Fax: (901) 495-3122
Bone marrow recurrence after initial intensive treatment for childhood acute lymphoblastic leukemia
Article first published online: 14 DEC 2004
Copyright © 2004 American Cancer Society
Volume 103, Issue 2, pages 368–376, 15 January 2005
How to Cite
Rivera, G. K., Zhou, Y., Hancock, M. L., Gajjar, A., Rubnitz, J., Ribeiro, R. C., Sandlund, J. T., Hudson, M., Relling, M., Evans, W. E. and Pui, C.-H. (2005), Bone marrow recurrence after initial intensive treatment for childhood acute lymphoblastic leukemia. Cancer, 103: 368–376. doi: 10.1002/cncr.20743
- Issue published online: 5 JAN 2005
- Article first published online: 14 DEC 2004
- Manuscript Accepted: 10 SEP 2004
- Manuscript Revised: 8 SEP 2004
- Manuscript Received: 19 JUL 2004
- American Lebanese Syrian Associated Charities (ALSAC). Grant Number: CA21765
- recurrent lymphoblastic leukemia;
- prognostic factors;
The authors studied the clinical outcome of 106 children with acute lymphoblastic leukemia (ALL) who developed a bone marrow recurrence as the first adverse event after contemporary intensified therapy.
Endpoints were the rates and lengths of second remission, the cumulative incidence of second hematologic recurrence, second event-free survival (EFS), and survival.
Bone marrow recurrences were isolated in 79 patients, and combined with an extramedullary site in 27 patients. The median time to recurrence was 2.6 years (range, 0.3–11.6 years). Seventy-six patients (71.7%) attained a second remission (median length, 0.7 year; range, 0.03–13.3 years). The 5-year survival probability among all patients was 24.2% ± 4.2% (standard error). On multivariate analysis, time to first disease recurrence and blast cell lineage were found to be independent predictors of a second EFS (P = 0.008 and P = 0.028, respectively). The 5-year EFS estimate in patients with an initial disease remission of ≥ 36 months was 42.6% ± 7.8% but was only 12.5% ± 3.9% among children with a short duration of disease remission (< 36 months). These estimates were 28.7% ± 4.9% and 5.0% ± 3.4%, respectively, for B blast and T blast cell lineages.
Despite acceptable long-term second EFS rates for certain subgroups, overall bone marrow recurrence after intensified first-line therapy for childhood ALL signals a poor outcome. Cancer 2005. © 2004 American Cancer Society.
Although the proportion of patients developing bone marrow recurrence of childhood acute lymphoblastic leukemia (ALL) has been reported to have decreased steadily,1, 2 this complication remains the most frequent cause of failure in frontline clinical trials, and the most difficult to treat. Clearly, the successes attained in treating patients with newly diagnosed ALL have not been reproduced in children with bone marrow recurrence.3, 4 To our knowledge, few studies to date have focused on the long-term results of treatment for patients whose disease recurs in the bone marrow after receiving intensified chemotherapy during first disease remission. Therefore, we analyzed prognostic factors and outcome data for patients who were treated intensively initially and at the time of disease recurrence. The major aims of the current study were to determine: 1) the rates and durations of second remission, 2) the patterns and cumulative incidence of second bone marrow recurrence, and 3) event-free survival (EFS) and overall survival rates after disease remission retrieval therapy. We also examined the impact of well defined biologic risk features on the duration of second EFS.
MATERIALS AND METHODS
Between 1984–1994, 711 patients were enrolled in 3 consecutive front-line studies of newly diagnosed ALL conducted at St. Jude Children's Research Hospital: Study 11 was conducted between 1984–1988, and reported a 5-year EFS rate of 72% ± 2.4; Study 12 was conducted between 1988–1991, and reported a 5-year EFS rate of 68% ± 3.4; and Study 13A was conducted between 1991–1994, and reported a 5-year EFS rate of 77% ± 3.3.5 Children with all subtypes of ALL, excluding mature B-cell leukemia, were eligible for these trials, regardless of age (infants included), initial leukocyte count, immunophenotype, or genotype.
