. Second, since non-significant results are inconclusive, one cannot directly conclude greater strength of association from one significant and one non-significant result. The interaction analysis circumvents both of these issues. Event-free survival curves were constructed by the method of Kaplan and Meier 36 with standard errors of Peto et al. 34 Note that no formal statistical analysis of clinical features was conducted. It is our belief that the standard for clinical significance vastly exceeds that of statistical significance for these types of comparisons. Since we have individual interest in each factor, our analyses are restricted to univariate rather than multivariate methods. Even with this large number of T-ALL patients registered on the same protocol, treatment influence must still be taken into account, since one of the two regimens fared significantly better. 12 Therefore, some of the outcome comparisons have been adjusted for treatment regimen. However, since the patients were randomized (except for very early in the pilot study) inferences about prognostic factors are valid whether or not one stratifies for treatment. We employed a cutoff of April 1997, rather than the most recent available at the time of this writing (April 1998) for two reasons. First, this is the cutoff used to produce our findings. Second, there have been no additional failures in T-ALL patients and only seven in B-precursor ALL patients in the intervening year. Results Clinical and laboratory characteristics by immunophenotype The median age for children with T-ALL was 8.2 years (quartiles 5.2 years; 12.3 years), as compared to 4.5 years (quartiles 3.0 years; 7.8 years) for those with B-precursor ALL. Those with T-ALL had a median presenting WBC of 75000/ l, (quartiles 19900/ l; / l) as compared to 8000/ l (quartiles 4000/ l; 26000/ l) for those with B-precursor ALL.

4 Table 2 shows other presenting clinical and laboratory characteristics for the patients with T-ALL, as compared to those with B-precursor ALL. Prognostic factor analyses comparisons between T-ALL and B-precursor ALL Consensus risk grouping: Using the NCI consensus risk group designations for age and WBC defined prognostic groups, (ie age 10 years and/or WBC 50000/ l confers poor prognosis), 8 outcome was compared between standard and poor risk patients with T-ALL, treated on the POG T3 regimens. Five year EFS was 56.9 ± 5.0% for the standard risk subgroup and 48.6 ± 3.0% for the poor risk subgroup (P = 0.11) (Figure 1). Although the trend was in the expected direction, the small difference was in sharp contrast to the wide difference in outcome between standard risk (5 year EFS 77.9 ± 1.2%) and poor risk (5 year EFS of 53.0 ± 2.1%) patients with B-precursor ALL, treated on the POG 8602 (ALinC 14) treatment regimens during the same time period (P 0.001) (Figure 1). When the T-ALL consensus risk group comparisons were adjusted for treatment regimen, the P value became 0.083; when adjusted for gender (see below), the P value was Overall 5 year EFS rates were 50.7 ± 2.5% for the patients with T-ALL and 70.4 ± 1.1% for those with B-precursor ALL. Age: Looking at age separately, children with B-precursor ALL who were 10 years of age at diagnosis fared decidedly better (5 year EFS 74.5 ± 1.1%) than those 10 years at diagnosis (5 year EFS 51.5 ± 2.8%) (P 0.001). However, in patients with T-ALL, the younger subgroup did not fare significantly better (5 year EFS 52.6 ± 3.2%) than the older subgroup (5 year EFS 47.5 ± 4.2%) (P = 0.15) (Figure 2). WBC: Similarly, WBC 50000/ l was considerably more useful for predicting the outcome of patients with B-precursor ALL than for those with T-ALL. Five year EFS rates for patients Table 2 Clinical and laboratory features by immunophenotype Presenting T cell B-precursor characteristics Positive/n % Positive/n % WBC 50000/ l 262/ / Age 10 years 166/ / Male gender 331/ / Race African American 64/ / Hispanic 34/ / Caucasian 320/ / CNS disease 50/ / DI / / CALLA expression 143/ / CD24 expression 37/ / t(1;19) 0/ / t(9;22) 5/ / t(4;11) 2/ / Any abnormal karyotype 201/ / Any translocation 122/ / with T-ALL and WBC below or above 50000/ l were 53.0 ± 4.0% and 49.2 ± 3.3%, respectively (P = 0.36). However, patients with B-precursor ALL and WBC 50000/ l had 74.4 ± 1.1% 5 year EFS, as compared to only 48.4 ± 3.0% for those with presenting WBC 50000/ l (P 0.001). Other breakpoints of WBC values were analyzed within T- ALL to determine whether a more definitive correlation could be found with outcome. Neither the / l WBC nor the / l WBC cutoff points were associated with significant differences in outcome for the patients with T-ALL; whereas in the B-precursor group, both of these breakpoints were significantly associated