Leukemia Research 109 (2021) 106612 Contents lists available at ScienceDirect Leukemia Research journal homepage: www.elsevier.com/locate/leukres Adverse prognostic impact of complex karyotype (≥3 cytogenetic alterations) in adult T-cell acute lymphoblastic leukemia (T-ALL) Eulàlia Genescà a, *, Mireia Morgades a, b, 1, Celia González-Gil a, 1, Francisco Fuster-Tormo a, Claudia Haferlach c, Manja Meggendorfer c, Pau Montesinos d, Pere Barba e, Cristina Gil f, Rosa Coll g, María-José Moreno h, Daniel Martínez-Carballeira i, Irene García-Cadenas j, Susana Vives a, b, Jordi Ribera a, José González-Campos k, Marina Díaz-Beya l, Santiago Mercadal m, María-Teresa Artola n, Antonia Cladera o, Mar Tormo p, Arancha Bermúdez q, Ferran Vall-llovera r, Pilar Martínez-Sánchez s, María-Luz Amigo t, Silvia Monsalvo u, Andrés Novo v, Marta Cervera w, Antonio García-Guiñon x, Juana Ciudad y, José Cervera d, Jesús-María Hernández-Rivas z, Isabel Granada a, b, Torsten Haferlach c, Alberto Orfao y, Francesc Solé a, Josep-Maria Ribera a, b a Josep Carreras Leukaemia Research Institute (IJC), Campus ICO-Hospital Germans Trias i Pujol, Universitat Autònoma de Barcelona (UAB), Badalona, Spain b Clinical Hematology Department, ICO-Hospital Germans Trias i Pujol, Badalona, Spain c MLL Munich Leukemia Laboratory, Munich, Germany d Hospital Universitari Politècnic La Fe & CIBERONC, Instituto Carlos III, Madrid, Spain e Clinical Hematology Service, Hospital Universitari de la Vall d’Hebron, Barcelona, Spain f Clinical Hematology Service, Hospital General de Alicante, Alicante, Spain g Clinical Hematology Service, Hospital Josep Trueta-ICO, Girona, Spain h Clinical Hematology Service, Hospital Virgen de la Victoria, Málaga, Spain i Clinical Hematology Service, Hospital Universitario Central de Asturias, Oviedo, Spain j Clinical Hematology Service, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain k Clinical Hematology Service, Hospital Virgen del Rocío, Sevilla, Spain l Clinical Hematology Service, Hospital Clínic de Barcelona, Barcelona, Spain m Clinical Hematology Service, Hospital Duran i Reynals-ICO, Hospitalet del LLobregat, Spain n Clinical Hematology Service, Hospital Universitario de Donostia, Donostia, Spain o Clinical Hematology Service, Hospital Son Llàtzer, Palma de Mallorca, Spain p Clinical Hematology Service, Hospital Clínico de Valencia, Valencia, Spain q Clinical Hematology Service, Hospital Marqués de Valdecilla, Santander, Spain r Clinical Hematology Service, Hospital Mútua de Terrassa, Terrassa, Spain s Clinical Hematology Service, Hospital 12 de Octubre, Madrid, Spain t Clinical Hematology Service, Hospital Morales Meseguer, Murcia, Spain u Clinical Hematology Service, Hospital Gregorio Marañón, Madrid, Spain v Clinical Hematology Service, Hospital Son Espases, Palma de Mallorca, Spain w Clinical Hematology Service, Hospital Joan XXIII, Tarragona, Spain x Clinical Hematology Service, Hospital Arnau de Vilanova, Lleida, Spain y Centro de Investigación del Cáncer (IBMCC-CSIC/USAL) (CIC), Hospital Clínico Universitario de Salamanca (HUS), Instituto Bio-Sanitario de Salamanca (IBSAL), CIBERONC, Salamanca, Spain z Hospital Clínico Universitario de Salamanca (HUS), Instituto de Biología Molecular y Celular del Cáncer (IBMCC), Centro de Investigación del Cáncer (CIC), Spain A R T I C L E I N F O A B S T R A C T Keywords: The potential prognostic value of conventional karyotyping in adult T-cell acute lymphoblastic leukemia (T-ALL) Prognosis remains an open question. We hypothesized that a modified cytogenetic classification, based on the number and Cytogenetics type of cytogenetic abnormalities, would allow the identification of high-risk adult T-ALL patients. Complex Adult T-ALL * Corresponding author at: ALL Research Group, Josep Carreras Leukaemia Research Institute (IJC), Camí de les Escoles s/n, Edifici IJC, 08916 Badalona, Spain. E-mail address: egenesca@carrerasresearch.org (E. Genescà). 1 These authors contributed equally. https://doi.org/10.1016/j.leukres.2021.106612 Received 24 February 2021; Received in revised form 3 May 2021; Accepted 4 May 2021 Available online 8 June 2021 0145-2126/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). E. Genescà et al. Leukemia Research 109 (2021) 106612 NGS karyotype defined by the presence of ≥3 cytogenetic alterations identified T-ALL patients with poor prognosis in Therapy this study. Karyotypes with ≥3 abnormalities accounted for 16 % (22/139) of all evaluable karyotypes, corre­ sponding to the largest poor prognosis cytogenetic subgroup of T-ALL identified so far. Patients carrying kar­ yotypes with ≥3 cytogenetic alterations showed a significantly inferior response to therapy, and a poor outcome in terms of event-free survival (EFS), overall survival (OS) and cumulative incidence of relapse (CIR), inde­ pendently of other baseline characteristics and the end-induction minimal residual disease (MRD) level. Addi­ tional molecular analyses of patients carrying ≥3 cytogenetic alterations showed a unique molecular profile that could contribute to understand the underlying molecular mechanisms of resistance and to evaluate novel tar­ geted therapies (e.g. IL7R directed) with potential impact on outcome of adult T-ALL patients. 