Prognostic Impact of Copy Number Alterations’ Profile and AID/RAG Signatures in Acute Lymphoblastic Leukemia (ALL) with BCR::ABL and without Recurrent Genetic Aberrations (NEG ALL) Treated with Intensive Chemotherapy

Simple Summary

Adult ALL is a highly aggressive blood cancer. Two classes of genetic aberrations are responsible for ALL: primary aberrations followed by secondary aberrations. Currently, primary aberrations are used for estimating patients’ risk in adult ALL. In this study, we reassessed the importance of primary and secondary copy number alterations (CNA) aberrations in intensively treated adult ALL patients in correlation to RAG/AID mutator enzyme expression. Primary aberrations alone specified the risk of 30% of patients. To define the prognosis of the remaining 70%, we identified high-risk and low-risk CNA profiles. We found the CNA profiles correlated with differential RAG/AID expression profiles. Furthermore, the outcome of CNA^neg adult ALL was stratified by AID expression. Thus, we suggested mechanisms linking secondary aberrations with patients’ outcomes and mutator enzymes. Finally, we propose a revised version of risk stratification in adult ALL patients which incorporates primary and secondary genetic lesions.

Abstract

Adult acute lymphoblastic leukemia (ALL) is associated with poor outcomes. ALL is initiated by primary aberrations, but secondary genetic lesions are necessary for overt ALL. In this study, we reassessed the value of primary and secondary aberrations in intensively treated ALL patients in relation to mutator enzyme expression. RT-PCR, genomic PCR, and sequencing were applied to evaluate primary aberrations, while qPCR was used to measure the expression of RAG and AID mutator enzymes in 166 adult ALL patients. Secondary copy number alterations (CNA) were studied in 94 cases by MLPA assay. Primary aberrations alone stratified 30% of the patients (27% high-risk, 3% low-risk cases). The remaining 70% intermediate-risk patients included BCR::ABL1^pos subgroup and ALL lacking identified genetic markers (NEG ALL). We identified three CNA profiles: high-risk bad-CNA (CNA^high/IKZF1^pos), low-risk good-CNA (all other CNAs), and intermediate-risk CNA^neg. Furthermore, based on RAG/AID expression, we report possible mechanisms underlying the CNA profiles associated with poor outcome: AID stratified outcome in CNA^neg, which accompanied most likely a particular profile of single nucleotide variations, while RAG in CNA^pos increased the odds for CNA^high/IKZF1^pos development. Finally, we integrated primary genetic aberrations with CNA to propose a revised risk stratification code, which allowed us to stratify 75% of BCR::ABL1^pos and NEG patients.

1. Introduction

B-cell acute lymphoblastic leukemia (B-ALL) is a biologically heterogeneous disease caused by specific genetic lesions leading to aberrant differentiation and proliferation of lymphoid progenitor cells [1]. Adult B-ALL is characterized by poor outcomes and a high rate of mortality when compared to children [2,3,4,5]. This is due to a particular genetic background and a higher rate of age-related comorbidities [6,7,8,9,10]. Although new therapies, as well as risk-adapted therapy protocols, led to a significant improvement in the outcome of adult ALL patients [11], the main reason for therapy failure in most cases is the emergence of new leukemic subclones leading to a relapse of the disease [12,13,14]. Thus, besides the standard risk group definition, the determination of the potential for leukemia’s evolution at the initial stage of the disease could be a key prognostication marker for those patients who would need further stratification.

A known hallmark of B-ALL development is the sequential acquisition of new genetic aberrations as a result of genetic instability and clonal evolution [13,15]. Chromosomal rearrangements and fusion genes, identified as primary aberrations, are the initiating events and key drivers of leukemogenesis, however, in most cases, they are insufficient for leukemia development. Secondary genetic events targeting B cell development genes are therefore an essential requirement for overt ALL [16,17]. These mutations, like copy number alterations (CNA; particularly gene deletions) and single-nucleotide variant (SNV) aberrations, have been frequently found within genes such as IKZF1, CDKN2A/B, and PAX5 and have been reported to cooperate with each other and with primary lesions, leading to a more aggressive phenotype. Importantly, except for rare genetic subtypes, e.g., MLL rearrangements, the progression of high-risk B-ALLs depends mostly on mutator mechanisms introducing new CNA/SNV-type aberrations [14,18]. Not surprisingly, disease relapse is associated with both clonal diversification and a higher level of CNA mutation burden [19,20,21].

Recently, two lymphocyte-specific mutator enzymes, RAG1/2 (recombination-activating gene) and AID (activation-induced cytidine deaminase), have been extensively studied in relation to genetic instability and oncogenicity in ALL. RAG1 and RAG2 genes encode recombinases, which introduce double-strand breaks into DNA during immunoglobulin/T-cell receptor genes’ rearrangement. However, aberrant targeting of RAG1/2 to non-IGH sites contributes to the development of driver mutations and the clonal evolution of ALL [22]. The AID gene encodes cytidine deaminase, which generates point mutations in immunoglobulin genes in a process known as somatic hypermutations in germinal B-cells. AID was documented to be a driver of oncogenic mutations in lymphomas [23,24]. However, recent studies have shown that aberrant activation of AID by infectious signals may also accelerate mutagenic processes leading to childhood ALL [13,25,26,27]. The above-presented mutational mechanisms have been well described in pediatric ALL, but the data are insufficient for adult ALL. Although recent studies suggest the involvement of RAG2-mediated aberrant recombinations in the evolution of adult B-ALL: t(9;22)/BCR::ABL1^pos and BCR::ABL-like [28,29,30,31]; however, no comprehensive analyses on the expression of both enzymes across different genetic subgroups, in correlation to CNA status and clinical outcome, have been carried out in adult ALL so far.

Previous studies documented that both the mutation burden corresponding to clonal heterogeneity, and the type of the mutated gene (IKZF1, CDKN2A/B, or PAX5) may provide prognostic information [3,4,5,32,33,34]. However, controversial results have also been reported, particularly in adult ALL [2,35,36]. Although the factors that contribute to these discrepancies are not fully understood, they may be related to the complex nature of interactions between coexisting primary and/or secondary aberrations, different inclusion criteria (e.g., age, genetic background), or treatment protocols between independent trials. Because of these ambiguities, the fifth edition of WHO, as well as the ICC classification did not include a broad range of CNAs, as a diagnostic criterion for disease entities [37,38]. Importantly, most of these controversies apply to adult ALL. Whereas in pediatric ALL, the clinical significance of CNA has been well recognized and incorporated, e.g., into the integrated risk scoring system widely used in clinical trials, with the particular role of IKZF1 deletions coexisting with other concomitant CNAs. This CNA profile was first described by Stanulla et al. as IKZF^plus and indicates a subgroup of patients with particularly poor outcomes (in our study, referred to as CNA^high/IKZF1^pos) [5,39,40]. As childhood ALL presents a cure rate of 85–90%, while most adult ALLs are poor-risk leukemias, it is important to understand better the possible role of mutational processes behind high-risk leukemias in adults.

Therefore, we carried out a detailed characterization of CNA profiles: CNA mutation burden, the type of mutated genes, and association with mutator enzyme AID/RAG1/2 expression—to establish their correlation with prognosis in an adult B-ALL population treated with a standard intensive protocol according to PALG. Based on the obtained results, we proposed a combined revision of the genetic risk classification integrating CNA data with well-established primary aberrations, which enabled further prognostic stratification of BCR::ABL1^pos and NEG ALL patients. Parallelly, a comprehensive evaluation of RAG1/2 and AID expression signatures in correlation with CNA profiles allowed for further characterization of mutational processes behind selected ALL subtypes associated with poor prognosis.

2. Materials and Methods

2.1. Patients’ Characteristics, Treatment, Material Collection, and Detection of Fusion Genes and Mutations by Molecular Analysis

A total of 166 patients with B-ALL were enrolled in the present study. The patients were diagnosed, and 161 patients met the criteria to be treated according to the Polish Adult Leukemia Group (PALG)-ALL5 or PALG-ALL6 protocols between 2007 and 2017 in six Polish hematology centers. The remaining 5 patients were disqualified from intensive therapy and received palliative treatment; their outcomes were not analyzed in this study. Allogenic hematopoietic stem cell transplantation (alloHSCT) from either an HLA-matched sibling or an unrelated donor was performed in 77/161 (47%) patients out of the whole group.

The entire population was analyzed for common fusion genes: BCR::ABL1, TEL::AML1, E2A::PBX1, and MLL::AF4, using the RT-PCR protocol and primers designed according to the consensus of the European BIOMED-1 Concerted Action [41]. Subsequently, patients negative for the fusion genes listed above (102 patients) were screened for markers or surrogates of the BCR::ABL1-like phenotype (see details in Supplementary Materials). In parallel, cytogenetic analysis was performed on diagnostic bone marrow samples by PALG laboratories as previously described [42].

2.2. Treatment Protocols

The PALG ALL6 protocol encompassed all patients with newly diagnosed ALL. Patients were categorized for one of the first-line treatment options based on the subtype of the disease (ALL with or without the Philadelphia chromosome/BCR::ABL1 rearrangement) and their age (either up to 55 years old or older). Younger patients adhered to a “pediatric” approach consisting of a pre-treatment phase, intensive induction, and consolidation therapy, followed by extended maintenance. Additionally, the protocols incorporated intensive and prolonged central nervous system prophylaxis. AlloHSCT was considered for patients with BCR::ABL1-positive disease and those with post-remission positive minimal residual disease (MRD) assessed by the multiparameter flow cytometry (MFC) method (MRD ≥ 0.1% after induction and/or >0.01% after consolidation). Compared to PALG ALL6, PALG ALL5 was a similar MRD-driven regimen, except for a less individualized approach to the treatment of older patients over 55 years old. Further information on the treatment protocols is provided in Supplementary Materials.

