|Home | About | Journals | Submit | Contact Us | Français|
Although the majority of children with acute lymphoblastic leukemia (ALL) are cured with current therapy, the event-free survival (EFS) of infants with ALL, particularly those with mixed lineage leukemia (MLL) gene rearrangements, is only 30% to 40%. Relapse has been the major source of treatment failure for these patients. The parallel Children's Cancer Group (CCG) 1953 and Pediatric Oncology Group (POG) 9407 studies were designed to test the hypothesis that more intensive therapy, including dose intensification of chemotherapy, and hematopoietic stem-cell transplantation (HSCT) would improve the outcome for this group of patients.
One hundred eighty-nine infants (CCG 1953, n = 115; POG 9407, n = 74) were enrolled between October 1996 and August 2000. For infants with the MLL gene rearrangement and an appropriate donor, HSCT was the preferred treatment on CCG 1953 and investigator option on POG 9407 after completion of the second phase of therapy. Fifty-three infants underwent HSCT.
The 5-year EFS rate was 48.8% (95% CI, 33.9% to 63.7%) in patients who received HSCT and 48.7% (95% CI, 33.8% to 63.6%) in patients treated with chemotherapy alone (P = .60). Transplantation outcomes were not affected by the preparatory regimen or donor source.
Our data suggest that routine use of HSCT for infants with MLL-rearranged ALL is not indicated. However, limited by small numbers, this study should not be considered the definitive answer to this question.
Despite excellent cure rates in most children with acute lymphoblastic leukemia (ALL),1,2 infants with ALL continue to have poor event-free survival (EFS).3 The prognosis of patients with ALL with mixed lineage leukemia (MLL) gene rearrangements is particularly dismal, with failure related primarily to early relapse.4–8
The parallel Children's Cancer Group (CCG) 1953 and Pediatric Oncology Group (POG) 9407 studies tested whether intensification of early therapy would improve the outcome of infants with ALL. In addition to intensifying chemotherapy by calculating doses of chemotherapy by body-surface area (BSA) rather than weight, these protocols investigated the role of hematopoietic stem-cell transplantation (HSCT) in early first complete remission (CR1) for infants with MLL-positive ALL.
In this report, HSCT results are compared with the outcome of infants with MLL-positive ALL treated with the intensified chemotherapy protocol of studies CCG 1953 and POG 9407. Our data suggest that HSCT does not improve EFS or overall survival for these children.
The CCG 1953 and POG 9407 protocols were approved by the National Cancer Institute and by the institutional review boards at participating institutions. Informed consent was obtained from parents or legal guardians according to Department of Health and Human Services guidelines. CCG 1953 enrolled 115 infants and POG 9407 enrolled 74 infants (< 1 year of age) with newly diagnosed ALL between October 1996 and August 2000 (Fig 1).
The diagnosis of ALL was based on morphology, cytochemistry, and immunophenotyping performed at the individual institutions. Leukemic cell karyotype was determined at individual institutions as previously described.9
For CCG patients, molecular analyses were performed in a Clinical Laboratory Improvement Amendments–certified laboratory (University of Minnesota Molecular Diagnostics Lab, Minneapolis, MN). Detection of MLL gene rearrangement by Southern analysis and detection of t(4;11), t(11;19), or t(9;11) fusion transcripts by reverse transcriptase polymerase chain reaction were performed as previously described.10–12 For POG patients, analysis for MLL gene rearrangement was performed in reference laboratories at St Jude Children's Research Hospital (Memphis, TN) and the University of New Mexico (Albuquerque, NM) using marrow or peripheral blood in patients presenting with elevated WBC counts. Reverse transcriptase polymerase chain reaction was used to identify the MLL-AF4 common translocation fusion gene partner. Southern blot hybridization was used for detecting the rarer MLL gene rearrangements.
Patients on CCG-1953 received induction/intensification (phase 1) followed by reinduction (phase 2), reintensification (phase 3), consolidation (phase 4), three cycles of an intensified maintenance (phase 5), and four cycles of routine maintenance (phase 6; Table 1). Phases 1 and 2 of the POG 9407 protocol were identical to those on CCG 1953 to ensure that patients had been treated identically before HSCT. During phases 3 and 4, patients enrolled onto CCG 1953 received 24-hour infusions of methotrexate 33.6 g/m2, whereas patients enrolled onto POG 9407 received 24-hour infusions of methotrexate 4 g/m2 given only in phases 1 and 3. Patients enrolled onto POG 9407 did not receive phase 6 maintenance therapy (Table 1). The total duration of therapy on CCG 1953 was 110 weeks, whereas therapy on POG 9407 lasted only 46 weeks.