1. Introduction ALL-HR-2003 trial the response to induction chemotherapy was evalu­ ated by cytomorphology and flow cytometry. Good responders -<5 % Genetic information in acute lymphoblastic leukemia (ALL) has long blasts (cytologic CR), and MRD ≤ 0.1 %- proceeded to consolidation been obtained via conventional cytogenetics in clinical practice. Kar­ chemotherapy, and whenever a good MRD response was maintained yotyping performed at diagnosis in childhood and adult B-cell precursor (MRD ≤ 0.05 %) they followed maintenance chemotherapy treatment. ALL (BCP-ALL) patients has allowed to identify recurrent abnormalities Poor responders (> 5 % blasts and/or MRD ≥ 0.1 %) received intensi­ with prognostic relevance [1–4]. This is particularly true for primary fication of induction treatment, followed by allogeneic-hematopoietic genetic events such as t(9;22)(q34;q11.2), t(4;11)(q21;q23), t(12;21) stem cell transplantation (Allo-HSCT). Poor responders after consoli­ (p13;q22) or t(1;19)(q21;p13.3) [1–4], and numerical chromosomal dation treatment (≥ 0.05 %) were also allocated to allo-HSCT. MRD changes like low hypodiploidy/near triploidy, high hyperdiploidy [1,3, assessment by flow cytometry was partially centralized in the 5], and complex karyotypes (CK) [1–4]. In contrast, widely accepted ALL-HR-03 protocol. In the ALL-HR-11 trial the treatment allocation was classification based only conventional cytogenetics is still missing for exclusively based on fully centralized flow cytometry (MRD cut-off of T-ALL. Moreover, the cut-off set in ≥5 cytogenetic abnormalities to ≤0.1 % for good responders after induction treatment and MRD cut-off define CK in BCP-ALL, might not be appropriate for T-ALL, since in other of ≤0.01 % for good responders after consolidation treatment), using the leukemias such as acute myeloid leukemia (AML) and myelodysplastic EuroFlow guidelines, as previously described [18]. syndromes, a different number of cytogenetic lesions is used for the definition of CK associated with adverse patient outcomes [6–8]. 2.2. Karyotyping Few attempts have been made in T-ALL to assess the prognostic impact of conventional karyotyping, based on large cohorts of homo­ Bone marrow (BM) and/or peripheral blood (PB) samples were geneously treated patients. In a study from the Pediatric Oncology processed at the institution-of-origin laboratories using standard cyto­ Group only a normal karyotype and t(10;14)(q24;q11) were associated genetic methods, and karyotypes were reported according to the Inter­ with a better prognosis in childhood T-ALL [9]. In turn, CK defined ac­ national System for Human Cytogenetic Nomenclature [19]. cording to the Moorman’s criteria [1], was associated with a poorer Cytogenetic data were compiled and centrally analyzed. Karyotypes overall survival (OS) in a large study (n = 356 adult cases) by the UKALL with ≥15 metaphases without abnormalities were classified as normal XII/ECOG Group [10]. Lastly, the Nordic study showed that only karyotypes (NK), whereas cases with <15 normal metaphases or with no infrequent TCR rearrangements were associated with a poorer prognosis growth of leukemic cells were considered to be non-evaluable (NE). Any in pediatric T-ALL [11]. A major challenge in developing a cytogenetic additional genetic data coming from FISH analysis, was not considered classification with prognostic significance for T-ALL is the rarity of the in the genetic classification. Monosomal karyotype (MK) was defined disease. In addition, while many recurrent abnormalities are rather according to Breems et al. definition [20], and the number of karyotypic infrequent, others are better detected by genomic techniques, including abnormalities was assessed according to the criteria of Chu et al. [21]. some recurrent primary alterations in TAL1, LMO2 and TLX3 tran­ Conventional CK included cases with ≥5 unrelated chromosomal ab­ scription factors, that have been associated with prognostic value in normalities [1] and hyperdiploidy was defined as >46 chromosomes pediatric T-ALL patients [12]. without structural abnormalities. Here we investigated the prognostic impact of cytogenetic data in a large series of 216 adult T-ALL patients treated within two consecutive minimal residual disease (MRD)-oriented trials from the Spanish 2.3. Next generation sequencing (NGS) PETHEMA (Programa Español de Tratamientos en Hematología) group. Our goal was to assess the potential usefulness of karyotypic data to T-ALL cases with available DNA obtained at diagnosis (n = 30) from identify high-risk adult T-ALL patients, and to further dissect the po­ the PETHEMA cohort, were analyzed by target deep sequencing (TDS) tential underlying molecular alterations. using a NGS panel (NGSp) designed at the Agilent SureDesing platform (Agilent Technologies, Santa Clara, CA), from which 5 DNA corre­ 2. Methods sponded to patients with ≥3 CK and 25 to patients with <3 alterations. The genes included in the panel are listed in Supplemental Table 1. To 2.1. Patient cohort increase genomic data of cases with ≥3 CK, we took advantage of the genomic data generated at the MLL Munich Leukemia Laboratory A total of 216 patients diagnosed with T-ALL according to the WHO (Munich, Germany) and we included 38 T-ALL cases from the MLL criteria and treated within two consecutive MRD-oriented high-risk cohort from which whole genome sequencing (WGS) data was available. adults ALL protocols (ALL-HR-2003 [from 2003 to 2012, Main patient characteristics are shown in Supplemental Table 2. We NCT00853008], [2011 to 2019, NCT01540812]) were analyzed. The select adult patients in the same age range as the Spanish cohort [15;60]. immunological T-ALL subtypes were defined according to the European We distributed them