Outcomes and prognostic factors in adolescents and young adults with ALL treated with a modified BFM-90 protocol

Acute lymphoblastic leukemia (ALL) in adolescents and young adults (AYAs) accounts for less than one-fourth of the total ALL cases but leads to 80% of ALL-related deaths.¹  The AYA cohort has inferior outcomes compared with the younger children, who achieve cure rates of >90%.^2-6  They fall at the transition between pediatric and adult populations because a uniform treatment strategy has never been followed and they are underrepresented in clinical trials.⁷  An increase in the incidence of high-risk subtypes, such as early thymic precursor (ETP) ALL, Philadelphia-positive (Ph⁺), and Ph-like ALL, also accounts for the inferior outcomes.^8,9  Pediatric protocols deliver higher cumulative doses of nonmyelosuppressive drugs like asparaginase, vincristine, and steroids, whereas the adult protocol uses a higher dose of cytarabine and hematopoietic stem cell transplantation (HSCT). Treatment in this age group with pediatric protocols resulted in superior outcomes compared with adult protocols.^7,10-14  In addition to the inferior outcomes with the adult protocols in the resource-poor setting,^15-19  a major hindrance is its reliance on HSCT as a modality of dose intensification. HSCT frequencies are much lower in the developing world in comparison with the developed world.^20,21  On the other hand, the use of pediatric dose-intensification strategies in adults is often compounded by increased risk for infections and other complications such as avascular necrosis, hepatitis, and pancreatitis.^22-26 

Herein, we report the single-center experience with a modified BFM-90 protocol, a pediatrics-inspired regimen, in AYA ALL patients aged 15 to 25 years. We aimed to evaluate the efficacy and safety and identify the impact of prognostic factors (minimal residual disease [MRD], baseline white blood cell [WBC] count, cytogenetics, cerebrospinal fluid [CSF] status, toxicities, and infections) on survival outcomes.

We included treatment-naive patients (between 15 and 25 years of age) with ALL (B-cell [B-ALL] and T-cell [T-ALL]) who were treated at the Tata Memorial Centre, a tertiary cancer care center in India, between January 2013 and December 2016. The study was approved by the Institutional Ethics Committee (IEC/0319/3209/001). Patients who had received up to a week of steroids or 1 dose of vincristine were also included in the study. Data were retrieved from the electronic medical records.

The diagnosis of ALL was established by peripheral blood and /or bone marrow morphologic assessment and flow cytometry (FCM). Cytogenetics by fluorescent in situ hybridization (FISH) and ploidy analysis was performed for risk stratification. For ploidy analysis, Giemsa-stained metaphases were captured using the Applied Spectral Imaging (Netser Sereni, Israel) GenASIs platform; modal chromosome counting was done on 20 to 30 metaphases and reported according to the International System for Human Cytogenetic Nomenclature (ISCN). Central nervous system (CNS) involvement was ascertained by CSF examination with both cytomorphology and FCM; the CNS status was defined as per the standard criteria.²⁷  A mediastinal mass or lymphadenopathy with <25% lymphoblasts in the bone marrow was defined as T-lymphoblastic lymphoma (T-LBL).²⁸ 

Patients were risk-stratified based on the baseline WBC count (>30 × 10⁹/L for B-ALL and >100 × 10⁹/L for T-ALL), cytogenetics [high-risk cytogenetics included t(9;22), t(4;11), t(1;19)], and ploidy (hypodiploidy).^29,30  Patients with any of these factors were risk-stratified as high risk and others as standard risk.

