Duration of hospitalization as a measure of cost on Children's Cancer Group acute lymphoblastic leukemia studies.

PURPOSE: We used duration of hospitalization as a surrogate for cost and event-free survival as a measure of effectiveness to estimate the cost-effectiveness ratios of various treatment regimens on Children's Cancer Group trials for acute lymphoblastic leukemia. PATIENTS AND METHODS: The analyses included 4,986 children (2 to 21 years of age) with newly diagnosed acute lymphoblastic leukemia enrolled onto risk-adjusted protocols between 1988 and 1995. Analyses were based on a model of 100 patients. The marginal cost-effectiveness ratio (hospital days per additional patient surviving event-free) was the difference in total duration of hospitalization divided by the difference in number of event-free survivors at 5 years for two regimens. Relapse-adjusted marginal cost of frontline therapy was the difference in total duration of hospitalization for frontline therapy plus relapse therapy divided by the difference in number of event-free survivors at 5 years on the frontline therapy for two regimens. RESULTS: One or two delayed intensification (DI) phases, augmented therapy, and dexamethasone all improved outcome. Marginal cost-effectiveness of these regimens compared with the control regimens was 133 days per patient for DI, 117 days per patient for double DI, and 41 days per patient for augmented therapy. Dexamethasone resulted in 17 fewer days per patient. Relapse-adjusted marginal costs were 68 days per patient for DI and 52 days for double DI. Augmented therapy and dexamethasone-based therapy resulted in 16 and 82 fewer hospital days, respectively. The estimated cost-effectiveness for treating any first relapse was 250 days per patient. CONCLUSION: DI, double DI, augmented therapy, and dexamethasone-based therapy are cost-effective strategies compared with current treatment of first relapse.

Plasma pharmacokinetics and cerebrospinal fluid penetration of thioguanine in children with acute lymphoblastic leukemia: a collaborative Pediatric Oncology Branch, NCI, and Children's Cancer Group study.

PURPOSE: In preclinical studies, thioguanine (TG) has been shown to be more potent than the standard acute lymphoblastic leukemia (ALL) maintenance agent, mercaptopurine (MP), suggesting that TG may be more efficacious than MP in the treatment of childhood ALL. As part of a pilot trial in which TG was used in place of MP, we studied the plasma pharmacokinetics of oral TG and measured steady-state plasma and CSF TG concentrations during a continuous intravenous infusion (CIVI) in children with newly diagnosed standard-risk ALL. METHODS: Nine plasma samples were collected after each patient's first 60 mg/m2 oral TG dose during maintenance. CIVI TG (20 mg/m2/h over 24 h) was administered during the consolidation phase of therapy, and simultaneous plasma and CSF samples were collected near the end of the infusion. TG was measured by reverse-phase HPLC with ultraviolet detection. Erythrocyte TG nucleotide (TGN) concentrations were measured 7 days after a course of CIVI TG and prior to the start of each maintenance cycle. RESULTS: After oral TG (n = 35), the mean (+/- SD) peak plasma concentration was 0.46 +/- 0.68 microM and the AUC ranged from 0.18 to 9.5 microM.h (mean 1.5 microM.h). Mean steady-state plasma and CSF TG concentrations during CIVI (n = 33) were 2.7 and 0.5 microM, respectively. The mean (+/- SD) TG clearance was 935 +/- 463 ml/min per m2. Plasma TG concentrations did not correlate with erythrocyte TGN concentrations after oral or CIVI TG. The 8-OH-TG metabolite was detected in plasma and CSF. CONCLUSIONS: TG concentrations that are cytotoxic to human leukemia cell lines can be achieved in plasma after a 60 mg/m2 oral dose of TG and in plasma and CSF during CIVI of TG.

Abnormalities of chromosome bands 13q12 to 13q14 in childhood acute lymphoblastic leukemia.