2.3. Gene Expression Analysis of Mutator Enzymes by Quantitative Real-Time PCR

Gene expression of RAG1, RAG2, and AID mutator enzymes was determined in the entire B-ALL cohort by real-time polymerase chain reaction (qPCR; see details in the Supplementary Methods). The whole group was segregated into 2 subgroups based on RAG1, RAG2, and AID expression levels at diagnosis, i.e., higher or lower than the median. As in our setting, there was a highly positive correlation between RAG1 and RAG2 enzymes, we used only the RAG2 mRNA expression level for further analyses (Spearman’s rank correlation coefficient rho = 0.6; p < 0.05) [43].

2.4. CNA Detection: Multiplex Ligation-Dependent Probe Amplification (MLPA) and RT-PCR

CNAs were evaluated using a multiplex ligation-dependent probe amplification (MLPA) assay. The SALSA MLPA P335-B2 ALL-IKZF1 (MRC-Holland, Amsterdam, The Netherlands) was used according to the manufacturer’s protocol in a group of 94 cases whose genetic material was available. The deletions in the loci of the following 7 genes were scored as deleted or non-deleted: IKZF1, CDKN2A/B, PAX5, RB1, EBF1, ETV6, and BTG1 (see details in Supplementary Materials). Additionally, a cohort of 163 patients was evaluated for expression of 2 isoforms of IKZF1 mRNA (Δ3–6, Δ1–7; primers for RT-PCR protocol described by Iacobucci I. et al. [44].

2.5. Statistical Analysis

Complete remission (CR) rate and probability of overall survival (OS), relapse-free survival (RFS), and disease-free survival (DFS) were the study endpoints. The CR rate was defined according to previously published criteria [4]. DFS was defined as the time from the achievement of CR to hematological relapse, death, or the last follow-up. OS was defined as the time from diagnosis to death or the last follow-up.

Statistical differences between groups were tested with non-parametric tests. The Mann-Whitney U test was used for comparing continuous variables, the chi-squared test or Fisher exact test for categorical variables, and Spearman’s correlation test for correlating continuous data. The patients’ survival rates (OS, RFS, DFS) were compared with the log-rank test. In multivariate analyses, the hazard ratio was computed from the Cox proportional hazards model for survival rates (OS, RFS, DFS) and from the general linear model for categorical variables (CR). All statistical analyses were performed using the Statistica 13.3 software (TIBCO Software Inc., Palo Alto, CA, USA).

3. Results

3.1. Patients’ Characteristics; Frequency and Clinical Correlates of Primary Chromosomal Abnormalities

In the present study, 166 adults with B-ALL were investigated for the presence of fusion genes using the diagnostic workup according to the BIOMED-1 protocol. Comparisons of molecular results with cytogenetic data allowed for the identification of genetic subgroups according to the established primary chromosomal aberrations. The median age of the studied population was 37.5 years (range 18–70 years). According to the international criteria, 55 patients were classified into the very high-risk group, 94 into the high-risk group, and 17 into the standard-risk group. The demographic and clinical characteristics of the patients’ cohort are summarized in

Table 1. Patients’ characteristics according to the primary aberrations, secondary aberrations and mutator enzyme expression. (A) Patients’ characteristics according to the primary aberration. (B) Patients’ characteristics according to RAG2 and AID expression. (C) Patients’ characteristics according to the number of CNAs and IKZF1 mutation status.

(A)
	Total		NEG		BCR::ABL1	“Bad Primary”						E2A::PBX
						BCR::ABL1-like	MLL::AF4		Complex	Hyperdiploid	All “Bad Primary” ‡
	n = 166		n = 63 (38%)		n = 55 (33%)	n = 16 (10%)	n = 5 (3%)		n = 15 (9%)	n = 6 (4%)	n = 44 (27%)	n = 4 (2%)
Age
≤37 years (%)	83 (50%)		33 (52%)		24 (44%)	6 (38%)	5 (100%)		8 (53%)	3 (50%)	23 (52%)	3 (75%)
>37 years (%)	83 (50%)		30 (48%)		31 (56%)	10 (62%)	0 (0%)		7 (47%)	3 (50%)	21 (48%)	1 (25%)
Gender
Female (%)	57 (34%)		15 (24%)		24 (44%)	7 (44%)	0 (0%)		5 (33%)	3 (50%)	16 (36%)	2 (50%)
Male (%)	109 (66%)		48 (76%)		31 (56%)	9 (56%)	5 (100%)		10 (67%)	3 (50%)	28 (64%)	2 (50%)
Immunological subtype
prepreB (%)	22 (13%)		11 (17%)		3 (5%)	1 (6%)	3 (60%)		3 (21%)	1 (17%)	8 (19%)	0 (0%)
preB (%)	42 (25%)		17 (27%)		13 (24%)	4 (25%)	2 (40%)		4 (29%)	1 (17%)	11 (26%)	1 (25%)
common (%)	101 (62%)		35 (56%)		39 (71%)	11 (69%)	0 (0%)		7 (50%)	4 (67%)	24 (56%)	3 (75%)
Median of WBC, ×109/L	17.0		13.5		26.0	28.7	127.9 #		9.2	4.6	10.9	24.3
Rate of CD20 positive, %	18.0		14.0		28.0	31.0	0.5		35.0	25.0	25.0	9.5
Rate of CD52 positive, %	18.0		18.0		22.0	43.0	8.7		12.0	5.5	14.7	62.0
Risk group according to PALG
very high risk (%)	55 (33%)		0 (0%)		55 (100%)	0 (0%)	0 (0%)		0 (0%)	0 (0%)	0 (0%)	0 (0%)
high risk (%)	94 (57%)		54 (86%)		0 (0%)	16 (100%)	5 (100%)		13 (87%)	3 (50%)	38 (86%)	2 (50%)
standard risk (%)	17 (10%)		9 (14%)		0 (0%)	0 (0%)	0 (0%)		2 (13%)	3 (50%)	6 (14%)	2 (50%)
MRD status after first induction
positive (%)	45 (38%)		13 (30%)		19 (53%)	5 (56%)	0 (0%)		6 (43%)	1 (20%)	13 (37%)	0 (0%)
negative (%)	73 (62%)		31 (70%)		17 (47%)	4 (44%)	5 (100%)		8 (57%)	4 (80%)	22 (63%)	3 (100%)
Induction regimen
PALG5 (%)	15 (9%)		4 (6%)		6 (11%)	0 (0%)	0 (0%)		2 (13%)	1 (17%)	3 (7%)	2 (50%)
PALG6 (%)	145 (88%)		58 (92%)		46 (84%)	15 (100%)	5 (100%)		13 (87%)	5 (83%)	40 (93%)	1 (25%)
PALG modified (%)	5 (3%)		1 (2%)		3 (5%)	0 (0%)	0 (0%)		0 (0%)	0 (0%)	0 (0%)	1 (25%)
AlloHSCT in first complete response
yes (%)	77 (47%)		27 (44%)		29 (53%)	7 (44%)	5 (100%)		6 (40%)	1 (17%)	19 (43%)	2 (50%)
no (%)	88 (53%)		35 (56%)		26 (47%)	9 (56%)	0 (0%)		9 (60%)	5 (83%)	25 (57%)	2 (50%)
(B)
	Total		RAG2 Expression						AID Expression
			Low			High	p		Low		High	p
	n = 166		n = 83			n = 78			n = 78		n = 79
Median age, years	37.5		40.0			37.0	0.4817 *		35.5		41.0	0.2802 *
Gender							0.5191 †					0.1440 †
Female (%)	57		28 (34%)			27 (35%)			31 (40%)		24 (30%)
Male (%)	109		55 (66%)			51 (65%)			47 (60%)		55 (70%)
Primary aberration
NEG	63		37 (62%)			23 (38%)	0.0478 a		22 (38%)		36 (62%)	0.0242 a
BCR::ABL1	55		21 (39%)			33 (61%)	0.0224 b		30 (57%)		23 (43%)	0.2156 b
BCR::ABL1-like	16		8 (53%)			7 (47%)			7 (50%)		7 (50%)
MLL::AF4	5		2 (40%)			3 (60%)			5 (100%)		0 (0%)
Complex	15		8 (53%)			7 (47%)			7 (47%)		8 (53%)
Hyperdiploid	8		5 (83%)			1 (17%)			3 (50%)		3 (50%)
E2A::PBX	4		1 (25%)			3 (75%)			3 (75%)		1 (25%)
Immunological subtype							0.4641 c					0.0420 c
prepreB (%)	22		13 (59%)			9 (41%)			14 (67%)		7 (33%)
preB (%)	42		18 (46%)			21(54%)			14 (36%)		25 (64%)
common (%)	101		51 (52%)			48 (48%)			50 (52%)		46 (48%)
Median of WBC, ×109/L	17.0		10.2			23.0	0.0559 *		21.0		12.9	0.0589 *
Rate of CD20 positive, %	18.0		22.0			20.0	0.5278 *		11.0		30.0	0.3524 *
Rate of CD52 positive, %	18.0		22.5			17.5	0.5950 *		22.5		17.0	0.1320 *
Risk group according to PALG							0.0681 †					0.1691 †
very high risk (%)	55		21 (39%)			33 (61%)			30 (57%)		23 (43%)
high risk (%)	94		52 (57%)			39 (43%)			38 (43%)		50 (57%)
standard risk (%)	17		10 (63%)			6 (37%)			10 (63%)		6 (37%)
MRD status after first induction							0.4792 †					0.1463 †
positive (%)	45		24 (55%)			20 (45%)			18 (43%)		24 (57%)
negative (%)	73		36 (52%)			33 (48%)			38 (55%)		31 (45%)
Induction regimen
PALG5 (%)	15		8 (57%)			6 (43%)			4 (29%)		10 (71%)
PALG6 (%)	145		73 (52%)			68 (48%)			71 (52%)		66 (48%)
PALG modified (%)	5		1 (20%)			4 (80%)			3 (60%)		2 (40%)
AlloHSCT in first complete response
yes (%)	77		36 (49%)			38(51%)			34 (47%)		38 (53%)
no (%)	88		47 (54%)			40 (46%)			44 (52%)		41 (48%)
(C)
	Total	CNA Presence			CNA Mutation Burden			IKZF1 Mutation Status			IKZF1pos Patients Only d
		CNAneg	CNApos	p	CNAlow (0–1 CNA)	CNAhigh (≥2 CNAs)	p	IKZFneg	IKZFpos e	p	1 CNA	CNAhigh	p
	n = 94	n = 28 (30%)	n = 66 (70%)		n = 53 (56%)	n = 41 (44%)		n = 53 (52%)	n = 49 (48%)		n = 12 (29%)	n = 29 (71%)
Median age, years	37.0	33.0	39.5	0.0227 *	36.0	39.0	0.1128 *	35.0	41.0	0.0152 *	41.0	45.0	0.6466 *
Gender				0.5383 †			0.3658 †			0.3789 †			0.5365 †
Female (%)	36	11 (31%)	25 (69%)		19 (53%)	17 (47%)		21 (58%)	17 (42%)		4 (27%)	11 (73%)
Male (%)	58	17 (29%)	41 (71%)		34 (59%)	24 (41%)		32 (50%)	32 (50%)		8 (31%)	18 (69%)
Primary aberration
NEG	35 (37%)	11 (31%)	24 (69%)	0.7887 a	20 (57%)	15 (43%)	0.9089 a	21 (60%)	14 (40%)	0.2402 a	4 (29%)	10 (71%)	0.9437 a
BCR::ABL1	30 (32%)	6 (20%)	24 (80%)	0.1554 b	14 (47%)	16 (53%)	0.1934 b	10 (27%)	27 (73%)	0.0001 b	6 (30%)	14 (70%)	0.9200 b
BCR::ABL1-like	12 (13%)	2 (17%)	10 (83%)		5 (42%)	7 (58%)		6 (50%)	6 (50%)		2 (33%)	4 (67%)
MLL::AF4	3 (3%)	2 (67%)	1 (33%)		3 (100%)	0 (0%)		3 (100%)	0 (0%)		0 (0%)	0 (0%)
Complex	6 (6%)	2 (33%)	4 (67%)		3 (50%)	3 (50%)		5 (71%)	2 (29%)		0 (0%)	1 (100%)
Hyperdiploid	6 (6%)	3 (50%)	3 (50%)		6 (100%)	0 (0%)		6 (100%)	0 (0%)		0 (0%)	0 (0%)
E2A::PBX	2 (2%)	2 (100%)	0 (0%)		2 (100%)	0 (0%)		2 (100%)	0 (0%)		0 (0%)	0 (0%)
Immunological subtype				0.2555 c			0.2765 c			0.8579 c			0.9534 c
prepreB (%)	11	5 (45%)	6 (55%)		7 (64%)	4 (46%)		9 (82%)	2 (18%)		0 (0%)	2 (100%)
preB (%)	26	10 (38%)	16 (62%)		17 (65%)	9 (35%)		16 (53%)	14 (47%)		3 (30%)	7 (70%)
common (%)	57	13 (23%)	44 (77%)		29 (51%)	28 (49%)		28 (46%)	33 (54%)		9 (31%)	20 (69%)
Median of WBC, ×109/L	15.5	13.7	17.9	0.8178 *	13.6	23.6	0.0939 *	13.8	19.0	0.3394 *	7.2	24.4	0.1188 *
Rate of CD20 positive, %	18.0	11.0	18.0	0.7008 *	10.0	20.0	0.2409 *	18.0	20.0	0.9968 *	6.2	19.0	0.1939 *
Rate of CD52 positive, %	22.5	15.0	26.2	0.2507 *	26.0	21.5	0.4730 *	22.0	18.0	0.6487 *	82.0	18.0	0.0502 *
Risk group acc. to PALG				0.0567 †			0.0561 †			0.0005 †			0.7786 †
very high risk (%)	30	6 (20%)	24 (80%)		14 (47%)	16 (53%)		10 (27%)	27 (73%)		6 (30%)	14 (70%)
high risk (%)	54	16 (30%)	38 (70%)		30 (56%)	24 (44%)		35 (64%)	20 (36%)		5 (26%)	14 (74%)
standard risk (%)	10	6 (60%)	4 (40%)		9 (90%)	1 (10%)		8 (80%)	2 (20%)		1 (50%)	1 (50%)
MRD after 1st induction				0.0486 †			0.0003 †			0.0019 †			0.0149 †
positive (%)	25	5 (20%)	20 (80%)		9 (36%)	16 (64%)		9 (31%)	20 (69%)		2 (13%)	14 (87%)
negative (%)	42	18 (43%)	24 (57%)		34 (81%)	8 (19%)		30 (68%)	14 (32%)		7 (58%)	5 (42%)
Induction regimen
PALG5 (%)	8	2 (25%)	6 (75%)		6 (75%)	2 (25%)		3 (38%)	5 (63%)		3 (60%)	2 (40%)
PALG6 (%)	83	25 (30%)	58 (70%)		46 (55%)	37 (45%)		49 (54%)	42 (46%)		9 (26%)	25 (74%)
PALG modified (%)	2	0 (0%)	2 (100%)		0 (0%)	2 (100%)		0 (0%)	2 (100%)		0 (0%)	2 (100%)
AlloHSCT in first CR
yes (%)	40	14 (35%)	26 (65%)		29 (73%)	11 (27%)		25 (56%)	20 (44%)		8 (53%)	7 (47%)
no (%)	54	14 (26%)	40 (74%)		24 (44%)	30 (56%)		28 (49%)	29 (51%)		4 (15%)	22 (85%)