Patients were administered a modified Berlin-Frankfurt-Münster 90 (BFM-90) protocol.³¹  The following modifications were made: native Escherichia colil-asparaginase was administered at a dose of 10 000 U/m² either as a deep intramuscular (IM) injection or as an IV infusion. During the protocol M phase, high-dose methotrexate (HDMTx) was administered at a dose of 3 g/m² along with leucovorin rescue. During phase 2 reinduction, all patients received cranial radiation therapy (RT) at a dose of 18 Gy per 10 fractions for CSF⁺ cases or 12.6 Gy per 7 fractions for CSF⁻ cases. Patients with the diagnosis of T-LBL did not receive cranial or mediastinal irradiation. Intrathecal methotrexate (12 mg) was administered during phase 1a, 1b, protocol M, and reinduction phase 2. In comparison with the original BFM protocol in the ALL-BFM-90 trial, we used reduced doses of 6-mercaptopurine (6MP) during phase 1b and maintenance because of the poor tolerance with the use of standard doses (60 mg/m²) of 6MP in our patients. The starting dose of 6MP was determined by the body surface area (BSA) of the patient (Table 1) and the doses were subsequently titrated based on tolerance. Similarly, with the concerns of increased toxicity with 5 g/m² HDMTx, 3 g/m² was used for our patients. Standard dosing for HDMTx ranges from 2 to 5 g/m^232-34  During reinduction phase 2, in comparison with the original protocol, either 6 thioguanine (6TG) or 6MP were used. Change to 6MP was due to the nonavailability of the 6TG in India from 2017 onward.

Treatment protocol for patients with Ph⁺ ALL included tyrosine kinase inhibitor (TKI), either imatinib at a dose of 600 mg once daily or dasatinib at a dose of 100 mg once daily after the cytogenetic confirmation. Given the documented evidence of increased toxicity associated with l-asparaginase and TKI combination, l-asparaginase was omitted from all phases in patients with Ph⁺ ALL.^35,36  Further details of the protocol can be found in Table 1.

As part of the protocol, cotrimoxazole prophylaxis was initiated from the third week of phase 1a induction. Given the interactions between azoles and vinca alkaloids, fungal prophylaxis with azole was not administered. Antibacterial prophylaxis with quinolones was not used as a standard practice because of the high incidence of quinolone resistance in our population.³⁷ 

Response assessment was performed at the end of phase 1a induction with bone marrow morphology and MRD assessment by FCM and cytogenetics studies (those with baseline abnormalities). Bone marrow response was assessed using the M bone marrow criteria for ALL; if M0 to M1 status (blast cells <5%) was achieved, the patient was considered to be a responder (in clinical remission) as reported by Stock et al.³⁸  MRD assessment was performed at the end of phase 1a induction on the first-pull bone marrow aspirate sample using 8- to 10-color FCM assays.^39-41  Further details of the MRD assessment are provided in supplemental Appendix 1. An MRD level of <0.01% was considered as negative and ≥0.01% was considered as positive.⁴²  In case MRD was positive at the end of phase 1a induction, repeat MRD assessment was performed at subsequent time points (either phase 1b, M phase, or reinduction). In patients with Ph⁺ ALL, in addition to FCM-based MRD, MRD assessment was also performed using real-time quantitative polymerase chain reaction (qPCR) testing for BCR-ABL transcripts. High-risk patients with Ph⁺ ALL, MLL translocation, or postinduction MRD⁺ status were counseled for allogeneic HSCT. The treatment was not modified in poor prednisolone responders or patients with positive MRD postinduction. The treatment was delivered as an outpatient except for HDMTx. Patients were also admitted for the management of grade 3/4 toxicities that warranted hospitalization.

Toxicities that were captured include infections (bacterial or fungal or viral), pancreatitis, hyperglycemia, avascular necrosis, posterior reversible encephalopathy syndrome (PRES), and seizures. Infections leading to discontinuation of therapy, change of protocol, or death of the patient were recorded. In patients suspected of having pneumonia (fungal or bacterial), noncontrast computed tomography (NCCT) imaging of the thorax was performed along with serum galactomannan. Antifungals were initiated in those with radiologic suspicion or positive galactomannan. Whenever feasible, bronchoalveolar lavage fluid analysis was performed to confirm the etiology of the infection. Fungal infections were defined as possible, probable, and proven fungal infection as per the revised European Organisation for Research and Treatment of Cancer (EORTC) and Mycoses Study Group Education and Research Consortium (MSGERC) definitions.⁴³ 

Relapse was diagnosed in patients who had achieved complete remission (CR) and subsequently developed a recurrence of the disease. Relapses were classified as medullary, extramedullary (CNS, testicular, or other sites), or combined. Time to relapse was calculated from the date of CR. Relapse was classified as very early relapse, early relapse, and late relapse according to the BFM classification of relapsed ALL.⁴⁴ 

The primary end points of the study were event-free survival (EFS) and overall survival (OS) for the overall population and separately within B-ALL and T-ALL subsets. Secondary end points were Complete response (CR) rate, MRD-positive rate, the impact of MRD on EFS and OS, the impact of known poor-risk features (like elevated WBC count, high-risk cytogenetics, and hypodiploidy) on EFS and OS, and the impact of toxicities on EFS and OS.