Infants with documented abnormalities in chromosome band 11q23 or MLL-positive ALL gene rearrangement were eligible for HSCT after phase 2 if they had a suitable donor. Suitable donors included an HLA-A, HLA-B, or HLA-DRB1 genotypic identical sibling; an HLA-A, HLA-B, or HLA-DRB1 phenotypic identical relative; a single HLA-A, HLA-B, or HLA-DRB1 antigen-mismatched relative; an HLA-A, HLA-B, or HLA-DRB1 identical unrelated individual (including cord blood); or a single HLA-A, HLA-B, or HLA-DRB1 antigen-mismatched unrelated individual (including cord blood). Class I antigens were determined by standard serologic methods, and class II antigens were determined by low-resolution DNA typing.
The protocol guidelines on CCG 1953 indicated that HSCT was the preferred treatment for patients with abnormalities in chromosome band 11q23 or MLL-positive ALL if an appropriate donor could be identified within the allowed time frame. On POG 9407, transplantation was an investigator option. All patients were to proceed to HSCT at completion of reinduction.
The protocol-prescribed transplantation regimen included cytarabine (100 mg/kg/dose every 12 hours, days −8, −7, and −6), cyclophosphamide (45 mg/kg/dose, days −7 and −6), methylprednisolone (33 mg/kg twice per day starting at midnight on days −3, −2, and −1), and total-body irradiation (TBI; given in two 1.5-Gy fractions daily to a total of 12 Gy, on days −3 to 0; Table 2).
The recommended dose of the HSCT infusion was 1 to 3 × 108 nucleated, unmanipulated bone marrow cells per kilogram of patient weight. Graft-versus-host disease (GVHD) prophylaxis was single-agent cyclosporine.
Enrollment was suspended in July of 1997 because of a 25% induction mortality rate predominantly resulting from infections associated with severe skin and mucosal breakdown. Because this was thought to be related to the dose of daunorubicin calculated by BSA rather than weight, an amendment was passed in November 1997 to calculate the dose based on weight. Despite some reduction in toxicity, the protocol was further amended in July 1998 changing the daunorubicin to an intravenous bolus on days 0 and 1 rather than a 48-hour infusion, resulting in a reduction in induction toxicity.
Protocol compliance on the transplantation arm was a major problem, with more than half the patients receiving nonprotocol preparative regimens. In the overall study analysis, these patients were censored at the time of the off-study transplantation.13 Designed as a prospective trial, this lack of compliance severely hindered the analysis of the trial as prospectively defined. Therefore, two statistical approaches were used to analyze the outcome of HSCT. The first approach was life-table comparisons of infants who received transplantation in CR1 both on (HSCT-ON) and off (HSCT-OFF) protocol compared with a control, or chemotherapy-only, group using the log-rank test and giving the EFS outcome in life-table form. For this evaluation, the control group consisted of 47 patients who were identified with MLL-positive ALL who were treated with chemotherapy and survived 143 days, the median time to CR1 transplantation. The second approach was a statistical regression model that allowed each individual transplantation patient to remain in the chemotherapy control group up to exactly the time that they underwent HSCT and then transfer into the HSCT group for the remaining follow-up time. Life-table estimates were calculated using the Kaplan-Meier method,14 with SEs of the estimates obtained using the method of Peto et al.15 The relative hazard rate (RHR) was estimated by the ratio of observed to expected events in the comparison groups.15,16 EFS events included leukemic relapse at any site and death during remission, whichever occurred first.