with outcome. A surprising finding was noted for the small subgroup of patients with T-ALL who had WBC 10000/ l at diagnosis (n = 67). They fared unexpectedly poorly with a 5 year EFS of only 37.2 ± 6.8%. This was significantly worse than for T- ALL patients with initial WBC 10000/ l (P = 0.024) (Figure 3). After adjustment for treatment regimen, the small subgroup of patients with WBC 10000/ l at diagnosis still fared less well than those with WBC 10000/ l, but the difference was no longer significant (P = 0.16). Gender: Although the degree of prognostic importance of both age and WBC was much less for patients with T-ALL than for those with B-precursor ALL, the association of gender with outcome was similar for both immunophenotypes, with boys faring significantly worse than girls. Figure 4 shows that boys with T-ALL have a 5 year EFS of only 46.7 ± 2.9%, as compared to 62.6 ± 4.9% for girls (P = 0.004). Likewise, within the B-precursor ALL group girls fared significantly better than boys with 5 year EFS rates of 75.8 ± 1.5% for girls and 65.9 ± 1.5% for boys (P 0.001). Race: Within patients with B-precursor ALL the 5 year EFS was 72.3 ± 1.2% for Caucasians, compared to 63.4 ± 3.7% for African American children (P = 0.008) and 55.1 ± 4.5% for Hispanic patients (P 0.001). Similarly, in T-ALL, Caucasians fared better (5 year EFS 52.8 ± 3.0%), compared to either African American patients (5 year EFS 48.3 ± 6.7%) (P = 0.41) or to Hispanic children (5 year EFS 41.2%) (P = 0.10). CNS disease at diagnosis: As shown in Table 2, the T-cell ALL patients were more likely to have CNS disease at diagnosis (11.5%) when compared to the patients with B-precursor ALL (2.8%). While B-precursor ALL patients with CNS disease at diagnosis had a significantly inferior 5 year EFS (59.0 ± 6.9%), compared to those without CNS involvement at diagnosis (70.7 ± 1.1%) (P = 0.049). There was no evidence of a similar trend in the T-ALL group, however, with 5 year EFS values of 52.0 ± 7.4% for those with CNS disease and 50.7 ± 2.7% for those without CNS disease at diagnosis (P = 0.96). Ploidy: While patients with B-precursor ALL and a leukemic cell DNA Index 1.16 have a superior prognosis, hyperdiploidy is extremely rare in T-ALL (Table 2). Only seven T- ALL patients had a DI 1.16, and five of these were neartetraploid (DI 1.80). If second leukemic lines are also considered, 17 T-ALL patients had a clone with DI 1.16, including 15 (88%) with DI Conversely, of the 335 B- precursor ALL patients demonstrating a leukemic line 1.16, 1699

5 1700 Figure 1 Event-free survival for NCl consensus (age/wbc) risk groups for B-precursor ALL and for T-ALL. (P for B-precursor ALL.) (P = 0.11 for T-ALL (adjusted for treatment regimen P = 0.083; adjusted for gender P = 0.13).) Figure 2 Event-free survival, relationship to age, for B-precursor ALL and for T-ALL. (P for B-precursor ALL and P = 0.15 for T-ALL).

6 1701 Figure 3 Event-free survival for patients with T-ALL, relationship to WBC subgroups. P = for WBC 10000/ l vs 10000/ l (adjusted for treatment regimen, P = 0.16 or for gender, P = 0.010). Figure 4 Event-free survival, relationship to gender, for B-precursor ALL (P 0.001) and for T-ALL (P = 0.004).

7 1702 only 20 (6%) had a clone with DI Table 3 shows the incidence of cytogenetically determined ploidy subgroups for the B-precursor and T-ALL groups, respectively, as well as 5 year EFS rates. Whereas B-precursor ALL patients with 52 chromosomes in the leukemic cells (approximately equivalent to DI 1.16) fare by far the best of any ploidy subgroup, this is not true within T-ALL, in which the high ploidy group consisted predominately of near tetraploid cases. Translocations: Patients with T-ALL were less likely than those with B-precursor ALL to have karyotypes with demonstrated abnormalities. However, for those patients with demonstrable abnormalities in the leukemic cell karyotype, T- ALL patients were more likely to have translocations (59.3% for T-ALL vs 39.4% for B-precursor ALL). While leukemic cell translocations other than the cryptic t(12;21) are significantly associated with inferior outcome in patients with B-precursor ALL, this association was not seen at all in our patients with T-ALL. In T-ALL, the 5 year EFS for patients with translocations in the blast cell karyotypes was 46.0 ± 5.0%, compared to 40.4 ± 6.0% in those lacking translocations (P = 0.52). In B- precursor ALL, those lacking translocations had a 74.4 ± 1.9% 5 year EFS, compared to 63.1 ± 2.5% if a translocation was present (P 0.001). It should be noted that only patients with demonstrable leukemic cell cytogenetic abnormalities were included in the cytogenetic analyses, since patients with either unsatisfactory or normal leukemic cell karyotypes