PURPOSE: Little is known about nonrandom deletions of chromosome bands 13q12 to 13q14 (13q12-14) in acute lymphoblastic leukemia (ALL). We determined the prognostic significance of cytogenetically identified breakpoints in 13q12-14 in children with newly diagnosed ALL treated on Children's Cancer Group protocols from 1988 to 1995. PATIENTS AND METHODS: Breakpoints in 13q12-14 were identified in 36 (2%) of the 1,946 cases with accepted cytogenetic data. Outcome analysis used standard life-table methods. RESULTS: Seventeen patients (47%) with an abnormal 13q12-14 were classified, according to the National Cancer Institute (NCI), as poor risk, and 15 patients (42%) were standard risk; four (11%) were infants less than 12 months of age. Eight cases had balanced rearrangements of 13q12-14, 27 patients had a partial loss of 13q, and one had both a partial gain and a partial loss. The most frequent additional abnormalities among these patients were an abnormal 12p, a del(6q), a del(9p), a 14q11 breakpoint, and an 11q23 breakpoint. Nineteen patients were pseudodiploid, 10 were hyperdiploid, and seven were hypodiploid. Patients with an abnormal 13q12-14 had significantly worse event-free survival than patients lacking such an abnormality, with estimates at 6 years of 61% (SD = 14%) and 74% (SD = 1%), respectively (P =.04; relative risk = 1.74). Overall survival, however, was similar for the two groups (P =.25). The prognostic effect of an abnormal 13q was attenuated in a multivariate analysis adjusted for NCI risk status and ploidy (P =.72). CONCLUSION: Aberrations of 13q12-14 may contribute to leukemogenesis of childhood ALL and confer increased risk of treatment failure but are associated with other poor-risk features.

Clinical significance of deletions of chromosome arm 6q in childhood acute lymphoblastic leukemia: a report from the Children's Cancer Group.

We have compared outcome for 167 (9.0%) children with a del(6q) and 1713 (91%) children without a del(6q) treated on Children's Cancer Group (CCG) risk-adjusted treatment protocols for acute lymphoblastic leukemia (ALL). Thirty-three patients had a del(6q) as the sole aberration; 22 patients had a del(6q) only as a secondary abnormality. Thirty-six cases had a del(6q) and high hyperdiploidy (>50 chromosomes). Six patients with a del(6q) also had +16 and 8 patients had loss of a sex chromosome. Frequent recurring breakpoints were q13, q15, q21, q23, and q25. Patients with a del(6q) were more likely to have T-lineage ALL (p < 0.001), a mediastinal mass (p = 0.01), and higher WBC counts (p = 0.04), although only half of these patients were classified as poor risk. Event-free survival at 6 years was similar for patients with or without a del(6q), with estimates of 77% (SD = 5%) and 74% (SD = 2%), respectively (p = 0.44). This finding was also observed within NCI poor and standard risk groups. Thus, cytogenetically detectable del(6q) is not associated with adverse risk in pediatric ALL.

Prognostic impact of trisomies of chromosomes 10, 17, and 5 among children with acute lymphoblastic leukemia and high hyperdiploidy (> 50 chromosomes).

PURPOSE: Children with acute lymphoblastic leukemia (ALL) and high hyperdiploidy (> 50 chromosomes) have improved outcome compared with other ALL patients. We sought to identify cytogenetic features that would predict differences in outcome within this low-risk subset of ALL patients. MATERIALS AND METHODS: High-hyperdiploid ALL patients (N = 480) were enrolled between 1988 and 1995 on Children's Cancer Group (CCG) trials. Karyotypes were determined by conventional banding. Treatment outcome was analyzed by life-table methods. RESULTS: Patients with 54 to 58 chromosomes had better outcome than patients with 51 to 53 or 59 to 68 chromosomes (P = .0002). Patients with a trisomy of chromosome 10 (P<.0001), chromosome 17 (P = .0002), or chromosome 18 (P = .004) had significantly improved outcome compared with their counterparts who lacked the given trisomy. Patients with a trisomy of chromosome 5 had worse outcome than patients lacking this trisomy (P = .02). Patients with trisomies of both chromosomes 10 and 17 had better outcome than those with a trisomy of chromosome 10 (P = .09), a trisomy of chromosome 17 (P =.01), or neither trisomy (P<.0001). Multivariate analysis indicated that trisomy of chromosome 10 (P = .001) was the most significant prognostic factor for high-hyperdiploid patients, yet trisomy of chromosome 17 (P =.02) or chromosome 5 (P = .01) and modal chromosome number (P = .02) also had significant multivariate effects. CONCLUSION: Trisomy of chromosomes 10 and 17 as well as modal chromosome number 54 to 58 identify subgroups of patients with high-hyperdiploid ALL who have a better outcome than high-hyperdiploid patients who lack these cytogenetic features. Trisomy of chromosome 5 confers poorer outcome among high-hyperdiploid patients.

Association of chromosome arm 9p abnormalities with adverse risk in childhood acute lymphoblastic leukemia: A report from the Children's Cancer Group.