^‡ Two hypodiploid cases are included in the “bad genetics” group but are not included in the preceding subgroups. ^† computed by chi-squared or Fisher exact test. * computed by Mann-Whitney U test. ^# WBC in MLL::AF4 significantly higher than in NEG, Complex, and Hyperdiploid patients (ANOVA p = 0.0038, Tukey HSD test p = 0.0163, p = 0.0470 and p = 0.0271, respectively. Other comparisons not significant). ^a computed by chi-squared test: NEG vs. all other cases. ^b computed by chi-squared test: BCR::ABL1 vs. all other cases. ^c computed by chi-squared test: preB vs. all other cases. ^d “MLPA group”: this series includes IKZF1^pos cases identified using MLPA (37) and RT-PCR (2) in a group of 94 patients, for whom other gene deletions were analyzed using MLPA; this series was used for evaluation of patients’ characteristic (see also Table S2). ^e “MLPA+RT-PCR group”: this series includes IKZF1^pos cases identified using MPLA (37) and RT-PCR (10); This series was used for survival analyses. Abbreviations: RAG2, Recombination Activating Gene 2; AID, Activation Induced Cytidine Deaminase; WBC, White Blood Cell count; PALG, Polish Adult Leukemia Group; CNA, copy number alterations.

.

Out of the total 166 B-ALL patients analyzed, 55 (33%) had BCR::ABL1, 5 (3%) had KMT2A::AFF1 (MLL::AF4), 4 (2%) had TCF3::PBX1 (E2A::PBX1), and none had ETV6::RUNX1 (TEL::AML1) fusion genes (Figure 1). One hundred-two patients (62%) were found negative for the common fusion genes analyzed according to the BIOMED-1 protocol. A subsequent analysis of these 102 patients identified 16 cases accompanied by the markers or surrogates of the BCR::ABL1-like phenotype (10% of the total population). All of the BCR::ABL1-like ALL represented JAK::STAT2–class aberrations: 8 patients with overexpression of CRLF2 gene, 7 with P2RY8::CRLF2 rearrangement, 1 with CRLF2 gene point mutation, 6 with JAK2 (see details in Table S1).

When we matched the remaining group of 86 patients, who had no identified molecular aberrations, with cytogenetic data, 23 patients could be defined as a subgroup with poor risk genetic characteristics: 15 with complex karyotype, 6 hyperdiploid, 2 hypodiploid, and 1 with another aberration of poor risk. Out of the remaining 63 cases, 42 presented normal karyotypes, 1 had other aberrations of unknown significance, and 20 lacked the cytogenetic analysis. As this latter subgroup represented a “real world” population of ALL patients for whom the initial risk stratification is unknown, for the sake of further analyses in this manuscript, we grouped these patients under the label of “NEG ALL” (38%).

3.2. Clinical and Biological Characteristics of RAG2 and AID Mutator Enzymes’ Expression—In Correlation with Primary Genetic Subgroups

In order to evaluate the potential for RAG2- and AID-mediated genetic instability, all patients were assigned into subgroups based on both enzymes’ mRNA expression levels that were higher or lower than the median (see Table 1B and Table S2). Thus, using the integrated AID/RAG2 profile, we could identify 4 signatures of differential AID and RAG2 expression (referred to here as “sig. 1–4): sig. 1 AID^low/RAG2^low was found in 18.6% of ALL patients, who showed low expression levels of both enzymes; sig. 2 AID^high/RAG2^low accompanied 32.7% of ALL population with high AID but low RAG2; sig. 3 AID^low/RAG2^high was found in 31.4% of ALL, that showed high RAG2 while little or no AID. Sig. 4 AID^high/RAG2^high represented 17.3% of ALL who had both enzymes in abundance, suggesting co-synergic involvement of AID with RAG2. In summary, the highest frequency of sig. 3 (AID^low/RAG2^high) was found in the BCR::ABL1^pos subgroup, while the lowest one was in the NEG B-ALL (43% vs. 22%, p = 0.005). In contrast, a higher incidence of AID^high/RAG2^low (sig. 2) was reported for the NEG B-ALL subgroup when compared to the BCR::ABL1^pos population (47% vs. 21.6%, p = 0.005). Detailed data on AID/RAG2 profiles and their distribution in correlation to primary genetic aberrations are presented in the Supplementary Results.