were considered to have non-informative results. Also, patients with the t(9;22)(q34;q11) or the t(4;11)(q21;q23) were excluded from the analyses because of the known extremely poor prognosis associated with these specific translocations for patients with B-precursor ALL. CALLA expression: As a predictor of favorable outcome, CALLA expression was more important in B-precursor than in T-cell ALL. Five year EFS comparisons of CALLA expression vs lack of CALLA expression (infants 12 months of age excluded) show 53.6 ± 4.5% vs 48.4 ± 3.3% for patients with T-ALL (P = 0.15), compared with 71.0 ± 1.2% vs 50.5 ± 5.5% for those with B-precursor ALL (P 0.001). CD24 expression: CD24 expression was inconclusive for patients with T-ALL (5 year EFS 54.1 ± 8.4% for the CD24- positive subgroup vs 50.5 ± 2.8% for the CD24-negative subgroup (P = 0.57)). However, it was significantly prognostic for patients with B-precursor ALL (5 year EFS 70.2 ± 1.2% for the Table 3 Outcome by chromosome number a B-precursor T-ALL n 5 year s.e. n 5 year s.e. EFS % % EFS % % 52 chrm chrm chrm chrm a Only patients with abnormal cytogenetics are included. chrm, chromosomes. CD24-positive subgroup vs 51.4 ± 7.8% for the CD24- negative subgroup (P 0.001)). Interactions between phenotype and prognostic factors: Table 4 investigates the relative strength of the prognostic importance of features in T-ALL vs B-precursor ALL. The P values are defined for the test of equal hazard ratios (T-ALL vs B-precursor). The estimated hazard ratio (favorable:unfavorable) was significantly lower (more significant) in several features for B-precursor ALL (consensus risk, WBC, age, translocation and CALLA). There were no variables in Table 4 where prognostic importance was significantly greater in T-ALL. T cell ALL maturational stage relationship to outcome: Using POG s modified Ludwig definitions 33 for T- ALL maturational stage, 32.5% of patients with T-ALL were placed in the early, 43.9% in the intermediate, and 23.6% in the mature subgroups. Our results show a significant difference in outcome (P = 0.033) among patients with early, intermediate, and mature maturational stages of T-ALL (Figure 5). Patients with intermediate stage T-ALL fare the best (5 year EFS 55.2 ± 4.1%), and patients with early stage T-ALL fare better ( %) than those with mature maturational stage disease (37.8 ± 5.3%). The maturational stage prognostic significance holds when stratified for treatment regimen (P = 0.022, 3-way) and for gender (P = 0.016, 3-way). CD2 expression was not useful as a prognostic factor, with 50.0 ± 2.9% 5 year EFS for those patients with CD2-positive T-ALL, as compared to 47.2 ± 7.3% for those with CD2- negative T-ALL (P = 0.71). Using a 30%, instead of the 20% cut-off, to define CD2 expression gave a similar result. Discussion There is a great deal of evidence that the major immunophenotypes of ALL (B, B-precursor and T) have innate clinical, laboratory and pharmacologic differences. Most groups now consider that patients with sig-positive, - or -positive, B cell ALL (B-ALL) should be treated separately from those with non- B-ALL. Indeed, the intrinsic pharmacologic differences in responsiveness demand therapeutic regimens for patients with B-ALL or advanced stage B-NHL that are different from those used for B-precursor ALL. There is much less universal agreement, however, as to whether patients with T-ALL should be categorized and managed differently from those with B-precursor ALL. The differences in clinical features at presentation between patients with T-ALL and B-precursor ALL are well recognized. For example, the T-ALL patients tend to be older and to have a higher male:female ratio, higher WBC, higher incidence of mediastinal mass, and increased frequency of CNS disease. 2,3,7,22 Analysis of presenting characteristics of the patients in the current report confirms the reproducibility of these well established clinical features of T cell ALL. A few papers have addressed possible prognostic factors among patients with T-ALL, 9,37 39 including those reporting the previous POG experience. 16,22 The current report analyzes outcome with respect to certain possible risk factors for 441 eligible patients with T-ALL, all treated on the same POG protocol for T cell leukemia. The paper addresses whether the risk factors, which proved significantly prognostic for B-pre-