Cytogenetic abnormalities of chromosome arm 9p occur frequently in children with acute lymphoblastic leukemia (ALL). We analyzed 201 such cases (11%) in 1,839 children with newly diagnosed ALL treated between 1989 and 1995 on risk-adjusted protocols of the Children's Cancer Group (CCG). The majority of patients (131; 65%) with a 9p abnormality were classified as higher risk. Nearly all patients had complex karyotypes; most cases had deletions of 9p, add/der(9p), a dicentric involving chromosome arm 9p, and/or balanced translocations and inversions involving 9p. Event-free survival (EFS) estimates at 6 years for patients with and without a 9p aberration were 61% (standard deviation [SD] = 5%) and 76% (SD = 2%; P <.0001). In addition, patients with a 9p abnormality had an increased cumulative incidence of both marrow (P =.04) and central nervous system (P =.0001) relapses. Overall survival also was significantly worse for patients with an abnormal 9p (P <.0001). These effects were most pronounced in standard-risk patients (age 1 to 9 years with white blood cell count <50,000/microL): 6-year EFS of 61% (SD = 9%) versus 80% (SD = 2%; P <.0001). Also, a 9p aberration was an adverse risk factor for B-lineage, but not T-lineage patients. The effect of 9p status on EFS was attenuated, but maintained in a multivariate analysis of EFS after adjustment for Philadelphia chromosome status, age, white blood cell (WBC) count, sex, race, and ploidy group (P =.01). Thus, abnormalities of chromosome arm 9p identify a subgroup of standard-risk patients with increased risk of treatment failure.

High-dose carboplatin, thiotepa, and etoposide with autologous stem-cell rescue for patients with recurrent medulloblastoma. Children's Cancer Group.

PURPOSE: Medulloblastoma is a highly lethal disease when it recurs. Very few patients survive with conventional treatment. This study evaluated the use of high-dose carboplatin, thiotepa, and etoposide with autologous stem-cell rescue (ASCR) in patients with recurrent medulloblastoma. METHODS: Chemotherapy consisted of carboplatin 500 mg/m2 (or area under the curve = 7 mg/mL x min via Calvert formula) on days -8, -7, and -6; and thiotepa 300 mg/m2 and etoposide 250 mg/m2 on days -5, -4, and -3; followed by ASCR on day 0. In addition to the study-prescribed therapy, 21 patients received other treatment: neurosurgical resection in seven, conventional chemotherapy in 17, and external-beam irradiation in 11 cases. RESULTS: Twenty-three patients with recurrent medulloblastoma, aged two to 44 years (median, 13 years) at ASCR, were treated. Three patients died of treatment-related toxicities within 21 days of ASCR; multiorgan system failure in two, and Aspergillus infection with venoocclusive disease in one. Seven of 23 patients (30%) are event-free survivors at a median of 54 months post-ASCR (range, 24 to 78 months). Kaplan-Meier estimates of event-free (EFS) and overall survival are 34% +/- 10% and 46% +/- 11%, respectively, at 36 months post-ASCR. CONCLUSION: This strategy may provide long-term survival for some patients with recurrent medulloblastoma.

Early response to therapy and outcome in childhood acute lymphoblastic leukemia: a review.

BACKGROUND: Early response to therapy is defined as the initial response prior to Day 28 of treatment, the conventional time of marrow evaluation. The number of reports linking early response to therapy with the ultimate outcome of childhood acute lymphoblastic leukemia is substantial and growing. When this study began, these experiences had yet to be comprehensively reviewed. METHODS: A comprehensive search of the published literature yielded contributory reports of 14 trials conducted in the United States and Europe. In addition, unpublished data from one Children's Cancer Group trial were made available. Outcome measures were standardized by conversion to ratios of the incidence of adverse events among poorer and better responders. RESULTS: Early response to therapy was an independent prognostic factor in each of the 15 trials, which together included more than 10,000 patients. The incidence of slower early response ranged from 2-33%, with various measures and criteria used in different trials. Patients with a slower early response were 1.5-6.1 times (median, 2.7) more likely to have an adverse event than patients with a more rapid early response, however defined. Early response maintained prognostic significance after the exclusion of induction failure and within risk strata defined by age, white blood cell count, and/or immunophenotype. Its significance was also maintained in multivariate analyses where performed. CONCLUSIONS: Early response to therapy, whether determined by evaluation of bone marrow or peripheral blood, is a consistent, independent prognostic factor in childhood acute lymphoblastic leukemia. Slower early response may serve as a useful surrogate for outcome, a more complex end point, in investigations of the cellular and molecular determinants of resistance to therapy. It may also allow early identification of a patient subpopulation for whom current therapy is less effective and alternative strategies may be justified.

