Long-Term Outcomes of Childhood Acute Lymphocytic Leukemia Treated with Adapted Berlin–Frankfurt–Münster (BFM) Protocols: A Multicentric Analysis from a Developing Country

Simple Summary Berlin–Frankfurt–Münster (BFM) protocols are widely used outside clinical trials to treat pediatric acute lymphoblastic leukemia patients. However, local specificities might demand treatment adaptations, like the reduction of high-dose methotrexate due to lack of pharmacokinetic monitoring, the substitution of conventional asparaginase for peg-asparaginase because of the unavailability of the latter, or flow-cytometry-based measurable residual disease instead of a PCR-based one. Here, we report the results of a 22-year period of children treated with BFM protocols in a developing country. The results were somewhat comparable to the BFM reports, and we conclude that BFM protocol adaptations can be safely implemented in developing countries, accounting for local specificities. Abstract Introduction: The objective of the current study was to determine the survival probabilities of children and adolescents with acute lymphocytic leukemia treated with adapted Berlin–Frankfurt–Münster (BFM) protocols and compare our results with the original BFM reports. Methods: This retrospective study included 695 patients up to 19 years old treated with adapted BFM protocols between 1997 and 2018 in four hospitals in Rio de Janeiro. The 1997–2007 and 2008–2018 cohorts were analyzed separately. Results: More than half of the patients were stratified into the high-risk BFM classification. Overall and event-free survivals were, in the 1997–2007 period, respectively, 88% and 80% (BFM standard risk group—SRG), 75% and 67% (intermediate risk group—IRG), and 48% and 33% (high-risk group—HRG). The corresponding numbers for the 2008–2018 period were 93% and 84% (SRG), 75% and 63% (IRG), and 64% and 57% (HRG). In the second period, both the OS (HR = 0.71, p = 0.011) and EFS (HR = 0.62, p < 0.001) were higher. Except for the intermediate-risk group, the latter results are comparable to the BFM. Conclusion: The BFM protocol adaptations can be safely implemented in developing countries, accounting for local specificities.


Introduction
Acute lymphoblastic leukemia (ALL) is characterized by genetic modifications that impair differentiation and promote the proliferation of lymphoid precursor cells, which is important for risk stratification [1].Although ALL incidence is lower in adolescents and young adults (AYA) compared with children, survival and long-term prognosis are poorer for AYA [2].Likewise, and despite novel drugs and newer protocols, age still has an important prognostic value, and children older than 10 years fare poorer compared with younger ones.Current pediatric ALL protocols achieve a 5-year overall survival (OS) of up to 90%.
The incidence of ALL varies depending on the country, and it seems to be higher among Hispanics [3].In Costa Rica, for example, the incidence of ALL was 56 per million children younger than 15 years, and Brazil, in 2022, reported an incidence of 53 per million in young adults [3,4].Moreover, Mexico, California, and Florida have the highest ALL incidence rates in North America [5].
BFM protocols have long been used in Brazil.Dismal results [6] have been reported in adolescents and young adults (AYA).Poor results have also been achieved in children in Brazil with BFM protocols [7].
In the 1980s, pediatric hospitals in Rio de Janeiro, Brazil, adopted BFM-based protocols seeking a more effective treatment since, by that time, survival was very poor.Four hospitals in Rio de Janeiro have been trying to unify the BFM-adapted protocols to improve survival and to build a local ALL network.Here, we report a 22-year analysis of a large cohort of children with ALL treated at these four institutions.Our aim was to estimate the survival probabilities and characterize the main prognostic factors for children and adolescents with ALL in Rio de Janeiro, as well as indirectly compare those results with other BFM studies.This is the first multicentric survival analysis study for ALL in children at Rio de Janeiro, and we report the results of this collaboration here.

Methods
This is a multicentric retrospective study of children and adolescents younger than 19 years who began treatment with BFM-based protocols between 1997 and 2018 in any of the following four institutions in Rio de Janeiro, Brazil: Hospital Federal da Lagoa (HFL), Hospital Universitário Pedro Ernesto (HUPE/UERJ), Instituto Nacional de Cancer (INCA/MS), and Hospital de Pediatria e Puericultura Martagão Gesteira (IPPMG/UFRJ).Patients treated with protocols other than BFM-based ones were excluded.All but one institution treated all patients with BFM-based protocols.
The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by all local Ethics Committees, which waived informed consent.

Statistical Analysis
Continuous variables were reported as mean and standard deviation or median and interquartile range and range and compared with a Student's t test or Mann-Whitney, as appropriate.We reported the frequency and percentages for the categorical variables and compared them with Pearson's chi-square or Fisher's exact test.Overall (OS) and event-free survival (EFS) curves were built with the Kaplan-Meier method and compared with the logrank test.OS was defined from diagnosis to death or last follow-up.EFS was defined as death, relapse, or secondary neoplasm, whichever occurred first.Univariable analyses were also performed with Cox models.The Cox proportional hazard assumption was checked.Multivariable Cox models were built based on the lowest Akaike information criteria (AIC).Multicollinearity was checked based on the variance inflation factor or whether the variable was already included in another collected scoring system.Cranial radiotherapy, which is a time-dependent variable, was analyzed separately using landmark analysis (in which only patients scheduled for the variable were included).All analyses were performed with R (R Foundation for Statistical Computing, Vienna, Austria), version 4.3.2.