8 Table 4 Favorable Analysis of interaction between phenotype and prognostic factor with respect to event-free survival Factor Estimated quotient a 95% confidence P value of hazard ratios limits (2-sided) Unfavorable 1703 Consensus good Consensus poor , WBC 50.0 WBC , Age Age , Female Male , Caucasian African American , Caucasian Hispanic , CNS CNS , No translocation b Translocation b , CALLA+ CALLA , CD24 + CD , a Estimated hazard ratios favorable:unfavorable. B-precursor:T cell. A value of 1.00 would suggest equivalent prognostic magnitude. Values below 1.00 imply stronger prognostic magnitude in B-precursor ALL. b 4;11 and 9;22 excluded: restricted to patients with abnormal clone. Figure 5 Event-free survival for POG patients with T-ALL, according to the modified Ludwig maturational staging system. P = (three way); when adjusted for treatment regimen, P = and for gender, P = (The 59 patients not shown on these curves had technically unsatisfactory phenotyping or had an inadequate volume of sample to permit testing for the entire battery of antigens.) cursor ALL patients treated during the same period of time on a different protocol, are of lesser or greater importance for T-ALL patients. Our data show that several of the clinical and laboratory prognostic factors, which are used commonly and reliably for B-precursor ALL, are much less predictive in T-ALL. Our study demonstrates that the B-precursor ALL mainstay prognostic factors of age and of WBC, respectively, are significantly less important in T-ALL risk group assignment. Moreover, when both age and WBC were taken into account in the NCI consensus risk group designations, the resulting stratification was significantly less predictive in T-ALL than in B-precursor ALL, even after adjustment for treatment regimen. Thus, the most widely used tool for assigning prognosis in B-precursor ALL may not be useful for children with T-ALL. Our analysis of age as a prognostic factor in the preceding POG T2 study showed that children 5 years or 7 years of age with T-ALL fared significantly better in univariate, but not in multivariate analyses. 16 Crist et al, 40 reporting a still earlier group of POG T-ALL patients, noted that patients 11 years of age at diagnosis fared significantly worse. Pui et al 37 found age 15 years to confer an increased risk of treatment failure for pediatric patients with T-ALL, although Arico et al 9 found no significant difference in outcome for T-ALL patients 10

9 1704 years, as compared to younger patients, in the AIEOP report. Our current data show that age 10 years at diagnosis is of significantly less prognostic importance for patients with T-ALL, than for those with B-precursor ALL. The eligibility of the POG ALL protocols (both B-precursor and T) reported here included patients up to years of age. Even though some pediatric groups ALL treatment protocols do not include patients 18 to 21 years of age, this does not appear to affect significantly comparison of treatment results. Our trials included only nine patients with T-ALL and 13 patients with B-precursor ALL who were 18 to 21 years of age at diagnosis, and the basic conclusions are unaltered if these patients are excluded from the analysis. The finding that WBC 50000/ l, / l or / l was an inconclusive prognostic indicator for our patients with T-ALL on the T3 protocol may be partially due to the fact that the small (15%) subgroup of patients with WBC 10000/ l at diagnosis fared surprisingly poorly, even worse than those with WBC 50000/ l at diagnosis. We looked closely at this small subset, but were unable to identify any associated detrimental clinical or laboratory characteristics. It should be noted that POG s results with the earlier T1 22 and T2 16,23 treatment protocols for T-ALL showed WBC 50000/ l to be significantly associated with inferior outcome. Also, Garand et al 38 and Arico et al 9 found WBC to have prognostic significance for pediatric T-ALL patients. CCG found WBC to be of univariate, but not multivariate, significance in patients with T-ALL. 39 Pui et al 37 found presenting WBC to be of no prognostic significance for 120 consecutive T-ALL patients at SJCRH. The primary point of the current T3 WBC analysis is to show that WBC 50000/ l or / l or / l is of significantly less prognostic import for our patients with T-ALL than for those with B-precursor ALL. As discussed recently by Amylon et al, 12 patients with T- ALL and 50000/ l WBC at diagnosis showed proportionately more improvement in EFS between the POG T2 and T3 protocols, than did those with 50000/ l at diagnosis. Whereas all patients received cranial radiation as part of CNS preventive treatment on the T2 protocol, those with 50000/ l WBC did not receive cranial radiation on the T3 protocol. It is possible that this contributed to the fact that WBC is not a prognostic indicator in the T3 results. Although both age and WBC had significantly stronger prognostic import within B-precursor ALL than within T-ALL, it is of interest that no significant interaction was demonstrated between phenotypes as to either race or gender. Boys fared better than girls and