in cases with ≥3 CK or <3 alterations, similarly as Group for the Immunological Characterization of Leukemias (EGIL) we did for the PETHEMA cohort, taking only into account cytogenetic criteria [13], and the WHO [14] criteria was used to identify early T-cell data. Briefly, WGS libraries were prepared with the TruSeq PCR free precursor (ETP)-ALL, as previously described [15]. Treatment protocol library prep kit. Hundred-fifty base pair (bp) paired-end sequences were schedules have been described elsewhere [16,17]. Briefly, in the generated on a NovaSeq 6000 or HiSeqX instrument (Illumina, San Diego, CA) with 100x coverage. NGSp libraries were prepared with the 2 E. Genescà et al. Leukemia Research 109 (2021) 106612 SureSelect XT HS (Agilent technologies) chemist. Pooled libraries were Table 1 sequenced 2 × 75bp paired-end sequences on a Miseq instrument Demographic, clinical, laboratory and genetic features of adult T-ALL patients (Illumina) at 300X mean depth. Fastq files from WGS and TDS were included in this study. aligned to the hg19 reference genome. The Illumina tumor/unmatched Patient distribution normal workflow was used for the WGS variant calling. Here, a mixture Clinical and laboratory features of genomic DNA from multiple anonymous donors (n = 6) was used as Age (yrs)* 32 [15; 60] normal controls. TDS variant calling was performed using Samtools4 Female/male ratio 54/153 (version 1.10, http://www.htslib.org/) and VarScan2 (version 2.4.0, htt Lymph node enlargement 109/183 (60 %) Splenomegaly 72/201 (36 %) ps://dkoboldt.github.io/varscan/). Exonic and splice-site variants ob­ Hepatomegaly 47/198 (24 %) tained by WGS and TDS were annotated using ANNOVAR (version 2018- Mediastinal mass 96/205 (47 %) 04-16, http://annovar.openbioinformatics.org/en/latest/). Poly­ WBC count (x109/L)* 40.8 [0.6; 842] morphisms described in population databases (1000Genomes, ExAC, CNS involvement 20/198 (10 %) gnomAD and Exome Variant Server) with a calculated mean population Immunophenotype Unclassified 19 (9 %) frequency > 0.1 % were excluded from further analyses. Annotated ETP-ALL 35 (17 %) variants were selected after filtering out calls according to the following Pre-T 35 (17 %) criteria: coverage <30X, <8 alternative reads or variant allelic fraction Cortical 79 (38 %) (VAF) <5 %. Finally, selected variants were annotated using DBNSFP in Mature 39 (19 %) T-ALL genetic hallmarks silico predictors (SIFT, Polyphen2, LRT, MutationTaster, Muta­ TLX1 3/207 (1.5 %) tionAssessor, FATHMM, PROVEAN, VEST3, MetaSVM and MetaLR), TLX3 1/207 (0.5 %) excluding benign predicted variants (>5 predictor scores considering a TAL1 4/207 (2 %) variant as benign). LMO2 2/207 (1 %) KMT2A rearranged 3/207 (2 %) TCR rearranged 1/207 (0.5 %) 2.4. Outcome variables and statistical analysis del(6q) 11/207 (5 %) CDKN2A/B gene deletion 31/53 (58 %) Overall survival (OS) was measured from the time of diagnosis to the TCRG gene deletion 24/31 (77 %) time of death or last follow-up. Event-free survival (EFS) was calculated Treatment Chemotherapy 97/139 (70 %) from diagnosis to last follow-up, considering failure, relapse or death by Allo-HSCT 42/139 (30 %) any cause as events. Cumulative incidence of relapse (CIR) was Response to treatment measured from first CR date to relapse or last follow-up, being non- Slow response at day +14 90/186 (48 %) relapse mortality the competing event. Main demographic, clinical, Nº of induction cycles to CR: laboratory and genetic characteristics were described. Comparisons One induction 156/187 (83 %) Two inductions 31/187 (17 %) between groups were performed by the chi-square test or Fisher’s exact CR 187/207 (90 %) test (categorical variables) and the median test (numerical variables). MRD day +35 < 0.1 % 125/162 (77 %) EFS and OS curves were plotted by the Kaplan-Meier method and were MRD day +35 < 0.01 % 90/154 (58 %) compared by the log-rank test. Estimates of CIR were calculated using Results expressed as number of cases (percentage) or * as a median [range]. cumulative incidence functions and compared using Gray’s test. Multi­ Specific alterations detected included: TLX1, 3 cases with t(10;14); TLX3 1 case variable analyses were performed using the Cox proportional hazards with t(5;7)(q35;q21); TAL1 1 case with t(1;14)(p32;q21) and t(1;12)(p32;q13), model for OS and EFS, and the Fine and Gray model for CIR. The sig­ and 2 cases with del(1)(p32); LMO2, 2 cases with t(11;14)(p13;q11). nificance level was set at P < .05. All statistical analyses were carried out MRD values were considered only for those patients that reached CR. Yrs: years; with the SPSS package v24.0 and the R 3.6.3 software. WBC: white blood cells; CNS: central nervous system; ETP-ALL: early T-cell precursor acute lymphoblastic leukemia; CR: complete remission; MRD: mini­ 3. Results mal residual disease; d+14: fourteen days after starting induction treatment; d+35: thirty-five days after starting induction treatment; Allo-HSCT: allogeneic hematopoietic stem cell transplantation. From a total of 216 treated adults, 207 were included in the study (100 from the ALL-HR-2003 and 107 from the ALL-HR-11 trials). Lack of karyotype (n = 5), lack of follow-up (n = 2), protocol violation (n = 1) were also classified according to the number of chromosomal alter­ and T-cell lymphoma (n = 1) were the reasons of exclusion. Detailed ations, independently of their nature (Supplemental Table 3). patient characteristics at diagnosis and at follow-up are described in Based