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.

PURPOSE: Leukemic cells from T-lineage acute lymphoblastic leukemia (ALL) patients are thought to originate from T-lymphocyte precursors corresponding to discrete stages of T-cell ontogeny. Here we sought to determine the influence of leukemic cell apparent maturational stage on treatment outcomes in pediatric T-lineage ALL. PATIENTS AND METHODS: From 1983 through 1993, 407 pediatric T-lineage ALL patients were enrolled onto two sequential series of risk-adjusted treatment protocols of the Children's Cancer Group. In the current analysis, T-lineage ALL patients were immunophenotypically classified as follows: CD7+ CD2- CD5- pro-thymocyte leukemia (pro-TL), CD7+ (CD2 or CD5)+ CD3- immature TL, and CD7+ CD2+ CD5+ CD3+ mature TL. RESULTS: Similar induction outcomes of 91.4%, 97.1%, and 98.3% were obtained by the pro-, immature, and mature TL groups, respectively. Four-year event-free survival (EFS) was lower for pro-TL patients (57.1%; SD = 8.4%,) compared with immature and mature TL patients (68.5%; SD = 3.5%; and 77.1%; SD = 4.0%, respectively) with an overall significance of .05 (log-rank test) or .04 (log-rank trend test). Relative hazards rates (RHR) were 2.11 and 1.22 for pro-TL and immature TL versus mature TL, respectively. Highly significant differences were found for overall survival (P = .005, log-rank test; P = .009, log-rank trend test). Multivariate analysis confirmed that the prognostic influence of ontogeny grouping was independent of that of other prognostic factors. CONCLUSION: Leukemic cells of the pro-TL maturation stage identify a small subgroup of T-lineage ALL patients who have a significantly worse EFS outcome than patients whose cells are of a more mature stage of development.

Expression of BCR-ABL, E2A-PBX1, and MLL-AF4 fusion transcripts in newly diagnosed children with acute lymphoblastic leukemia: a Children's Cancer Group initiative.

We used reverse transcriptase polymerase chain reaction (RT-PCR) assays to examine primary leukemic cells in on-study diagnostic bone marrow specimens from 642 children with newly diagnosed acute lymphoblastic leukemia (ALL) for the expression of MLL-AF4, E2A-PBX1, and BCR-ABL fusion transcripts. All PCR assays were performed centrally in the Children's Cancer Group ALL Biology Reference Laboratory. MLL-AF4 transcript was found in only 0.7% of the study population which excluded infants. E2A-PBX1 transcript was found in 2.5% of the study population and 3.3% of B-precursor cases. Expression was associated with massive hepatomegaly. BCR-ABL transcript was found in 2.3% of cases and correlated with older age, induction failure, and inferior event-free survival (EFS). RT-PCR assays allow rapid identification of patients with MLL-AF4 and BCR-ABL positive ALL. These patients have a poor outcome with contemporary therapy and rapid identification facilitates timely allocation to innovative treatment programs.

Treatment of childhood lymphangiomas with interferon-alpha.

PURPOSE: Nonsurgical treatment of lymphangiomas has shown limited efficacy and often carries unacceptable toxicities, demonstrating the need for a more effective, less toxic therapy. PATIENTS AND METHODS: We describe two patients with lymphangiomatosis treated for 12 to 40 months with recombinant interferon-alpha. RESULTS: Both patients demonstrated stabilization or marked improvement of disease, based on clinical and radiologic findings, with minimal toxicity. CONCLUSIONS: The favorable responses to interferon-alpha therapy in these two cases suggest that this is an effective and well-tolerated treatment for lymphangiomas in children.

Prospective, randomized, double-blind, placebo-controlled comparison of metoclopramide and ondansetron for prevention of posttonsillectomy or adenotonsillectomy emesis.