Caucasians better than either African American or Hispanic children, regardless of immunophenotype. There are too few patients with CNS disease at diagnosis in either the B-precursor or the T-ALL groups to draw meaningful comparisons. Nevertheless, it is of interest that, although the T-ALL patients were more likely to have CNS disease at diagnosis, the T-ALL patients with CNS disease at diagnosis did not have shorter EFS than the other T-ALL patients. Conversely, B- precursor ALL patients with CNS disease at diagnosis fared worse than those without CNS disease. Concerning leukemic cell laboratory characteristics, the presence of translocations and the lack of CALLA expression were significantly stronger predictors of poor outcome for patients with B-precursor ALL than for those with T-ALL. Differing results have been reported previously as to whether CALLA expression is favorable in T-ALL. 14,37,39,41,42 The favorable cytogenetic subgroup with 52 but 74 chromosomes (1.16 DI 1.60) that comprises about 20% of B-precursor ALL cases is rarely identified in T cell ALL. As we have noted previously, 43 hyperdiploid T-ALL cases are generally neartetraploid, and our results do not suggest that these patients have a more favorable prognosis than those with near-diploid blast cell karyotypes (Table 2). Since some of the most dependable factors for predicting risk for patients with B-precursor ALL are of limited value for patients with T-ALL, we have been interested in identifying other biologic features of the leukemic cells that might be predictive of prognosis for T-cell ALL. One of these is the maturational stage of the T lymphoblasts. POG s earliest attempt to correlate T-ALL outcome with maturational stage was reported by Crist et al 44 in 1988, and suggested that the intermediate stage, as then defined, fared best, although the number of patients was small and significance was not demonstrated. The POG thereafter attempted to categorize its patients with T-ALL according to several different models of maturational staging, but some patients failed to fit into any one of the defined categories. However, after Ludwig et al 33 described their staging system and showed a significant difference in outcome among early, intermediate, and mature subgroups, we decided to employ a modified Ludwig system for our patients. Using this modified Ludwig maturational staging system, we were able to categorize every patient with T-ALL who had satisfactory immunophenotyping. The distribution for our patients was 32.5%, 43.9% and 23.6% in the early (E), intermediate (I), and mature (M) subgroups, respectively, as compared to 39% (E), 48% (I) and 11% (M) for Ludwig s 93 patients. It should be noted that Ludwig s subgroupings are unquestionably more precise than ours, since he utilized cytoplasmic CD3 as part of the system. However, the Ludwig system, modified by POG for use without cytoplasmic CD3, also appears to define a prognostic separation by maturational stage. Similar to Ludwig s results, the POG outcome data show that the intermediate subgroup (also called the cortical group, per the designation of the European Group for the Immunological Classification of Acute Leukemias (EGIL) 45 ) has a superior outcome. However, our mature group fared worst, while Ludwig s early group fared worst. The Ludwig system is based on the expression of CD1 and CD3, in CD7 and cy CD3-positive patients. 33 Interestingly, Pui et al s 37 results showed that SJCRH patients with leukemic cell expression of surface CD3, corresponding to a mature stage phenotype, had a significantly poorer treatment outcome than did cases with blasts lacking the CD3 antigen. This is similar to our results showing adverse outcome for the mature (CD1- negative, CD3-positive) subgroup. CCG has described prognostic value for CD2 antigen expression in their patients with T-ALL. 39 However, our data with the T3 regimen did not show prognostic value for CD2. CCG has also recently reported prognostic significance for a T-ALL maturation staging defined differently from the Ludwig system. 46 The CCG definitions subgrouped 407 patients, resulting in 8.6% (E), 60.9% (I) and 30.5% (M), with outcome best for the mature and worst for the early group. However, as pointed out in their paper, CCG s early subgroup was limited to patients with CD7 as the only T-lineage antigen expressed, whereas with Ludwig s system and POG s modification of the Ludwig system, the definitions allow for the possibilities of CD2 or CD5 expression in the early subgroup. However, because the CCG also did not assay cytoplasmic CD3, it is possible that a few of their CD7 only positive patients actually had undifferentiated (M0) AML. While this is also true of our patients, our broader definition of early T-ALL