EFS was defined as the time from registration to any event defined as induction failure, relapse, or death. OS was defined as the time from registration to death resulting from any cause. Patients who were deemed palliative after relapse and subsequently lost to follow-up were counted as an event for OS analysis. Patients lost to follow-up but in remission were censored on the date of last follow-up. EFS and OS were analyzed using the Kaplan-Meier method. Univariate analyses by log-rank tests and multivariate analyses by the Cox proportional hazards model was used to assess the impact of known prognostic variables on EFS and OS.

We registered 463 patients with ALL during the study period, of whom 349 (273 male [78.2%], 76 female [21.7%]) received the modified BFM-90 protocol. The remaining 114 patients could not be included as they opted to take treatment elsewhere or were lost to follow-up before treatment initiation. The baseline parameters between these 2 groups of patients are similar and are presented in Table 2.

All the 349 patients were treated with the modified BFM-90 protocol (Table 1). The course of treatment is summarized in Figure 1. Of the 37 patients diagnosed with Ph⁺ ALL, imatinib was used in 28 patients and dasatinib in 9 patients.

Among the 349 patients, 285 patients (81.6%) achieved CR at the end of induction. There were 21 deaths (6%) and 43 discontinued treatment before phase 1b (12%). Almost half of the induction deaths and discontinuation could be attributed to life-threatening infections (30 [47%]). Other reasons included patients not in remission (22 [34%]), CNS bleed (1 [1.5%]), and lost to follow-up (9 [(14%]). Twenty-five patients did not achieve CR even after completion of phase 1b. Of the 25 patients, 12 were T-ALL and 13 were B-ALL. Of these 12 patients with T-ALL, 8 were ETP/near ETP-ALL.

Among the 349 patients, MRD assessment was available for 272 patients. MRD was negative in 167 patients (61.39%) at the end of phase 1 induction. An additional 61 patients achieved MRD⁻ status subsequently (phase 1b, 34; postprotocol M, 21; postreinduction, 6). Details of the time points of the repeat MRD assessment are summarized in Table 3. There was no difference in the postinduction MRD⁺ rates between B-ALL (30.2%) and T-ALL (30.4%). Molecular response in patients with Ph⁺ ALL is summarized in Table 4.

After a median follow-up of 41 months, 79 patients relapsed (22.6%) and 82 patients died (23.5%). Among the 79 patients who relapsed, 54 had a medullary relapse (68%), 18 had CNS relapse (23%), 5 had both medullary and CNS relapse (6.3%), 1 had both medullary and testicular relapse (1.2%), and 1 had other site of relapse. The median time to relapse was 14months (interquartile range [IQR], 8-23 months). Very early relapse (<18 months from diagnosis) occurred in 48 patients (61%), early relapse (>18 months from diagnosis but <6 months from completion of therapy) occurred in 27 patients (34%), and late relapse (>6 months after completion of therapy) occurred in 4 patients (5%). Among the 82 patients who died, 60 patients were not in CR (73%) and 22 patients were in CR (26.8%).

The survival outcomes are summarized in Table 5. Median OS and EFS were not reached. After a median follow-up of 41 months, the 3-year EFS and OS of the cohort were 59.4% and 61.8%, respectively. ETP-ALL and near ETP-ALL had inferior EFS and OS in comparison with the non-ETP T-ALL. Survival curves are represented in Figures 2 and 3.

Among these 349 patients, only 15 underwent HSCT. Details of these patients are described in Table 6.

A univariate Cox regression analysis identified that baseline CNS involvement (CNS 2 or 3), positive postinduction MRD, positive MRD beyond induction, and development of infections were associated with poorer outcomes (EFS and OS). On multivariate analyses, positive postinduction MRD and positive MRD beyond induction were associated with poorer survival. Univariate and multivariate analyses are summarized in Tables 7 and 8.

