We report on the treatment of children and adolescents with acute lymphoblastic leukemia (ALL) in first relapse. The protocol focused on: (1) Intensive chemotherapy preceding allogeneic stem cell transplantation (SCT) in early bone marrow relapse; (2) Rotational chemotherapy in late relapse, without donor; (3) Postponement of cerebro-spinal irradiation in late isolated CNS relapse; and (4) Treatment in very late bone marrow relapse with chemotherapy only.
From January 1999 until July 2006 all 158 Dutch pediatric patients with ALL in first relapse were recorded. Ninety-nine patients were eligible; 54 patients with early and 45 with late relapse. Eighteen patients had an isolated extra-medullary relapse; 69 patients had bone marrow involvement only.
Five-years EFS rates for early and late relapses were 12% and 35%, respectively. For early relapses 5 years EFSs were 25% for patients transplanted; 0% for non-transplanted patients. For late relapses 5 years EFS was 64% for patients treated with chemotherapy only, and 16% for transplanted patients. For very late relapses EFS was 58%.
The introduction of multidrug chemotherapy in the treatment of children suffering from acute lymphoblastic leukemia (ALL) has been one of the major successes in pediatric oncology. Long-term event-free survival rates currently range from 75% upwards. However, this means that still a substantial proportion of children will eventually suffer from a relapse. Although most patients achieve a second remission; maintaining this remission is often less successful 1. In current relapse strategies re-induction therapy is often quite similar to induction therapy at primary manifestation of the disease. Treatment strategies to consolidate a second remission status are less uniform. In designing an effective treatment protocol for ALL relapse patients the duration of the first remission is important. Chemotherapy resistance is probably the most important factor in early relapse. Considering these factors the Dutch Childhood Leukemia Study Group (DCLSG, currently DCOG) designed a protocol to treat all ALL patients in first relapse, that is, Relapse Acute Lymphoblastic Leukemia 1998 Protocol (Rel-ALL'98 protocol). The aims of the protocol were: (1) to offer patients with an early isolated or combined bone marrow relapse of ALL an intensive treatment preceding allogeneic bone marrow transplantation using either an HLA-identical sibling donor or a matched unrelated bone marrow donor, (2) to offer patients with a late relapse chemotherapy followed by (if available) bone marrow from an HLA-matched sibling donor, (3) to offer patients with an early extra-medullary relapse without bone marrow involvement chemotherapy followed by (if available) a bone marrow from an HLA-matched sibling donor, (4) to offer, in order to avoid intolerance for methotrexate chemotherapy, patients with a late isolated relapse in the central nervous system prolonged chemotherapy and to postpone cranio-spinal irradiation until chemotherapy was finished, (5) to treat patients with a very late bone marrow relapse with chemotherapy only. The protocol was based on the most favorable treatment results, published by that time: St. Jude Protocol R11 2. In the Rel-ALL'98 Protocol, the following modifications were introduced: (1) Substitution of prednisone by the more effective dexamethasone during re-induction treatment; (2) All children in 2nd complete remission were eligible for stem cell transplantation (SCT) with marrow from an HLA identical sibling donor if available, except patients with a very late bone marrow relapse and patients with a late isolated CNS or testicular relapse. If an HLA-matched sibling donor was not available, patients with an early bone marrow relapse were eligible for transplantation with a matched unrelated donor or haplo-identical donor; (3) For patients with early extramedullary relapse without an HLA identical donor, maintenance treatment of Protocol R11 was substituted by the more intensive maintenance treatment of the POG Simal9 Pilot Protocol; (4) Boys with testicular relapse received high-dose methotrexate before the start of re-induction treatment and during maintenance treatment. Testicular irradiation was only used in patients with early relapse and no suitable donor; (5) For patients with late isolated CNS relapse, cranio-spinal irradiation was postponed until the end of maintenance treatment, in order to avoid intolerability for chemotherapy. In this manuscript, we report on the treatment results of this protocol.
All patients up to the age of 18 years experiencing a first relapse in the period from January 1999 until July 1, 2006 were eligible for the Rel-ALL'98 protocol. In the protocol early relapse was defined as a relapse within 30 months after attaining the first complete remission (CR1); late and very late relapses were defined as relapses between 30–60 months and more than 60 months duration of CR1, respectively. CR was defined as absence of clinical signs of disease in combination with a blast percentage <5% in a recovered bone marrow and blast negative CSF. Recovery of the bone marrow was assumed in case the white blood cell (WBC) count >2.0 × 109/L and platelets were >50 × 109/L.
