Philadelphia‐like acute lymphoblastic leukemia: Characterization in a pediatric cohort in a referral center in Colombia

Abstract Background Philadelphia‐like (Ph‐like) acute lymphoblastic leukemia (ALL) is a subtype of pediatric leukemia with high risk factors and poor outcome. There are few reports of its prevalence in Latin America. Aim This study evaluated the frequency and clinical and biological characteristics of Ph‐like ALL in a pediatric cancer center in Colombia. Methods The Ph‐like genetic profile was analyzed by a low‐density array (LDA). Samples from patients with Ph‐like ALL were analyzed by fluorescent in situ hybridization for cytokine receptor like factor 2 (CRLF2) and ABL proto‐oncogene 1, non‐receptor tyrosine kinase (ABL1) rearrangements. Copy number variations were assessed by multiplex ligation probe amplification. Results Data from 121 patients were analyzed. Fifteen patients (12.4%) had Ph‐like ALL, and these patients had significantly higher leukocyte counts at diagnosis and higher levels of minimal residual disease on days 15 and 33 of induction than patients without the Ph‐like subtype. There were no significant differences in sex, age, or response to prednisone at day 8 between the two groups. CRLF2 rearrangements were identified in eight patients, and ABL1 rearrangements were identified in two patients. Other genetic alterations alone or in combination were identified in 77% of patients, including deletions in cyclin dependent kinase inhibitor 2 A/B (46.2%), IKAROS family zinc finger 1 (38.3%), and paired box 5 (30.8%). Conclusions Ph‐like ALL had a 12.4% prevalence in our cohort of patients with pediatric ALL. The identification of this group of patients has importance for risk stratification and future targeted therapy.


| INTRODUCTION
Acute lymphoblastic leukemia (ALL) is the most common pediatric neoplasm, in developed countries, long-term survival close to 85% has been achieved with treatment strategies based on disease-risk stratification. 1,2 Genetic alterations confer distinctive biological and clinical behaviors, which influence the response to treatment and are considered in the risk classification of modern protocols.
More than a decade ago, two academic groups, COG/St. Jude in the United States and DCOG in the Netherlands, simultaneously described a B-ALL subtype with a gene expression profile similar to the one already recognized for BCR activator of RhoGEF and GTPase (BCR)-ABL proto-oncogene 1, non-receptor tyrosine kinase (ABL1) but without the BCR-ABL1 fusion protein, which is known as Philadelphia-like (Ph-like) or BCR-ABL1-like. 3,4 The Ph-like subtype was recognized as a provisional entity in the last acute leukemia classification of the World Health Organization in 2016. 5 This subtype is characterized by multiple genetic alterations in cytokine receptors and kinase signaling pathways. More than 70 alterations have been described, including translocations, cryptic rearrangements, single nucleotide variants, and copy number variations (CNVs). 6 Alterations are also detected in the transcription factors involved in B lymphocyte development and maturation, including IKAROS family zinc finger 1 (IKZF1), EBF transcription factor 1 (EBF1), ETS variant transcription factor 6 (ETV6), ETS transcription factor ERG, transcription factor 3 (TCF3), and paired box 5 (PAX5), in 86% of the cases and with a frequency very similar to what is observed in Philadelphia-positive B-ALL cases. 7 However, there is currently no consensus on how to diagnose and characterize the Ph-like expression profile in patients with B-ALL. 6,8 The most frequent genetic alterations in patients with Ph-like ALL are rearrangements of cytokine receptor like factor 2 (CRLF2), 9 which are strongly associated with adverse clinical and treatment response factors and a higher risk of relapse with conventional therapy. Such rearrangements are found more frequently in the Hispanic population 10 within cohorts of patients treated in North America. Unfortunately, there are little data on the frequency of Ph-like B-ALL in the Hispanic pediatric population outside the United States. 11 The aim of this study was to identify the frequency of Ph-like B-ALL and examine its genomic and clinical characteristics in a cohort of patients under 18 years of age from a referral center in Colombia.

| METHODS
This was a prospective, descriptive study, nested in a cohort of patients between 1 and 18 years of age with a confirmed diagnosis of B-ALL from July 1, 2018 to August 1, 2020. Participants were selected at convenience from patients with a de novo confirmed diagnosis of B-precursor ALL at the HOMI Fundaci on Hospital Pediátrico La Misericordia who received the ALL-IC 2009 treatment protocol. Of 137 admitted patients, 10 were excluded due to lack of sample to perform Ph-like expression profiling and 6 were excluded because they were BCR-ABL1-positive. Therefore, 121 patients were included in the analysis.

