by S.M. Ng, H.P. Lin, W.A. Arif n, A.K. Zainab, S.K. Lam and L.L. Chan Department of Paediatrics, University Hospital, Kuala Lumpur 59100, Malaysia

Age, Sex, Haemoglobin Level, and White Cell Count at Diagnosis are Important Prognostic Factors in Children with Acute Lymphoblastic Leukemia Treated with BFM-type Protocol by S.M. Ng, H.P. Lin, W.A. Arif n, A.K. Zainab, S.K. Lam and L.L. Chan Department of Paediatrics, University Hospital, Kuala Lumpur 59100, Malaysia Summary The presenting features and treatment outcome for 575 Malaysian children (# 12 years of age) with newly diagnosed acute lymphoblastic leukemia (ALL), admitted to the University Hospital, Kuala Lumpur, Malaysia between 1 January 1980 and 30 May 1995 were evaluated to determine their prognostic signi cance. Two-year overall survival was achieved in 67 per cent of all patients and 55 per cent of patients were relapse-free at 2 years. All except 10 patients, with identi ed French± American±British L3 morphology were treated with the modi ed Berlin±Frankfurt±Munster 78 treatment protocol. Univariate analyses of failure rate conferred age, sex, white cell count and hemoglobin level as potentially signi cant prognostic factors. All four presenting features retained their prognostic strength in a multivariate analysis. Race, platelet count, morphological subtype, liver/spleen size, lymphadenopathy, central nervous system and mediastinal mass involvement did not show any signi cant effect on treatment outcome. The 2-year survival rate was signi cantly different with regard to age, white cell count and hemoglobin level. However, sex was not signi cantly related to overall survival. These prognostic factors may have implications on future strati cation of risk-adjusted initial treatment in the management of childhood ALL. Our analysis of Malaysian children is similar to what could be predicted based on previous studies in other populations. Introduction Acute lymphoblastic leukemia (ALL) is the most common malignant disorder in childhood. It results from the clonal proliferation by successive division of lymphoid progenitor cells with arrested maturation. 1 The accumulation of these abnormal cells results in the replacement of the normal hemopoietic precursor cells of the bone marrow leading to bone marrow failure. With current treatment, up to 80 per cent of children with ALL are expected to remain in continuous complete remission. 3 Various studies have shown that prognosis of childhood ALL is dependent on age, sex, race, immunophenotype, morphology, leucocyte count, central nervous system (CNS), and mediastinal mass involvement at presentation. 2±6 However, such studies of prognostic factors in childhood ALL have also yielded controversial results and are widely debated. Prognostic factors should allow the separation and identi cation of risk groups at diagnosis for more effective initial and less toxic treatment of childhood ALL. Herein, we report the results of a retrospective Correspondence: Dr Ng Sze May, 83, Ashby, High Street, Drayton, Oxon OX14 4JW, UK. E-mail <ngszemay@hotmail. com>. study of presenting features and treatment outcome of 575 Malaysian children with newly diagnosed ALL who were admitted to University Hospital, Kuala Lumpur (UHKL), Malaysia. Patients and Methods Patients The study sample included 601 paediatric patients less than 12 years of age, with newly diagnosed ALL admitted between 1 January 1980 and 31 May 1995 to UHKL, one of four major paediatric cancer centres in Malaysia. Morphological and cytochemical features of leukemic cells of standard criteria were used as the basis for diagnosis. However, immunophenotyping was also used in 228 (40 per cent) patients for diagnosis when available. All patients, except 10 (4 per cent), with identi ed French±American±British (FAB) L3 morphological subtype were treated with the modi ed Berlin±Frankfurt±Munster (BFM) 78 treatment protocol. Patients with FAB L3 morphology were treated with the modi ed ALL-BFM 81/86 treatment protocol and were excluded from the analysis due to their different biology and treatment needs. Patients were analysed for sex, age, race, hemoglobin (Hb) level, white cell count (WCC), 338 q Oxford University Press 2000 Journal of Tropical Pediatrics Vol. 46 December 2000