All patients up to the age of 18 years experiencing a first ALL relapse in the Netherlands in the period from January 1999 until July 1, 2006 were eligible for treatment according to the Rel-ALL'98 protocol. The relapse had to be confirmed by the DCOG Central Laboratory. Patients were treated in the Pediatric Oncology and the Bone Marrow Transplantation Centers of the Academic Hospitals in the Netherlands. Informed consent was given.
Diagnosis and Definitions
Isolated bone marrow (medullary) relapse was defined as ≥25% blasts in the bone marrow and/or blasts cells in the peripheral blood without evidence of ALL in the CNS or testicle(s).
Combined bone marrow (medullary) relapse was defined as: ≥5% blasts in the bone marrow in combination with extramedullary ALL. Isolated CNS relapse was defined as: ≥5 cells/mm3 in the cerebro-spinal fluid with leukemic blasts (cytomorphological) without major blood contamination (≤20 erythrocytes/mm3) in two consecutive CSF samples taken with an interval of at least 24-hr, <5% blasts in the bone marrow, no blasts in the peripheral blood and absence of leukemic infiltrations elsewhere. Isolated relapse elsewhere (testicle, skin, bone orbita, mediastinum, lymph nodes, and tonsils): leukemic infiltrations demonstrated by biopsy (both microscopically and immunologically), with <5% blasts in the bone marrow, no blasts in the peripheral blood and absence of leukemic infiltrations elsewhere. Second complete remission was defined as: <5% blasts in a recovered bone marrow (WBC >2.0 × 109/L, platelets >50 × 109/L) and absence of leukemic infiltrations elsewhere. Immunophenotyping, cytogenetic analysis, and DNA index were performed as described previously 3.
Chemotherapy in late and very late relapses was based on the St. Judes R11 protocol 2. For early relapses the post-induction chemotherapy was based on the POG Simal9 Pilot Protocol. Data on the cytostatic scheme are given in Table I. In order to enable full-dose chemotherapy, irradiation in patients with central nervous system involvement was given after finishing all chemotherapy (24 and 15 Gy to cranium and spine, respectively). This strategy was based on the data of Mandell et al. 4 and van den Berg et al. 5. Patients with late testicular relapses were treated without irradiation and without surgery, as based on the results of van den Berg et al. 6. Patients with early testicular relapse received 24 Gy testicular irradiation assuming primary chemo resistance of leukemia cells in these cases. SCT was planned for any patient with an early relapse. At the start of the protocol only patients with HLA-matched sibling donors were eligible for transplantation. From September 1999, also early relapse patients with HLA-matched unrelated donors or with haplo-identical donors were eligible for transplantation. For patients with late relapses, SCT was still only planned in case a matched sibling donor was available. Very late relapse patients (i.e., a relapse >60 months after attaining first CR) were not considered for transplantation, as based on data reported by Niethammer et al. 7.
Table I. Chemotherapy in Relapse ALL 1998 Protocol
Days of administration
Maximum dosage 2 mg irrespective of body surface area.
In case of CNS involvement.
In case of testicular involvement.
Given in every second course in case of testicular relapse instead of oral methotrexate.
Cumulative dosage maximized to prevent cardiotoxicity.
Maintenance/consolidation treatment courses in (very) late relapse (given seven times)
Methotrexate, cytosar, prednisolone intrathecally
Dosages according to age
Day 1 (only in 1st and 3rd maintenance course)
Evaluation Criteria and Statistical Analyses
All results were updated until July 2008. The primary endpoints were overall and event-free survival. All analyses were performed on the basis of intention-to-treat. Survival rates were calculated according to the Kaplan–Meier method. Overall survival was calculated as the time from diagnosis of first relapse to date of death or latest follow-up (censored observations). Event-free survival was defined as the time from first relapse to death, re-induction failure, second relapse, or the occurrence of a second tumor. Patients not achieving second CR (re-induction failures) were included and considered as having an event on day 0. For patients alive at the latest follow-up (censored observations), event-free survival was calculated until this latest follow-up. Event-free survival and overall distributions were compared by using the log-rank test. Cox regression analysis was used for univariate and multivariate analysis in relation to event-free survival. A time-dependent Cox regression analysis was applied in order to compare transplanted versus non-transplanted patients. Statistical analyses with time-dependent Cox regression were performed with mstate library 8.
