Sánchez et al. Blood Cancer Journal (2020)10:43 https://doi.org/10.1038/s41408-020-0308-3 Blood Cancer Journal ARTICLE Open Access A novel targeted RNA-Seq panel identifies a subset of adult patients with acute lymphoblastic leukemia with BCR-ABL1-like characteristics Ricardo Sánchez 1,2,3, Jordi Ribera4, Mireia Morgades4, Rosa Ayala1,2,3,5,6, Esther Onecha1,2, Yanira Ruiz-Heredia1, Alexandra Juárez-Rufián1, Rodrigo de Nicolás1, José Sánchez-Pina1, Susana Vives4, Lurdes Zamora4, Santiago Mercadal7, Rosa Coll8, Marta Cervera9, Olga García4, Josep-Maria Ribera4 and Joaquín Martínez-López1,2,3,5,6 Abstract BCR-ABL1-like B-cell precursor acute lymphoblastic leukemia (BCP-ALL) remains poorly characterized in adults. We sought to establish the frequency and outcome of adolescent and adult BCR-ABL1-like ALL using a novel RNA-Seq signature in a series of patients with BCP-ALL. To this end, we developed and tested an RNA-Seq custom panel of 42 genes related to a BCR-ABL1-like signature in a cohort of 100 patients with BCP-ALL and treated with risk-adapted ALL trials. Mutations related to BCR-ABL1-like ALL were studied in a panel of 33 genes by next-generation sequencing (NGS). Also, CRLF2 overexpression and IKZF1/CDKN2A/B deletions were analyzed. Twenty out of 79 patients (12–84 years) were classified as BCR-ABL1-like (25%) based on heatmap clustering, with significant overexpression of ENAM, IGJ, and CRLF2 (P ≤ 0.001). The BCR-ABL1-like subgroup accounted for 29% of 15–60-year-old patients, with the following 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; molecular characteristics: CRLF2 overexpression (75% of cases), IKZF1 deletions (64%), CDKN2A/B deletions (57%), and JAK2 mutations (57%). Among patients with postinduction negative minimal residual disease, those with the BCR-ABL1- like ALL signature had a higher rate of relapse and lower complete response duration than non-BCR-ABL1-like patients (P = 0.007). Thus, we have identified a new molecular signature of BCR-ABL1-like ALL that correlates with adverse prognosis in adult patients with ALL. Introduction instrumental in identifying new ALL subtypes, and have The 2016 World Health Organization (WHO)classifi- aided in the discovery of new leukemogenic mechanisms. cation of acute leukemias recognizes nine different enti- It has been known for many years that a subset of patients ties within B-cell precursor acute lymphoblastic leukemia/ with ALL (~25% of BCP-ALL) have no established lymphoma (BCP-ALL) and two new provisional entities, abnormalities, commonly referred to as B-other ALL. The including BCR-ABL1-like. These 11 subtypes are based on genomic landscape of B-other ALL is becoming increas- specific molecular alterations, mainly chromosome rear- ingly clear, and the proportion of unclassifiable patients rangements, that promote the formation of aberrant chi- has declined significantly2,3. In this context, BCR-ABL1- meric proteins and aneuploidies1. Next-generation like B-ALL has emerged as one of the most relevant new sequencing (NGS) and array technologies have been subtypes due to its frequency and the potential benefit of targeted therapies (i.e., ABL and JAK inhibitors). Philadelphia chromosome-positive ALL is defined by Correspondence: Ricardo Sánchez (ricardsanchez.hdoc@gmail.com) or the t(9;22)(q34;q11) translocation that encodes BCR- Joaquín. Martínez-López (jmarti01@med.ucm.es) 1 ABL1 oncogene, a constitutively active kinase. This Hematology Department, Hospital Universitario 12 de Octubre, Madrid, Spain 2 aberration is present in >95% of patients with chronic Instituto de Investigación Hospital 12 Octubre (i+12), Madrid, Spain Full list of author information is available at the end of the article. myeloid leukemia, and in 3–5% and 25% of pediatric and These authors contributed equally: Josep-Maria Ribera, Joaquín Martínez-López © The Author(s) 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 2 of 11 adult ALL cases, respectively. In 2009, the DCOG/Eras- Total patients initially analyzed mus and COG/St. Jude groups independently discovered a BCP-ALL (n=115) treated in Philadelphia positive -ALL PETHEMA trials high risk BCR-ABL1-negative subgroup in children with (n=15) Eosinophilic BCP-ALL B-ALL, exhibiting a gene expression signature similar to and Undifferentiated BCR-ABL1 negative Low completion Acute Leukemia (n=2) clinical data (n=2) that of BCR-ABL1 positive-ALL but lacking the BCR- BCP-ALL (n=100) Rearrangements ABL1 rearrangement4. This subtype is associated with Burkitt Lymphoma (n=4) Clinical data (n=84) Hyperdiploid (n=8) Hypodiploid downregulation of B-cell development genes and over- Low Quality RNA-Seq Analysis RNA-Seq Analysis expression of stem- and progenitor-cell genes. Clinically, Age 15 - 60 (n=56) Age 12-84 (n=79) NGS data (n=5) Age 12 -15 + this ALL subtype presents with high-risk clinical features elderly patients (>60 years old) (n=28) Non -Ph-like Non - Ph-like such as high white blood cell (WBC) count, poor response Ph-like (n=16) (n=40) Ph-like (n=20) (n=59) to induction chemotherapy, higher measurable residual Fig. 1 Study flow chart. Design and distribution of the patients. BCP- disease levels, and low probability of survival4–8. ALL: B-cell precursor acute lymphoblastic leukemia; NGS: Next- The frequency of BCR-ABL1-like ALL has been repor- generation sequencing; Ph: BCR-ABL1. ted as 20–30% in adults, with a peak of incidence in the adolescent and young adult population (up to 42%)9–12. The BCR-ABL1-like ALL subtype shows deletions in and 2012 was evaluated for the identification of the BCR- several transcription factors involved in B-cell develop- ABL1-like signature (n = 49), and a second series of