The diagnosis of ALL (at presentation and disease recurrence) was based on morphologic findings and immunophenotyping with monoclonal antibodies directed toward lineage-associated antigens, as previously described.6 Chromosomal characteristics were described according to the International System for Human Cytogenetic Nomenclature.7 Immunophenotyping and cytogenetic studies were repeated at the time of disease recurrence. Complete disease remission was defined as the absence of leukemia blasts in the blood or cerebrospinal fluid and ≤ 5% lymphoblasts in bone marrow aspirates, evidence of regeneration of normal cells, absence of symptoms and signs of leukemia, and a normal clinical performance status.8 An institutional review board approved each protocol, and written informed consent was obtained from all patients.
Briefly, primary treatment included seven-drug induction regimens, followed by periodic courses of high-dose methotrexate and leucovorin rescue. Postremission treatment was comprised of eight-drug rotational combination chemotherapy in Study 11,9 pharmacologically targeted chemotherapy in Study 12,10 and rotational chemotherapy augmented with a course of reinduction therapy in Study 13A.11 Central nervous system (CNS)-directed therapy was comprised of intrathecal chemotherapy (methotrexate, hydrocortisone, and cytarabine) for all patients, and 18-gray (Gy) cranial irradiation for higher-risk patients at 1 year of disease remission. Treatment for hematologic disease recurrence was administered according to institutional secondary protocols used between 1985–2000 (information available on request). Treatment was comprised of three phases: reinduction of disease remission, consolidation, and continuation therapy. Reinduction of disease remission treatment was given within 4–6 weeks and featured a 4-drug combination with prednisone or dexamethasone, vincristine, Escherichia coli or polyethylene glycol (PEG)-asparaginase, and daunorubicin (in 1 study etoposide was substituted for the anthracycline, and teniposide was substituted in another study). Consolidation therapy varied (i.e., teniposide plus cytarabine, cyclophosphamide plus mitoxantrone, and fludarabine plus high-dose cytarabine [3 weeks]). Continuation therapy was comprised of multiagent rotational chemotherapy with two-drug pairs (6-mercaptopurine plus methotrexate, teniposide plus cytarabine, etoposide plus cyclophosphamide, and prednisone or dexamethasone plus vincristine with or without intervening courses of PEG-asparaginase). In children with no evidence of CNS leukemia, triple intrathecal chemotherapy (methotrexate, hydrocortisone, and cytarabine) was used as preventive treatment. Patients with combined bone marrow and testicular recurrence received a 24-Gy dose of testicular irradiation. Patients with combined bone marrow and CNS disease recurrence received 24 Gy of cranial irradiation and 15 Gy of spinal irradiation at the time of completion of the second course of therapy. The total duration of therapy was 2–2.5 years. At the time of second remission, hematopoietic stem cell transplantation (HSCT) was offered. Those patients who lacked an human leukocyte antigen (HLA)-matched sibling donor received grafts from a compatible unrelated donor, if available.
The cumulative incidence of bone marrow disease recurrence was estimated using the method of Kalbfleish and Prentice12 and tested by the method of Gray.13 All EFS and survival curves in the current study were constructed by the method of Kaplan and Meier and, when compared, were tested with the Mantel–Haenszel test.14 The only exception was comparison of the EFS of patients who received chemotherapy alone with that of patients who received chemotherapy and HSCT, which was tested using the Mantel–Byar method.15
Presenting features commonly accepted or suspected to be predictive of disease recurrence were tested for their association with first and second hematologic and EFS experiences and the attainment of a second complete remission (CR) among the 685 patients who achieved a CR in Studies 11, 12, or 13A. Bone marrow recurrence (isolated or in conjunction with an extramedullary recurrence) was the event of interest, with death from any cause and second tumors considered to be competing events. The time to a bone marrow recurrence was calculated from the date of CR to the date of bone marrow recurrence, or to the date of a competing event, whichever occurred first. The durations of second EFS and survival after the first hematologic recurrence were defined as the time from initial recurrence (isolated bone marrow recurrence or bone marrow combined with another site) until the occurrence of any event or until the date of death, respectively. In patients who were alive and free of any type of event, the durations were calculated until the date of last contact, and were censored at that time. Failure to achieve disease remission was considered an event at time zero. The same features were examined with the Gray proportional hazards model, adapted for the presence of competing risks16 to determine their prognostic influence on the cumulative incidence of second hematologic recurrence, or by the Cox proportional hazards model17 when considering second EFS. The final analysis of the data was performed on June 24, 2004; all patients were contacted within 2 years of the date of the final analysis. P values were two-sided and were not adjusted for multiple comparisons.