In total, 158 patients were reported. Early relapse was diagnosed in 99 patients, late relapse in 41 patients and very late relapse in 18 cases. One hundred forty-nine patients were eligible; 9 patients were not eligible, 8 because of age (>18 years), and 1 patient due to the pre-existing Fanconi anaemia. Of these 149 relapsed patients 99; 54 (early), 32 (late), and 13 (very late) were eligible and treated according to the protocol. Fifty of the 149 eligible patients were not treated according to the protocol; 43 of them had an early relapse, 6 had a late relapse, and 1 had a very late relapse. In 41 cases this was done on basis of investigators choice. In 37 of these 41 patients an early relapse had been diagnosed. Fifteen were treated according to a protocol of another cooperative group, 16 according to institutional protocols, 9 received palliative treatment (among them 4 with imatinib), 3 were based on their age referred to adult hematology departments and treated accordingly and in 7 cases no treatment was given. Initial treatment of the 99 included patients varied and originated from the DCOG-ALL7 9, DCOG-ALL 8 3, DCOG-ALL9 10, DCOG-ALL10, and Interfant 99 11 protocols. The majority of the patients (81%) had initially been treated according to the DCOG-ALL 9 protocol. Protocol DCOG-ALL10 is based on the current MRD-stratified BFM-protocol. Thirty-nine (50%) of the bone marrow relapses (isolated or combined) and 12 (92.3%) of the isolated CNS relapses had occurred early. Immune-phenotyping of the patients with an isolated or combined bone marrow relapse revealed that the vast majority had a common-ALL or pre-B-ALL phenotype; that is, 90% of the 89 patients with reported immune-phenotyping. Immune-phenotyping was not statistically different for early versus (very) late relapses (P = 0.18; Fisher exact test). Cytogenetic data were obtained in 55 patients. Karyotyping revealed in one case a Philadelphia chromosome. In only a limited number of cases (n = 6) rearrangement studies were reported; one of them had a TEL-AML1 rearrangement.
Of the 99 included patients, 27 died during re-induction or failed to reach CR. Five patients were withdrawn before achieving CR; in four of them this was due to treatment-related toxicity and one patient with an isolated bone marrow relapse was found ineligible since no specimens for review of remission status after induction therapy were centrally reviewed. Twenty-three patients with an early relapse were transplanted with stem cells. Twenty-one patients with late and very late relapses had a stem cell transplant, despite the fact that the protocol did not foresee a transplant in the very late relapse patients. Stem cell sources were in 74% bone marrow cells, in 19% stem cells originated from peripheral blood, and in 7% they were collected from cord blood. After achieving a 2nd CR 9 patients were treated according other protocols as based on investigators choice. Thirty-four patients experienced a second relapse; 28 of these patients have died at time of analysis. Of the early relapse patients 14 out of 54 patients are alive in 2nd CR. Of the late relapse patients 11 out of 32 patients are alive. From the 13 very late relapse cases 5 patients are alive. Figures 1–4 give the overall and event-free survival curves for early, late (inclusive very) relapses; also in relation to SCT. Figures 3 and 4 are corrected for moment of transplantation.
From the 54 early relapse patients, 19 patients did not achieve a complete remission after re-induction treatment or had progressed during re-induction treatment. In 7 of these 19 patients a second remission was reached on alternative chemotherapy; 1 died due to infection, the remaining died due to a further relapses, in 4 of them SCT had been attempted. In four early relapse patients re-induction treatment was stopped due to toxicity; three of these four died; the single patient who survived despite severe toxicity in conjunction with a candida esophagitis had suffered from an early isolated testicular relapse.
From the 45 late and very late relapse patients, 7 patients did not achieve complete remission after re-induction treatment or had progressed during re-induction treatment. In four of these seven patients a second remission was reached on different chemotherapy; however, all died despite SCT in two of them. In one very late relapse patient induction treatment was stopped due to toxicity.
From patients with late relapses, who did not receive a SCT five died due to a re-induction failure, one died due to septicemia, one died due to intracerebral bleeding in conjunction with a systemic aspergillosis. From the late relapsing patients who had a transplant 7 patients had an event after SCT (all second relapses); from the remaining 14 cases 3 died from infection. The single Ph+ ALL cases is alive. This patient has been censored at 30 months.