In total, 164 patients had developed infections (all grades) and 104 developed fungal nodules. There were a total of 21 induction deaths, which were predominantly attributable to infection (19 [82.6%]). The occurrence of infections declined as we proceeded from phase 1a induction to protocol M. During the reinduction phase, there were a total of 6 episodes of life-threatening infections, 5 of which were fatal. The number and severity of the infections during the various phases are summarized in Table 9 and 10.⁴³  During the treatment, commonly encountered treatment-related toxicity included cortical venous thrombosis (6.3%), pancreatitis (4.8%), avascular necrosis (1.1%), hyperglycemia (2.5%), PRES (1.4%), and seizures (12.8%) (Table 11).

This study shows that AYAs with ALL treated with a pediatrics-inspired protocol achieved 3-year EFS and OS of 59.4% and 61.8%, respectively. Our study cohort belongs to the 15- to 25-year-old age group, which is underrepresented in literature. In previous trials utilizing pediatric protocols, the upper age limit varied from 18 years to 39 years, underlining the ambiguity in the definition of this age group in literature^45-47  (Table 12). Other studies performed in the 15- to 30-year-old age group showed EFS and OS ranging from 61% to 86% and 66% to 88%, respectively.^14,48-53 

Improvement in survival across the various age groups has been attributed to intensification of therapy after remission induction, and various approaches have been used over the last 50 years.^54,55  Various retrospective analyses and trials performed in the 15- to 30-year-old age group have used pediatric protocols with intensive postremission treatment consisting of vincristine, l-asparaginase, cytarabine, cyclophosphamide, and etoposide resulting in varied EFS and OS rates of 61% to 86% and 66% to 88%, respectively.^14,48-56  Strategies used to dose intensify in these studies are summarized in supplemental Table 1. Even though these trials have proven the benefit of dose intensification, the representation of the AYA population was minimal with many of the trials excluding or having underrepresentation of patients with high-risk features (baseline WBC count and cytogenetics).^14,48-53,55  So a clear strategy for dose intensification is not available from literature in the AYA population. In our patients, postinduction therapy consisted of phase 1b (4 blocks of cytarabine, 2 doses of cyclophosphamide, and intrathecal methotrexate) followed by protocol M (HDMTx with leucovorin rescue) followed by delayed intensification (vincristine, dexamethasone, adriamycin, and l-asparaginase). Dose intensification was kept uniform for standard risk as well as for high risk irrespective of the postinduction MRD status. We planned for all patients to receive prophylactic cranial RT except the patients with a diagnosis of T-LBL. Vincristine or prednisolone pulses were not used during the maintenance. With the use of intensive pediatric and pediatrics-inspired regimens, the need for HSCT has declined in the AYA population. In our cohort, only 15 patients (4.3%) underwent HSCT, whereas the other trials utilizing pediatric-inspired protocols had HSCT rates ranging from 3% to 30%. Children’s Cancer Group (CCG; 1882-1901) and Dutch Childhood Oncology Group (DCOG; 6-9) retrospective analyses were the 2 studies that had transplant rates less than our cohort. Utilizing a risk-adapted approach, the 7-year EFS and OS rates in the CCG 1882-1901 were 63% and 67%, respectively.^7,48,55  Using a dose-intensified approach, 5-year EFS and OS rates in the DCOG 6-9 were 69% and 79%, respectively.¹⁰ 

Our cohort comprised 39% T-ALL and 61% B-ALL subtype, substantiating the fact that with increasing age the incidence of T-ALL increases.^{14,38,48-50,57}  Hyperleukocytosis was found in 10% of our patients. With increasing age, good-risk cytogenetics like t(12;21) decreases and the poor-risk cytogenetics like t(9;22) and t(1;19) increases, which is evident from our cohort with the incidences being 0.6%, 11%, and 4.6%, respectively.^14,48,50  CNS involvement in our cohort is 5.5%, which is similar to the other studies with a reported incidence of 5% to 25%. The use of a pediatric protocol has uniformly resulted in CR rates >90% in the previous trials.⁵⁸  In our population, the CR rate was 81.6%. The low CR rate is due to the infections that led to induction deaths (6%) and treatment discontinuation (3.4%).