Of the 711 patients enrolled in St. Jude Studies 11, 12, and 13A, 685 (96.3%) attained a CR and were at risk for hematologic recurrence. At median follow-up times of 16.3 years, 12.1 years, and 9.5 years, respectively, 106 patients (15.5%) had developed a disease recurrence in the bone marrow as a first adverse event. Seventy-nine patients (74.5%) developed an isolated bone marrow recurrence, whereas 27 patients (25.5%) developed a combined disease recurrence (bone marrow and CNS in 18 patients; bone marrow and testes in 5 patients; bone marrow, CNS, and testes in 1 patient; and bone marrow and thymus in 3 patients). The similarity between the induction rates and 5-year cumulative incidences of bone marrow disease recurrence in Studies 11, 12, and 13A (13.2% ± 1.8% vs. 15.9% ± 2.7% vs. 11.1% ± 2.5%; P = 0.765) led us to combine these groups in subsequent analyses. Bone marrow recurrence was reported to develop at a median of 2.6 years (range, 0.3–11.6 years), and its 5-year cumulative incidence was 13.4% ± 1.3% (standard error). Table 1 shows the distribution of hematologic disease recurrences by study and cell lineage. Significant variation in phenotype and karyotype were not noted in the 103 patients classified at the time of diagnosis and disease recurrence. We analyzed a number of clinical and biologic risk factors that were predictive of hematologic recurrence during the first disease remission (data not shown). Briefly, subgroups with the lowest 5-year cumulative incidence rates (< 10%) of initial bone marrow recurrence included children with either hyperdiploid ALL (leukemic cell DNA index ≥ 1.16, equivalent to ≥ 53 chromosomes) or a TEL-AML1 genotype. Disease recurrences in these two subgroups of patients were usually late and were never preceded by an extramedullary disease recurrence.
|Initial treatment regimen|
|Study 11 (1984–1988) (n = 358)||Study 12 (1988–1991) (n = 188)||Study 13A (1991–1994) (n = 165)||Total (1984–1994) (n = 711)|
|No. (%) attaining a CR (%)||341 (95.3)||182 (96.8)||162 (98.2)||685 (96.3)|
|No. B-cell lineage ALL||284||155||141||580|
|No. T-cell lineage ALL||57||27||21||105|
|First hematologic recurrence|
|B-cell lineage ALL|
Response to Disease Remission Retrieval Treatment
Table 2 summarizes treatment results for all 106 patients who developed disease recurrence in the bone marrow. Seventy-six patients (71.7%) attained a second remission and could be followed for the duration of their second event-free remission, incidence of a second bone marrow recurrence, and overall survival. None of the 30 patients (28.3%) who failed to achieve a second remission survived. Forty-six patients received chemotherapy alone during the second remission and 30 received chemotherapy followed by HSCT (29 allogeneic transplants: 17 from an HLA-matched sibling, 10 from unrelated donors, 2 from mismatch family donors, and 1 autologous). The median time of HSCT from the second remission was 2.6 months (range, 0.7–13.07 months). The median duration of the second hematologic disease remission for all patients was 0.7 year (range, 0.03–13.3 years). A total of 32 patients developed a second bone marrow recurrence, and 31 had died at the time of last follow-up. No patient developed an extramedullary recurrence during the second remission. The 5-year survival estimate after the first bone marrow recurrence was 24.2% ± 4.2% overall (Fig. 1). The 5-year estimate of second EFS for patients who received versus those who did not receive HSCT were 38% ± 8.6% and 30% ± 6.8%, respectively (Fig. 2). Among the 30 patients who received HSCT, 12 were alive with no evidence of disease at the time of last follow-up.
|B-cell lineage (n = 86)||T-cell lineage (n = 20)||Total (n = 106)|
|Attained CR2 (%)||64 (74.4)||12 (60)||76 (71.7)|
|Chemotherapy vs. HSCT||37 vs. 27||9 vs. 3||46 vs. 30|
|Second bone marrow recurrence||24||8||32|
|Death in CR2||15||3||18|
|Died (%)||61 (70.9)||19 (95)||80 (75.5)|
|Alive (%)||25 (29.1)||1 (5.0)||26 (24.5)|
Prognostic Factors after Bone Marrow Recurrence
Table 3 lists the clinical and biologic presenting features that were examined for their impact on second remission induction, bone marrow recurrence, or second EFS. It is significant to note that, on multivariate analysis (not shown), the length of the first disease remission and blast cell lineage were found to be independent prognostic factors for both second bone marrow recurrence (P < 0.001 and P < 0.049, respectively) and second EFS (P = 0.008 and P = 0.028, respectively).