on the above main cytogenetic subgroups, we investigated the Table 1. Overall, 74/207 (36 %) showed a normal karyotype (NK) and potential clinical impact of this cytogenetic classification on patient 68/207 (33 %) had non-evaluable (NE) cytogenetic results. Patients outcome (OS, EFS and CIR). Overall, patients with NE karyotype had a with abnormal karyotypes (65/139, 47 %) were grouped as follows: 1) very similar outcome to cases with NK; in contrast, the outcome of NE structural abnormalities (SA) only, (38/139 [29 %]); 2) structural + and NK cases was clearly better than that of patients with an abnormal numerical abnormalities (SNA) (21/139 [15 %]; and 3) hyperdiploidy karyotype in terms of EFS, but not of OS and CIR (Supplemental Fig. 1). only (6/139 [4 %]). In addition, karyotypes were classified according to These findings suggest the absence of poor prognosis karyotypes among classical CK (≥5 abnormalities, 10/139 [7 %]) and MK (10/139 [7 %]). NE cases. Despite this, cases with NE karyotypes were excluded from Chromosome 14 was the most frequently lost chromosome (n = 3) and further prognostic analyses. In such analyses, specific cytogenetic sub­ no cases only showing chromosomal losses were detected in our cohort. groups such as MK and CK (≥5 cytogenetic abnormalities) were asso­ The percentages of NK, NE karyotypes, and CK, were similar to those ciated with poor OS (3y prob [CI 95 %] of 15 % [0 %; 40 %] for MK vs 51 reported by others [10]. Previously reported primary cytogenetic al­ % [41 %; 61 %] for non-MK cases, p = 0∙012; and of 13 % [0 %; 36 %] terations involving T-ALL-associated transcription factors such as for ≥5 CK vs 51 % [41 %; 61 %] for other cases with <5 chromosomal TLX1/TLX3 and LMO [22–25] were also identified at very low fre­ alterations, p = 0∙012), EFS (1y prob [CI 95 %] of 40 % [10 % ; 70 %] quencies (Table 1). Of note, patients carrying the TLX1/TLX3 and LMO2 for MK vs 64 % [55 %; 73 %] for non-MK, p=<0∙001; and of 40 % [10 alterations belonged to the SA category and the genetic event (rear­ %; 70 %] for ≥5 CK vs 64 % [55 %; 73 %] for those with <5 cytogenetic rangement) was mostly found as unique abnormality or it was associated alterations, p = 0∙004), and CIR (1y prob [CI 95 %] of 88 % [6 %; 99 %] with another single alteration. In turn, 7/11 (64 %) of cases with del(6q) for MK vs 43 % [33 %; 53 %] for non-MK, p = 0∙001; and of 56 % [16 %; were found as a single or as a partner alteration. In parallel, karyotypes 83 %] for ≥5 CK vs 45 % [34 %; 55 %] for those with <5 chromosomal 3 E. Genescà et al. Leukemia Research 109 (2021) 106612 changes, p = 0∙013), as previously reported by others [3,10]. Patients immunophenotype did not prove to be independent predictors of patient with MK and CK (≥5 abnormalities) were both included in the large SNA outcome in the multivariate analysis. No relationship was found be­ cytogenetic group (15 % T-ALL cases) and retained their adverse clinical tween ≥3 CK and other patient characteristics such as age, WBC count, outcome when compared to non-SNA patients (Supplemental Fig. 2). In CNS involvement and ETP-ALL. contrast, hyperdiploidy and SA did not show an adverse impact on T-ALL In order to analyze the underlying reasons for the poor prognosis of patient outcome (data not shown). Further investigation of the prog­ ≥3 CK in adult T-ALL, we investigate the potential association of ≥3 CK nostic impact of the number of chromosomal alterations independently with response to therapy. Despite the similar initial response to therapy of their nature, showed that the presence of ≥3 cytogenetic abnormal­ (day+14) between ≥3 CK and other T-ALL cases (p = 0∙167), ≥3 CK ities (≥3 CK) in our adult T-ALL patients, already conferred a poorer patients showed a significantly lower end-induction CR rate (77 % vs. 94 prognosis in terms of OS, EFS and CIR, similar to that of cases showing a % for the remaining T-ALL, p = 0∙032). Of note, 2/5 ≥ 3 CK patients classical CK with ≥5 cytogenetic alterations (Fig. 1 and Supplemental who did not reach CR at the end of induction therapy died due to toxicity Table 3). Of note, karyotypes with ≥3 abnormalities accounted for 16 % whereas the other 3/5 did not continue the protocol because of re­ of all evaluable karyotypes (22/139), which would represent the largest fractory disease. Among those patients that achieved CR, a similar poor prognosis cytogenetic subgroup identified so far in T-ALL. All cases number of induction cycles were administered for ≥3 CK vs other T-ALL (n = 4) in which chromosomal rearrangements or deletion involving the cases. Consistent with the lower CR rate observed for ≥3 CK patients, TAL1 transcription factor were detected, were included in this category, lack of end-induction MRD response (>0.01 %) was more frequently in addition to patients with MK and CK (≥5 abnormalities) (Supple­ observed among patients with ≥3 CK than the remaining patients (65 % mental Table 4). Other variables than the new ≥3 CK cytogenetic sub­ vs. 40 %, p = 0∙06). However, no differences in the frequencies of group, the MK, ≥5 CK and SNA cytogenetic category, that showed a chemotherapy or allo-HSCT allocation at the end of consolidation significant impact on patient outcome, included treatment protocol, age, therapy were observed on comparison of patients within ≥3 CK with the white blood cell (WBC) count, and ETP-ALL (Table 2). Multivariate remaining. analysis confirmed the independent adverse prognostic value of the In order to better