10 makes it likely that we have fewer M0 patients contaminating this patient population. Recently, Niehues et al 14 reported results of the COALL studies in pediatric patients with T-ALL. Their data suggest that T-ALL blasts with the phenotype of cortical thymocytes (CD1 positive and/or CD4 positive, CD8 positive) have a thymic selection-related (SR) phenotype, which may correlate with an increased susceptibility to apoptosis. This SR subgroup (comparable to the BFM and our intermediate subgroups), had a superior outcome, when compared to the other COALL T-ALL patients. Some groups, including the AIEOP, BFM, and CCG, 9,11,47 are now putting major emphasis on early response to therapy as the best predictor of eventual outcome for either the T-ALL or the B-precursor immunophenotypes. The BFM group has recently reported that peripheral blood blast count response, after a 7-day prednisone pre-phase, is even more important than maturational stage in determining prognosis in T-ALL. 42 Rapidity of response data were not routinely collected for the POG T3 protocol. POG is currently accumulating data concerning early response in its open B-precursor and T-ALL protocols. In summary, the prognostic factors that have proven most reliable in risk group assignment for patients with B-precursor ALL are much less useful for patients with T-ALL. T-ALL has innate differences from B-precursor ALL, not only in its characteristics at the time of presentation, but also in the usefulness of individual prognostic features. Thus, using traditional risk group criteria to stratify T-ALL patients for therapy may not be appropriate; one should not blindly apply stratification criteria, useful for patients with B-precursor ALL, to those with T-ALL. Lastly, our data concerning maturational stage lend further support to evidence that the intermediate or cortical maturational stage of T-ALL fares best. Acknowledgements We thank Jackie Geeslin, CCRA, for her assistance with manuscript preparation. This work was supported in part by the following grants from the National Cancer Institute: CA-15989, CA-29139, CA-33603, CA-28476, CA-25408, CA-31566, CA References 1 Sallan SE, Ritz J, Pesando J, Gelber R, O Brien C, Hitchcock S, Coral F, Schlossman SF. Cell surface antigens: Prognostic implications in childhood acute lymphoblastic leukemia. Blood 1980; 55: Greaves MF, Janossy G, Peto J, Humphrey K. Immunologically defined subclasses of acute lymphoblastic leukemia in children: their relationship to presentation features and prognosis. Br J Haematol 1981; 48: Pullen DJ, Boyett JM, Crist WM, Falletta JM, Roper M, Dowell B, van Eys J, Jackson JF, Humphrey GB, Metzgar RS, Cooper MD. Pediatric Oncology Group utilization of immunologic markers in the designation of acute lymphocytic leukemia subgroups: influence on treatment response. Ann NY Acad Sci 1984; 428: Crist WM, Grossi CE, Pullen DJ, Cooper MD. Immunologic markers in childhood acute lymphocytic leukemia. Semin Oncol 1985; 12: Rivera GK, Raimondi SC, Hancock ML, Behm FG, Pui CH, Abromowitch M, Mirro J Jr, Ochs JS, Look AT, Williams DL, Murphy SB, Dahl GV, Kalwinski DK, Evans WE, Kun LE, Simone JV, Crist WM. Improved outcome in childhood acute lymphoblastic leukemia with reinforced early treatment and rotational combination chemotherapy. Lancet 1991; 337: Crist W, Shuster J, Look T, Borowitz M, Behm F, Bowman P, Frankel L, Pullen J, Krance R, Steuber P, Camitta B, Amylon M, Link M, Land V. Current results of studies of immunophenotype-, age-, and leukocyte-based therapy for children with acute lymphoblastic leukemia. A Pediatric Oncology Group study. Leukemia 1992; 6: Pui C-H, Behm FG, Crist WM. Clinical and biologic relevance of immunologic marker studies in childhood acute lymphoblastic leukemia. Blood 1993; 82: Smith M, Arthur D, Camitta B, Carroll AJ, Crist W, Gaynon P, Gelber R, Heerema N, Korn EL, Link M, Murphy S, Pui C-H, Pullen J, Reaman G, Sallan SE, Sather H, Shuster J, Simon R, Trigg M, Tubergen D, Uckun F, Ungerleider R. Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. J Clin Oncol 1996; 14: Aricò M, Basso G, Mandelli F, Rizzari C, Colella R, Barisone E, Zanesco L, Rondelli R, Pession A, Masera G for the Associazione Italiana Ematologia Oncologia Pediatrica (AIEOP). Good steroid response in vivo predicts a favorable outcome in children with T- cell acute lymphoblastic leukemia. Cancer 1995; 75: Schorin MA, Blattner S, Gelber RD, Tarbell NJ, Donnelly M, Dalton V, Cohen HJ, Sallan SE. Treatment of childhood acute lymphoblastic leukemia: results of Dana-Farber Cancer Institute Children s Hospital Acute Lymphoblastic Leukemia Consortium Protocol J Clin Oncol 1994; 12: Reiter A, Schrappe M, Ludwig W-D, Hiddemann W, Sauter S, Henze G, Zimmermann M, Lampert F, Havers W, Niethammer D, Odenwald E, Ritter J, Mann G, Welte K, Gadner H, Riehm H. Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients. Results and conclusions of the multicenter trial ALL-BFM 86. Blood 1994; 84: Amylon MD, Shuster J, Pullen J, Berard C, Link MP, Wharam M, Katz J, Yu A, Laver J, Ravindranath Y, Kurtzberg J, Desai S, Camitta B, Murphy SB. Intensive high dose asparaginase consolidation improves survival for pediatric patients with T-cell acute lymphoblastic leukemia and advanced stage lymphoblastic lymphoma: a Pediatric Oncology Group study. Leukemia 1999; 13: Uckun FM, Sensel MG, Sun L, Steinherz PG, Trigg ME, Heerema NA, Sather HN, Reaman GH, Gaynon PS. Biology and treatment of childhood T-lineage acute lymphoblastic leukemia. Blood 1998; 91: Niehues T, Kapaun P, Harms DO, Burdach S, Kramm C, Körholz D, Janka-Schaub G, Göbel U. A classification based on T-cell selection-related phenotypes identifies a subgroup of childhood T-ALL with favorable outcome in the COALL studies. Leukemia 1999; 13: Clavell LA, Gelber RD, Cohen HJ, Hitchcock-Bryan S, Cassady JR, Tarbell NJ, Blattner SR, Tantravahi R, Leavitt P, Sallan SE. Four agent induction and intensive asparaginase therapy for treatment of childhood acute lymphoblastic leukemia. New Engl J Med 1986; 315: Shuster JJ, Falletta JM, Pullen J, Crist WM, Humphrey GB, Dowell BL, Wharam MD, Borowitz M. Prognostic factors in childhood T- cell acute lymphoblastic leukemia: a Pediatric Oncology Group study. Blood 1990; 75: Pui C-H. Childhood leukemias. New Engl J Med 1995; 332: Uckun FM, Reaman G, Steinherz PG, Arthur DC, Sather H, Trigg M, Tubergen D, Gaynon P. Improved clinical outcome for children with T-lineage acute lymphoblastic leukemia after contemporary chemotherapy: a Children s Cancer Group study. Leuk Lymphoma 1996; 24: Chessels JM, Bailey C, Richards SM. Intensification of treatment and survival in all children with lymphoblastic leukaemia: results of UK Medical Research Council trial UKALL X. Medical Resarch Council Working Party on Childhood Leukemia. Lancet 1995; 345: Gaynon PS, Bleyer WA, Steinherz PG, Finklestein JZ, Littman PS, Miller DR, Reaman GH, Sather HN, Hammond GD. Modified BFM therapy for children with previously untreated acute lymphoblastic leukemia and unfavorable prognostic features. Report 1705

11 1706 of Children s Cancer Study Group study CCG-193P. Am J Ped Hematol/Oncol 1988; 10: Cherlow JM, Steinherz PG, Sather HN, Gaynon PS, Grossman NJ, Kersey JH, Johnstone HS, Breneman JC, Trigg ME, Hammond GD. The role of radiation therapy in the treatment of acute lymphoblastic leukemia with lymphomatous presentation: a report from the Children s Cancer Group. Int J Radiat Oncol Biol Phys 1993; 27: Pullen DJ, Sullivan MP, Falletta JM, Boyett JM, Humphrey GB, Starling KA, Land VJ, Dyment PG, Vats T, Duncan MH. Modified LSA2-L2 treatment in 53 children with E-rosette positive T-cell leukemia: results and prognostic factors. (a Pediatric Oncology Group study.) Blood 1982; 60: Falletta JM, Shuster JJ, Crist WM, Pullen DJ, Borowitz MJ, Wharam M, Patterson R, Foreman E, Vietti TJ. Different patterns of relapse associated with three intensive treatment regimens for pediatric E- rosette positive T-cell leukemia: a Pediatric Oncology Group study. Leukemia 1992; 6: Crist W, Boyett J, Pullen J, van Eys J, Vietti T. Clinical and biologic features predict poor prognosis in acute lymphoid leukemias in children and adolescents: a Pediatric Oncology Group review. Med Pediatr Oncol 1986; 14: Ludwig W-D, Teichmann JV, Sperling C, Komischke B, Ritter J, Reiter A, Odenwald E, Sauter S, Riehm H. Incidence, clinical characteristics, and prognostic significance of immunological subtypes in childhood acute lymphoblastic leukemia: experience of therapy studies ALL-BFM 83 and 86. Klin Pädiatr 1990; 202: Trueworthy R, Shuster J, Look T, Crist W, Borowitz M, Carroll A, Frankel L, Harris M, Wagner H, Haggard M, Mosijczuk A, Pullen J, Steuber P, Land V. Ploidy of lymphoblasts is the strongest predictor of treatment outcome in B-progenitor cell acute lymphoblastic leukemia of childhood: a Pediatric Oncology Group study. J Clin Oncol 1992; 10: Borowitz MJ, Carroll AJ, Shuster JJ, Look AT, Behm FG, Pullen DJ, Land VJ, Steuber P, Crist WM. Use of clinical and laboratory features to define prognostic subgroups in B-precursor acute lymphoblastic leukemia: experience of the Pediatric Oncology Group. In: WD Ludwig, E Thiel (eds). Recent Results in Cancer Research, vol Springer-Verlag: Berlin, Heidelberg, 1993, pp Crist WM, Carroll AJ, Shuster JJ, Behm FG, Whitehead M, Vietti TJ, Look AT, Mahoney D, Ragab A, Pullen DJ, Land VJ. Poor prognosis of children with pre-b acute lymphoblastic leukemia is associated with the t(1;19)(q23;p13). A