platelet count, morphological subtype, hepatomegaly, splenomegaly, lymphadenopathy, CNS and mediastinal mass involvement at presentation. Standard cutpoints used to categorize high/low values for WCC, Hb, platelet count, liver and spleen size are shown in Table 1. Classi cation of morphological subtype was based on the FAB classi cation system. Sixteen (3 per cent) patients were determined to be ineligible and excluded from the study due to insuf cient data, inadequate follow-up or protocol violation. Karyotyping and immunophenotyping could only be performed in a minority of patients (38 per cent), therefore their ndings were not evaluated. Complete remission was de ned by < 5 per cent lymphoblast cells in the marrow, the absence of leukemic blast cells in the blood and cerebrospinal (CSF) uid, normal peripheral blood count and absence of localized disease. 6;8 Relapse was de ned as the recurrence of > 25 per cent blast cells in the marrow or the presence of localized leukemic in ltrates at any site after completion of induction therapy. 8 Treatment There were no risk strati cation among most patients. All patients except those identi ed with FAB L3 morphology were subject to the following treatment regime: induction therapy A was based on intrathecal (IT) methotrexate (MTX) (day 1, age-related dose), vincristine (VCR) (1.5 mg/m 2 /week 4), daunorubicin (DAUNO) (30 mg/m 2 /week 4) and prednisolone (PRED) (60 mg/ m 2 /day, tapered on week 4 to stop by week 5). From week 4 to week 6, patients received L-asparaginase (ASP) (10 000 units/m 2, three times/week for 3 weeks). Bone marrow samples were examined after completion of induction therapy A. Induction therapy B, beginning at week 7, consisted of TABLE 1 Distribution and in uence of presenting features on treament outcome Features Category No. of patients No. of failures (%) p value Age (years) $1 552 239 (43) 0.0001 < 1 23 18 (78) Sex Female 249 90 (36) 0.0003 Male 326 167 (51) Race Malay 211 103 (49) 0.48 Chinese 297 126 (42) Indian 52 21 (41) Other 15 7 (47) FAB morphology L1 170 73 (43) 0.27 L2 113 59 (52) WCC ( 10 9 /l) < 50 441 179 (41) 0.003 $50 126 70 (56) Hb level (g/dl) < 11 481 199 (41) 0.01 $11 72 41 (57) Platelet count ( 10 9 /l) < 100 435 191 (44) 0.86 $100 114 47 (41) Liver size (cm) a Normal 73 36 (49) 0.38 < 5 309 130 (42) $5 193 91 (47) Spleen size (cm) a Normal 169 76 (45) 0.28 < 5 255 106 (42) $5 151 75 (50) CNS involvement Absent 536 235 (44) 0.32 Present 12 7 (58) Mediastinal mass involvement Absent 508 224 (44) 0.68 Present 60 30 (50) Lymphadenopathy Absent 472 213 (45) 0.44 Present 99 44 (44) Data not available for some patients. a Edge palpable below costal margins. FAB, French±American±British; WCC, white cell count; Hb, hemoglobin; CNS, central nervous system. Journal of Tropical Pediatrics Vol. 46 December 2000 339

cyclophosphamide (CYCLO: 1000 mg/m 2, two doses, 4 weeks apart), cytosine arabinoside (ARA-C: 75 mg/m 2 four times/week for 4 weeks) and IT MTX (weekly, on age-related dose). CNS prophylaxis by cranial irradiation was administered with 1800 rad in 10 fractions for patients $2 years old. Patients <2 years of age were given IT MTX (age-related dose) every 6 weeks until the age of 2 when cranial irradiation was given. Interim maintenance from week 13 to 16 consisted of 6- mercaptopurine (6-MP): (70 mg/m 2 /day 21) and oral MTX (20 mg/m 2 /week 3). The reintensi cation phase, beginning at week 17, was based on VCR (1.5 mg/m 2 /week 4), adriamycin (ADR: 30 mg/m 2 /week 4), ASP (10 000 units/m 2 /twice weekly for 2 weeks), PRED (60 mg/m 2 /day, tapered to stop by week 21), CYCLO (1000 mg/m 2 1, on week 22), and ARA-C (75 mg/m 2 /day 4, on weeks 22 and 23). Maintenance chemotherapy was based on 6-MP (70 mg/m 2 /day) and MTX (20 mg/m 2 /week). Maintenance chemotherapy was discontinued 24 months from the start of treatment. Statistical methods Failure was de ned as a lack of response to induction therapy, a relapse during rst complete remission, or death due to any cause. 9 The relationship of each presenting feature to treatment outcome (failure) was analysed by the Pearson chi-square statistics. Correlations between these factors were analysed and multivariate analyses were performed in a forward stepwise fashion using the logistic regression model which permits comparison of treatment outcome between variables while adjusting for the effects of other factors (covariates). Event-free-survival (EFS) was de ned as the interval between date of diagnosis and the earliest of the following events: induction failure, leukemic relapse, death from any cause, last contact, development of a second malignant neoplasm, and bone marrow transplant. 10;11 Survival was calculated from the time of diagnosis to the last follow-up or time of death. 