understand the molecular mechanisms underlying presence of ≥3 cytogenetic abnormalities at diagnosis, for all the three the poorer outcome of ≥3 CK patients, molecular studies by NGS were prognostic end-points evaluated (Table 2). Importantly, karyotypes with performed. T-ALL cases with available DNA obtained at diagnosis (n = ≥3 abnormalities retained their adverse prognostic impact also when 30) from the PETHEMA cohort, were analyzed by TDS. To increase MRD data obtained at the end of induction therapy (day+35) were genomic data of cases with ≥ 3 CK, we included additional 38 T-ALL included in the multivariate analyses (Table 3). Of note, in our cohort cases from the MLL cohort from which WGS data was available. Selec­ having ≥3 CK was the only risk factor, together with age, with an in­ tion criteria of the T-ALL cases from the MLL cohort included in the dependent predictive value to identify at diagnosis patients that will study and homogenization and standardization of NGS data analysis is relapse (Table 2). In turn, the ≥3 CK together with the WBC count at described in material and methods section. Overall, we observed that the diagnosis and MRD status at day+35, were the only independent pre­ mutational profile identified in the T-ALL patients analyzed was similar dictors for OS and EFS (Tables 2 and 3). Other variables such as the ETP to those previously reported by others (Fig. 2) [26,27]. Interestingly, Fig. 1. Impact of karyotypes with ≥3 vs 0-2 cytogenetic alterations for OS (A), EFS (B) and CIR (C) at 3-year probability in the PETHEMA cohort. 4 E. Genescà et al. Leukemia Research 109 (2021) 106612 Table 2 Univariate and multivariate analysis of prognostic factors for OS, EFS and CIR in adult T-ALL patients. N (positive/ OS N (positive/ EFS N (positive/ CIR P value P value P value evaluable) HR (95 %CI) evaluable) HR (95 %CI) evaluable) HR (95 %CI) Univariate analysis ALL-HR2003 100/207 1.62 (1.05;2.5) 0.028 100/207 1.43 (0.97;2.09) 0.069 89/187 1.18 (0.73;1.89) 0.500 protocol Age* 207 1.01 (0.99;1.02) 0.456 207 0.99 (0.98;1.01) 0.897 /87 0.97 (0.95;0.99) 0.022 WBC count (x109/ 207 1.002 0.028 207 1.002 0.032 187 1.001 0.480 L)* (1.000;1.004) (1.000;1.003) (0.999;1.003) CNS involvement 20/198 1.16 (0.56;2.41) 0.681 20/198 1.24 (0.66;2.3) 0.498 16/178 1.08 (0.47;2.48) 0.850 ETP-ALL 35/188 1.83 (1.11;3.01) 0.018 35/188 1.67 (1.05;2.66) 0.029 27/170 1.27 (0.69;2.33) 0.440 CK (≥3 22/139 3.04 (1.67;5.51) <0.001 22/139 2.96 (1.68;5.2) <0.001 17/127 2.76 (1.41;5.4) 0.003 alterations) CK (≥5 10/139 2.52 (1.19;5.31) 0.016 10/139 2.72 (1.34;5.54) 0.006 9/127 2.57 (1.05;6.3) 0.038 alterations) MK 10/139 2.53(1.19;5.35) 0.015 10/139 3.23 (1.64;6.35) 0.001 8/127 3.67 (1.73;7.77) <0.001 SNA 21/139 2.30 (1.27;4.19) 0.006 21/139 3.01 (1.75;5.17) <0.001 17/127 3.23 (1.81;5.75) <0.001 SA 38/139 1.23 (0.72;2.09) 0.441 38/139 1.14 (0.69;1.82) 0.595 35/127 0.95 (0.50;1.79) 0.860 Multivariate analysis ALL-HR2003 – – NS – – NS – NA NA protocol (0.75) (0.29) Age* – NA NA – NA NA – 0.97 (0.94;0.99) 0.032 WBC count (x109/ – 1.003 0.001 – 1.004 <0.001 – NA NA L)* (1.001;1.005) (1.002;1.005) ETP-ALL – – NS – – NS – – – (0.33) (0.33) CK (≥3 – 3.37 (1.76;6.46) <0.001 – 3.43 (1.86;6.31) <0.001 – 2.91 (1.48;5.5) 0.002 alterations) * Age and WBC were considered as continuous variables. HR: Hazard ratio; CI: confidence interval; OS: overall survival; EFS: event-free survival; CIR: cumulative incidence of relapse; WBC: white blood cell; ETP-ALL: early T-cell precursor ALL; CNS: central nervous system; CK: complex karyotype; SNA: structural + numerical abnormalities; SA: only structural alterations. Table 3 Univariate and multivariate analysis of prognostic factors for OS and CIR in adult T-ALL patients including end-induction MRD level. OS CIR N (positive/evaluable) P value N (positive/evaluable) P value HR (95 %CI) HR (95 %CI) Univariate analysis ALL-HR2003 protocol 89/187 1.72 (1.05;2.82) 0.031 89/187 1.18 (0.73;1.89) 0.500 Age* 187 0.99 (0.98;1.02) 0.760 187 0.97 (0.95;0.99) 0.022 WBC count (x109/L)* 187 1.002 (1.000;1.004) 0.054 187 1.001 (0.999;1.003) 0.480 CNS involvement 16/178 0.98 (0.39;2.43) 0.960 16/178 1.08 (0.47;2.47) 0.850 ETP-ALL 27/170 1.47 (0.79;2.70) 0.219 27/170 1.27 (0.69;2.33) 0.440 CK (≥3 alterations) 17/127 2.64 (1.30;5.36) 0.007 17/127 2.76 (1.41;5.38) 0.003 CK (≥5 alterations) 9/127 2.81 (1.26;6.27) 0.012 9/127 2.57 (1.05;6.30) 0.038 MK 8/127 2.53 (1.07;5.96) 0.034 8/127 3.67 (1.73;7.78) <0.001 SNA 17/127 2.07 (1.03;4.16) 0.040 17/127 3.23 (1.81;5.75) <0.001 SA 35/127 1.32 (0.74;2.37) 0.348 35/127 0.95 (0.50;1.79) 0.860 MRD day +35 (≥0.1 %) 37/162 1.94 (1.11;3.37) 0.019 37/162 1.36 (0.76;2.44) 0.300 MRD day +35 (≥0.01 %) 64/154 2.30 (1.30;4.08) 0.004 64/154 1.20 (0.71;2.05) 0.490 Multivariate analysis ALL-HR2003 protocol – – NS (0.316) – NA NA Age* – NA NA – 0.97 (0.94;0.99) 0.032 WBC count (x109/L)* – 1.005 (1.002;1.009) 0.001 – NA NA ETP-ALL – – NS (0.807) – – – CK (≥3 alterations) – 4.17 (1.79;9.69) 0.001 – 2.91 (1.48;5.70) 0.002 MRD day +35 (≥0.01 %) . 