Pediatric Oncology Group study. Blood 1990; 76: Koehler M, Behm FG, Shuster J, Crist W, Borowitz M, Look AT, Head D, Carroll AJ, Land V, Steuber P, Pullen DJ. Transitional pre- B-cell acute lymphoblastic leukemia of childhood is associated with favorable prognostic clinical features and an excellent outcome: a Pediatric Oncology Group study. Leukemia 1993; 7: Bowman WP, Shuster JJ, Cook B, Griffin T, Behm F, Pullen J, Link M, Head D, Carroll A, Berard C, Murphy S. Improved survival for children with B-cell acute lymphoblastic leukemia and stage IV small noncleaved-cell lymphoma: a Pediatric Oncology Group study. J Clin Oncol 1996; 14: ISCN (1985). An International System for Human Cytogenetic Nomenclature. Harnden DG, Klinger HP (eds). S Karger: Basel, ISCN (1991). Guidelines for Cancer Cytogenetics, Supplement to an International System for Human Cytogenetic Nomenclature. Mitelman F (ed). S Karger: Basel, Ludwig W-D, Harbott J, Bartram CR, Komischke B, Sperling C, Teichmann JV, Seibt-Jung H, Notter M, Odenwald E, Nehmer A, Thiel E, Riehm H. Incidence and prognostic significance of immunophenotypic subgroups in childhood acute lymphoblastic leukemia: experience of the BFM study 86. In: WD Ludwig, E Thiel (eds). Recent Results in Cancer Research, vol Springer- Verlag: Berlin, Heidelberg, 1993, pp Peto R, Armitage P, Brewlow NE, Cox DR, Howard SV, Mantel N, McPherson K, Peto J, Smith PG. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br J Cancer 1977; 35: Cox DR. Regression models and life tables. J R Stat Soc [B] 1972; 34: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: Pui C-H, Behm FG, Singh B, Schell MJ, Williams DL, Rivera GK, Kalwinsky DK, Sandlund JT, Crist WM, Raimondi SC. Heterogeneity of presenting features and their relation to treatment outcome in 120 children with T-cell acute lymphoblastic leukemia. Blood 1990; 75: Garand R, Vannier JP, Béné MC, Faure G, Favre M, Bernard A and the Groupe d Etude Immunologique des Leucémies. Comparison of outcome, clinical, laboratory, and immunological features in 164 children and adults with T-ALL. Leukemia 1990; 4: Uckun FM, Steinherz PG, Sather H, Trigg M, Arthur D, Tubergen D, Gaynon P, Reaman G. CD2 antigen expression on leukemic cells as a predictor of event-free survival after chemotherapy for T- lineage acute lymphoblastic leukemia: a Children s Cancer Group study. Blood 1996; 88: Crist W, Pullen J, Boyett J, Falletta J, van Eys J, Borowitz M, Jackson J, Dowell B, Russell C, Quddus F, Ragab A, Vietti T. Acute lymphoid leukemia in adolescents: clinical and biologic features predict a poor prognosis a Pediatric Oncology Group study. J Clin Oncol 1988; 6: Dowell BL, Borowitz MJ, Boyett JM, Pullen DJ, Crist WM, Quddus FF, Russell EC, Falletta JM, Metzgar RS. Immunologic and clinicopathologic features of common acute lymphoblastic leukemia antigen-positive childhood T-cell leukemia. A Pediatric Oncology Group study. Cancer 1987; 59: Schrappe M, Sperling C, Reiter A, Ratei R, Zimmerman M, Goetz N, Riehm H, Ludwig W-D. In vivo sensitivity of childhood T-cell ALL is the most relevant prognostic marker independent of maturational stage. Blood 1998; 92: 397a. 43 Pui C-H, Carroll AJ, Head D, Raimondi SC, Shuster JJ, Crist WM, Link MP, Borowitz MJ, Behm FG, Land VJ, Nash MB, Pullen DJ, Look AT. Near-triploid and near-tetraploid acute lymphoblastic leukemia of childhood. Blood 1990; 76: Crist WM, Shuster JJ, Falletta J, Pullen DJ, Berard CW, Vietti TJ, Alvarado CS, Roper MA, Prasthofer E, Grossi CE. Clinical features and outcome in childhood T-cell leukemia lymphoma according to stage of thymocyte differentiation: a Pediatric Oncology Group study. Blood 1988; 72: European Group for the Immunological Characterization of Leukemias (EGIL): Bené MC, Castoldi G, Knapp W, Ludwig WD, Matutes E, Orfao A, van t Veer MB. Proposals for the immunological classification of acute leukemias. Leukemia 1995; 9: Uckun FM, Gaynon PS, Sensel MG, Nachman J, Trigg ME, Steinherz PG, Hutchinson R, Bostrom BC, Sather HN, Reaman GH. Clinical features and treatment outcome of childhood T-lineage acute lymphoblastic leukemia according to the apparent maturational stage of T-lineage leukemic blasts: a Children s Cancer Group study. J Clin Oncol 1997; 15: Steinherz PG, Gaynon PS, Breneman JC, Cherlow JM, Grossman NJ, Kersey JH, Johnstone HS, Sather HN, Trigg ME, Chappell R, Hammond D, Bleyer WA. Cytoreduction and prognosis in acute lymphoblastic leukemia the importance of early marrow response: report from the Children s Cancer Group. J Clin Oncol 1996; 14: Appendix Institution Alberta Children s Hospital All Children s Hospital Baylor Boston Floating Hospital Cancer Center of Hawaii Carolinas Medical Center Children s Hospital (San Diego) Grant number CA CA-69177