|Total (n = 106)||Second remission induction (n = 76)||Second EFS (%) (n = 106)|
|Yes (%) (n = 76)||P valuea||5-year estimate||P value|
|Age at diagnosis (yrs)|
|< 1||8||3 (4)||25.0 ± 12.5|
|1–9||58||43 (74)||26.1 ± 5.8|
|≥ 10||40||30 (75)||0.09||21.3 ± 6.3||0.06|
|Male||72||52 (72)||23.0 ± 5.0|
|Female||34||24 (31)||0.99||26.5 ± 7.2||0.51|
|White||86||61 (71)||25.1 ± 4.7|
|Black||18||14 (78)||22.2 ± 8.8|
|Leukocyte count (× 109/L)|
|< 50||73||56 (74)||25.3 ± 5.2|
|50–99||9||5 (56)||22.2 ± 11.3|
|≥ 100||24||15 (63)||0.23||20.8 ± 7.6||0.50|
|≥ 1.16||13||6 (46)||23.1 ± 10.1|
|Other||93||70 (75)||0.05||24.32 ± 4.5||0.55|
|No blast||66||50 (83)||27.5 ± 5.5|
|Any blast||40||26 (65)||0.27||18.3 ± 5.8||0.04|
|B-cell lineage||86||64 (74)||28.7 ± 4.9|
|T-cell lineage||20||12 (60)||0.27||5.0 ± 3.4||< 0.01|
|Absent||97||70 (72)||24.3 ± 4.4|
|Present||6||5 (83)||16.7 ± 10.8|
|Absent||100||73 (73)||24.5 ± 4.4|
|Absent||96||69 (72)||23.4 ± 4.4|
|Present||6||5 (83)||33.3 ± 15.7|
|Absent||43||33 (77)||31.2 ± 76.9|
|Present||7||7 (100)||66.7 ± 19.2|
|Standard risk||38||31 (82)||32.02 ± 7.7|
|High risk||41||30 (73)||0.43||25.3 ± 6.6||0.08|
|Length of first hematologic disease remission|
|≥ 36 mos||42||34 (81)||42.6 ± 7.8|
|< 36 mos||64||42 (66)||0.12||12.5 ± 3.9||< 0.01|
|Site of recurrence|
|Isolated||79||55 (70)||18.7 ± 4.4|
|Combined||27||21 (78)||0.42||40.7 ± 9.5||0.18|
Blast Cell Immunophenotype and Genotype
Results of second-line therapy according to blast cell lineage are shown in Table 2. Sixty-four of 86 patients (74.4%) with a B-lineage phenotype attained a second bone marrow remission, compared with 12 of 20 patients (60%) with T-lineage ALL (P = 0.197). Patients with T-lineage ALL experienced a worse outcome after bone marrow recurrence than patients with B-lineage disease (Table 2). The median duration of the second hematologic disease remission was 15 months (range, 0.5–161 months) versus 2.8 months (range, 0.9–140 months), respectively. Five-year estimates of the risk of second bone marrow recurrence (Table 3) were 37.1% ± 6.2% and 66.7% ± 15.7% for B-lineage and T-lineage cases, respectively (P = 0.026). Likewise, the 5-year second EFS estimate in patients with B-lineage ALL was found to be significantly better than that in the T-lineage subgroup (28.7% ± 4.9% vs. 5.0% ± 3.4%; P < 0.0005) (Fig. 3). In the current study, the prevalence of specific genotypes at the time of disease recurrence was: hyperdiploidy, 12.3% (13 of 106 patients tested); BCR-ABL, 5.8% (6 of 103 patients tested); MLL-AF4, 5.8% (6 of 102 patients tested); and TEL- AML1, 14.0% (7 of 50 patients tested). Five of the 6 patients with BCR-ABL-positive ALL attained a second remission, but at the time of last follow-up only 1 patient was still alive after > 9 years. Of the 5 children with MLL-AF4-positive ALL who attained second remission, 2 patients had survived for > 10 years. Second remissions were induced in 6 of 13 patients with hyperdiploid ALL, 3 of whom survived at least 9 years. All seven patients with recurrent TEL-AML1-positive ALL attained a second remission. At the time of last follow-up, 5 patients were still alive, 1 of whom had been reinduced recently; the remaining 4 patients had survived 5–12 years.