9;12 Life table analysis for EFS and overall survival were estimated using the Kaplan±Meier method, 13 with differences compared by the log-rank test. Results Presenting features and treatment outcome of patients are summarized in Table 1. In this study of the Malaysian population, infants < 1 year of age comprised 4 per cent of all patients and children $1 year of age comprised 96 per cent of all patients. Age was found to be a signi cant predictor of outcome, with the rate of failure signi cantly higher amongst the ages <1 year old (p ˆ 0:0001). Failure occurred in 51 per cent of 326 males and 36 per cent of 249 females. The difference was signi cant (p ˆ 0:0003). Race among the Malaysian population did not appear to in uence prognosis. The distribution showed 37 per cent of patients were Malays, 52 per cent Chinese, 9 per cent Indians and 2 per cent other races. The relationship of FAB morphological subtype to prognosis was not signi cant. WCC and Hb level demonstrated a signi cant relationship to prognosis in our study. Failure rate was higher (at 56 per cent) in the 126 patients with WCC $ 50 10 9 /l compared to 41 per cent in the low WCC patients (p ˆ 0:003). Of all patients, 22 per cent had WCC $ 50 10 9 /l at presentation. Hb level of $11 g/dl were also associated with a poorer outcome compared to an Hb level of < 11 g/dl ( p ˆ 0:01). At diagnosis, CNS involvement was present in 2 per cent of patients, mediastinal mass was reported in 11 per cent of patients and lymphadenopathy in 17 per cent of patients. The difference in treatment outcome for all three variables was not signi cant. Platelet count, liver and spleen size also showed no relationship to prognosis. Marked hepatomegaly and splenomegaly were reported in 34 and 26 per cent of all patients, respectively. In the univariate analyses, age, sex, WCC and Hb level emerged as potentially signi cant predictors of treatment outcome. When the same variables were entered stepwise in a multivariate analyses, all four factors retained their independent prognostic signi cance (see Table 2). Factors not signi cantly related to treatment outcome included race, platelet count, FAB morphology, liver/spleen size, presence of CNS involvement, mediastinal mass and lymphadenopathy. Life table analysis of survival is shown in Fig. 1. Overall, at 24 months, 67 per cent of patients were alive TABLE 2 Results of the multivariate analysis Features Worse category p value Odds ratio a 95% con dence interval Age < 1 year old 0.005 4.34 1.56±12.11 Sex Male 0.002 1.75 1.23±2.50 Hb level $11 g/dl 0.004 2.11 1.26±3.52 WCC $50 10 9 /l 0.007 1.77 1.17±3.52 a The proportional increase in the risk of failure at any time for a patient having the worse category of a factor relative to a patient having a better category. Hb, hemoglobin; WCC, white cell count. 340 Journal of Tropical Pediatrics Vol. 46 December 2000

FIG. 1. Life-table analysis of overall survival using the Kaplan±Meier method. FIG. 3. Overall survival according to age using the Kaplan±Meier method. and 55 per cent were relapse-free. Of 326 males, 64 per cent were alive and 71 per cent of females were alive at 2 years, the difference was not statistically signi cant (p ˆ 0:10, see Fig. 2). Children <1 year of age had a signi cantly shorter duration of survival. The survival rate was 30 per cent for children <1 year and 68 per cent for children $1 year old (p < 0:0001, see Fig. 3). In Fig. 4, 72 per cent of children with WCC < 50 10 9 /l and 53 per cent of children with WCC $ 50 10 9 /l were alive (p < 0:001). Our results also show that survival rates for children with an Hb level #11 g/dl were better than children with an Hb level $11 g/dl (p ˆ 0:02, see Fig. 5). Discussion In this study, we have shown that age group <1 year, male sex, high WCC of $50 10 9 /l and an Hb level of $11 g/dl are associated with a poorer treatment outcome in children with ALL. Notably, race, platelet count, FAB morphology, liver/spleen size, CNS, mediastinal mass or lymph node involvement did not appear to in uence the prognosis. The most important prognostic factor in ALL is the type of treatment give to patients. It is possible that the adverse prognostic impact of the presence of mediastinal mass and CNS leukemia at diagnosis has been eliminated by the intensity of early treatment given in the protocol. In many studies, age at diagnosis has been shown to be a consistently strong prognostic factor in childhood ALL. Patients <1 year old have higher relapse rates compared to other ages as reported in various studies. 