3.3. Clinical and Biological Characteristics of Secondary CNA Aberrations—In Correlation with Primary Genetic Subgroups

In the next step, we analyzed the characteristics of CNAs: CNA mutation burden and gene types in correlation with demographic data and B-ALL primary genetic aberrations in our study cohort (see Table 1C and Table S3). In summary, 1 or more deletions were observed in 66 out of 94 patients (70%), while 28 patients (30%) had no CNAs. The highest prevalence of CNAs was observed in the BCR::ABL1 and BCR::ABL1-like subgroups (80% and 83.3%), followed by an intermediate one in the NEG B-ALL (68.6%), and the lowest in MLL gene rearrangement (33%). The detailed data on secondary CNA characteristics and correlation with genetic subgroups are further presented in the Supplementary Results.

3.4. Correlation of Secondary CNA Aberrations with RAG2/AID Signatures

Furthermore, in order to establish a functional link, we correlated the integrated AID/RAG2 expression signatures with CNAs’ profiles (see

Table 2. Prevalence of markers of CNA-type genetic instability: mutational burden, gene type—in subgroups according to mutator enzyme expression.

Total B-ALL
	RAG2 Expression		AID Expression		p †
	Low n = 43	High n = 48	Low n = 54	High n = 36
CNA
CNAneg	18 (67%)	9 (33%)	15 (54%)	13 (46%)
CNApos	25 (39%)	39 (61%)	36 (61%)	23 (39%)	RAG2: CNAneg vs. CNApos: 0.0160
1 CNA	9 (39%)	14 (61%)	11 (50%)	11 (50%)
CNAhigh	16 (39%)	25 (61%)	25 (68%)	12 (32%)	RAG2: CNAneg vs. CNAhigh: 0.0257
IKZF1 deletion
IKZF1neg	31 (60%)	21 (40%)	24 (46%)	28 (54%)	RAG2: 0.0109
IKZF1pos	16 (34%)	31 (66%)	31 (72%)	12 (28%)	AID: 0.0108
CDKN2A/B deletion					RAG2: 0.0013
CDKN2A/Bneg	33 (56%)	26 (44%)	36 (63%)	21 (37%)
CDKN2A/Bpos	10 (31%)	22 (69%)	15 (50%)	15 (50%)
PAX5 deletion					NS
PAX5neg	36 (47%)	40 (53%)	41 (55%)	33 (45%)
PAX5pos	7 (47%)	8 (53%)	10 (77%)	3 (23%)
NEG B-ALL
	RAG2 Expression		AID Expression		p †
	Low n = 22	High n = 12	Low n = 12	High n = 21
CNA					RAG2: CNAneg vs. CNApos: 0.0464
CNAneg	9 (90%)	1 (10%)	3 (27%)	8 (73%)
CNApos	13 (54%)	11 (46%)	9 (41%)	13 (59%)
1 CNA	4 (44%)	5 (56%)	2 (25%)	6 (75%)
CNAhigh	9 (60%)	6 (40%)	7 (50%)	7 (50%)
IKZF1 deletion
IKZF1neg	16 (80%)	4 (20%)	3 (15%)	17 (85%)	RAG2: 0.0257
IKZF1pos	6 (43%)	8 (57%)	8 (67%)	4 (33%)	AID: 0.0029
BCR::ABL1pos B-ALL
	RAG2 Expression		AID Expression		p †
	Low n = 8	High n = 21	Low n = 22	High n = 6
CNA					NS
CNAneg	3 (50%)	3 (50%)	4 (67%)	2 (33%)
CNApos	5 (22%)	18 (78%)	18 (82%)	4 (18%)
1 CNA	1 (14%)	6 (86%)	5 (71%)	2 (29%)
CNAhigh	4 (25%)	12 (75%)	13 (87%)	2 (13%)
IKZF1 deletion					NS
IKZF1neg	4 (40%)	6 (60%)	6 (60%)	4 (40%)
IKZF1pos	7 (27%)	19 (73%)	20 (80%)	5 (20%)

^† computed by chi-squared or Fisher exact test. Abbreviations: RAG2, Recombination Activating Gene 2; AID, Activation Induced Cytidine Deaminase; CNA, Copy Number Alterations.

and Table S2). In summary, although we could confirm the presence of CNAs in each subgroup according to the integrated AID/RAG2 signature profile, we report that the abundance of RAG2 alone correlated rather with higher CNA levels and IKZF1 deletions, particularly in the BCR::ABL1^pos context (p = 0.001), while a lower CNA mutation number or CNA^neg was more frequently accompanied by AID^high/RAG2^low, particularly in NEG ALL context, suggesting involvement of secondary mutations other than CNA, e.g., SNV (see Table S2 for details). Significantly, sig. 4, i.e., with parallel AID and RAG2 abundance, was totally absent in CNA^neg patients, again emphasizing the link of both enzymes with CNA-type mutagenesis. Altogether, these results demonstrate a correlation of different CNA profiles with differential mutagenic properties linked to mutator enzymes or other mutagenic stimuli (AID^low/RAG2^low). The detailed description of correlation data between the integrated AID/RAG2 expression signatures with CNAs’ profiles is further provided in the Supplementary Results.

3.5. The Outcome of Intensively Treated B-ALL Patients in Relation to Established Primary Aberrations

In the next part of the study, we verified the clinical outcomes of primary and secondary aberrations in the adult ALL population treated according to PALG protocol. Among the 166 patients analyzed for primary aberrations using molecular testing, 161 were eligible for intensive induction therapy, and 152 out of 161 were evaluated for response to induction, while 9 patients died before the remission evaluation. In the whole cohort, complete remission (CR) was achieved in 131/152 (86%) of the study population, which is consistent with the previous observations [45]. The median follow-up of the study group was 37.9 months and the median survival reached 20.5 months, which is comparable to results in other studies of similar populations. The estimated 4-year overall survival (OS) for the whole population was 34%, with a standard error of ±4%.

The impact of primary chromosomal aberrations on survival is shown in

Table 3. Patients’ outcomes according to the primary aberration. (A) Univariate analysis of patients’ outcomes according to the primary aberration. (B) Multivariate analysis of patients’ outcomes according to the primary aberrations and demographic data.

(A)
End Point and Variables	Total n = 161	NEG		BCR::ABL1 n = 54	“Bad Primary”					E2A::PBX n = 3	p
		Total NEG * n = 61	NK Only n = 41		BCR::ABL1-like n = 15	MLL::AF4 n = 5	Complex n = 15	Hyperdiploid n = 6	All “Bad Primary” ‡ n = 43
CR											NEG vs. BCR::ABL1: 0.5454 †
No. of patients	131/152	49/57	34/41	47/54	10/12	5/5	11/14	4/5	32/38	3/3	NEG vs. BCR::ABL1-like: 0.5555 †
(%)	(86%)	(86%)	(83%)	(87%)	(83%)	(100%)	(79%)	(80%)	(78%)	(100%)	NEG vs. “Bad”: 0.5177 †
OS											NEG vs. BCR::ABL1:0.0895 #
No. of patients	161	61	41	54	15	5	15	6	43	3	NEG vs. BCR::ABL1-like: 0.1080 #
4-year rate ± SE	34 ± 4%	32 ± 8%	25 ± 8%	54 ± 8%	10 ± 9%	53 ± 25%	20 ± 10%	0%	16 ± 7%	100%	NEG vs. “Bad”: 0.0192 #
RFS											NEG vs. BCR::ABL1: 0.0356 #
No. of patients	129	49	34	46	10	5	10	6	31	3	NEG vs. BCR::ABL1-like: 0.3400 #
4-year rate ± SE	42 ± 6%	28 ± 9%	19 ± 9%	66 ± 8%	18 ± 16%	20 ± 18%	24 ± 15%	60 ± 22%	22 ± 10%	100%	NEG vs. “Bad”: 0.0365 #
DFS											NEG vs. BCR::ABL1:0.0978 # NEG vs. BCR::ABL1-like: 0.5055 # NEG vs. “Bad”: 0.0319 #
No. of patients	126	46	32	46	10	5	11	5	31	3
4-year rate ± SE	37 ± 5%	25 ± 8%	18 ± 9%	58 ± 8%	16 ± 15%	20 ± 18%	22 ± 13%	40 ± 22%	17 ± 8%	100%
(B)
				Primary Aberration
End Point and Variables	Age	WBC		NEG n = 61	BCR::ABL1 n = 54	All “Bad Primary” n = 43	BCR::ABL1-like n = 15	MLL::AF4 n = 5	Complex n = 15	Hyperdiploid n = 6	E2A::PBX n = 3
CR
OR (95% CI)	0.83 (0.70–0.98)	1.08 (0.91–1.29)		1.00	0.99 (0.71–1.39)	0.90 (0.66–1.21)	0.92 (0.64–1.31)	1.07 (0.67–1.71)	0.86 (0.61–1.21)	0.86 (0.55–1.35)	1.11 (0.65–1.92)
p *	0.0294	0.3593		(reference)	0.9726	0.4782	0.6280	0.7738	0.3726	0.5156	0.6936
OS
HR (95% CI)	1.04 (1.03–1.06)	1.00 (1.00–1.00)		1.00	0.55 (0.32–0.96)	1.67 (1.03–2.73)	1.39 (0.70–2.75)	1.10 (0.24–4.99)	1.62 (0.83–3.14)	3.91 (1.49–10.31)	2.62 (0.61–11.18)
p #	<0.0001	0.8974		(reference)	0.0338	0.0261	0.3434	0.8984	0.1549	0.0058	0.1929
RFS											-
HR (95% CI)	1.05 (1.03–1.07)	1.00 (1.00–1.01)		1.00	0.34 (0.16–0.72)	1.81 (0.98–3.37)	1.21 (0.49–3.02)	4.40 (1.37–14.13)	1.72 (0.71–4.14)	3.61 (0.81–16.18)
p #	<0.0001	0.0929		(reference)	0.0049	0.0597	0.6782	0.0129	0.2259	0.0934