Sites of Disease Recurrence
The 5-year second EFS probability for patients with combined sites of disease recurrence (40.7% ± 9.5%) was twice that of patients with an isolated bone marrow recurrence (18.7% ± 4.4%), but this difference did not attain statistical significance (P = 0.181).
Duration of First Disease Remission
Among the 42 patients with a first CR of ≥ 36 months, 34 (81%) achieved a second CR, compared with 42 of 64 patients with shorter first CRs (66%) (P = 0.123). Similarly, longer first CRs were associated with a lower cumulative incidence of second bone marrow recurrence (19.0% ± 7.2% vs. 59.5% ± 7.8%; P < 0.001). Figure 4 depicts the EFS advantage (P < 0.0001) conferred by a longer first disease remission (5-year estimates of 42.6% ± 7.8% vs. 12.5% ± 3.9%). A predictive model based on whether patients had none, one, or more adverse events is shown in Table 4.
|Length of first hematologic disease remission (mos)||Lineage||Patient no.||5-year EFS (%)|
|≥ 36||B||39||46.3 ± 8.2|
|< 36||B||47||14.9 ± 4.9|
|< 36||T||17||5.9 ± 4.0|
Outcome after Disease Remission Retrieval Treatment
Thirty-two of the 76 patients who attained a second hematologic disease remission developed a second bone marrow recurrence. Of these, four patients attained a third disease remission but at the time of last follow-up, only one patient had survived. A total of 26 of the 106 patients with a hematologic disease recurrence (24.5%) were still alive at the time of last follow-up, 25 of whom were in a second remission (median time of 9.6 years) and 1 of whom was in a third disease remission (4 years).
After a long 13-year follow-up, the results of the current study document the poor outlook for children with ALL who develop a disease recurrence in the bone marrow while on contemporary protocols of intensive therapy. In contrast to historic second remission induction rates of 80–90%,18, 19 only 72% of the patients in the current study attained a new disease remission. The 5-year survival estimate after bone marrow recurrence was 24.2% ± 4.2% overall. Investigators of the Children's Cancer Group20 reported comparable results for patients with isolated (n = 642; 6-year survival rate of 20% ± 2%) or combined (n = 120; 6-year survival rate of 29% ± 5%) bone marrow recurrences. It is significant to note that in that analysis, as in the current one, the specific protocol of frontline therapy did not appear to be correlated with response to secondary treatment.
The 106 patients who developed a disease recurrence in the bone marrow were not necessarily classified initially as having a higher risk of disease recurrence. Indeed, patients assigned to the National Cancer Institute standard-risk group and those considered low risk by virtue of a given genotype also were among the patients who developed a recurrence, emphasizing the need to improve risk models at the time of diagnosis. To this end, several groups21–24 currently are addressing the question of whether individual treatment adjustments early in therapy, based on the detection of minimal residual disease, may avert clinical recurrence. In principle, similar studies performed early during second remission might aid in the selection of postinduction treatments that are more likely to succeed when the tumor burden is minimal. It is interesting to note that studies of minimal residual leukemia in bone marrow have shown a significantly higher prevalence of detectable minimal residual disease and a higher level at the time of second remission as opposed to first disease remission of ALL.25 Levels of minimal residual disease at the time of induction of a second remission also have been reported to be associated with subsequent outcome.25, 26
Among the various factors analyzed for their influence on the risk of a second disease recurrence and survival experience, only the length of first disease remission and blast cell lineage demonstrated independent prognostic significance. In our experience27 and that of others20, 28 increasingly better treatment responses usually are related to longer initial disease remission lengths. Taken together, leukemia therapists have used the strength of late bone marrow recurrence as a predictor of subsequent outcome to design treatment strategies after disease recurrence.3, 29 Approximately two-thirds of the 106 patients in the current study developed early disease recurrences (length of first disease remission of < 36 months) and one-third developed a late disease recurrence (length of first disease remission of ≥ 36 months). Approximately 81% of the patients in the latter subgroup entered second remission, compared with only 66% in the former subgroup, and had a second EFS probability at 5 years that was > 3 times better than the estimate for patients with early disease recurrence (42.6% ± 7.8% vs. 12.5% ± 3.9%.