The covariates employed in the multivariate analysis were gender, age, WBC count at first relapse, immunophenotype, and time to occurrence of relapse. Time to relapse was divided in three categories: early, late, or very late relapse. While each of this relapse category had a significant impact on event-free survival (P-value equal to 0.004, 0.003, and 0.031, respectively), the remaining covariates showed no significant results. Transplanted versus non-transplanted patients were analyzed by employing a time-dependent Cox regression model. Time to relapse was then divided in two categories: early and late relapse. In the latter category late and very late relapse were collapsed together. For the category early relapse a risk factor equal to 0.466 (P-value 0.048) and 0.542 (P-value 0.12) was estimated for OS and EFS, respectively, implying that transplanted patients perform better. However, for the late relapse category, results do not follow the same trend: risk factors 6.91 (P < 0.001) and 3.02 (P = 0.012) for OS and EFS, respectively. Separate assessment for non-isolated extramedullary relapses were done, but were not essentially different.
Despite current cure rates above 75%, relapse is still considered as the most important obstacle in definite cure of children with ALL. In case of relapse second remissions are obtained in the vast majority of patients; for example, 95% in the UKALLR1 study 12. However, the chance to experience a second relapse is still high. The most important factor for a second relapse is the duration of the first remission. A cut-off point between early and late relapses is often made at 3–6 months after treatment cessation. Studies report for early relapses EFS rates ranging from 5% to 11% and for late relapses ranges from 19% to 57% 1, 13. Our findings are in line with these data. The exclusion of 50 patients from our study reflects that there are generally serious doubts on effective treatment for early relapses. The large majority of patients not entered in our study had an early relapse. In even one-third of these cases no treatment or palliative treatment was given. This is weakness in our study, but it is most probable that these patients had an even poorer chance of cure. Adding them to our study would not have improved our bad results in this subcategory of early relapse patients.
There is some debate in the literature on the prognostic factor of the WBC count and the presence of blasts in the peripheral blood at the time of relapse. T-cell immunophenotype is known to be a poor prognostic factor. A twofold rise in second relapse risk is claimed 14, 15. Minimal residual disease of leukemia (MRD; as measured by real-time PCR) both during second CR and after transplantation, has been reported to be a very strong prognostic factor for ultimate outcome 16. An additional prognostic factor for survival is the site of relapse and the combination with bone marrow involvement. The outcome of isolated CNS relapses is better than a CNS relapse combined with bone marrow involvement; whereas isolated testicular relapses even have a better prognosis. Chessells 17 reported survival rates of about 15%, 25%, and 65% for (combined) bone marrow, isolated CNS, and isolated testicular relapses, respectively. Our strategy for the various types of relapse was in line with the strategies in the UK and Germany; that is, early relapse cases are high risk, irrespective of localization of the disease at relapse, and abstaining from transplantation in isolated late relapse 13, 18.
Bone Marrow Relapses and Combined Relapses
According to our protocol, all patients with an HLA-identical sibling donor were advised to have a SCT, with the exception of patients with very late relapses and late isolated extramedullary relapses. In patients with early relapse, SCT treatment in CR2 showed a favorable outcome compared to non-transplant patients (OS rate of 37% and EFS of 25%, vs. 0% OS and 0% EFS, respectively). This difference (P < 0.001 for both OS and EFS) illustrates that treatment with only chemotherapy fails in these patients. We were not able to identify whether in all patients qualifying for bone marrow transplantation HLA-typing had been done; which may bias our results and conclusions. Statistical correction was done for interval between diagnosis and transplantation in order to make comparisons with non-transplanted patients. Unidentified differences, such as aggressiveness of the malignancy might still bias our conclusions. In our patients with a late relapse undergoing a transplant the outcome is poorer considering the risk factor of 6.91 and 3.02 for OS and EFS, respectively. Our data confirm reports in literature that SCT in late relapses is not beneficial. Since the majority of these patients die from a relapse of leukemia the benefits of the conditioning and the possible graft-versus-leukemia effect do not outweigh the benefit of the prolonged, rotational chemotherapy. The observed high number of early events in the stem cell transplanted group was in the majority of cases caused by infectious complications (often fungal; i.e., aspergillus).