(A) * NEG includes patients negative for all molecular markers with NK, or other aberrations of unknown significance, or without cytogenetic data. ^‡ Two hypodiploid cases are included in all the “bad genetics” groups but are not included in the preceding subgroups. ^† computed by Fisher’s exact test. ^# computed by the log-rank test. Abbreviations: NK, normal karyotype; CR, complete remission; OS, overall survival; RFS, relapse-free survival; DFS, disease-free survival; SE, standard error. (B) * computed from General Regression Model. ^# computed by Cox proportional hazard regression. Abbreviations: WBC, white blood count; CR, complete remission; OS, overall survival; RFS, relapse-free survival; OR, odds ratio; HR, hazard ratio; 95% CI, 95% confidence interval.

and Figure 2A,B. Our data confirmed the poor 4-year outcome for patients harboring MLL rearrangements (OS 53 ± 25%), complex karyotype (OS 20 ± 10%), hyperdiploid karyotype (OS 0%), and in the BCR::ABL1-like subgroup (OS 10 ± 9%). As the last subgroup represented JAK::STAT pathway abnormalities, the poor prognosis of these patients has already been described in other cohorts [46]. All the cases with E2A::PBX exhibited a good outcome. Interestingly, in our study cohort, a relatively good outcome was also documented for the BCR::ABL1^pos subgroup (54 ± 8%). The remaining patients, for whom neither molecular nor cytogenetic abnormalities with established prognostic importance were detected, thus indicating obscure prognosis, were referred to here as NEG ALL and showed an intermediate risk (OS 32 ± 8%; Table 3).

In multivariate analysis, the BCR::ABL1 fusion gene independently predicted a lower risk of death, and hyperdiploid karyotype was independently associated with an increased death rate (OS for BCR::ABL1: p = 0.034 with HR: 0.55, 95% CI, 0.32–0.96; for hyperdiploid: p = 0.006 with HR: 3.91, 95% CI, 1.49–10.31). BCR::ABL1 fusion independently predicted a lower incidence of relapse, while MLL::AF4 independently associated with a higher risk of relapse (RFS for BCR::ABL1: p = 0.005 with HR: 0.34, 95% CI, 0.16–0.72; for MLL rearrangements: p = 0.013 with HR: 4.40, 95% CI, 1.37–14.13) after adjustment for WBC, age, and other primary aberrations in the total ALL series (Table 3B).

Subsequently, based on the obtained data, we have assigned B-ALL patients to four categories: (1) “bad” primary aberrations, which grouped patients with all markers of poor prognosis (i.e., MLL rearrangements, BCR::ABL-like aberrations, complex, hyperdiploid, and hypodiploid karyotype), (2) “good” ones (E2A::PBX^pos), as well as “intermediate” ones (IM), which included 2 subgroups: (3) BCR::ABL1^pos and (4) NEG ALL (Figure 2C,D; Table 3A). As the prognoses of the first and second categories of aberrations are well established (either “bad” or “good”), the prognoses of the 2 remaining IM subgroups: BCR::ABL1^pos and NEG ALL needed further clarification. Thus, we asked the question of whether CNA markers may help to identify cases with a high and low risk of disease progression. As BCR::ABL1^pos and NEG ALL genetic subgroups may constitute up to 70% of adult ALL, it was important for us to find relatively easily accessible prognostic markers like CNA for more relevant therapeutic decisions. In parallel, we evaluated the prognostic impact of CNA on the total ALL population as a historical group to compare with the results obtained by the others.

3.6. Prognostic Relevance of Secondary CNA Mutations in BCR::ABL1^pos and NEG ALL

When we compared survival between CNA^pos and CNA^neg patients, we did not observe any significant differences (

Table 4. Patients’ outcomes according to the secondary aberrations. (A) Univariate analysis of patients’ outcomes according to the secondary aberrations. (B) Multivariate analysis of patients’ outcomes according to the secondary aberrations, adjusted for age and WBC.

(A)
End Point and Variables	Total	CNA Presence				CNA Mutation Burden				IKZF1 Mutation Status				Bad CNA (CNAhigh/IKZFpos)				Good CNA (CNApos Other than Bad CNA)
		CNAneg	CNApos	p		CNAlow	CNAhigh	p		IKZFneg	IKZFpos	p		CNAhigh IKZFpos	All Other Patients	p		Good CNA	All Other Patients	p
TOTAL B-ALL
CR				0.5283 †				0.0422 †				0.3746 †				0.0514 †				0.0880 †
No. of patients	73/88	22/26	51/62			45/50	28/38			41/48	38/47			20/28	53/60			31/34	42/54
(%)	83%	85%	82%			90%	74%			85%	85%			71%	88%			91%	78%
OS				0.4921 #				0.0179 #				0.2734 #				0.0045 #				0.0427 #
No. of patients	92	27	65			52	40			52	47			28	64			37	55
4-year rate ± SE	36 ± 6%	43 ± 10%	28 ± 6%			42 ± 8%	20 ± 7%			41 ± 8%	26 ± 7%			9 ± 6%	42 ± 7%			43 ± 9%	25 ± 6%
RFS				0.1460 #				0.0412 #				0.0320 #				0.0035 #				0.2096 #
No. of patients	73	22	49			44	27			40	37			19	52			30	41
4-year rate ± SE	41 ± 7%	43 ± 10%	34 ± 8%			53 ± 9%	20 ± 10%			48 ± 10%	31 ± 9%			14 ± 9%	49 ± 8%			45 ± 10%	40 ± 9%
NEG B-ALL
CR				0.6317 †				0.1710 †				0.5419 †				0.3105 †				0.4581 †
No. of patients	28/32	9/10	19/22			18/19	10/13			17/19	11/13			7/9	21/23			12/13	16/19
(%)	88%	90%	86%			95%	77%			89%	85%			78%	91%			92%	84%
OS				0.2117 #				0.0016 #				0.0055 #				0.0051 #				0.1513 #
No. of patients	34	10	24			19	15			20	14			10	24			14	20
4-year rate ± SE	28 ± 9%	53 ± 17%	22 ± 9%			45 ± 15%	7 ± 6%			46 ± 13%	7 ± 7%			0%	40 ± 12%			28 ± 13%	25 ± 10%
RFS				0.3322 #				0.0171 #				0.0118 #				0.0037 #				0.1042 #
No. of patients	28	9	19			18	10			17	11			7	21			12	16
4-year rate ± SE	25 ± 10%	42 ± 20%	19 ± 11%			43 ± 14%	0%			36 ± 15%	10 ± 10%			0%	34 ± 14%			31 ± 17%	22 ± 12%
BCR::ABL1pos B-ALL
CR				0.3438 †				0.3954 †				0.5441 †				0.2613 †				0.0653 †
No. of patients	24/30	4/6	20/24			12/14	12/16			8/10	22/26			10/14	14/16			10/10	14/20
(%)	80%	67%	83%			86%	75%			80%	85%			71%	88%			100%	70%
OS				0.3019 #				0.2129 #				0.6468 #				0.0406 #				0.0025 #
No. of patients	29	6	23			14	15			10	25			13	16			10	19
4-year rate ± SE	47 ± 10%	33 ± 19%	51 ± 11%			62 ± 13%	33 ± 13%			60 ± 15%	45 ± 11%			20 ± 12%	67 ± 12%			89 ± 10%	25 ± 11%
RFS				0.8073 #				0.1157 #				0.0875 #				0.0231 #				0.0404 #
No. of patients	23	4	19			12	11			8	21			9	14			10	13
4-year rate ± SE	65 ± 11%	75 ± 22%	63 ± 12%			82 ± 12%	44 ± 17%			88 ± 12%	48 ± 12%			29 ± 17%	85 ± 10%			89 ± 10%	44 ± 15%
(B)
End Point and Variables				CNAhigh					IKZFpos				Bad CNA (CNAhigh/IKZFpos)				Good CNA (CNApos Other than Bad CNA)
TOTAL B-ALL
CR
OR (95% CI)				0.83 (0.67–1.02)					0.97 (0.78–1.21)				0.83 (0.67–1.04)				1.20 (0.97–1.48)
p *				0.0791					0.8156				0.1054				0.0952
OS
HR (95% CI)				1.81 (1.04–3.17)					0.91 (0.51–1.61)				1.92 (1.06–3.48)				0.56 (0.31–1.01)
p ‡				0.0374					0.7393				0.0318				0.0521
RFS
HR (95% CI)				1.97 (0.99–3.92)					1.48 (0.74–2.97)				2.89 (1.37–6.10)				0.57 (0.28–1.16)
p ‡				0.0538					0.2695				0.0054				0.1189
NEG B-ALL
CR
OR (95% CI)				0.68 (0.47–0.98)					0.87 (0.58–1.31)				0.74 (0.50–1.10)				1.15 (0.77–1.71)
p *				0.0397					0.4921				0.1305				0.4897
OS
HR (95% CI)				4.95 (1.70–14.39)					4.12 (1.37–12.33)				4.85 (1.56–15.09)				0.42 (0.14–1.20)
p ‡				0.0033					0.0114				0.0065				0.1050
RFS
HR (95% CI)				4.11 (1.30–12.99)					4.59 (1.32–15.97)				9.80 (2.32–41.35)				0.32 (0.09–1.08)
p ‡				0.0160					0.0165				0.0019				0.0660
BCR:ABL1pos B-ALL
CR
OR (95% CI)				0.97 (0.68–1.40)					1.16 (0.84–1.62)				0.94 (0.65–1.38)				1.28 (0.88–1.87)
p *				0.8860					0.3595				0.7485				0.1871
OS
HR (95% CI)				1.84 (0.60–5.63)					0.88 (0.27–2.91)				2.36 (0.76–7.27)				0.11 (0.01–0.90)
p ‡				0.2860					0.8326				0.1359				0.0399
RFS
HR (95% CI)				6.29 (0.89–44.47)					3.52 (0.40–31.35)				7.65 (1.21–48.31)				0.14 (0.02–1.22)
p ‡				0.0654					0.2564				0.0304				0.0756

^† computed by chi-squared or Fisher’s exact test. ^# computed by log-rank test. * computed from General Linear Model. ^‡ computed by Cox proportional hazard regression. Abbreviations: WBC, white blood count; CR, complete remission; OS, overall survival; RFS, relapse-free survival; SE, standard error; CNA, copy number alteration; HR, hazard ratio; OR, odds ratio; 95% CI, 95% confidence interval.