The subgroup with shorter initial disease remissions included a high proportion (approximately 27%) of patients with T-lineage ALL, whose 5-year survival estimate was only 5%. German investigators reported that a T-cell phenotype at the time of disease recurrence is associated with very poor prognosis.30 Therefore, in the design of future protocols for recurrent ALL, it may be prudent to consider different therapeutic strategies for patients with recurrent T-lineage versus B-lineage ALL. Equally poor were the results for children with BCR-ABL-positive recurrent ALL. Beyermann et al.31 reported a second remission rate of 60% and a survival probability of < 10% among 20 patients with recurrent BCR-ABL-positive ALL who were treated in the ALL-REZ BFM 87 and 90 trials. Conversely, of seven patients with recurrent TEL-AML1-positive ALL, five were still alive at the time of last follow-up, four of whom were long-term survivors. The prevalence and prognostic significance of the TEL-AML1 fusion gene at the time of first recurrence of ALL remains controversial.32, 33 Seeger et al.34 reported what to our knowledge is one of the largest patient subgroups with this genotype. They found a TEL-AML1 prevalence of 18.7% among 53 patients with B-lineage ALL in first recurrence, and a trend toward a better outcome, but concluded that this finding lacked independent prognostic significance.
The findings of the current study confirmed that children with combined bone marrow and extramedullary disease recurrences fare better than those with an isolated bone marrow disease recurrence, but unlike other authors35 we were unable to demonstrate a statistically significant difference, most likely because of the small size of the combined-recurrence subgroup. Likewise, although HSCT may be an effective treatment during second remission4 we were unable to demonstrate a statistical advantage in the current study. Clearly, however, the current study was not designed to compare the efficacy of two different therapies.
Advances in front-line therapy for pediatric ALL demonstrate that a cure may be obtained with a variety of treatments.36–41 Nevertheless, regardless of the regimen selected, bone marrow recurrence remains the most common and most ominous adverse event. The findings of the current study demonstrate that considerably less progress has been made in the treatment of recurrent ALL, particularly in certain high-risk subsets of patients. What, then, are the viable options? Further intensification of chemotherapy appears to be a hazardous and likely unrewarding choice. Rather, we should most likely view the majority of patients with ALL in second remission, especially those with an early first recurrence, as candidates for innovative therapies, including drugs that target critical oncogenic pathways, as have been developed for acute promyelocytic leukemia42 and chronic myeloid leukemia.43 In this regard, FLT3 inhibitors currently are being tested in cases with MLL rearrangements.44 Additional help may come from the use of emerging biotechnologies to identify novel therapeutic targets.45–47
The authors thank Mr. John R. Gilbert, senior medical editor, for his valuable input during the preparation of this article.
- 3Six-year experience with a comprehensive approach to the treatment of recurrent childhood acute lymphoblastic leukemia (ALL-REZ BFM 85): a relapse study of the BFM group. Blood. 1991; 78: 166–1172., , , et al.
- 7MitelmanF, editor. International System for Human Cytogenetic Nomenclature (ICSN): Recommendations of the International Standing Committee on Human Cytogenetic Nomenclature, Memphis, Tenn., October 1994. Basel: S. Karger, 1995: 114.
- 12The statistical analysis of failure time data. New York: John Wiley & Sons, Inc., 1980: 163–188., .
- 17Regression models and life tables. J R Stat Soc B 1972; 34: 187–220..
- 30Outcome after relapse of T-cell acute lymphoblastic leukemia in childhood. A report from the BFM Relapse Study Group [abstract]. Med Pediatr Oncol. 2000; 35: 250., , , et al.
- 38Intermediate-dose intravenous methotrexate with intravenous mercaptopurine is superior to repetitive low-dose oral methotrexate with intravenous mercaptopurine for children with lower-risk B-lineage acute lymphoblastic leukemia: a Pediatric Oncology Group Phase III Trial. J Clin Oncol. 1998; 16: 246–254., , , et al.