2.27 (1.12;4.60) 0.023 – NA NA * Age and WBC were considered as continuous variables. HR: Hazard ratio; CI: confidence interval; OS: overall survival; CIR: cumulative incidence of relapse; WBC: white blood cell; ETP-ALL: early T-cell precursor ALL; CNS: central nervous system; CK: complex karyotype; SNA: structural + numerical abnormalities; SA: only structural alterations; MRD: minimal residual disease. three distinct genetic signatures were identified manually, when for ≥3 CK vs. 22 % for other T-ALL cases; p = 0∙26) gene. A subset of comparing ≥3 CK vs. <3 CK patient groups. Thus, there were genes that genes that was found to be non-mutated in ≥3 CK patients while were not differentially mutated in ≥3 CK vs other T-ALL cases. These mutated in other T-ALL cases, such as genes involved in the included mutations in the NOTCH1/FBXW7 signaling pathway (53 % for JAK1/3/AKT1 signaling pathway (0 % for ≥3 CK vs. 25 % for other ≥3 CK vs. 67 % for other T-ALL patients; p = 0∙31) and in PHF6 (35 % T-ALL cases, p = 0∙02) and genes related to DNA methylation, e.g. the 5 E. Genescà et al. Leukemia Research 109 (2021) 106612 Fig. 2. Mutational spectrum of ≥3 CK and <3 CK adult T-ALL patients. Genomic data from the PETHEMA (n = 30) and MLL (n = 38) cohorts are shown together. Global mutational frequency for each gene is shown on the left. Genes are grouped according the three signatures identified and ordered by frequency. Each colored square indicates a positive mutation for the indicated gene. Blue dark squares identify patients with ≥3 abnormalities. Light blue squares identify patients <3 CK. DNMT3A (0 % for ≥3 CK vs. 10 % among other T-ALL cases: p = 0∙18) BCP-ALL, a lower genetic complexity is required in adult T-ALL patients and TET2 (0 % for ≥3 CK vs. 6 % for other T-ALL patients; p = 0∙31) to define adverse complex karyotypes. Thus, complex karyotype with gene. These genes have been previously associated with an immature ≥3 alterations is a marker of inferior outcome in terms of OS, EFS and T-ALL immunophenotype [28]. Finally, a third group of genes that were CIR, independently of other prognostic features such as age, WBC count mutated at greater frequencies in ≥3 CK patients vs other T-ALL cases, at diagnosis and the MRD status after induction therapy. The adverse including the KMT2C (35 % vs 10 %, respectively; p = 0∙01), IL7R (29 % prognostic impact of ≥3 CK together with age were the only indepen­ vs 8 %, respectively; p = 0∙02) and GATA3 (12 % vs 0 %, respectively; p dent predictors for an impending relapse after starting therapy, ruling = 0∙01) genes. out the predictive value of MRD to identify patients with early relapse. This is probably due to the fact that while MRD could identify those 4. Discussion patients at lower (MRD negative) vs high (MRD positive) risk of relapse, ≥3 CK discriminated among the MRD positive patients those at risk of Conventional cytogenetics is still a widely used technique in the earlier recurrence of the disease. These results are of clinical relevance diagnostic of ALL for the fast and robust identification of Ph+ ALL or because in contrast to MRD results, karyotypic data are already available KMT2A/AFF1 patients with important therapeutic consequences. How­ at diagnosis, which facilitates early risk stratification of the disease for ever, with the irruption of high-throughput genomic techniques, the early risk-adapted treatment (intensification) decisions. usefulness of conventional cytogenetics has been questioned, particu­ In order to better understand the mechanisms underlying the poorer larly for T-ALL patients, due to the low number of cases with informative prognosis of patients with ≥3 CK we investigated the reasons for their abnormalities (30 %) and the lack of a widely-accepted standard cyto­ dismal outcome and the associated gene mutational profiles. Patients genetic classification. This, together with the low prevalence of patients with ≥3 CK showed a significantly higher rate of refractoriness to in­ with T- vs BCP-ALL, has translated into a very limited information on the duction therapy used in the two PETHEMA protocols, together with a prognostic value of karyotyping, particularly among adult T-ALL. poorer clearance of blast cells, as reflected by lower rates of both CR and The prognostic impact of conventional cytogenetics was assessed in a MRD negativity after induction therapy. Treatment intensification by large cohort of adult T-ALL patients treated with two sequential trails allo-HSCT was not more prevalent in patients with ≥3 CK, possibly due from the PETHEMA group. Our results show for the first time, that to premature exclusion of these patients due to refractoriness or early karyotype is a strong independent prognostic marker that can be ob­ relapse. Treatment resistance and tumor genetics have both been tained already at diagnosis to predict outcome of adult-TALL patients. recurrently associated with specific gene mutational profiles in ALL [29, Specifically, the presence of ≥3 cytogenetic alterations was the most 30]. Of note, genomic analysis performed combining NGS data revealed informative adverse prognostic factor for OS, EFS and CIR when T-ALL the presence of unique mutational profiles among ≥3 CK cases vs other karyotypes were classified according to the number of cytogenetic al­ T-ALL patients. These profiles consisted in a predominance of mutations terations, using a well standardized method [21]. Karyotypes with ≥3 in the KMT2C, GATA3 and IL7R genes, and absence of mutations abnormalities accounted for 16 % of all evaluable karyotypes, thereby involving genes of the JAK1/3/AKT1 signaling pathway and genes representing the largest poor prognosis cytogenetic subgroup identified related to DNA methylation, such as the DNMT3A and TET2. Such so far in T-ALL. However, it should be pointed out that the classification mutational profile might contribute to explain the poorer outcome for of adult T-ALL patients from this