We can confirm the initial data of Rivera et al., on which our protocol was based, claiming a 65% survival rate in cases of late relapse treated with only chemotherapy. Although a difference in outcome as compared with the patients in the Rivera et al. study might have been likely due to different treatment schemes given after initial diagnosis. Our patients were in a mix of BFM protocol-based treatment (49%) and Dutch protocol treated patients (36%), the latter often not treated with an anthracycline 2.
In extramedullary relapses, also a clear distinction has to be made for early relapses versus late relapses. Most authors claim that absence of bone marrow involvement is a favorable factor 19, 20. Relapses in extramedullary sites are often considered as relapses from malignant cells treated with suboptimal drug levels; due to their homing in these sanctuaries. However, it was shown that by increasing the dosages of methotrexate, outcome was not improved 21. In case of a testicular relapse, isolated relapse patients fare better with an EFS of 0.58 versus 0.28 for combined relapses. For CNS involvement, no statistical difference on outcome has been reported whether the bone marrow was involved or not 14. Since isolated extramedullary relapses are often late relapses; we are reluctant to generalize such a statement for patients with early relapse leukemia. Of note is a single report that patients with early isolated ALL in the CNS may have a poorer outcome than those with an early combined bone marrow/CNS relapse 13. In our study, separate analysis on non-isolated extramedullary relapses was done, but results were not essentially different. This is possibly explained by the fact that since only 3 out of 18 cases were late or very late relapses.
Only treatment results between autologous transplantation and chemotherapy have been reported. Results of autologous transplantation were not better than of only chemotherapy 22, 23. Some of these studies have flaws in the randomization of patients 12. The role of allogeneic transplantation has been reported to be superior to chemotherapy-only treatment options in various studies 24, 25. However, biases are not uncommon in these studies due to, for example, selective entry in the studies, length of interval between diagnosis and actual moment of transplantation and intervening complications. For late and very late isolated extramedullary relapses there is no indication for allogeneic bone marrow transplantation 26. Early relapse patients may benefit most from the transplantation strategy probably due to the lack of adequate chemotherapy 27, 28. Based on the similar outcome of SCT with HLA-identical donors versus chemotherapy in patients with a lower risk for a later relapse the role of transplantation is, however, debated 29. From the BFM group, Borgmann et al. 30 reported that intermediate risk patients had EFS rates after transplantation of 0.39 for; for non-transplanted patients this was 0.49; whereas transplanted high-risk patients had an EFS of 0.44 versus non-transplanted patients of 0.00. In line with such a statement are the data in a UK study including 256 patients, who were analyzed on basis of HLA-matched donor availability; no statistical benefit in outcome was seen 27.
Assessing transplantation data, the choice of the donor is of major importance considering survival data merits full attention, especially in relation with transplantation-related survival. Another impediment in comparing reports is the more favorable outcomes of non-sibling donors over the years. At the time our protocol was designed, transplants with matched unrelated donors were associated with severe toxicity and mortality. However, already in the first year after starting the protocol results of SCT with matched unrelated donors had significantly improved; as a consequence, the protocol was amended and early relapse patients became eligible for SCT with matched unrelated donors 14, 31. Confirming this change in policy are data from Bunin et al. reporting for children <15 years of age and a relapse within 6 months after CR1, the leukemia-free survival rate was 25%, for children with a longer CR1 the leukemia-free survival rate was 39% after transplantation with stem cells from unrelated donors. A matched-pair analysis of unrelated donor SCT versus chemotherapy revealed that only high-risk patients, defined as early isolated T-ALL (combined) bone marrow relapse benefited from unrelated donor transplantations 30. Comparing the outcome data in relation to the origin of the stem cell graft; that is, originating from HLA-matched siblings or HLA-matched non-sibling donors, no significant differences in outcome were noted in 124 patients with early bone marrow relapse leukemia 32. This is in line with the survival rate in our cohort showing better survival rates in chemotherapy only treated patients.
We conclude from our data that SCT shows improved outcomes for patients with early relapse leukemia. For patients with late relapse leukemia patients a more advantageous outcome is achieved for non-transplanted patients using this rotational chemotherapy scheme. Infections and relapses are the major causes of death in patients who have undergone transplantation and then suffered a disease relapse. The most important factor identified to date for survival is the duration of the first remission.
The authors want to thank Dr. M. Fiocco and Prof. Dr. T. Stijnen for their advise on statistical analysis and Dr. A. van der Does-van den Berg for critical review of the manuscript.