Of 189 patients enrolled, 132 were MLL positive. Of these patients, 53 underwent HSCT, and 47 received chemotherapy alone (control arm). Thirty-two patients did not survive to day 143, with deaths primarily caused by infection (26 deaths during induction; six deaths after induction). Twenty-five of 132 patients (seven from POG institutions, 18 from CCG institutions) identified as having MLL-positive ALL underwent transplantation in CR1 using the protocol-specified transplantation regimen. Twenty-eight patients (eight from POG institutions, 20 from CCG institutions) underwent transplantation in CR1 using non–protocol-specified transplantation regimens. Tables 3 and and44 list the clinical information of the patients in these two groups. Table 5 lists the characteristics of the following three patient groups: HSCT-ON, HSCT-OFF, and the chemotherapy-only control group. EFS comparing the 53 patients who underwent HSCT in CR1 compared with the 47 control patients treated with chemotherapy and who survived 143 days is displayed in Figure 2A. The 5-year EFS rate is similar for the HSCT group (48.8%; 95% CI, 33.9% to 63.7%; RHR, 1.15) and the chemotherapy control group (48.7%; 95% CI, 33.8% to 63.6%; P = .60). There were no statistically significant differences in EFS or survival for infants treated with chemotherapy compared with HSCT in subgroups defined by age, WBC count at diagnosis, or CD10 expression using either the life-table analysis or the statistical regression model.
There were more events in the first 6 months in the HSCT group (20 total events; 11 deaths as a result of transplantation-related morbidity and nine relapses) compared with the chemotherapy control group (nine total events; two toxic events and seven relapses). Overall survival at 5 years was 53.08% in the HSCT group versus 59.36% in the control group (P = .26).
The analysis using the statistical regression model includes all 132 patients enrolled onto CCG 1953 and POG 9407 with leukemia identified as being MLL-positive ALL. In this analysis, the RHR for HSCT is 1.454 (90% CI, 0.911 to 2.318; P = .19). However, 10% of the HSCT patients received transplantation more than 209 days from study entry.
The following exploratory analyses of EFS comparing transplantation-related factors were performed using life-table analysis. The 5-year EFS rate of HSCT-ON patients (40.0%; 95% CI, 19.8% to 60.2%) compared with HSCT-OFF patients (56.9%; 95% CI, 35.8% to 78.1%) was not statistically significant (log-rank P = .17; Fig 2B). The 5-year EFS rate for HSCT-OFF patients (56.9%; 95% CI, 35.8% to 78.1%) compared with the chemotherapy control group (48.7%; 95% CI, 33.8% to 63.6%) was also not statistically significant (log-rank P = .18; Fig 2B). Further analyses of the data suggest that the protocol-specified preparative regimen was not optimal (Tables 3 and and44).
When evaluating outcomes in patients who received HSCT on and off protocol and patients who received chemotherapy alone, the toxicities observed were quite varied. The incidence of grade 3 or 4 GVHD was 58% in HSCT-ON patients versus 21% in HSCT-OFF patients. There were twice as many toxic deaths during the first 100 days after transplantation in HSCT-ON patients (12 of 25 patients) than in HSCT-OFF patients (five of 11 patients). Toxic deaths were also more common in patients who received TBI (14 of 36 patients) compared with patients who did not receive TBI (five of 17 patients), although relapse was less common in patients who received TBI (four of 36 patients) compared with those who did not (five of 17 patients). The toxic death rate in remission was quite high in HSCT patients (17 of 53 patients) compared with the control arm (five of 47 patients).
There was no significant difference in EFS rate between patients who received TBI (50.0%; 95% CI, 31.6% to 68.4%) compared with patients who did not (46.3%; 95% CI, 21.2% to 71.4%; log-rank P = .91; Fig 2C) and no statistical difference in 5-year EFS based on the source of donor cells. The 5-year EFS rate was 60.0% (95% CI, 29.6% to 90.4%) for patients who received transplantation using a family donor compared with 50.0% (95% CI, 15.4% to 84.7%) for patients who received transplantation with unrelated donor bone marrow and 36.5% (95% CI, 3.6% to 69.4%) for patients who received transplantation with cord blood (log-rank P = .51; Fig 2D). All of these subgroup analyses are underpowered, and interpretation of the data should be made with caution.
The parallel CCG 1953 and POG 9407 studies were designed to test the hypothesis that intensification of early therapy would improve the outcome of infants with ALL. This was achieved by two approaches, dose intensification of chemotherapy and HSCT for infants with MLL-positive ALL early in CR1.