ment, including IKZF1, E2A, EBF1, and PAX513. Addi- patients treated between 2012 and 2017 was used for tionally, the main molecular characteristics of BCR-ABL1- validation (n = 100). like ALL are the multiple translocations in different The study design is shown in Fig. 1. One hundred cytokine receptor and kinase signaling genes such as patients with BCR-ABL1-negative B-ALL were selected ABL1 (excluding BCR association), JAK2, ABL2, PDGFRB, for molecular studies. Of those, 16 were discarded [Bur- TYK2, CSF1R, CRLF2, and EPOR. These mutations trigger kitt Lymphoma (n = 4), undifferentiated acute leukemia the activation of growth promoting kinase or cytokine (n = 2), ALL with MLL rearrangements (n = 2), TCF3- signaling pathways14. CRLF2 translocation and mutations PBX1 (n = 1), ETV6/RUNX1 (n = 1), hyperdiploidy (n = in the JAK family genes are recurrent and result in the 3) and hypodiploidy (n = 1) not classified as BCR-ABL1- activation of JAK-STAT pathways in patients with BCR- like based on the RNA-Seq signature, and two patients ABL1-like B-ALL15–17. with low quality clinical data]. The remaining 84 patients, Several different approaches have been employed for classified as B-other ALL, were considered valid for BCR- the characterization of BCR-ABL1-like ALL. Patients ABL1-like signature identification by RNA-Seq. Of these, were classified using hierarchical clustering on a gene 28 were discarded for survival analysis due to the het- expression array in early studies6,8. A simpler approach erogeneity of treatment protocols (including palliative was subsequently optimized based on Low Density therapy) and the absence of MRD assessment. Therefore, Arrays carrying a small number of genes selected by the prognostic relevance of the BCR-ABL1-like signature microarray prediction analysis15,18. In the present study, was investigated in a cohort of 56 homogeneously treated we designed a targeted NGS RNA-Seq panel of 42 genes patients (median age 34 years; range 16–59 years). The to classify patients with BCR-ABL1-like ALL. We sought main clinical characteristics and prognosis of patients to identify the BCR-ABL1-like ALL signature by targeted with BCR-ABL1-like B-ALL are shown in Table 1. expression in a series of adolescent and adult patients Details for patient samples are provided in Supple- with BCP-ALL, but without recurrent genetic abnorm- mentary Table S1. The study was conducted in accor- alities defined by the WHO classification (henceforth B- dance with the principles of the Declaration of Helsinki, other ALL), and to evaluate its clinical, prognostic and and the protocols were approved by the appropriate therapeutic relevance. institutional review boards. All patients provided written informed consent for the analysis of their biological Patients and methods specimens. Patients and study design We examined bone marrow (BM) or peripheral blood MRD assessment by multiparametric flow cytometry (PB) samples from adolescent and adult patients newly BM MRD levels were centrally assessed at the end of diagnosed with B-other ALL and treated between 2003 induction (weeks 5–6) in complete remission (CR) and 2017 in several Spanish hospitals. Patients received patients and at the end of the third consolidation cycle frontline chemotherapy according to PETHEMA (Pro- (weeks 16–18) by multiparameter flow cytometry. MRD grama Español de Tratamientos en Hematología) ALL levels at this latter time point were used to assign post- risk and minimal residual disease (MRD)-oriented consolidation therapy (continuation chemotherapy or trials19,20. A first series of patients treated between 2002 allogeneic hematopoietic stem cell transplantation). The Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 3 of 11 Table 1 Summary of clinical characteristics of the patients included in the study. Whole series BCR-ABL1-like ALL Remaining B-ALL P-value (N = 56) (n = 16) (n = 40) Age Median [range] 34 [16; 59] 22 [17; 59] 38.5 [16; 58] 0.310 Age range (years) 15–35 30 (54%) 11 (69%) 19 (47%) 0.150 36–59 26 (46%) 5 (31%) 21 (53%) Sex Male 27 (48%) 10 (63%) 17 (42%) 0.176 Female 29 (52%) 6 (37%) 23 (58%) WBC·10−3/µL (n = 52) Median (range) 7.87 [0.60-393.30] 13.40 [1.90-388] 5.01 [0.60-393.30] 0.071 CRLF2/GAPDH Negative (≤0,1%) 35 (63%) 4 (25%) 31 (78%) <0.001 Positive (>0,1%) 21 (37%) 12 (75%) 9 (22%) PB Blasts (%) (n = 56) Median [range] 41 [0; 95] 72 [24; 95] 34 [0; 90] 0.002 BM Blasts (%)(n = 30) Median [range] 93 [60; 100] 96,5 [93; 98] 92 [60; 100] 0.069 Phenotype Common 37 (66%) 10 (63%) 27 (68%) 0.417 Pre-B 12 (21%) 5 (31%) 7 (17%) Pro-B 7 (13%) 1 (6%) 6 (15%) Karyotype (n = 49; n = 15 BCR-ABL1-like; n = 34 remaining B-ALL) Normal 12 (25%) 3 (20%) 9 (27%) 0.428 IGHr 4 (8%) 1 (7%) 3 (9%) Complexa 2 (4%) 0 (0%) 2 (6%) No Growth 14 (29%) 3 (20%) 11 (32%) b Other 17 (35%) 8 (53%) 9 (27%) Risk classification High 46 (82%) 13 (81%) 33 (83%) 1.000 Intermediate 10 (18%) 3 (19%) 7 (17%) MRD postinduction (n = 41; BCR-ABL1-like n = 11; Remaining B-ALL n = 30) c Positive 18 (44%) 8 (73%) 10 (33%) 0.036 Negative 23 (56%) 3 (27%) 20 (67%) MRD post-consolidation (n = 36; BCR-ABL1-like n = 8; Remaining B-ALL n = 28) c Positive 10 (28%) 2 (25%) 8 (29%) 1.000 Negative 28 (72%) 6 (75%) 20 (71%) IKZF1 gene deletiond Positive 23 (52%) 9 (64%) 14 (47%) 0.276 Negative 21 (48%) 5 (36%) 16 (53%) Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 4 of 11 Table 1 continued Whole series BCR-ABL1-like ALL Remaining B-ALL P-value (N = 56) (n = 16) (n = 40) CDKN2A/B gene deletiond Positive 19 (43%) 8 (57%) 11 (37%) 0.202 Negative 25 (57%) 6 (43%) 19 (63%) d IKZF1 and CDKN2A/B codeletion Positive 11 (25%) 5 (36%) 6 (20%) 0.287 Negative 33 (75%) 9 (64%) 24 (80%) BM bone marrow, IGHr rearrangements of IGH@ gene, MRD minimal residual disease, PB peripheral blood, WBC white blood cells. a Complex karyotype is defined as more than four chromosomic alterations. b Other: del 12p (n = 1); del6q (n = 1); del7p plus del9p (n = 1); del9p (n = 2); del6q plus del 12p (n = 1); other deletions (n = 3); other rearrangements (n = 4); other alterations (n = 4). c Positive MRD ≥ 0.01 by multiparametric flow cytometry. d IKZF1 and CDKN2A/B deletions n = 44; BCR-ABL1-like n = 14; remaining B-ALL n = 30. detection limit of the method was 1 × 10−4. MRD level A barcoded cDNA library was then generated by amplifi- was considered positive when exceeded 0.01% or 1 × 10−4 cation using Ion AmpliSeq® (Thermo Fisher Scientific) at the end of induction and after consolidation. technology to precisely maintain expression levels of the targeted genes. A targeted RNA-Seq customized panel was Molecular biology analyses designed with 38 genes plus 4 housekeeping genes, gen- CRLF2 overexpression erating 42 primer pair amplicons. The tested genes were as One microgram of RNA was retrotranscribed to eval- follows (in alphabetical order): ABCA9, ARHGEF12, uate the expression levels of CRLF2 relative to GAPDH, by BMPR1B, CA6, CD99, CHN2, CRLF2, DCTN4, DENND3, real-time quantitative PCR (qPCR). The probes were ECM1, ENAM, GAB1, GBP5, GPR110, IFITM1, IGJ, IL7R, acquired from Gene Expression Taqman Assays (Thermo LDB3, MDFIC, MMP28, MUC4, NRXN3, PON2, RBM47, Fisher, Palo Alto, CA): Hs00845692_m1 (CRLF2), and RNF157, S100Z, SEMA6A, SCHIP1, SH2B3, SLC2A5, Hs02786624_g1 (GAPDH). Overexpression was deter- SLC37A3, SOC2, SPATS2L, TAF5L, TMEM154, TP53INP1, mined by means of 2−ΔΔCt method21, and was defined as TTYH2, and WNT9A22. The control genes were GUSB, positive when it was ≥0.1% relative to GAPDH gene JUN, PBGD, and TBP. Two of the 42 genes (SCHIP1 and expression. IFITM1) were discarded from analysis due to the low number of reads in each of the runs. Multiplex ligation-dependent probe amplification (MLPA) The quality of the amplified cDNA libraries was eval- One hundred ng of DNA was used for MLPA according uated using Bioanalyzer High Sensitivity chips (Agilent to manufacturer instructions (MRC Holland, Amsterdam, Technologies, Santa Clara, CA) and quantified with Ion the Netherlands). Two different kits have been used, one Library TaqMan™ Quantitation Kits (Thermo Fisher Sci- of them SALSA MLPA P335-C1 ALL-IKZF1 probemix kit entific). Libraries were diluted to 100 pM and pooled contains 57 MLPA probes with amplification products equally, assigning 150,000 reads per sample. Pooled between 120 and 504 nucleotides of the main exons of libraries were amplified using the Ion Chef System with EBF1, JAK2, CDKN2A, CDKN2B, PAX5, ETV6, BTG1, the Ion 540 Sequencing Kit (Thermo Fisher Scientific). RB1, IKZF1/2/3 genes, and genes of the X/Y PAR1 region Enriched libraries on a chip were sequenced on the Ion (CRLF2, CSF2RA, IL3RA and P2RY8). The second Kit GeneStudio S5 System using the Ion S5 Sequencing Kit SALSA MLPA P202 probemix kit contains 59 MLPA (Thermo Fisher Scientific) with 500 flows. The absolute probes with amplification products between 118 and 504 normalized Reads Per Million (RPM) matrix was obtained nucleotides, which span IKZF1 and CDKN2A-B genes (21 from RNA-Seq Analysis plug-in (v5.4.0.1) within Torrent and 3 probes, respectively). Suite software (v5.10) (Thermo Fisher Scientific). The matrix was subsequently normalized intra-patient relative Targeted RNA-sequencing to GUSB because it was the housekeeping gene with the cDNA was obtained after reverse transcription of 1 µg of higher Spearman correlation coefficient (ρ) of the reads RNA. cDNA integrity was checked by qPCR with a GUSB between patients within the matrix data, compared to the Taqman probe (Hs00939627_m1) (Thermo Fisher Scien- other three genes. A final analysis was performed using the tific), discarding cDNA with a Ct>25 at a threshold of 0.1. web-based tool Morpheus (https://software.broadinstitute. Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 5 of 11 org/morpheus/), a matrix visualization and analysis plat- cumulative incidence rates and were compared by Gray’s form, obtaining an unsupervised hierarchical cluster test. Two-sided P-values < 0.05 were considered statisti- heatmap using one minus Pearson correlation as a metric, cally significant. Multivariable analyses were performed and an average of the data as the linkage method. The raw using the Cox proportional hazards model for OS and RNA-Seq sequencing data were uploaded to NCBI with DFS, and the Fine and Gray model for CIR. The statistical BioProject ID: PRJNA613841. package SPSS version 24.0 (Statistical Package for Social Sciences Inc., Chicago, IL) and R 3.4.2 software were used Targeted DNA-sequencing for all analyses. A total of 33 lymphoid-related genes were sequenced by targeted NGS using an Ion Ampliseq® On Demand panel Results (Thermo Fisher Scientific), consisting of a custom mixture BCR-ABL1-like signature identification by RNA-Seq of oligonucleotides that generated 892 amplicons in two To identify a specific molecular signature for BCR- pools, covering 182 kb. The design includes whole exons ABL1-like B-ALL, we examined a dataset from patients of JAK1, NRAS, XPO1, CXCR4, SF3B1, MYD88, KLHL6, with B-other ALL (n = 49) by targeted RNA-Seq that, WHSC1, FBXW7, IRF4, IKZF1, CRLF2, BRAF, EZH2, after clustering, grouped a subset of patients with over- JAK2, CDKN2A, PAX5, NOTCH1, ATM, KRAS, KMT2D, expression of CRLF2 and mutations in several genes CREBBP, TP53, STAT5B, STAT3, TYK2, and JAK3 genes, related to the BCR-ABL1-like signature. Thirteen of these and selected