A). As the CNA^pos subgroup was heterogeneous with respect to the mutation burden and the gene type, we stratified patients according to the number of CNAs and compared their survival. After several intermediate analyses, which are presented in detail in the Supplementary Results section, we found rationale to stratify patients according to genetic instability level: the first subgroup was labeled as “CNA^low” and included both CNA^neg and 1CNA patients, while the second subgroup was described as “CNA^high” and included patients who harbored ≥2CNAs (see Table 1C for details). Comparing these two groups, we observed significant differences in survival in the NEG cohort as well as in the whole series: patients with CNA^high had lower OS and RFS rates than CNA^low patients (see Figure 3, Table 4A and Supplementary Results for details). Only a trend towards significance was observed in the BCR::ABL1^pos subgroup. The data suggested that a high mutation burden together with the cooperation of concomitant gene mutations may play a predominant role in prognosis rather than a single CNA.

Thus, in the next step, we wanted to verify if any of the particular genes had a higher impact on survival than the others. We started with IKZF1, as it was the most frequent deletion in our study cohort, and found that its presence significantly correlated with the inferior outcomes of patients from NEG B-ALL. A trend toward statistical significance was observed for the BCR::ABL1^pos subgroup (see Figure 4 and Table 4A). For a detailed description of the obtained data, see Supplementary Results. In contrast, gene deletions like CDKN2A/B and PAX5 had no prognostic impact in subgroup analysis or the entire series.

Since the total IKZF1^pos population remained heterogeneous regarding the CNA burden, we assessed the difference between IKZF1 deletion alone and IKZF1 mutation coexisting with other CNAs (1CNA/IKZF1^pos vs. CNA^high/IKZF1^pos). Of note, Stanulla et al. were the first to describe the IKZF1^pos profile that co-occurred with deletions in CDNK2A/B, PAX5, or PAR1 in the absence of ERG deletion, referring to it as the IKZF^plus subset, which defined poor prognostic subgroups among pediatric B-ALL [39]. In turn, in our study, the CNA^high/IKZF1^pos subgroup included all IKZF1^pos patients accompanied by other gene deletions. As expected, the CNA^high/IKZF1^pos patients showed worse survival when compared to the 1CNA/IKZF1^pos series. This difference was statistically significant in the BCR::ABL1^pos cohort and the entire series, while only a trend towards significance was found in the NEG subpopulation, probably due to a small number of 1CNA/IKZF1^pos patients. This suggests that the negative effect of IKZF1 deletions within the total IKZF1^pos population was found exclusively in patients with CNA^high/IKZF1^pos profile (see Figure S1, Table S4 for further data). A detailed description of these results is provided in the Supplementary Results.

Then, we attempted to verify if IKZF1 had an adverse impact on other genes in the CNA^high population. Interestingly, when we compared the outcome of CNA^high/IKZF1^pos vs. CNA^high/IKZF1^neg series, the presence of IKZF1 deletions conferred lower survival. Despite a relatively small number of cases in the subgroup analysis, these differences were statistically significant for NEG ALL, BCR::ABL1^pos cohort, and whole ALL series. These data may suggest again that the negative outcomes observed in the total CNA^high population may be mostly attributed to patients with CNA^high/IKZF1^pos (see Table S4 and Figure S2 for further results). A detailed description of these data is provided in the Supplementary Results section.

Furthermore, as some studies reported [47,48] on the negative prognostic impact of CDKN2A/B deletions when accompanied by other deletions, particularly in BCR::ABL1^pos context, we compared the impact of IKZF1 deletions in the CNA^high/CDNK2A/B^pos cohort. Interestingly, CDNK2A/B^pos patients accompanied by IKZF1 deletion showed worse outcomes when compared to the subgroup without IKZF1. Despite the relatively small number of patients, this effect was particularly evident in the BCR::ABL1^pos series (p = 0.01), suggesting that the coexistence of a CDNK2A/B gene deletion alone may not affect the outcome of the CNA^high population, but rather through cooperation with IKZF1 deletions (Figure S3). A detailed description of these data is presented in the Supplementary Results section.

In summary, these data document that CNA^high/IKZF1^pos status is the predominant CNA profile with a clear negative impact on the treatment outcome in our series. As the dominant negative Ik6 isoform of IKZF1 deletion was found in other study cohorts to be associated with the worst outcome, we compared the Ik6 prognostic power within the CNA^high/IKZF1^pos population [4,33]. Despite a small group of patients, we found a negative impact of Ik6 on relapse rate among the poor-risk CNA^high/IKZF1^pos subgroup.

Finally, we noticed that the prognostic impact of CNA^high/IKZF1^pos varied in different age groups; a statistically significant difference in outcomes was obtained for younger patients (<40 y.), while no difference was observed for the older population (Figure S6). This suggests age-related comorbidities confounding the power of genetic markers in elderly patients, which can contribute to discrepancies in the prognostic impact of CNAs between clinical trials.

Thus, based on the above CNA data, we proposed a revised coding for risk stratification of adult B-ALL by CNA profile, which comprised 3 key subgroups, accompanied by different clinical impact: (1) “bad”-CNAs comprising CNA^high/IKZF1^pos cases (OS 9 ± 6%; RFS 14 ± 9%); (2) relatively „good”-CNAs composed of 1CNA/IKZF1^pos or 1-4CNA/IKZF1^neg cases (OS 43 ± 9%; RFS 45 ± 9%); (3) the remaining CNA^neg ALL with unresolved prognosis–labeled as “intermediate”-CNAs (OS 43 ± 10%; RFS 59 ± 14%; p = 0.003 for bad-CNA vs. good-CNA; see Figure 5, Table 4A).

The revised CNA-risk index was valid for the whole series, but most importantly it stratified the prognosis of the two intermediate subgroups: 29% of NEG and 44% of BCR::ABL1^pos ALL were accompanied by bad-CNA, while 41% of NEG and 34% of BCR::ABL1^pos were associated with good-CNA profile (Figure 5C–F; Table 4A). Though the bad-CNA profile had equally poor prognosis both in the NEG and BCR::ABL1^pos subgroup, the good-CNA profile was associated with much better survival in the BCR::ABL1^pos patients (OS 89 ± 10%; RFS 89 ± 10%) when compared to the NEG ALL (n = 14, OS 28 ± 13%; RFS 31 ± 17%). It is possible that at least some good-CNA^pos cases from NEG ALL were in fact BCR::ABL-like patients, which are frequently accompanied by CNAs with poor prognosis by themselves.

Finally, the results of the multivariate Cox regression analysis were consistent with the results of the univariate analysis both in the NEG and BCR::ABL subgroups. CNA^high mutation burden, IKZF1 deletion as well as the bad-CNA^high/IKZF1^pos status were independent prognostic factors associated with worse OS and RFS in the NEG subgroup (OS p = 0.003 with HR: 4.95, 95% CI, 1.70–14.39 for CNA^high; p = 0.011 with HR: 4.12, 95% CI, 1.37–12.33 for IKZF1^pos; p = 0.007 with HR: 4.85, 95%CI, 1.56–15.09 for CNA^high/IKZF1^pos; Table 4B). In the BCR::ABL1^pos series, CNA^high/IKZF1^pos had an independent prognostic impact on an increased risk of relapse (p = 0.030 with HR:7.65, 95% CI, 1.21–48.31), while good-CNAs, which lacked CNA^high/IKZF1^pos, independently predicted a lower risk of death (p = 0.04 with HR: 0.11, 95% CI, 0.01–0.90; Table 4B).

3.7. Combined Genetic Risk Classification by Revised Coding of Primary and Secondary Aberrations in Adult B-ALL

Therefore, taking into account the data shown above, we could propose a novel combined risk-adapted classification for adult B-ALL: by incorporating the stratification of NEG and BCR::ABL1^pos subtypes by secondary CNA into the existing genetic classification based on primary aberrations with well-established prognostic effect (Figure 2C,D and Figure 5C–F) we could further refine prognosis of the entire adult ALL population. The entire cohort was divided into 3 subgroups, which were created by co-segregating different categories with comparable prognostic impact: intermediate-, low-, and high-risk features. The intermediate one (1) included intermediate-CNA^neg cases from both NEG and BCR::ABL1^pos subgroups as well as good-CNA cases from NEG ALL (1CNA/IKZF1^pos or 1-4CNA/IKZF1^pos) identifying 33% adult ALL from our series; (2) the low-risk subgroup comprised E2A::PBX^pos cases and good-CNAs from BCR::ABL1^pos series (1CNA/IKZF1^pos and 1-4CNA/IKZF1^neg) and contained 14% patients from our cohort; (3) while the 3rd one included both bad primary aberrations (MLL::AF4, BCR::ABL-like, complex, hyper/hypodiploid karyotype) and bad-CNAs (CNA^high/IKZF1^pos) from NEG or BCR::ABL1^pos ALL, and was associated with the worst outcome in 53% of total adult ALL (see Figure 6).

These data indicate that more than half of adult ALL patients demonstrated poor outcomes, which has been revised compared to the frequency of 27% bad-risk ALL identified using only the well-established primary aberrations. The univariate and multivariate analyses for the combined revised risk stratification of adult ALL treated with PALG protocol are presented in

Table 5. Outcome of B-ALL patients according to the revised code for risk stratification.