Dosing of chemotherapy for infants in previous protocols was determined by weight, not BSA, resulting in significantly lower doses of drugs for infants. To intensify the protocol therapy, BSA was used to calculate doses of all agents except vincristine and, after protocol amendment, daunorubicin. Dexamethasone was chosen as the corticosteroid based on reports of improved outcome with dexamethasone.17
At the time these protocols were developed, there were no reported large prospective trials of HSCT in infants with ALL. The promising results from the Fred Hutchinson Cancer Research Center (FHCRC) of 11 infants (seven of 11 infants were MLL positive) who received transplantation in first or second remission were the basis for evaluating this treatment modality. On the basis of this experience, a major objective of the CCG 1953 and POG 9407 studies was to prospectively evaluate the role of HSCT in the highest risk infants with MLL-positive ALL.4,18,19
The decision to include TBI in the pretransplantation preparative regimen was based on the favorable outcome of the infants treated at the FHCRC. Two reports of acceptable neurocognitive outcomes for infants treated with TBI-containing transplantation regimens were available at the time, although more recent reports do cite neurocognitive and hormonal disturbances as late effects associated with TBI.20–22 Cytarabine was incorporated based on superior results with regimens incorporating either hyperfractionated TBI (1.2 Gy three times daily for 4 days)23 or high-dose cytarabine.24 The combination of cytarabine with cyclophosphamide and TBI had been demonstrated to be tolerable in older children.25
A major impediment to successful transplantation in these patients was the lack of suitable family donors. To increase the number of patients with suitable donors, acceptable stem-cell sources included family or unrelated donors with up to a one antigen mismatch including unrelated cord blood units. Single-agent cyclosporine was the protocol-specified GVHD prophylactic regimen with the hope of an enhanced graft-versus-leukemia effect.
To our knowledge, this experience represents the largest evaluation of infants with MLL-rearranged ALL who received transplantation in CR1 compared with a concomitantly treated control group who received intensive chemotherapy. The results suggest that HSCT offered no advantage compared with intensive chemotherapy regardless of the choice of preparative regimen. The 5-year EFS rates for the 53 infants who received transplantation in CR1 and for the 47 infants treated with intensive chemotherapy were not dissimilar. Potential reasons for the lack of a difference in outcome between patients who did and did not received transplantation were the improved outcome of patients treated with the chemotherapy regimens and the suboptimal transplantation regimen.
The overall 3-year EFS rates for CCG 1953 (33.6%)13 and POG 9407 (40%)26 are comparable to those reported in the multi-institution Japan Infant Leukemia Study Group trial in which the 3-year EFS rate for infants with MLL-positive ALL was 44% (95% CI, 28.5% to 58.7%). However, for the 29 infants who underwent HSCT, the EFS rate was 64%.27
The CCG 1953/POG 9407 results contrast somewhat with the results of selected single-center experiences of HSCT in infants. The FHCRC reported a 3-year disease-free survival rate of 76% among 17 infants (three without documented MLL rearrangement) who received transplantation in CR1.22 The Children's Memorial Hospital experience included 16 infants who received transplantation (age, 1 to 15 months at diagnosis; five infants without MLL rearrangement) and demonstrated a 1-year EFS rate of 64% (95% CI, 40% to 88%) and a 3-year overall survival rate of 75% (95% CI, 47% to 92%).28 Differences in patient characteristics, such as older age, lower WBC count, and inclusion of infants without the MLL rearrangement, preclude strict comparison with our data.
The recently published international study Interfant-99 demonstrated a disease-free survival advantage for HSCT (n = 16) over chemotherapy alone (n = 71) in patients with MLL-positive ALL identified as high risk based on age less than 6 months and either poor response to corticosteroid at day 8 or WBC greater than 300 × 103/μL. This advantage was more evident with each successive year.3 However, an earlier review of 497 children, including infants and noninfants, with ALL and 11q23 rearrangements failed to prove an advantage for HSCT over intensified chemotherapy.4
More than half the infants included in this report did not receive transplantation according to the protocol specifications. When comparing HSCT with chemotherapy, bias is introduced when patients receive transplantation after prolonged periods in remission, resulting in unbalanced allocation of patients who experience relapse early to the chemotherapy arm. To address this problem, the protocol specified that infants with MLL-positive ALL and suitable donors were eligible for transplantation, which was to be scheduled within 4 months of enrollment. However, 60% of patients treated with the protocol-specified preparative regimen and only 29% of the patients treated with non–protocol-specified preparative regimens received transplantation within 4 months of enrollment (Tables 3 and and44).