exons of RAB39A, CUL5, EXPH5, DLEU1, patients (27%) were classified as BCR-ABL1-like based on SAT2, and EFNB3. heatmap clustering, with all equally showing over- Libraries were prepared following the Ampliseq® pro- expression of ENAM and IGJ. Also, most of the patients tocol using at least 10 ng of template DNA per reaction. with high CRLF2 expression (CRLF2+) were classified Multiple indexed libraries were pooled and sequenced on into the BCR-ABL1-like ALL subgroup: 11/13 (85%) ver- the Ion GeneStudio S5 System using Ion S5 Sequencing sus 7/36 (19%) non-BCR-ABL1-like, P < 0.001. Mutations Kit, with 500 flows. Samples were sequenced to an average of IKZF1, KRAS, JAK2, CRLF2, NRAS, JAK1, TYK2, and 1900× coverage. Alignment and variant detection were PAX5 genes related to BCR-ABL1-like genotype were performed using Ion Reporter v5.10 using the human present in 7/11 BCR-ABL1-like patients (64%) versus 3/20 reference genome (hg19). Variants were manually non-BCR-ABL1-like patients (15%), P = 0.013. This sig- reviewed in Integrated Genome Viewer v2.3.81 (Broad nature identified BCR-ABL1-like patients who showed Institute, Cambridge, MA). Variants were classified as significantly shorter DFS probability at 5 years [95%con- benign, unknown significance, pathogenic or likely fidence interval (CI): 20% (0%; 45%) vs 54% (95% CI: 35%; pathogenic according to VARSOME software23. The raw 73%), P = 0.047]. DNA-Seq sequencing data were uploaded to NCBI with BioProject ID: PRJNA614523. BCR-ABL1-like signature by RNA-Seq, validation cohort We next validated the reproducibility of the identified Statistical analyses signature by extending the study cohort to 100 BCR- Baseline characteristics were reported as frequency and ABL1-negative patients (Fig. 1). Targeted RNA- percentage for categorical variables and as median and sequencing was performed on cDNA from 84 patients. range for quantitative variables. Comparisons of propor- Those patients with WHO cytogenetic subtypes (i.e., MLL tions and the medians of variables between different rearrangements, ploidies, Burkitt ALL), with other groups were performed using the χ2 test, Fisher’s exact hemopathies, or with BCP-ALL with incomplete follow- test, or the nonparametric median test as appropriate. up data were excluded. A normalized matrix (input for Overall survival (OS) was measured from the time of heatmap clustering) was obtained for 79/84 of these diagnosis to the time of death from any cause. Disease- patients after removing data for five patients due to the free survival (DFS) was measured from the date of low quality of the generated reads. One of the groups achievement of the first remission until the date of relapse generated in the heatmap was classified as BCR-ABL1-like or death from any cause. Cumulative incidence of relapse and accounted for 20/79 patients (25%) (Fig. 2). The (CIR) was calculated from the date of achievement of the expression of ECM1, ENAM, IGJ, and CRLF2 was sig- first remission until the date of relapse. Patients who died nificantly higher in the BCR-ABL1-like subgroup than in without relapse were counted as a competing risk. the non-BCR-ABL1-like subgroup (P ≤ 0.015). By contrast, Patients not known to have relapsed ordied at last follow- MDFIC was underexpressed in 100% of patients with the up were censored on the date they were last examined. OS BCR-ABL1-like signature (P = 0.001). In addition, this and DFS curves were performed using the Kaplan–Meier group was enriched inpatients who were CRLF2+ [15/20 estimation, and the log-rank test was used for compar- (75%) vs 16/59 (27%), P < 0.001]; and in mutations related isons between groups. CIR curves were estimated using to the BCR-ABL1-like signature16 such as JAK2 and RAS, Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 6 of 11 Color Key – Expression levels BCR-ABL1-like signature Fig. 2 Unsupervised hierarchical cluster with dendrogram of 79 B-other-ALL patients based on gene expression of 40 genes. The 20 BCR- ABL1-like patients are marked with a blue square. or PAX5 and IKZF1 genes [14/18 (78%) vs 16/42 (38%), groups (72% vs 34%, P = 0.002), and WBC counts P = 0.005]; and both alterations [13/18(72%) vs 6/42 [13.40 × 109/L (95% CI: 1.9; 388.0) vs 5.01 × 109/L (95% (14%), P < 0.001]. Most of the BCR-ABL1-like patients CI: 0.6; 393.3), P = 0.071]. agree on the result of CRLF2 overexpression (17/20: 85%) Although there were no significant differences in CR by both techniques qPCR and RNA-Seq. achievement, BCR-ABL1-like patients had a tendency for a poorer response to induction treatment than non-BCR- Clinical characteristics and prognosis of patients with ABL1-like patients [3/16 (19%) non-responders in BCR- BCR-ABL1-like ALL ABL1-like vs 2/40 (5%) in non-BCR-ABL1-like patients, Outcome analysis was performed in patients between 15 P = 0.135]. In addition, BCR-ABL1-like patients showed a and 60 years of age treated in PETHEMA trials with higher proportion of positive MRD than non-BCR-ABL1- intermediate (10/56, 18%) and high-risk (46/56, 82%) like patients [8/11 (73%) vs 10/30 (33%), P = 0.036] at the protocols, in order to ensure homogeneity between end of induction phase. Interestingly, among patients with treatments. With a median follow up of 3.8 years [range, postinduction negative MRD (MRD-) (n = 26), BCR- 0.82–14.94], the 4- and 8-year OS of the 56 patients with ABL1-like patients (n = 6) had a higher rate of relapse BCP ALL were 58% (95% CI: 44%; 72%) and 37% (95% CI: and lower CR duration than non-BCR-ABL1-like patients 17%; 57%), respectively; the 4- and 8-year DFS were 47% (n = 20) (P = 0.007) (Fig. 3). Nine out of 51 CR patient- (95% CI: 32%; 62%) and 39% (95% CI: 20%; 