STUDY OBJECTIVE: To compare the antimetic efficacy of prophylactic ondansetron, metoclopramide, and placebo for prevention of postoperative vomiting in pediatric tonsillectomy or adenotonsillectomy patients. DESIGN: Prospective, randomized, double-blind, placebo controlled study. SETTING: Children's hospital. PATIENTS: 132 ASA status I and II children, ages 2 to 12 years, undergoing tonsillectomy or adenotonsillectomy. INTERVENTIONS: Patients received intravenous (IV) melodopramide 0.25 mg/kg, IV on dansetron 0.15 mg/kg, or IV saline placebo after induction of standardized halothane, nitrous oxide, and oxygen anesthesia. Muscle relaxants and their antagonists were allowed. Patients received postoperative analgesics as needed. MEASUREMENTS AND MAIN RESULTS: Incidence of postoperative vomiting, time of vomitting onset, and hospital length of stay (LOS) were measured. Patients who were admitted were excluded from LOS analysis. The postoperative incidence of vomiting was 54% for patients receiving metoclopramide, 26% for patients receiving ondansetron, and 69% for the placebo group. These differences were significant for ondansetron versus metoclopramide (p = 0.008) and placebo (p = 0.001). The mean (SD) LOS was significantly shorter for patients not vomiting 488 (88) minutes for vomiters versus 435 (65) minutes for non-vomiters. CONCLUSIONS: Prophylactic ondansetron is more effective than metoclopramide or placebo for the prevention of vomiting after tonsillectomy or adenotonsillectomy. Patients who do not vomit postoperatively have shorter LOS.

Juvenile xanthogranuloma: forms of systemic disease and their clinical implications.

OBJECTIVE: Juvenile xanthogranuloma (JXG) with systemic (extracutaneous) involvement is a rare histiocytic disorder in which significant morbidity and occasional deaths may occur. The objective of this study was to characterize the spectrum of anatomic involvement, associated clinical problems, and management considerations in children with systemic JXG. STUDY DESIGN: Two current cases and literature reports of 34 children with various forms of systemic AG were analyzed with respect to age, clinical presentation, site(s) of involvement, therapy, and outcome. RESULTS: The median age of the 36 patients was 0.3 years (range, birth to 12 years). Symptoms were usually referable to bulky or infiltrative disease. Twenty patients had disease in two or more sites. Cutaneous lesions were present in fewer than half the patients. The most frequent extracutaneous sites of disease were the subcutaneous soft tissue (12); central nervous system (8); liver/spleen (8); lung (6); eye/orbit, oropharynx, and muscle (4 each); with three or fewer instances of disease in each of several other sites. Most patients were treated with excision or had spontaneous regression (some with organ involvement). However, 12 patients received treatment that included radiation or systemic chemotherapy. Survivors, some with long-term disabilities, included young children who had received radiation therapy to the brain, eye, skin, or heart. Two patients died of disease. CONCLUSIONS: Systemic AG may involve varying numbers and combinations of extracutaneous sites. The extent of disease should be determined in patients with AG who are suspected to have systemic involvement. In contrast to the cutaneous form, systemic AG may be associated with significant complications requiring aggressive medical care. When feasible, surgical excision of lesions may be curative. Optimal treatment for symptomatic, unresectable disease is currently undefined but should be selected to minimize toxic effects in these children who are typically younger than 1 year old at presentation.

Pharmacokinetics of fluconazole in immune-compromised children with leukemia or other hematologic diseases.

STUDY OBJECTIVE: To describe the pharmacokinetics of fluconazole in immune-compromised children with leukemia or other hematologic disease. DESIGN: Prospective. SETTING: Children's Health Care-Minneapolis hematology/oncology inpatient ward and outpatient clinic. PATIENTS: Ten immune-compromised children (mean +/- SD age 7.4 +/- 4.0 yrs, weight 31.6 +/- 25.9 kg) with leukemia or other hematologic disease. INTERVENTIONS: Serum was sampled before and after a single 6-mg/kg intravenous dose and after seven oral 3-mg/kg doses of fluconazole. MEASUREMENTS AND MAIN RESULTS: Mean (SD) pharmacokinetics were distribution half-life 1.67 (1.25) hours, elimination half-life 15.62 (3.21) hours, total body clearance 0.63 (0.19) ml/min/kg, volume of distribution for the central compartment 0.56 (0.10) L/kg, volume of distribution at steady state 0.77 (0.12) L/kg, absorption half-life 0.41 (0.26) hour, and oral bioavailability 0.92 (0.09). Volume of distribution for the central compartment was highly correlated with body surface area (r2 = 0.891) and weight (r2 = 0.949). Volume of distribution at steady state correlated with body surface area (r2 = 0.986), and total body clearance correlated with body surface area (r2 = 0.867). CONCLUSIONS: Fluconazole elimination was well described using a two-compartment model. Oral absorption was rapid and nearly complete. Children have a larger volume of distribution for the central compartment and faster elimination rate than adults. Body surface area and weight are important factors in determining pharmacokinetics in these patients.

Reversed-phase high-performance liquid chromatographic approach to determine total lymphocyte concentrations of 6-thioguanine, methylmercaptopurine and methylthioguanine in humans.