Additionally, more than half of the patients were not treated with the protocol-specified preparative regimen. Five-year EFS rate was 56.9% (95% CI, 35.8% to 78.1%) for HSCT-OFF patients compared with 40.0% (95% CI, 19.8% to 60.2%) for HSCT-ON patients (P = .17). The median time to transplantation was 6 months for the HSCT-OFF group compared with 4 months for the HSCT-ON group. This difference in time to transplantation may partially explain the nonsignificantly better outcome of the off-protocol transplantation cohort.
Decreased transplantation-related morbidity also may have contributed to the better outcome of the off-protocol cohort. Grade 3 or 4 GVHD occurred more often in the HSCT-ON group. A more intensive GVHD prophylactic regimen may have improved EFS for this group. The preparative regimen, which incorporated high-dose cytarabine and corticosteroids, may have contributed to the higher transplantation-related morbidity and mortality of the protocol-specified transplantation regimen. TBI does not seem to have affected EFS (Fig 2C). The transplantation-related morbidity was quite high in this study compared with the chemotherapy-only control group and compared with other studies of HSCT in infants. With a less morbid transplantation preparative regimen, the EFS in the patients who received transplantation may have been more favorable compared with the control group.
Additional limitations to this study include the broad heterogeneity of preparative regimens and GVHD prophylaxis in patients who received transplantation with non–protocol-specified therapy. In addition, because the decision for HSCT was investigator option on POG 9407, investigator bias must be included as a potential limitation.
The conclusions of our study and the recent report of Interfant-99 establish several important points regarding the treatment of infant ALL. First, there is no role for the routine use of HSCT in infant ALL, and most infants clearly do not benefit from this approach. Interfant-99 demonstrated a modest advantage for a small subgroup of particularly high-risk infants. We did not see an advantage for HSCT in a similar population in our trial, but numbers were limited, making subgroup analyses difficult. Second, in both studies, there is a high early failure rate for the highest risk group, particularly those younger than 3 months of age. A primary goal of future trials must be to improve complete remission rates, decrease early relapse, and lessen the toxic death rate for this patient population.
Additional studies, such as gene expression profiles or minimum residual disease, might be useful for identification of subgroups most likely to benefit from HSCT (or, conversely, who would have a high cure rate with chemotherapy alone). As chemotherapy and transplantation regimens change, the comparison of HSCT and chemotherapy in infants with MLL-rearranged ALL will require re-evaluation.
Supported by Chair Grant No. U10 CA98543 and Statistics and Data Center Grant No. U10 CA98413 to the Children's Oncology Group from the National Cancer Institute, National Institutes of Health, Bethesda, MD.
Presented in part at the 43rd Annual Meeting of the American Society of Clinical Oncology, June 1-5, 2007, Chicago, IL.
The views expressed in this article are those solely of the authors and not necessarily those of the US Food and Drug Administration.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
Clinical trial information can be found for the following: NCT00002785.
The author(s) indicated no potential conflicts of interest.
Conception and design: ZoAnn E. Dreyer, Patricia A. Dinndorf, Bruce Camitta, Joanne M. Hilden, Jean E. Sanders, Franklin O. Smith, William G. Woods, Kamar Godder
Provision of study materials or patients: ZoAnn E. Dreyer, Patricia A. Dinndorf, Bruce Camitta, Nyla A. Heerema, Fred Behm
Collection and assembly of data: ZoAnn E. Dreyer, Patricia A. Dinndorf, Bruce Camitta, Harland Sather, Meenakshi Devidas, Joanne M. Hilden, Ron McGlennen, Cheryl L. Willman, Andrew J. Carroll, Fred Behm
Data analysis and interpretation: ZoAnn E. Dreyer, Patricia A. Dinndorf, Bruce Camitta, Harland Sather, Mei K. La, Meenakshi Devidas, Joanne M. Hilden, Jean E. Sanders, Ron McGlennen, Cheryl L. Willman, Fred Behm, Franklin O. Smith, William G. Woods,Gregory H. Reaman
Manuscript writing: All authors
Final approval of manuscript: All authors