Results
Table 1 describes the patients' characteristics.In brief, there was a slight male predominance, and most patients were between 1 and 9 years old.As expected, B-ALL was by far the most frequent immunophenotype.More than half of the patients in both periods were classified as high risk according to the BFM risk stratification.The 285 patients treated between 1997 and 2007 were followed for a median of 11.6 years, while the 410 patients included from the 2008-2018 period were followed for a median of 4.2 years.
There was a trend towards inferior EFS in patients who underwent cranial (HR = 1.37, p = 0.07), although relapse was significantly higher (HR = 1.83, p = 0.0017) in both multivariable analyses.
The protocol quality indicators are detailed in Table 5.Of note, induction death rates were 3.1% for the 1997-2007 period and 1.4% for the 2008-2018 period.Death in first complete remission was 12.1% in the first period (1997-2007) and 6.1% in the second one (2008-2018).

Discussion
Our study shows that adapted BFM protocols, originally designed and implemented in high-income countries, can also achieve excellent outcomes in developing countries, even though the frequency of high-risk patients was higher than previously reported.We also validate previously reported risk factors in a Brazilian socioeconomically vulnerable sample.
In the analysis of our study's last period (2008-2018), the 5 y EFSs were 84%, 63%, and 57% for the BFM-defined standard, intermediate, and high-risk groups, respectively, which is far better than we would expect given the socioeconomic profile of our sample.For comparison, the most recent BFM-ALL-09 trial [8] has reported 90%, 76%, and 56% 5 y EFSs for the same.Most of the discrepancies can be seen in the intermediate group, where our results were poorer, with an absolute difference of 13%.Here, some of the adaptations, reducing the methotrexate dose from 5 g/m 2 to 2 g/m 2 and substituting conventional asparaginase for peg-asparaginase, which happened in the former because in vivo methotrexate dosing was seldom available and, in the latter, because the drug was only recently approved in Brazil, might have taken its toll.The impact of a socioeconomic deprivileged population cannot be underestimated.Also, given the unexpectedly good results in the high-risk group, comparable to the BFM, we cannot rule out some sort of misclassification (affecting the intermediate-and high-risk groups) or survival bias in the high-risk group, in which some high-risk patients with more aggressive diseases die before being diagnosed.Despite that, our results have improved from the 1997-2007 to the 2008-2018 periods, especially in the BFM high-risk group, reflecting either better supportive care in the participating hospitals or the implementation of more recent and adapted BFM protocols.
In our study, either between 1997-2007 or 2008-2018, the BFM high-risk group accounted for more than half of the patients, obviously leading to poorer results.This is surprising since, in the BFM-ALL-09 publication [8], less than 25% were high-risk patients.Even though some weak evidence points to genetic differences between Latin American and other populations [9][10][11] and that racial background can impact ALL risk classification [12], Rio de Janeiro is considered a highly miscegenated city, we actually believe that delayed diagnoses or an overestimated D8 blast count can have played a major role.Indeed, the Brazilian public healthcare structure is known to be suboptimal and inefficient [13].We must underscore that the patients treated in these four hospitals in Rio de Janeiro have a very unfavorable socioeconomic profile, and the impact of low levels of education of patients and caregivers in ALL outcomes has already been reported [9][10][11]14].Regardless, our induction and CR1 mortality, overall survival, and event-free survival are very close to those reported by the BFM when analyzed within the protocol risk groups.
We validated most of the classical ALL independent prognostic factors [15], like age, sex, and BFM risk group, as well as the lack of independent prognosis of T-ALL.In the landmark analysis, cranial radiotherapy, which was usually indicated in patients with CNS+ or at high risk of CNS relapse, was associated with higher relapse rates.Since cranial radiotherapy is scheduled for 6 months after diagnosis, methods like landmark analysis are needed to not incur immortal-time bias [16].Although most modern protocols tend to avoid cranial radiotherapy, its omission should be carefully implemented in our population, particularly if adaptations of the original BFM protocols are expected to be used.
Even though it is recommended to strictly adhere to treatment protocols, this is not always possible.In our country, peg-asparaginase was only approved in 2016 and, therefore, incorporated.Likewise, methotrexate pharmacokinetics monitoring was not available in all four centers until 2017, limiting the maximum dose in centers to 2 g/m 2 instead of 5 g/m 2 until then.Moreover, socioeconomic disparities and significant income inequality hinder rapid access to reference centers.However, collaborative efforts among multiple healthcare stakeholders have the potential to improve this access and speed up the implementation of more effective primary care early diagnosis strategies and regulations.Despite that, acceptable results have been achieved, suggesting that local adaptations can be implemented, given that patients are included in the rigorous assessment of the results.It should be underscored that the BFM protocols are a part of studies having centralized quality control.
This study has several limitations.This is a retrospective study.Moreover, many different BFM protocols have been used without a central review of D8 and D15 assessments and no MRD standardization at D33 and week 12.However, we estimated that this study included about 85% of all predicted pediatric ALL cases diagnosed in Rio de Janeiro state (Rio de Janeiro city is the capital of Rio de Janeiro state) during the period.Another strength of our study is that all four centers used were familiar with BFM-adapted protocols.

Conclusions
In summary, we have shown that the BFM protocols can be adapted to developing countries' conditions, and excellent results, close to those reported by the BFM study group, can be achieved in ALL patients.Developing countries should focus on establishing local ALL networks and standardizing treatments and procedures.

Figure 1 .
Figure 1.Overall and event-free survival and relapse rate according to period.OS (A), EFS (C), and relapse (E) in the 1997-2007 period; OS (B), EFS (D), and relapse (F) in the 2008-2018 period.In the relapse rate figures, the Y axis ranges from 0 to 50%.

Figure 1 .
Figure 1.Overall and event-free survival and relapse rate according to period.OS (A), EFS (C), and relapse (E) in the 1997-2007 period; OS (B), EFS (D), and relapse (F) in the 2008-2018 period.In the relapse rate figures, the Y axis ranges from 0 to 50%.

Table 3 .
Multivariable analyses for overall and event-free survival.