A reversed-phase high-performance liquid chromatographic (HPLC) procedure was developed to quantify intracellular lymphocyte 6-thioguanine, methylmercaptopurine and methylthioguanine. The free base of each metabolite was obtained by acid hydrolysis, which allowed for a total determination of thiopurine metabolites. 6-Thioguanine was analyzed on an octadecylsilane column using acetonitrile-10 mM sodium phosphate (11:89), pH 7, containing 0.06% tetrabutylammonium chloride. 6-Thioguanine was oxidized with potassium permanganate, and fluorescence was measured at 330 nm excitation and 410 nm emission. Methylmercaptopurine and methylthioguanine were separated on a cyanopropylsilane column using methanol-40 mM sodium phosphate (22:78), pH 2.7, and detected by ultraviolet absorbance at 314 and 290 nm, respectively.

A reversed phase high performance liquid chromatography approach in determining total red blood cell concentrations of 6-thioguanine, 6-mercaptopurine, methylthioguanine, and methylmercaptopurine in a patient receiving thiopurine therapy.

A reversed phase high performance liquid chromatographic procedure was developed to quantify 6-thioguanine, 6-mercaptopurine, methylthioguanine, and methylmercaptopurine in red blood cells. The free base of each thiopurine was liberated from the respective nucleoside and nucleotide moiety by acid hydrolysis, which allowed for a determination of the total thiopurine present. 6-Thioguanine and 6-mercaptopurine were analyzed on an octadecylsilane column using methanol + 20 mM sodium phosphate (15:85), pH 7.5, containing 0.07% tetrabutylammonium chloride. Detection was by potassium permanganate oxidation and fluorescence detection at 290 nm excitation and 400 nm emission. Methylmercaptopurine and methylthioguanine were analyzed on a cyanopropylsilane column using methanol + 40 mM sodium phosphate (18:82), pH 2.7, and then ultraviolet absorption at 314 nm and 290 nm, respectively. The method was used to quantify the four primary thiopurines present in red blood cells of an acute lymphoblastic leukemia patient. The procedure may be a therapeutic monitoring technique that quantifies the cytotoxic drug burden in patients receiving azathioprine or 6-mercaptopurine therapy.

Allogeneic bone marrow transplantation for acute nonlymphocytic leukemia in first remission.

Seventy-three patients with acute nonlymphocytic leukemia in first complete remission (CR) have received allogeneic bone marrow transplantation (BMT) with non-T-lymphocyte-depleted marrow obtained from matched sibling donors. The first 36 patients received a preparative regimen consisting of cyclophosphamide, 60 mg/kg/d (days -6 and -5), and 750 cGy single-dose total-body irradiation (TBI) (day -1). Subsequently, 37 patients received cyclophosphamide 60 mg/kg/d (days -6 and -5), and 165 cGy fractionated TBI administered twice daily for a total dose of 1,320 cGy (days -4, -3, -2, and -1). Survivors have been followed from 9 to 124 months (median, 40 months). The 61% (95% confidence interval [CI], 45% to 77%) projected disease-free survival (DFS) of 41 children less than 18 years old does not differ significantly from the 62% (95% CI, 49% to 73%) projected DFS of 32 adults at 84 months (P = .89). Similarly, the 15% (95% CI, 1% to 29%) projected relapse rate seen in children does not differ from the 9% (95% CI, 0% to 21%) seen in adults (P = .69). Multivariate Cox regression analysis of presenting features demonstrates that a presenting WBC count greater than 20,000/m3 is associated with decreased DFS (P = .01). When compared with other French-American-British (FAB) subtypes, presentation with FAB M4 or M5 morphology is significantly associated with relapse in multivariate analysis (P = .014). Other presenting features such as preparation with single-dose or fractionated TBI, interval from diagnosis to CR or CR to BMT, donor or recipient sex, and donor or recipient cytomegalovirus serology do not correlate independently with either DFS or relapse. When included in the stepwise multivariate analysis of presenting patient features, two posttransplant events, development of grades 2 to 4 acute graft-v-host disease (GVHD) (P less than .03) and development of interstitial pneumonitis (P less than .001), also correlate independently with poor DFS. Allogeneic BMT provides equivalent, prolonged DFS in both children and young adults when performed in first CR and should be considered the therapy of choice for all first CR patients under 45 years of age with a suitable donor. Continued efforts to prevent and treat acute GVHD and pneumonitis as well as efforts designed to prevent relapse in patients presenting with FAB M4 and M5 morphology should further improve outcome.