Our analyses prove that postinduction MRD trumps all other baseline disease-related characteristics in determining the overall outcome of the patients. Among the 272 patients who achieved CR at the end of phase 1 induction, 167 (61.4%) were MRD⁻ and 105 (38.6%) were MRD⁺, which is similar to what is described in literature.^59,60  The EFS and OS of the MRD⁻ (compared with MRD⁺) patients in our cohort were 77.4% (vs 41%) and 79.2% (vs 44.8%), respectively (Table 5). The odds ratios for occurrence of relapse or death in a patient with a positive postinduction MRD was 2.4 and2.583, respectively (Table 8). Similar findings were reported from the US intergroup trial, the MD Anderson Cancer Center (MDACC) trial, and the German Multicenter ALL Study Group (GMALL) July 2003 trial.^61-63 In the GMALL June 1999 study, which included only standard-risk adult (15-64 years) patients (MLL, Ph⁺, WBC count >30 × 10⁹/L were excluded), patients with MRD⁺ status(>0.01%) had a 2.4-fold higher risk of relapse in comparison with the patients with MRD <0.01%.⁴²  MRD assessment performed in 256 pediatric T-ALL patients treated with the MCP841 protocol in our institute also showed that MRD⁺ status had a hazard ratio (HR) of 2.01 and 1.64 for 3-year EFS and OS.⁶⁴  MRD assessment in 167 patients (1-41 years of age) treated with the MCP-841 protocol from the Regional Cancer Centre in western India also showed that MRD negativity was associated with longer survival (29 months) in comparison with MRD⁺ patients (22 months).⁶⁵  As per the literature, MRD negativity rates increased as the intensive therapy continues from 6% on day 11 to 36% on day 26 to 70% on day 71.^66,67  Even in our cohort additional patients achieved MRD negativity by the end of phase 1b induction or consolidation (Table 3). In the end, only 38 patients (10.8%) remained MRD⁺. Achievement of MRD negativity during and after consolidation does confer a survival advantage in comparison with those who remain MRD⁺, however, they fare worse than those who became MRD⁻ postinduction (Table 5). Similar to the pediatric population, postinduction MRD can be used as a tool to further intensify the subsequent treatment in patients with positive MRD. Also, as the achievement of negative MRD results in better outcomes, we must develop strategies to improve MRD clearance.

Risk stratification based on the WBC count, baseline cytogenetics, and baseline ploidy analysis did not impact the outcome in our cohort. Previous studies have identified each of these factors to be detrimental in the outcome of the patients.^{38,48,62,68-70}  This could be explained by the increased representation of the high-risk cytogenetics with Ph⁺ ALL and lesser representation with MLL and t(1;19). Outcomes of Ph⁺ ALL have improved after the utilization of TKIs along with multiagent chemotherapy.⁷¹  The EFS and OS of the Ph⁺ subgroup were considerably lesser (51.4% and 61.9%) in comparison with the rest of the B-ALL subgroup. Previous studies have shown that intensive chemotherapy along with TKI results in 5 year OS ranging from 40% to 50%.^35,72-75  Among the 37 patients with Ph⁺ ALL, 7 patients underwent HSCT. Despite such low HSCT rates, the EFS and OS at 3 years were comparable to cohorts with much greater HSCT rates. So, the practice must be to achieve remission using chemotherapy with TKI backbone. HSCT must be considered only when a fully matched donor is available. Furthermore, the patients also must be risk-stratified based on MRD clearance using real-time PCR for BCR-ABL1 transcripts and the presence of IKZF1 at diagnosis.⁷⁶  Achievement of complete molecular response (CMR) at 3 months has already been proven to be a significant factor in the overall outcome of these patients.⁷⁷  In our cohort, 8 patients achieved CMR at 3 months.