14±16 In our study of Malaysian children, age emerged as the strongest predictor of prognosis (p ˆ 0:0001). The poor outcome may be related to the common occurrence of other poor prognostic factors within these age groups. Infants less than 1 year of age and children more than 10 years of age have been shown to have a higher incidence of karyotypic abnormalities. 17;18 For example, the 11q23/MLL gene arrangement, e.g. t(4;11) translocation, is found in 68 per cent of infants with ALL. 19 Sex had a signi cant in uence on outcome with a higher failure rate in males than in females (p ˆ 0:0003). These results were in agreement with earlier studies. 5;20;21 This could be accounted for by the different distribution FIG. 2. Overall survival according to sex using the Kaplan±Meier method. FIG. 4. Overall survival according to white cell count using the Kaplan±Meier method. Journal of Tropical Pediatrics Vol. 46 December 2000 341

per cent and relapse-free rate of 55 per cent in this population compares less favourably than other reported studies. This may be affected by the lack of adequate strati cation of treatment. High-risk patients may have been under-treated and low risk patients may have been subjected to toxic treatment. Due to inadequate data, we were unable to con rm the negative prognostic value of immunophenotype and karyotype as described in other studies of childhood ALL. 25±28 FIG. 5. Overall survival according to hemoglobin level using the Kaplan±Meier method. of other factors, as well as by the higher testicular relapse rate in this study. The difference in 2-year survival between males and females was not signi cant (p ˆ 0:10). The relationship of WCC to prognosis is rmly established in many studies of childhood ALL 2±6 and was con rmed in this study of Malaysian children. WCC $ 50 10 9 /l has been shown to be an adverse risk factor in childhood ALL. This value has been proposed as the value by which to de ne patients with a poor prognosis by the National Cancer Institute sponsored workshop. 22 The adverse impact of the high WCC in this study was not removed by the intensity of the treatment, unlike that found in the BFM study. Our results showed that an Hb level $11 g/dl was signi cantly associated with a poor treatment outcome (p ˆ 0:01). Our ndings, however, contrast with other studies which showed no proven association between Hb level and prognosis. Various studies on the prognostic in uence of FAB morphological subtype in childhood ALL have yielded controversial results. A Pediatric Oncology Group Study reported that there was no association between FAB morphological subtype and prognosis. 23 In contrast, a Children's Cancer Study Group found L2 morphology to be associated with a poor treatment outcome in childhood ALL. 24 Our studies found that FAB morphological subtype did not show any signi cant effect on treatment outcome in childhood ALL. In this study, L1 was the most common variety in children, representing 58 per cent of all cases, L2 occurred in 38 per cent, and L3 represented 4 per cent of all cases. This distribution representing Malaysian children differed with other studies where L1 represented 85 per cent of ALL cases, L2 less than 15 per cent and L3 less than 5 per cent of all cases. Additional studies are needed to clarify the biological signi cance of this factor. Our analysis con rms that Malaysian children up to 12 years of age have similar prognostic factors to children treated all over the world with a multitude of protocols for the last 20±30 years. The 2-year survival rate of 67 Conclusions In summary, the results of our univariate analyses and multivariate analyses showed that the prognosis of childhood ALL within a Malaysian population was dependent on age, sex, WCC and Hb level. Prognostic factors are important in the appropriate strati cation of proper risk-adjusted initial management of childhood ALL, and more studies are needed to evaluate them further. References 1. Hoffbrand AV, Pettit JE (eds). Essential Haematology. Blackwell Science Ltd, London, 1995. 2. Cortes JE, Kantarjian HM. Acute lymphoblastic leukemia: a comprehensive review with emphasis on biology and therapy. Cancer 1995; 76: 2393±409. 3. Pui CH, Evans WE. Acute lymphoblastic leukemia. N Engl J Med 1998; 339: 605±15. 4. Genter H, Rudiger F, Reinhard H, et al. Six-year experience with a comprehensive approach to the treatment of recurrent childhood acute lymphoblastic leukemia (ALL- REZ BFM 85). Blood 1991; 78: 1166±172. 