End Point and Variables	Intermediate Risk	Low Risk	High Risk	Total
CR
No. of patients	24/28	13/13	44/55	81/96
(%)	86%	100%	80%	85%
p *	(reference)	0.1514	0.5224
OS
No. of patients	29	13	58	100
4-year rate ± SE	41 ± 10%	75 ± 13%	13 ± 5%	30 ± 5%
p ‡	(reference)	0.0148	0.0011
RFS
No. of patients	26	13	42	81
4-year rate ± SE	39 ± 12%	90 ± 9%	23 ± 7%	38 ± 7%
p ‡	(reference)	0.0254	0.0037
DFS
No. of patients	24	13	43	80
4-year rate ± SE	36 ± 11%	90 ± 9%	17 ± 6%	34 ± 6%
p ‡	(reference)	0.0183	0.0010
Multivariate analysis adjusted for age and WBC
CR
OR (95% CI)	1.00	0.92 (0.73–1.16)	1.12 (0.89–1.40)
p †	(reference)	0.3259	0.4713
OS
HR (95% CI)	1.00	0.13 (0.02–0.99)	2.12 (1.14–3.93)
p #	(reference)	0.0484	0.0176
RFS
HR (95% CI)	1.00	0.16 (0.02–1.23)	2.59 (1.25–5.33)
p #	(reference)	0.0787	0.0100

* computed from the chi-squared test. ^‡ computed from log-rank test. ^† computed from a General Linear Model. ^# computed by Cox proportional hazard regression. All p-values are calculated as a comparison to the intermediate risk group. Abbreviations: CR, complete remission; OS, overall survival; RFS, relapse-free survival; DFS, disease-free survival; SE, standard error; HR, hazard ratio; OR, odds ratio; 95% CI, 95% confidence interval. Intermediate risk: BCR::ABL1^pos: CNA^neg; NEG: CNA^neg or 1 CNA/IKZF1^pos or CNA^pos/IKZF1^neg. Low risk: E2A::PBX^pos; BCR::ABL1^pos: 1 CNA/IKZF1^pos or CNA^pos/IKZF1^neg. High risk: MLL::AF4^pos; Complex; Hyperdiploid; Hypodiploid; BCR::ABL1-like; CNA^high/IKZF1^pos.

.

Finally, Figure 7 summarizes the flowchart of revised risk classification based on both primary and secondary aberrations, leading to a hierarchical definition of genetic subcategories. The identified CNA-based risk subgroups can be used to refine patient treatment according to a more detailed description of their leukemia.

3.8. Prognostic Relevance of RAG2 and AID Expression in CNA^neg and CNA^pos B-ALL Population

Interestingly, the three identified CNA profiles were associated not only with different outcomes but also with differential AID/RAG2 signatures, suggesting a possible link between disease progression and the causal mutagenic mechanisms. Importantly, markers of CNA-based mutagenesis could not stratify CNA^neg cases, and these patients still constituted 31% and 20% of cases from the NEG and BCR::ABL1^pos subtypes, respectively. Thus, we tested whether the AID/RAG2 signature was associated with disease progression in CNA^neg patients. Unfortunately, in this study, we did not have a chance to investigate SNVs or larger genomic lesions, which probably could drive the leukemogenesis of this subgroup. However, as SNV mutagenesis is related to AID activity, we speculated that a high AID expression may correlate with disease progression in CNA^neg cases. Indeed, despite a small number of cases, we report here a positive correlation between the AID^high/RAG2^low expression signature and the incidence of adverse events in the CNA^neg subtype; patients who relapsed in 71% (5/7) were accompanied by high AID expression when compared to 20% (3/15) of those who remained in CR in the entire B-ALL series (p = 0.03; Table S5A). Such correlation was reversed for the CNA^pos series: only 20% (5/25) of AID^high/RAG2^low signature was found among patients who eventually relapsed (p = 0.09). Interestingly, opposite data were reported for RAG2 abundance in the CNA^neg series: a negative correlation with relapses (0/7) while a positive one with remission status (8/14; p = 0.01).

These data prompted us to compare survival in the CNA^neg subgroup according to both AID and RAG2 expression levels. Interestingly, high AID and low RAG2 expressors were associated with low RFS and OS in this subgroup (Figure S4, Table S5B). In multivariate analysis, high AID and low RAG2 independently predicted a higher incidence of relapse in the CNA^neg subset (Table S5C). These data, although based on a limited series of patients, allowed us to speculate that genomic markers other than CNA, which associate with AID^high/RAG2^low signature may contribute to the outcome of CNA^neg patients. Thus, although we did not directly analyze the genetic background of CNA^neg ALL, we can assume that CNA^neg cases with AID^high/RAG2^low signature, along with CNA^high/IKZF1^pos cases may also represent a profile of “bad” secondary mutations. This again may help in the selection of new poor- and low-risk subgroups among CNA^neg patients, that can be used to refine patients’ treatment (about 17% of total ALL; see Figure 7).

Finally, in the same manner, we decided to verify whether both enzymes correlate with clinical outcomes in the CNA^pos series, although an inverse pattern of correlation was assumed. This time, due to larger subgroups, the analysis could be performed separately for the NEG ALL and BCR::ABL1^pos series. Indeed, in the CNA^pos subgroup, AID^high was found to correlate with better survival in NEG ALL, while AID^low/RAG2^high profile was associated with poor outcomes in BCR::ABL1^pos (Figure S5). Importantly, both AID and RAG2 did not impact the outcome when evaluated independently from CNA status in the whole series, as well as in genetic subgroups.

4. Discussion

In this study, we reassessed the value of primary and CNA-type secondary aberrations in adult ALL treated with standard remission induction protocol according to PALG in relation to mutator enzyme AID/RAG1/2 signatures. Our principal aim was to identify risk-defined subgroups and mutational processes underlying adult leukemias with poor outcomes.

4.1. Revised Risk Index Based on Primary Aberrations Characterized the Prognosis of 30% of ALL Patients; Rationale for Further Stratification of NEG and BCR::ABL1^pos Subgroups

We have confirmed that established primary aberrations allowed 30% of adult ALL to be assigned to well-defined risk subgroups, either to the bad prognosis subgroup (28%; KMT2A::AFF1 (MLL::AF4), BCR::ABL1-like phenotype, complex karyotype, and hyper-/hypodiploid karyotype); or the good prognosis subgroup with TCF3::PBX1; E2A::PBX1 (2%). The remaining patients with BCR::ABL1 or lacking identified genetic markers (NEG ALL) represented an intermediate outcome in need of further stratification.

We proved a relatively good prognosis of BCR::ABL1^pos patients [49,50,51,52,53], however, we questioned whether all BCR::ABL1^pos patients may respond well to therapy. NEG ALL, which was negative for all investigated fusion genes and BCR::ABL1-like surrogate markers, was composed mainly of cases with normal karyotype (70%), or lacking cytogenetic analysis (30%). This subgroup thus represented a “real world” population of ALL patients with unknown prognoses. Importantly, although these two intermediate prognosis subgroups: NEG and BCR::ABL1^pos ALL covered 70% of adult ALL, their treatment outcome could not be further improved just by using conventional markers for initial risk stratification. Classic prognostic factors like age, WBC, fusion genes, and MRD are used for risk groups’ assignment [54]. However, routine practice shows that recurrent disease is not restricted only to the high-risk group. Additionally, the application of high-resolution techniques revealed that the genomic landscape particularly of NEG ALL is complex and heterogeneous [1,7,35,55]. Yet, the full spectrum of diagnostic workup according to the recent WHO/ICC classification is not always available, with many single aberrations lacking proven prognostic significance [37,38]. In relation to poor treatment outcomes of adult ALL, all these facts present a challenge for genetic classification based on risk.

Therefore, we tested our ALL cohort for hidden genomic lesions: secondary copy number alterations (CNA), which cooperate with primary aberrations, and thus may impact the prognosis, and the choice of risk-adapted therapy. Their incidence reaches 60–70% across both childhood and adult ALL populations, representing an easily accessible marker for risk stratification [16,33]. However, as some results from different clinical trials were contradictory [2,3,4,36], we aimed to verify their clinical impact for intensively treated ALL patients according to PALG. As RAG and AID—known mutator enzymes—were documented to increase clonal heterogeneity [13,22], we decided to verify whether RAG2/AID expression signatures, may relate to CNA characteristics, their prognostic impact, and thus provide additional information on relapse potential.

Using the MLPA technique we detected CNAs in seven genes in 94 patients. Furthermore, RAG2 and AID expression signatures were measured on a broader population of 166 patients with available RNA material. The distribution and frequencies of CNAs were generally consistent with those from other study cohorts in the whole series, as well as across genetic subgroups, with the highest CNA frequency in BCR::ABL1^pos (80% CNA) and BCR::ABL1-like subgroups (83% CNA), intermediate in NEG B-ALL (69% CNA), while the lowest in patients with MLL rearrangements [2,33]. This differential distribution already suggests a possible correlation with subtype-specific mutational processes and/or cooperation of concomitant gene aberrations.

4.2. RAG Is Associated with High CNA Mutation Burden, While AID Is Frequently Observed in CNA^low Patients

In order to elucidate a functional link between CNAs and the mutagenic process, we identified four different signatures of AID and RAG2 expression and found their correlation with particular CNA profiles. Importantly, the signature with the highest expression of both enzymes was totally absent in the CNA^neg subgroup, re-emphasizing AID and RAG2 involvement in CNA formation. The highest CNA mutation burden (≥3), IKZF1 deletions, and particularly CNA^high/IKZF1^pos profile were associated with high RAG2 and low AID (AID^low/RAG2^high) signature. This suggests an increased RAG-mediated clonal diversity involved in the progression of these ALL subtypes. Interestingly, the highest incidence of the above CNAs as well as high RAG2 expression was observed in BCR::ABL1^pos and BCR::ABL-like ALLs from our series. Increased RAG2 expression has been already described as a result of constitutive VDJ recombination activation by ABL and JAK-STAT kinase-activating lesions [30,31]. Additionally, our data indicate that IKZF1 deletion by itself activates RAG1/2 expression, thus increasing the risk for further deletions, which can lead to disastrous consequences [56].