58%), sunderwent allogeneic-HSCT with no significant differ- respectively; and the CIR was 45% (95% CI: 30%; 59%) for ences in the proportion of transplanted patients between both time points. A summary of the clinical data is shown BCR-ABL1-like and the remaining B-other ALL patients in Table 1. [2/13(15%) vs 7/38 (18%), respectively, P = 1.000]. Sixteen of the 56 patients (29%) were classified as BCR- BCR-ABL1-like patients showed poorer OS than the ABL1-like in the heatmap (Supplementary Fig. 1). remaining BCP-ALL patients [47% (95% CI: 21%; 73%) vs There were no significant differences in baseline char- 63% (95% CI: 48%; 78%), P = 0.105], with a higher fre- acteristics such as age, gender, ALL phenotype, % BM quency of relapses [66% (95% CI: 29%; 87%) for BCR- blasts, cytogenetics, and risk classification in BCR-ABL1- ABL1-like vs 38%(95% CI: 21%; 55%) for B-other ALL, like patients in comparison with the remaining patients in P = 0.067] and the consequent lower DFS [26% (95% CI: the B-other ALL subgroup. However, significant differ- 1%; 51%) vs 54% (95% CI: 36%; 72%), P = 0.069] as the ences were observed in the median percentage of blasts in main reasonsfor the poorer OS (Fig. 4). The limited PB between BCR-ABL1-like and non-BCR-ABL1-like number of patients in this series and the relatively low Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 7 of 11 ratio of mutations did not allow us to assess the possible overexpression clustered in the BCR-ABL1-like subgroup prognostic significance of JAK/STAT pathway mutations, [12/16(75%) vs 9/40(22%) non-BCR-ABL1-like patients, N/KRAS, IKZF1, or PAX5 point mutations. The univari- P < 0.001]. There were available samples for NGS mole- able and multivariable models for the 56 patients are cular studies of DNA for 42/56 patients (14/16 BCR- shown in Supplementary Tables S2 and S3. ABL1-like and 24/40 non-BCR-ABL1-like). A total of 57 non-synonymous variants affecting 15 genes were iden- CRLF2 overexpression: association with BCR-ABL1-like, tified among the 42 patients. At least one variant could be JAK/STAT mutation pathway and prognosis identified in 79% (33/42) of the patients, and at least one Among the 56 B-other ALL patients with clinical data, pathogenic mutation was identified in 57% (24/42). The most of the BCR-ABL1-like patients showed CRLF2 distribution of the mutated genes is shown in Fig. 5a and in more detail in Supplementary Table S4. JAK2 muta- tions (more recurrently c.2047A > G and p.R683G) were enriched in BCR-ABL1-like ALL patients [9/14(64%) vs non-BCR-ABL1-like 3/28(11%), P = 0.001]. The distribu- tion of JAK2 mutations across the protein domains is shown in Fig. 5b. Significant differences in overexpression of CRLF2 were found between BCR-ABL1-like and non-BCR-ABL1-like groups (see Supplementary Table S5, P < 0.001), and additional mutations were found in JAK-STAT, RAS, and transcription factors such as IKZF1 or PAX5. Statistically significant differences were found in the rate of JAK/ STAT mutations between BCR-ABL1-like patients and the remaining B-other ALL patients [BCR-ABL1-like 9/14 (64%) vs 3/28 (11%), P = 0.001], but not for RAS genes or lymphoid transcription factors. We found significant differences in DFS for CRLF2 overexpression [4-year DFS 28% (95% CI: 6%; 50%) vs 59% (95% CI: 40%; 78%) and 8-year DFS 28% (95% CI: 6%; Fig. 3 Cumulative incidence of relapse for BCR-ABL1-like and 50%) vs 47% (95% CI: 22%; 72%), P = 0.006] (Supple- non-BCR-ABL1-like patients within the post-induction negative- mentary Fig. 2b). CRLF2+ patients had higher 4- and MRD population. BCR-ABL1-like patients are denoted by the green line and non-BCR-ABL1-like patients by the blue line. 8-year CIR than patients with no CRLF2 overexpression [61% (95% CI: 33%; 81%) vs 35% (95% CI: 17%; 54%), Fig. 4 Kaplan–Meier survival curves for BCP-ALL patients: BCR-ABL1-like patients are denoted by the green line and non-BCR-ABL1-like patients by the blue line. a Overall survival; b disease-free survival. Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 8 of 11 a) IKZF1 PAX5 STAT5B JAK2 KRAS TYK2 NRAS JAK1 EZH2 CREBBP KLHL6 ATM TP53 CRLF2ov CDKN2A/Bdel IKZF1del EMR_POS ** ** ** ** ** RELAPSE DEATH SURV.TIME Ph-like Non Ph-like SURVIVAL TIME (days): <500; 500-1000; >1000 CRLF2/GAPDH ≥0.1 Yes/No Yes/No ** NO COMPLETE REMISSION AT ALL b) RELAPSE Yes/no DEATH Yes/no VARIANTS Yes/No Non-available data Fig. 5 Main alterations identified in this study with a detail of JAK2 mutations. a Molecular, genetic, and clinical alterations in BCR-ABL1-like and non-BCR-ABL1-like patients. b Schematic representation of JAK2 primary sequence. Point mutations encountered in the studied population are depicted in green; p.R683G was observed in five patients. P = 0.018; at both time points] (Supplementary Fig. 2c). deletion (n = 6) showed lower OS than equivalent Although CRLF2+ patients showed lower OS than those patients without loss of CDKN2A/B (n = 8) [4-year OS without overexpression, the differences were not statisti- 17% (95% CI: 0%; 46%) vs 83% (95% CI: 53%; 100%), P = cally significant [4-year OS 52% (95% CI: 29%; 75%) vs 0.041]. However, these differences showed a trend for DFS 62% (95% CI: 46%; 78%) and 8-year OS 20% (95% CI: 0%; and were not statistically significant for CIR, in which 51%) vs 45% (95% CI: 20%; 70%), P = 0.234]. patients with CDKN2A/B loss had higher relapse inci- Of note, no differences were observed in CR achieve- dence in both subgroups individually. By contrast, the ment, MRD clearance or outcome between CRLF2+/BCR- prognostic differences on OS and DFS observed for BCR- ABL1-like and CRLF2+/non-BCR-ABL1-like patients ABL1-like were not observed within the