Cranial irradiation was planned to be administered to all the patients as part of the protocol except in T-LBL patients. Our patients received intrathecal chemotherapy (methotrexate 12 mg), HDMTx (3 g/m²), and cranial irradiation as CNS-directed therapy. In pediatric ALL, the use of cranial radiotherapy has reduced especially because intrathecal and systemic chemotherapy provide similar results and due to the risk of neurocognitive impairment with cranial irradiation.⁷⁸  However, there is no consensus regarding the utilization of cranial irradiation in the AYA population. No uniform pattern is seen in previous studies, some of them using the cranial irradiation for patients with T-ALL and CNS⁺ leukemia and others using for the whole cohort.^38,61,68,69  In the Total Therapy Study performed at the St. Jude Children Research Hospital, cranial irradiation was omitted for all its patients. The incidence of CNS relapse was 4%.⁵⁰  In our cohort, CNS relapse occurred in 23 patients (6.5%). On analyzing the baseline CNS status of these patients, 3 patients had CNS 2 or 3 disease and 20 had CNS 1 disease. Thus, future studies must aim to objectively define the place of cranial irradiation in the treatment protocol for patients with AYA ALL.

In terms of the complications, unlike the western studies that predominantly documented complications related to steroids and l-asparaginase, our analyses revealed an increased incidence of infections both bacterial as well as fungal. Our induction mortality was 6.5% similar to other studies from India.^19,20,79,80  Of the 21 induction deaths, 19 (82%) were due to infections. The high induction death rate is a major concern and future efforts must be made to prevent infection-related induction deaths with the use of prophylactic antifungals during induction, rigorous implementation of preventive measures, and early deintensification of therapy in those patients who are likely to improve. Despite the outpatient-based treatment and increased number of induction deaths in our cohort, 178 patients (51%) have either completed maintenance (177 patients) or continuing maintenance (1 patient) at the time of data analysis.

One of the major strengths of this study was the use of a single regimen uniformly in all of the patients. Another strength of this study is also its major flaw, we did not use a risk-stratified or response-adapted strategy. HSCT was the only intensification used but limited to a few patients. Although this is not considered an optimal strategy in the present age, it provides an opportunity to understand the prognostic factors and also question the place of HSCT in this setting. The major limitation of this study is that it is a single-center experience. There have been recent efforts in conducting multicenter analyses in India, like the Hematology Cancer Consortium of India, which presented the retrospective analyses of 1454 patients in the AYA age group (15-29 years) from 9 centers across India over 6 years using both pediatrics-inspired and adult protocols. This analysis showed that Ph⁺ status (HR, 1.56; P = .044) and use of adult protocol (HR, 1.6; P < .001) were the only factors associated with inferior outcome. We await complete analyses of these data.⁸¹  Another limitation would be that newer molecular higher-risk features like Ph-like features and IKZF1 deletions for B-ALL, NOTCH1/FWBX7 mutation, and RAS/PTEN expression were not assessed in our patients to help in further risk stratification. This is especially important as these baseline factors seemed to influence the outcome significantly in recent studies.^38,45  Finally, due to the retrospective nature of the data collection, the OS data are not accurate because of the losses to follow-up of patients who have relapsed or have not responded. Future research must be focused on better risk stratification, the use of targeted therapies, and response adaptation.

This study illustrates that it is possible to treat patients in the 15- to 25-year-old age group with a pediatrics-inspired modified BFM-90 protocol, with low treatment-related mortality and comparable outcomes as that of the published literature.

The authors acknowledge the help of Neha Sharma, Monisha Pillai, Seena Porathur, Gauri Patil, and Himanshi Gupta in data capturing.

Contribution: A.R., Hasmukh Jain, V.N.A.B., L.N., J.T., and M.S. designed and implemented the study; A.R., Hasmukh Jain, V.N.A.B., L.N., D.S., J.T., Hemani Jain, N.K., A.G., S.P., B.B., and M.S. contributed to the acquisition of the data and the documentation process; A.R., Hasmukh Jain, V.N.A.B., L.N., P.T., D.S., J.T., Hemani Jain, P.G.S., N.P., G.C., S.M., and M.S. planned and conducted data analysis; A.R., Hasmukh Jain, V.N.A.B., L.N., P.T., D.S., J.T., and M.S. drafted the manuscript; A.R., Hasmukh Jain, V.N.A.B., L.N., P.T., D.S., J.T., Hemani Jain, P.G.S., N.P., G.C., and M.S. contributed to the critical revision of the manuscript; and all authors reviewed the manuscript and approved the final version for publication.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Manju Sengar, Adult Hematolymphoid Unit, Tata Memorial Centre, affiliated with Homi Bhabha National University, E Borges Rd, Mumbai, 400 012, Maharashtra, India; e-mail: manju.sengar@gmail.com.