In contrast to the above data, low CNA mutation burden (0–1), CNAs lacking IKZF1, and the absence of CNA (CNA^neg) correlated with a higher representation of the AID^high/RAG2^low signature, which signifies a possible role of AID-mediated mutagenesis underlying progression of these leukemias. AID is a member of the family of cytidine deaminases leading to substitution mutations, therefore its involvement in SNV-type rather than CNA-type mutagenesis may be expected. However, a co-synergic involvement of AID with RAG in CNA formation has been described as well [13]. Previous studies already reported higher AID expression in ALL lacking common primary aberrations (our NEG ALL subgroup) [57]. Here we expand these data by reporting prevalent AID^high/RAG^low signature not only in NEG ALL but also in the CNA^neg series, suggesting its involvement in mutagenic processes other than CNA rearrangements in the absence of RAG.

Interestingly, in contrast to RAG2, AID was found to be also upregulated by environmental factors, like repeated exposure to inflammatory stimuli, paralleling chronic infections in childhood [13,22,25,27]. Recent studies also describe a depleted microbiome as an accelerator of ALL development, while metabolites of a healthy microbiome prevented leukemia through AID inhibition [58,59,60]. Importantly, this interplay of AID with inflammation may be a factor stimulating clonal heterogeneity, e.g., in the infectious environment of immunodeficient ALL patients. Thus, while RAG has a well-established role in pro-B cells, expression of AID represents a recently discovered threat to lymphoid precursor genome integrity, but also a link to other cancers associated with chronic inflammation [13,26,61,62,63].

4.3. Prognostic Impact of CNAs Aberrations in Correlation with Data from Other Studies

Most importantly, CNA profiles in correlation with various AID/RAG2 ratios—impacted treatment outcomes in our series. Thus, on the basis of the clinical data, we have proposed a revised CNA-based index stratifying prognosis of the intermediate subgroups from our series: NEG and BCR::ABL ALL. The CNA-based classification defined 3 subgroups with different CNA profiles and prognoses: bad-CNA (CNA^high/IKZF1^pos), relatively good-CNA (all other CNAs, lacking CNA^high/IKZF1^pos), and CNA^neg with an intermediate prognosis.

We report here that bad-CNA: CNA^high/IKZF1^pos is the most important CNA profile identifying patients with the worst outcome both in NEG and BCR::ABL subgroups of intensively treated adult ALL, and with the high RAG2 expression as a causal driver of the mutagenic process. Additionally, despite a small group of patients, we report a particularly unfavorable outcome for the dominant negative Ik6 variant within the CNA^high/IKZF1^pos subgroup. Our data are consistent with the observations of many authors [4,33]. Stanulla et al. were the first to describe the CNA^high/IKZF1^pos profile as defining very poor prognosis in pediatric ALL (so-called IKZF^plus subgroup) [39]. Since then, these data have been confirmed in several studies both in childhood and adult ALL [3,5,33,40,64]. However, opposite results have also been published. Discrepancies mainly concerned adult ALL cohorts, while in childhood ALL the prognostic impact of CNA profiles was consistently well documented [2,35]. This points to age-related factors frequently confounding the prognostic power of genetic lesions. The prevalence of IKZF1 deletions increases with age, thus studies omitting younger adults or adolescents will have a higher incidence of IKZF1^pos patients with an overlapping impact of age-related comorbidities [6,7]. Importantly, we confirmed this hypothesis in our study, showing that the prognostic impact of CNA^high/IKZF1^pos was restricted to the younger population only (<40 y).

Another factor contributing to clinical discrepancies may relate to different group definitions based on genetic background. For example, many studies have chosen to analyze the impact of CNA^high/IKZF1^pos among BCR::ABL1^neg patients [2,35]. In contrast, our NEG B-ALL population was not only negative for BCR::ABL1 fusion but also excluded all other known primary aberrations, e.g., the majority of BCR::ABL1-like cases.

4.4. Prognostic Impact of CNAs Aberrations in Particular Context of BCR::ABL1^pos Subgroup

Importantly, the BCR::ABL1^pos series from our cohort was stratified not only by bad-CNA but also by good-CNA profile with favorite prognosis. Thus, the presence of good-CNAs could contribute to a relatively good outcome of our BCR::ABL1^pos series as a whole group. This observation is of importance, as some studies did not show any correlation between CNA and outcome for adult BCR::ABL1^pos ALL in the final multivariate analyses [2,33,64]. This may be due to the complex nature of chemotherapy resistance in BCR::ABL1^pos leukemias. According to the literature, most BCR::ABL1^pos patients relapse due to kinase domain mutations, which represent SNV-type aberrations, while 30–40% of patients relapse without kinase domain mutations [31]. We assume that the latter population may correspond to our bad-CNA with CNA^high/IKZF1^pos profile, as IKZF1 itself was documented to confer poor responsiveness to ABL kinase inhibitor therapy in the absence of mutation [65]. Interestingly, 5/7 patients who relapsed in our BCR::ABL1^pos series presented CNA^high/IKZF1^pos deletions. Although we did not look for ABL1 kinase domain point mutations in this subgroup, we assume that RAG-mediated CNA-type mutagenesis is more relevant in this subtype than SNV-type ABL1 lesions, in line with high RAG2 expression. However, overlap of CNA^high/IKZF1^pos with SNV may also be considered, in line with observations of Koptyra et al. suggesting involvement of ROS-mediated SNV-type mutagenesis within ABL point mutations hotspot [66]. In contrast, good-CNA BCR::ABL1^pos patients from our series responded well to TKI treatment (see Figure 7).

Other studies report on CDKN2A/B deletions as a high-risk marker for adult BCR::ABL1^pos ALL [47,48]. Although on a limited series, we document, that CDKN2A/B deletions confer an adverse prognostic impact only when accompanied by IKZF1 deletions. CDKN2A/B deletions were frequently accompanied by IKZF1 deletions in BCR::ABL1^pos subgroups. Therefore, it is possible, that the poor prognosis of the CNA^high/CDKN2A/B^pos subgroup could be attributed to the frequent coexistence of IKZF1 with CDKN2A/B deletions as a part of CNA^high/IKZF1^pos population.

In the NEG ALL subgroup, the clinical impact of good-CNA was labeled as “good”, although remained unresolved. This was mostly due to the unknown nature of other mutations, whose interplay with good-CNA could affect the prognostic power.

4.5. Prognostic Impact of AID and RAG Expression in CNA^neg vs. CNA^pos Subgroups

Importantly, the evaluation of AID expression helped in the prognostic stratification of the CNA^neg subgroup (30% of total ALL). High AID expression has already been reported by Swaminathan et al. as a strong indicator of poor overall survival [13]. However, as part of this evaluation was performed on a pediatric group, here we present data for adult patients, specifying that the prognostic significance of AID was reserved only for CNA^neg patients. Although we do not know the mutation profile associated with the AID^high/RAG2^low signature, our data signify distinct mutagenic processes accompanying disease progression in these high-risk CNA^neg patients, probably related to SNV-type lesions. Their association with inflammatory signals is also possible.

In contrast, in the CNA^pos series high RAG2 marked patients with poor outcomes, which resulted from the presence of CNA^high/IKZF1^pos mutations. Thus, our data suggest that patients whose ALL involves increased RAG2 or AID-mediated clonal heterogeneity exhibit a more aggressive disease because RAG and AID increase the probability of acquiring high-risk mutations, distinct in the context of CNA^pos and CNA^neg subgroups.

4.6. Combined Revised Risk Classification of Adult ALL: CNA Data with Primary Aberrations Reclassify Prognostic Index of Adult ALL

Finally, on the basis of an extensive analysis of genetic data, we proposed a combined version of risk classification for intensively treated adult ALL, incorporating the CNA data into the risk index for primary aberrations. This is in contrast to the previous study by Moorman et al., who proposed a revised UKALL genetic risk classification based on key established primary aberrations as a stronger prognostic marker than secondary deletions [2]. Although the treatment protocol may affect prognostic factors, our study provides evidence that, at least for intensive treatment protocols such as PALG, besides MLL and hyperdiploid aberrations, secondary CNA aberrations stratify prognosis stronger than the established conventional factors. For example, these aberrations may confer a totally different outcome in patients with the same primary aberration.

Unfortunately, we did not have a chance to correlate our results with MRD data for all patients. However, independently from the prognostic power of MRD by itself, one can imagine, that the blast cells possess the potential to evolve, which can change their genotype and phenotype. Thus, the negative MRD results assessed using techniques following initial profiles—may not always recapitulate the risk of relapse. Whereas initial CNA profiles inform on the risk of adverse events related to clonal evolution and heterogeneity.

5. Conclusions

In summary, we present a pragmatic approach to select risk-defined ALL subgroups based on a simple categorization of CNA profiles. Integrating CNA data with the risk index by primary aberrations, allowed us to stratify the outcome of about 70% of adult ALL: into the poor-risk (53%), and favorable-risk populations (14%). This resulted in the reclassification of an additional 37% of patients who changed their status from IM-risk patients to favorable- or poor-risk, which may result in more relevant treatment stratification (see Figure 7).

Importantly, specific CNA profiles correlated with different AID/RAG2 signatures. We propose that these signatures contribute to the high-risk behavior of selected ALL subsets: RAG as an active driver increasing the odds for CNA^high/IKZF1^pos development among CNA^pos, while the abundance of AID indicates poor risk subset among CNA^neg ALL patients. This may be helpful to refine the treatment of CNA^neg patients. These data may also encourage the notion of risk-reducing antimutagenic interventions in the course of treatment (e.g., reinforcing anti-inflammatory background).

Given the limitations of this study: its retrospective nature, the lack of a comprehensive SNV mutational profile at diagnosis, and relatively small subgroups, as well as scant data on mutational processes in adult ALL compared to the vast data in childhood leukemias—further investigations in this field are warranted.