non-BCR-ABL1- (data not shown). like patients, although non-BCR-ABL1-like patients with CDKN2A/B deletion showed a trend towards higher CIR Prognostic impact of other molecular alterations: IKZF1 and than those without CDKN2A/B loss (data not shown). CDKN2A/B deletions IKZF1 deletions did not have prognostic significance for Analysis of copy number alterations by MLPA was any outcome parameter analyzed in the B-other ALL available for 44/56 B-other ALL patients. Results showed series. However, the concomitance of IKZF1 and that 75% of these patients (33/44) had at least one deletion CDKN2A/B deletions identified a subset of patients with in IKZF1 or CDKN2A/B, but no significant differences poor prognosis. Among the 44 patients with known were found between BCR-ABL1-like and non-BCR-ABL1- IKZF1 and CDKN2A/B status, those having both altera- like groups, or for individual deletions in the case of tions (n = 11) showed poorer OS than those without both codeletion of both genes (Table 1). deletions (n = 33) [4-year OS 46% (95% CI: 17%; 72%) vs As shown in univariable and multivariable analysis 67% (95% CI: 50%; 84%), P = 0.090]. The same scenario (Supplementary Tables S2 and S3), CDKN2A/B deletions was observed for DFS [3-year DFS 40% (95% CI: 10%; showed statistical significance for outcome prediction in 70%) vs 67% (95% CI: 50%; 84%), P = 0.014] and CIR [3- the 56 B-other ALL patients. However, we observed sig- year CIR 50% (95% CI: 16%; 77%) vs 30% (95% CI: 15%; nificant differences regarding the impact of these dele- 47%), P = 0.061]. tions within BCR-ABL1-like and non-BCR-ABL1-like Subgroup analysis was repeated for patients with both subgroups individually. Although the number of patients deletions, to assess whether the concomitant loss of IKZF1 was low, BCR-ABL1-like patients with CDKN2A/B and CDKN2A/B had adifferent prognosis in BCR-ABL1-like Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 9 of 11 and non-BCR-ABL1-like cohorts individually. Again, classified 73% of patients within the BCR-ABL1-like although the patient numbers were low, we observed that population as positive, St. Jude reported 81% of IKZF1 within the BCR-ABL1-like subset, patients harboring both losses, and we found 64% of mutated patients. In our deletions experienced significantly lower probability of study, we identified 75% of patients CRLF2+ and 57% with survival than those without both alterations (4/5 vs 5/9 mutated JAK2, similar to the published results. Indeed, deaths, P = 0.029), mainly due to a higher relapse rate (3/4 i 61% and 51% of CRLF2+ and 45% and 27% of mutated 5/8 relapses, P = 0.043) (Fig. 6). Within the remaining B- JAK2 have been reported within BCR-ABL1-like patients other ALL subgroup, patients with deletion of both genes in the St. Jude and MD Anderson studies, respectively. also experienced more relapses although the results were The prognostic impact of the BCR-ABL1-like ALL subtype not statistically significant (data not shown). in adults homogeneously treated within MRD-oriented trials remains still unclear. In our series, these patients had higher Discussion WBC and blast counts than non-BCR-ABL1-like patients, We sought to identify an RNA-Seq signature for BCR- suggesting a greater degree of cell proliferation with a strong ABL1-like ALL in a series of homogeneously treated capacity for dissemination—an essential characteristic rela- adolescent and adult patients with B-other ALL, and ted to a higher degree of clonal heterogeneity and treatment analyze its prognostic impact. We demonstrate the resistance. Interestingly, a high degree of cell cycle dereg- capacity of a simplified targeted RNA-Seq signature to ulation (57% CDKN2A/B deletion) and stem cell-like char- segregate BCR-ABL1-like patients within a BCP-ALL acteristics as a result of IKZF1 losses (64%) might play an adult population. This approach enabled us to confirm important role in the aggressiveness and resistance seen in that BCR-ABL1-like is a high-risk ALL subtype also in the present and other series of BCR-ABL1-like patients. As young, adolescent, and older adults despite treatment with mentioned, 75% of BCR-ABL1-like patients overexpressed modern MRD-oriented protocols. CRLF2 and 64% had mutations in the JAK-STAT pathway The prognostic impact of the BCR-ABL1-like subtype (and all CRLF2+ patients also had JAK2 mutations), sup- has been mainly reported in pediatric and adolescent porting the idea that this is an essential pathway in this populations, and studies in adult BCP-ALL are scarce6,8,16. aggressive subtype. Most JAK2 mutations identified in the The DCOG/Erasmus and St Jude groups used different present study were pathogenic and involved the tyrosine- gene profiles to define the BCR-ABL1-like subtype and, kinase domains 1 and 2, close to the nucleotide binding site consequently, there is a lack of standardization and com- for ATP-ADP exchange (residues 855–863), and structurally parability regarding the best strategy to identify these distant from the self-regulatory tyrosine 1007 and 1008. patients. The frequency (29%) and clinical outcome of Currently, several clinical trials are evaluating the efficacy of BCR-ABL1-like ALL found in this study is in accord with JAK and mTOR inhibitors (in addition to TKI inhibitors), that of the MD Anderson and St. Judereports9,11: Roberts alone or in combination with other drugs based on this and coworkers11 characterized 194/798 (24%) patients (age genetic rationale24–26. range 21–86 years) as BCR-ABL1-like, and Jain et al.9 IKZF1 and CDKN2A/B deletions are known to be asso- identified 49/148 (33%) patients (age range 15–71 years). ciated with poor prognosis in pediatric and adult Regarding IKZF1 deletions, the MD Anderson group B-ALL populations27–29, mainly in the BCR-ABL1 B-ALL Fig. 6 Log-rank comparison curves between IKZF1 and CDKN2A-B-codeletion positive and negative patients within the BCR-ABL1-like population. IKZF1/CDKN2A-B codeleted patients are denoted by the green line and patients with no codeletion of IKZF1/CDKN2A-B by the blue line. a Overall survival; b disease-free survival; c cumulative incidence relapse. Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 10 of 11 subpopulation. Our results suggest that, despite the low this entity make NGS an attractive option, although this number of cases analyzed, CDKN2A/B deletions are mar- requires complex bioinformatic analysis. Interestingly, the kers of poor survival in B-other ALL, and the association approach shown here enables the identification of the between CDKN2A/B and IKZF1 deletions might also BCR-ABL1-like signature by targeted RNA-Seq in 3 days contribute to the dismal prognosis of BCR-ABL1-like. upon sample arrival with a simple and fast NGS library Finally, the survival analysis suggests that all CRLF2+ protocol and subsequent sequencing. It is also repro- patients show poor response to standard treatment and bad ducible, as we have shown in the validation cohort. prognosis whether or not they are BCR-ABL1-like. Finally, a simple computer plug-in will output a normal- Unfortunately, we did not have enough samples to evaluate ized matrix that will predict patient outcome in the the prognostic significance of additional alterations (i.e., hierarchical clustering heatmap. JAK2 mutations, IKZF1, and CDKN2A/B deletions) in the In summary, our study demonstrates that targeted cohort of 56 B-other ALL patients or within the subgroups RNA-Seq correctly identifies BCR-ABL1-like ALL in BCP- of patients with and without CRLF2 overexpression. ALL patients. This methodology is inexpensive, available, Due to the limited number of BCR-ABL1-like patients and feasible, and could be introduced in the routine in our series, the difference in CR achievement between clinical workout for ALL patients. We have also con- BCR-ABL1-like and non-BCR-ABL1-like patients did not firmed the poor prognosis of BCR-ABL1-like ALL in the reach statistical significance, although there were four adult setting. An early diagnosis of BCR-ABL1-like ALL times as many patients not achieving CR in the BCR- could be critical to initiate appropriate treatment ABL1-like subgroup. Taken together with the low MRD depending on the kinase profile (i.e., JAK inhibitor, clearance at the end of induction seen in BCR-ABL1-like tyrosine-kinase inhibitor, etc.). These results endorse the patients (three-quarters of BCR-ABL1-like patients were inclusion of these molecular approaches in further clinical MRD-positive at the end of induction compared with one- protocols for adopting early clinical decisions with the third of non-BCR-ABL1-like patients), this indicates goal to better manage patients with ALL. higher treatment resistance. Given the small number of B- Supplementary information is available at Blood Cancer other patients (n = 10) MRD-positive at the end of con- Journal website. solidation, we could not evaluate the role of HSCT on BCR-ABL1-like and non-BCR-ABL1-like patients. Acknowledgements In addition to treatment resistance, the high degree of The authors would like to thank all the participants of the PETHEMA group. relapse observed in BCR-ABL1-like patients and, more This project was funded in part by CRIS CANCER FOUNDATION. importantly, those who were MRD-negative at the end of induction, suggests that standard treatments are not less Author details 1 Hematology Department, Hospital Universitario 12 de Octubre, Madrid, Spain. efficient for this ALL subtype. Also, while MRD has 2 Instituto de Investigación Hospital 12 Octubre (i+12), Madrid, Spain. become the most powerful outcome predictor in ALL, it is 3 Haematological Malignancies Clinical Research Unit, Spanish National Cancer not fully predictive, and other factors beyond MRD (e.g., Research Centre (CNIO), Madrid, Spain. 4Hematology Department, ICO – Hospital Germans Trias i Pujol. Josep Carreras Leukaemia Research Institute, genetic alterations) also impact patients’ prognosis, Universitat Autònoma de Barcelona, Badalona, Spain. 5Centro de Investigación especially in the B-other subpopulation. Biomédica en Red Cáncer (CIBERONC), Madrid, Spain. 6Universidad The BCR-ABL1-like signature methodology clearly dis- Complutense de Madrid, Madrid, Spain. 7Hematology Department, ICO – Hospital Duran i Reynals (Bellvitge), Barcelona, Spain. 8Hematology tinguishes the outcome of patients in whom no recurrent Department, ICO – Hospital Dr. Josep Trueta, Girona, Spain. 9Hematology genetic abnormalities could be identified by standard Department, ICO – Hospital Universitari Joan XXIII, Tarragona, Spain methods. Specifically, loss of CDKN2A/B identifies Conflict of interest patients at high risk of disease progression among those J.M.L. received research funding from Incyte, Celgene, Janssen, Novartis and with non-available genetic risk categorization. We also BMS. J.M.L. is also on the speaking bureau of Incyte, Celgene, Janssen, Novartis provide more detailed information on the prognostic and BMS. The remaining co-authors declare no competing financial interests. impact of IKZF1 and CDKN2A/B deletions (together and separately), specifically within the BCR-ABL1-like subtype, Publisher’s note where they seem to confer poor outcome. By contrast, the Springer Nature remains neutral with regard to jurisdictional claims in prognostic relevance of these alterations in the non-BCR- published maps and institutional affiliations. ABL1-like subgroup is less clear. Due to the paucity of Supplementary Information accompanies this paper at (https://doi.org/ samples, more studies are needed to address these issues. 