series according to previously defined ≥3 CK patients. As an example, activating mutations in IL7R, confer primary cytogenetic events (gene rearrangements), was not possible due glucocorticoid resistance and poor clinical outcome in childhood T-ALL to the low number of cases in each of these subgroups (i.e. t(10;14); t [31]. Thus, IL7R works as biomarker of reduced steroid response, which (5;7)(q35;q21); t(1;14)(p32;q21); t(1;12)(p32;q13); del(1)(p32) or t could contribute to explain, at least in part, the increased resistance to (11;14)(p13;q11)). Underrepresentation of the primary genetic events treatment and the poorer blast clearance observed in patients with high in our cohort might be due to differences in the prevalence of these al­ genetic complexity (≥3 CK). In this sense, a recent report from the terations in adult vs. childhood T-ALL, to the locally performed con­ GRAALL cooperative group showed that patients with IL7Rp mutated ventional cytogenetics and to the lower sensitivity of cytogenetics vs, T-ALL were slow-responders, with a high level of MRD in day-8 bone molecular techniques, among other reasons. marrow, compared to unmutated cases, although no correlation was Despite the above limitations, our results show that compared to observed with response to prednisone [32]. 6 E. Genescà et al. Leukemia Research 109 (2021) 106612 In summary, here we propose a definition for CK in adult T-ALL [6] C. Schoch, T. Haferlach, D. Haase, et al., Patients with de novo acute myeloid leukaemia and complex karyotype aberrations show a poor prognosis despite based on the presence of ≥3 cytogenetic alterations, with important intensive treatment: a study of 90 patients, Br. J. Haematol. 112 (1) (2001) implications for improved risk stratification at diagnosis. In addition, we 118–126. have shown that patients with ≥3 CK carry unique gene mutational [7] S.S. Farag, K.J. Archer, K. Mrozek, et al., Pretreatment cytogenetics add to other profiles, that might benefit in the future from novel targeted therapies prognostic factors predicting complete remission and long-term outcome in patients 60 years of age or older with acute myeloid leukemia: results from Cancer directed against the mutated genes (e.g. IL7R). Further studies in large and Leukemia Group B 8461, Blood 108 (1) (2006) 63–73. series of adult T-ALL patients treated with similar and novel therapies [8] H. Dohner, E.H. Estey, S. Amadori, et al., Diagnosis and management of acute are necessary to confirm our results. myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet, Blood 115 (3) (2010) 453–474. [9] N.R. Schneider, A.J. Carroll, J.J. Shuster, et al., New recurring cytogenetic Contributors abnormalities and association of blast cell karyotypes with prognosis in childhood T-cell acute lymphoblastic leukemia: a pediatric oncology group report of 343 cases, Blood 96 (7) (2000) 2543–2549. E. Genescà, designed the study and wrote the manuscript. M. Mor­ [10] D.I. Marks, E.M. Paietta, A.V. Moorman, et al., T-cell acute lymphoblastic leukemia gades performed statistical analysis. C. González-Gil and F. Fuster per­ in adults: clinical features, immunophenotype, cytogenetics, and outcome from the formed NGSp sequencing and analysis. M.M. performed WGS large randomized prospective trial (UKALL XII/ECOG 2993), Blood 114 (25) (2009) 5136–5145. sequencing and analysis. E. Genescà and F. Solé, performed cytogenetic [11] K. Karrman, E. Forestier, M. Heyman, et al., Clinical and cytogenetic features of a central review and karyotype analysis. E. Genescà, M. Morgades, I. population-based consecutive series of 285 pediatric T-cell acute lymphoblastic Granada, F. Solé and J.M. Ribera analyzed data. P. Montesinos, P. Barba, leukemias: rare T-cell receptor gene rearrangements are associated with poor outcome, Genes Chromosome Cancer 48 (9) (2009) 795–805. C. Gil, R. Coll, M.J. Moreno, D. Martínez-Carballeira, I. García-Cadenas, [12] Y. Liu, J. Easton, Y. Shao, et al., The genomic landscape of pediatric and young S. Vives, J. Ribera, J.González-Campos, M. Díaz-Beya, S. Mercadal, M.T. adult T-lineage acute lymphoblastic leukemia, Nat. Genet. 49 (8) (2017) Artola, A. Cladera, M. Tormo, A. Bermúdez, F. Vall-llovera, P. Martínez, 1211–1218. M.L. Amigo, S. Monsalvo, A. Novo, M. Cervera, A. García, J. Cervera and [13] M.C. Bene, G. Castoldi, W. Knapp, et al., Proposals for the immunological classification of acute leukemias. European Group for the Immunological JM Hernández-Rivas provided clinical data and C. Haferlach, M. Meg­ Characterization of Leukemias (EGIL), Leukemia 9 (10) (1995) 1783–1786. gendorfer and T. Haferlach, F.Solé, A. Orfao and JM Ribera reviewed the [14] D.A. Arber, A. Orazi, R. Hasserjian, et al., The 2016 revision to the World Health manuscript. All authors have read and approved the manuscript. Organization classification of myeloid neoplasms and acute leukemia, Blood 127 (20) (2016) 2391–2405. [15] E. Genescà, M. Morgades, P. Montesinos, et al., Unique clinico-biological, genetic Declaration of Competing Interest and prognostic features of adult early T cell precursor acute lymphoblastic leukemia, Haematologica 105 (6) (2020) e294–e297. [16] E. Genesca, A. Lazarenkov, M. Morgades, et al., Frequency and clinical impact of The authors report no declarations of interest. CDKN2A/ARF/CDKN2B gene deletions as assessed by in-depth genetic analyses in adult T cell acute lymphoblastic leukemia, J. Hematol. Oncol. 