5. Pui CH, Behm FG, Singh, et al, Heterogenity of presenting features and their relation to treatment outcome in 120 children with T-cell acute lymphoblastic leukemia. Blood 1990; 75: 174±79. 6. Baccarani M, Corbelli G, Amadori S, et al. Adolescent and adult acute lymphoblastic leukemia: prognostic features and outcome of therapy. A study of 293 patients. Blood 1982; 60: 677±84. 7. Williams, DL, Harber J, Murphy SB, et al. Chromosomal translocations play a unique role in in uencing prognosis in childhood acute lymphoblastic leukemia. Blood 1986; 68: 205±12. 8. Reiter, A, Scrappe M, Ludwig WD, et al. Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients: results and conclusion of the multicenter trial ALL-BFM 86. Blood 1994; 84: 3122±133. 9. Reiter A, Scrappe M, Ludwig WD, et al. Favorable outcome of B-cell acute lymphoblastic leukemia in childhood: a report of three consecutive studies of the BFM group. Blood 1992; 80: 2471±478. 10. Uckun FM, Sather HN, Gaynon PS, et al. Clinical features and treatment outcome of children with myeloid antigen positive acute lymphoblastic leukemia: a report from the Children's Cancer Study Group. Blood 1997; 90: 28±35. 11. Crist W, Caroll JS, Jackson J, et al. Philadelphia chromosome positive childhood acute lymphoblastic leukemia: clinical and cytogenic characteristics and treatment outcome. A Pediatric Oncology Study. Blood 1990; 76: 489±94. 342 Journal of Tropical Pediatrics Vol. 46 December 2000

12. Crist W, Boyett J, Jackson J, et al. Prognostic importance of pre-b cell immunophenotype and other presenting features in B-lineage childhood acute lymphoblastic leukemia: A Pediatric Oncology Group Study. Blood 1989; 74: 1252±259. 13. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1970; 58: 457±61. 14. Reaman GH, Sposto R, Sensel MG, et al. Treatment outcome and prognostic factors for infants with acute lymphoblastic leukemia treated on 2 consecutive trials of Children's Cancer Group. J Clin Oncol 1999; 17: 445±55. 15. Silverman LB, McLean TW, Gelber RD, et al. Intensi ed therapy for infants with acute lymphoblastic leukemia. Results from Dana-Farber Cancer Institute Consortium. Cancer 1997; 80: 2285±295. 16. Pui CH, Kane JR, Crist WM. Biology and treatment of infant leukemias. Leukemia 1995; 9: 762±69. 17. Pui CH, Crist WM, Look T. Biological and clinical signi cance of cytogenetic abnormalities in childhood acute lymphoblastic leukemia. Blood 1990; 76: 1449±463. 18. Bloom eld CD, Goldman AI, Berger GA, et al. Chromosomal abnormalities identify high-risk and low-risk patients with acute lymphoblastic leukemia. Blood 1986; 67: 415±20. 19. Pui CH, Behm FG, Downings JR, et al. 11q23/MLL rearrangement confers a poor prognosis in infants with acute lymphoblastic leukemia. J Clin Oncol 1994; 12: 909±15. 20. Pui CH, Boyett JM, Relling MV, et al. Sex differences in prognosis for children with acute lymphoblastic leukemia. J Clin Oncol 1999; 17: 818±24. 21. Shuster JJ, Wacker P, Pullen J, et al. Prognostic signi cance of sex in childhood B-precursor acute lymphoblastic leukemia: A Pediatric Oncology Group Study. J Clin Oncol 1998; 16: 2854±863. 22. Smith M, Arthur D, Camitta B, et al. Uniform approach to risk classi cation and treatment assignment for children with acute lymphoblastic leukemia. J Clin Oncol 1996; 14: 18±24. 23. Eys J, Pullen J, Head D, et al. The French±American± British classi cation of leukemia. A Pediatric Oncology Group experience with lymphocytic leukemia. Cancer 1986; 57: 1046±051. 24. Miller DR, Krailo M, Bleyer WA, et al. Prognostic implications of blast cell morphology in childhood acute lymphoblastic leukemia: A report from the Children's Cancer Study Group. Cancer Treat Rep 1985; 69: 1211±215. 25. Pui CH, Behm FG, Crist WM. Clinical and biological relevance of immunological marker studies in childhood acute lymphoblastic leukemia. Blood 1993; 82: 343±62. 26. Garand R, Vannier JP, Bene ML, Faure GL, Bernard A. Correlations between acute lymphoblastic leukemia immunophenotype and clinical laboratory data at presentation. A study of 350 patients. Cancer 1989; 64: 1437±445. 27. Beyerman B, Agthe AG, Adams HP, et al. Clinical features and outcome of children with rst marrow relapse of acute lymphoblastic leukemia expressing BCR-ABL fusion transcripts. Blood 1996; 87: 1532±539. 28. Pui CH, Williams DL, Raimondi SC, et al. Hypodiploidy is associated with poor prognosis in childhood acute lymphoblastic leukemia. Blood 1987; 70: 247±53. Journal of Tropical Pediatrics Vol. 46 December 2000 343