10.1038/s41408-020-0308-3). Currently, methodical characterization BCR-ABL1-like is expensive and laborious. The multiple rearrangements Received: 20 December 2019 Revised: 23 March 2020 Accepted: 1 April 2020 present in this subtype and the constellation of other molecular alterations characteristic but not exclusive of Blood Cancer Journal Sánchez et al. Blood Cancer Journal (2020)10:43 Page 11 of 11 References 17. Tasian, S. K., Loh, M. L. & Hunger, S. P. Philadelphia chromosome-like acute 1. Arber, D. A. et al. The 2016 revision to the World Health Organization classi- lymphoblastic leukemia. Blood 130, 2064–2072 (2017). fication of myeloid neoplasms and acute leukemia. Blood 127, 2391–2405 18. Harvey, R. C. et al. Development and validation of a highly sensitive and (2016). specific gene expression classifier to prospectively screen and identify 2. Gu, Z. et al. PAX5-driven subtypes of B-progenitor acute lymphoblastic leu- B-precursor acute lymphoblastic leukemia (ALL) patients With a Phila- kemia. Nat. Genet. 51, 296–307 (2019). delphia chromosome-like (“Ph-like” or “BCR-ABL1-Like”) signature for 3. Zaliova, M. et al. Genomic landscape of pediatric B-other acute lymphoblastic therapeutic targeting and clinical intervention. Blood 122, 826–826 leukemia in a consecutive European cohort. Haematologica 104, 1396–1406 (2013). (2019). 19. Ribera, J.-M. et al. Treatment of high-risk philadelphia chromosome—negative 4. Iacobucci, I. & Mullighan, C. G. Genetic basis of acute lymphoblastic leukemia. acute lymphoblastic leukemia in adolescents and adults according to early J. Clin. Oncol. 35, 975–983 (2017). cytologic response and minimal residual disease after consolidation assessed 5. Roberts, K. G. et al. Outcomes of children with BCR-ABL1-like acute lympho- by flow cytometry: final results of the PETHEMA ALL-AR-03 Trial. J. Clin. Oncol. blastic leukemia treated with risk-directed therapy based on the levels of 32, 1595–1604 (2014). minimal residual disease. J. Clin. Oncol. 32, 3012–3020 (2014). 20. Ribera, J.-M. et al. Post-remission treatment with chemotherapy or 6. Mullighan, C. G. et al. Deletion of IKZF1 and prognosis in acute lymphoblastic allogeneic hematopoietic stem cell transplantation (alloHSCT) of high- leukemia. N. Engl. J. Med. 360, 470–480 (2009). risk (HR) Philadelphia chromosome-negative (Ph-neg) adult acute 7. Roberts, K. G. et al. Genetic alterations activating kinase and cytokine receptor lymphoblastic leukemia (ALL) according to minimal residual disease signaling in high-risk acute lymphoblastic leukemia. Cancer Cell 22, 153–166 (MRD). preliminary results of the Pethema ALL-HR-11 Trial. Blood 126, (2012). 1333–1333 (2015). 8. Den Boer, M. L. et al. A subtype of childhood acute lymphoblastic leukaemia 21. Pfaffl, M. W. A new mathematical model for relative quantification in real-time with poor treatment outcome: a genome-wide classification study. Lancet RT-PCR. Nucleic Acids Res. 29, e45 (2001). Oncol. 10, 125–134 (2009). 22. Willman, C. L. et al. Gene expression signatures for detection of underlying 9. Jain, N. et al. Ph-like acute lymphoblastic leukemia: a high-risk subtype in Philadelphia chromosome-like (Ph-like) events and therapeutic targeting in adults. Blood 129, 572–581 (2017). leukemia. Patent Application WO 2013/090413. 10. Herold, T. et al. Adults with Philadelphia chromosome-like acute lymphoblastic 23. Kopanos, C. et al. VarSome: the human genomic variant search engine. leukemia in adults have frequent IGH-CRLF2 and JAK2 mutations, persistence Bioinformatics 35, 1978–1980 (2019). of minimal residual disease and poor prognosis. Haematologica 102, 130–138 24. Wells, J., Jain, N. & Konopleva, M. Philadelphia chromosome-Like acute lym- (2016). phoblastic leukemia: progress in a new cancer subtype. Clin. Adv. Hematol. 11. Roberts, K. G. et al. High frequency and poor outcome of philadelphia Oncol. 15, 554–561 (2017). chromosome-like acute lymphoblastic leukemia in adults. J. Clin. Oncol. 35, 25. Tran, T. H. & Loh, M. L. Ph-like acute lymphoblastic leukemia. Hematology 394–401 (2017). 2016, 561–566 (2016). 12. Herold, T., Baldus, C. D. & Gökbuget, N. Ph-like acute lymphoblastic leukemia in 26. Tanasi, I. et al. Efficacy of tyrosine kinase inhibitors in Ph-like acute lympho- older adults. N. Engl. J. Med. 371, 2235–2235 (2014). blastic leukemia harboring ABL-class rearrangements. Blood 134, 1351–1355 13. Terwilliger, T. & Abdul-Hay, M. Acute lymphoblastic leukemia: a comprehen- (2019). sive review and 2017 update. Blood Cancer J. 7, e577 (2017). 27. Patel, S. et al. Genomic analysis of adult B-ALL identifies potential markers of 14. Hunger, S. P. & Mullighan, C. G. Redefining ALL classification: toward detecting shorter survival. Leukemia Res. 56, 44–51 (2017). high-risk ALL and implementing precision medicine. Blood 125, 3977–3987 28. Braun, M. et al. Biallelic loss of CDKN2A is associated with poor response to (2015). treatment in pediatric acute lymphoblastic leukemia. Leukemia Lymphoma 58, 15. Reshmi, S. C. et al. Targetable kinase gene fusions in high risk B-ALL: a study 1162–1171 (2017). from the Children’s Oncology Group. Blood 129, 3352–3361 (2017). 29. Ou, Z. et al. The genomic landscape of PAX5, IKZF1, and CDKN2A/B. Altera- 16. Roberts, K. G. et al. Targetable kinase-activating lesions in Ph-like acute lym- tions in B-Cell precursor acute lymphoblastic leukemia. Cytogenet. Genome Res. phoblastic leukemia. N. Engl. J. Med. 371, 1005–1015 (2014). 150, 242–252 (2016). Blood Cancer Journal