11 (2018). Acknowledgements [17] P. Barba, M. Morgades, P. Montesinos, et al., Increased survival due to lower toxicity for high-risk T-cell acute lymphoblastic leukemia patients in two consecutive pediatric-inspired PETHEMA trials, Eur. J. Haematol. 102 (1) (2019) The authors would like to thank Rodrigo Martino, Raimundo García 79–86. Boyero, Pilar Herrera, Juan Bergua, MªJesús Peñarrubia, Eloy del Potro, [18] J.J. van Dongen, L. Lhermitte, S. Bottcher, et al., EuroFlow antibody panels for standardized n-dimensional flow cytometric immunophenotyping of normal, Magdalena Sánchez, Beatriz Soria, Aurelio López Martínez, Mª José reactive and malignant leukocytes, Leukemia 26 (9) (2012) 1908–1975. Sánchez Sánchez, Pilar Bravo, Pilar Vivancos, Alfons Serrano, José [19] L.G. Shaffer, M. Schmidt, JM-J, An International System for Human Cytogenetic Santiago Bermón, Maria Lourdes Amador Barciela and Matxalen Oli­ Nomenclature, 2013. vares Salaverri for providing retrospective clinical data for this study. [20] D.A. Breems, W.L. Van Putten, G.E. De Greef, et al., Monosomal karyotype in acute myeloid leukemia: a better indicator of poor prognosis than a complex karyotype, This project was supported by the AECC (GC16173697BIGA); ISCIII J. Clin. Oncol. 26 (29) (2008) 4791–4797. (PI19/01828) co-funded by ERDF/ESF "A way to make Europe"/ [21] K. Chun, A. Hagemeijer, A. Iqbal, M.L. Slovak, Implementation of standardized "Investing in your future", CERCA/Generalitat de Catalunya SGR 2017 international karyotype scoring practices is needed to provide uniform and systematic evaluation for patients with myelodysplastic syndrome using IPSS 288 (GRC)/ “La Caixa” P. Barba was supported by the Instituto de Salud criteria: an International Working Group on MDS Cytogenetics Study, Leuk. Res. 34 Carlos III FIS16/01433 and PERIS 2018-2020 from Generalitat de Cat­ (2) (2010) 160–165. alunya (BDNS357800). [22] U.R. Kees, R. Lukeis, J. Ford, O.M. Garson, Establishment and characterization of a childhood T-cell acute lymphoblastic leukemia cell line, PER-255, with chromosome abnormalities involving 7q32-34 in association with T-cell receptor- Appendix A. Supplementary data beta gene rearrangement, Blood 74 (1) (1989) 369–373. [23] I.D. Dube, S.C. Raimondi, D. Pi, D.K. Kalousek, A new translocation, t(10;14)(q24; q11), in T cell neoplasia, Blood 67 (4) (1986) 1181–1184. Supplementary material related to this article can be found, in the [24] M.A. Kennedy, R. Gonzalez-Sarmiento, U.R. Kees, et al., HOX11, a homeobox- online version, at doi:https://doi.org/10.1016/j.leukres.2021.106612. containing T-cell oncogene on human chromosome 10q24, Proc. Natl. Acad. Sci. U. S. A. 88 (20) (1991) 8900–8904. [25] T. Boehm, I. Lavenir, A. Forster, et al., The T-ALL specific t(11;14)(p13;q11) References translocation breakpoint cluster region is located near to the Wilms’ tumour predisposition locus, Oncogene 3 (6) (1988) 691–695. [1] A.V. Moorman, C.J. Harrison, G.A. Buck, et al., Karyotype is an independent [26] C. Vicente, C. Schwab, M. Broux, et al., Targeted sequencing identifies associations prognostic factor in adult acute lymphoblastic leukemia (ALL): analysis of between IL7R-JAK mutations and epigenetic modulators in T-cell acute cytogenetic data from patients treated on the Medical Research Council (MRC) lymphoblastic leukemia, Haematologica 100 (10) (2015) 1301–1310. UKALLXII/Eastern Cooperative Oncology Group (ECOG) 2993 trial, Blood 109 (8) [27] M. Neumann, S. Vosberg, C. Schlee, et al., Mutational spectrum of adult T-ALL, (2007) 3189–3197. Oncogene 6 (2015) 2754–2766. [2] A.V. Moorman, H.M. Ensor, S.M. Richards, et al., Prognostic effect of chromosomal [28] J. Zhang, L. Ding, L. Holmfeldt, et al., The genetic basis of early T-cell precursor abnormalities in childhood B-cell precursor acute lymphoblastic leukaemia: results acute lymphoblastic leukaemia, Nature 481 (7380) (2012) 157–163. from the UK Medical Research Council ALL97/99 randomised trial, Lancet Oncol. [29] C.L. Dieck, G. Tzoneva, F. Forouhar, et al., Structure and mechanisms of NT5C2 11 (5) (2010) 429–438. mutations driving thiopurine resistance in relapsed lymphoblastic leukemia, [3] C. Motllo, J.M. Ribera, M. Morgades, et al., Prognostic significance of complex Cancer Cell 34 (2018) 136–147. karyotype and monosomal karyotype in adult patients with acute lymphoblastic [30] K. Oshima, H. Khiabanian, A.C. da Silva-Almeida, et al., Mutational landscape, leukemia treated with risk-adapted protocols, Cancer 120 (24) (2014) 3958–3964. clonal evolution patterns, and role of RAS mutations in relapsed acute [4] M. Lafage-Pochitaloff, L. Baranger, M. Hunault, et al., Impact of cytogenetic lymphoblastic leukemia, Proc. Natl. Acad. Sci. U. S. A. 113 (2016) 11306–11311. abnormalities in adults with Ph-negative B-cell precursor acute lymphoblastic [31] P.P. Zenatti, D. Ribeiro, W. Li, et al., Oncogenic IL7R gain-of-function mutations in leukemia, Blood 130 (16) (2017) 1832–1844. childhood T-cell acute lymphoblastic leukemia, Nat. Genet. 43 (2011) 32–39. [5] C. Charrin, X. Thomas, M. Ffrench, et al., A report from the LALA-94 and LALA-SA [32] R. Kim, N. Boissel, A. Touzart, et al., Adult T-cell acute lymphoblastic leukemias groups on hypodiploidy with 30 to 39 chromosomes and near-triploidy: 2 possible with IL7R pathway mutations are slow-responders who do not benefit from expressions of a sole entity conferring poor prognosis in adult acute lymphoblastic allogeneic stem-cell transplantation, Leukemia 34 (7) (2020) 1730–1740. leukemia (ALL), Blood 104 (8) (2004) 2444–2451. 7