| Biopsy and bone marrow aspiration procedure
Once consent was obtained and information formats were reviewed and completed, a bone marrow aspirate sample was taken to process flow cytometry (FC) to identify recurrent translocations, including t(4; 11), t(12; 21), and t(9; 22). In addition, a sample was taken for DNA and RNA extraction, conventional and molecular cytogenetics tests, and CNV analysis by multiplex ligation probe amplification (MLPA).
Samples in which recurrent translocations were not found underwent low density array (LDA) analysis for the Ph-like expression profile.
Those with a positive LDA test underwent fluorescence in situ hybridization (FISH) for CRLF2, ABL1, and ABL2.

| Immunophenotype analysis by FC
The Euroflow Screening Panel was used to analyze the immunophenotype of different maturation stages of neoplastic B cells in the bone marrow samples, with a combination of eight different fluorescent markers that allow the identification and phenotypic characterization of B lineage acute leukemias. 12  The DNA obtained was stored at À20 C until use. was stored at À80 C, and the cDNA was stored at À20 C until use.

| Cytogenetics
Cell culture was performed to obtain metaphases for the chromosomal study with G and Q bands according to standardized protocols. 13 Chromosome visualization was performed using GenASIs  were analyzed, and the interpretation was performed by two independent observers using GenASIs. The nomenclature was described according to the recommendations of the ISCN 2020.   14 with some modifications, we used as a housekeeping gene glucose-6-phosphate dehydrogenase (G6PD), to normalize the data. Also, each sample was processed in triplicate, and the gene expression was obtained using the 2 ÀΔΔCt method. The LDA expression profile was considered positive when four or more genes (>0.5) had expression levels higher than log e 10 > 2.3 compared to the control gene. Additionally, samples with individual overexpression of CRLF2 in the LDA profile were analyzed by CRLF2 FISH.

| DISCUSSION
In this study, we examined the frequency of Ph-like B-ALL and assessed its genomic and clinical characteristics in a cohort of patients under 18 years of age from a referral center in Colombia. A Ph-like profile frequency of 12.4% was found in our cohort. This frequency is similar to that reported by other authors (between 10 and 19%). 7,15,16 It is important to note that our patients are a Latino/Hispanic population outside of North America; there is very little published information on the frequency of this variant of ALL in children in Latin America. Perez-Vera et al. published a summary reporting 53% of this subtype in two institutions in Mexico this is high compared with our findings, however their diagnostic methodology to consider a result positive was different from that used in this study. 11 The clinical characteristics found in the Ph-like group were similar to those found in other studies in children, such as higher white blood cell counts 17,18 and a higher frequency of failure of remission at the end of induction. [17][18][19] No differences were found by sex, unlike the predominance in males found by other authors. 19 MRD at the end of induction has become essential in many protocols to define risk and intensification of therapy; therefore, it is important to identify this subgroup of patients at an early point in treatment. In our study, a statistically significant difference was found in MRD at the end of induction between the Ph-like and non-Ph-like groups (60 vs. 17% p = .002). Other authors have also reported this association in children with Ph-like ALL. 18,19 In 20% of patients with Ph-like ALL, the risk classification was modified to a higher classifica- Studies in adults with B-ALL also report a higher frequency of CRLF2 rearrangements in patients with a Ph-like expression profile. 20 Other genomic alterations, such as deletions of IKZF1, were observed in both groups, albeit less frequently. Deletions of IKZF1 were observed in 38.6% of patients in the Ph-like group, which is similar to the findings of Roberts et al. 19 but lower than that reported in other pediatric cohorts (60-80%). 12,21 This result may be due to different sample sizes or patient populations, since some examined a selected high-risk population. We found a small number of patients with CRLF2 overexpression without an LDA-positive Ph-like expression profile. These patients did not harbor CRLF2 rearrangements and did not have adverse prognostic factors, as has been described by other authors. 22 Hyperdiploidy in patients with Ph-like ALL is not a consistent finding; Van Der Veer et al they did not find hyperdiploidy in a group of patients ALL Ph-like. 23  Identifying this B-ALL subtype may provide an opportunity to understand why some BCR-ABL1-negative patients without findings of poor prognosis have persistent MRD at the end of induction and require treatment intensification. The current diagnostic strategies are not uniform, and many may not be feasible in countries with limited resources. It is necessary to design straightforward methodologies that can be easily implemented to identify this subgroup of patients. 26 The best therapeutic strategy for this group of patients is still being examined, motivated by the impact of the revolutionary role of tyrosine kinase inhibitors such as imatinib in improving the outcomes of patients with Philadelphia-positive B-ALL. Several collaborative groups are currently conducting randomized clinical trials, whose results will be known over the next few years. 17,24 Among the questions to be resolved in this specific B-ALL subtype is the possibility of using targeted therapy associated with conventional chemotherapy, which allows improving current results without increasing toxicity and eliminating the need for consolidation with HPT. 27 In conclusion, the frequency of the Ph-like ALL subtype in our cohort (12.4%) was similar to that described in cohorts in North America. This subtype was associated with poor prognostic factors, as pre-