Clinical Relevance and Applicability of the 2022 World Health Organization Classification of Childhood B Lymphoblastic Leukemia in the Context of MRD-Directed Therapy

doi:10.21203/rs.3.rs-5303520/v1

Download PDF

Article

Clinical Relevance and Applicability of the 2022 World Health Organization Classification of Childhood B Lymphoblastic Leukemia in the Context of MRD-Directed Therapy

https://doi.org/10.21203/rs.3.rs-5303520/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

WHO5-2022 classification of B-lymphoblastic leukemia (B-ALL) incorporates several novel entities requiring high-throughput sequencing for their accurate characterization. The clinical relevance of this classification in the context of contemporary MRD-directed therapy is unclear. We analyzed 533 pediatric B-ALL uniformly treated with ICiCLe-ALL-14 protocol as defined by WHO2016 and reclassified them as per WHO5-2022 using targeted sequencing, FISH, and cytogenetics. Subtype-defining genetic abnormalities were identified in 81.2% of the cohort as per the WHO5 classification. Among the new subtypes, PAX5^alt, MEF2D-r, and BCR::ABL1-like(ABL-class) were associated with an inferior 2-year event-free survival (EFS) of 39.1% (p < 0.0001), 53.8% (p = 0.024) and 60.6% (p = 0.043), respectively. We developed a 3-tier genetic risk stratification model incorporating 15 genetic subtypes and the IKZF1 deletion. Children with standard, intermediate, and high genetic risk demonstrated 2-year EFS of 92.6%, 71.0%, and 50.7% (p < 0.0001), and 2-year overall survival of 94.3%, 81.9%, and 71.6% (p < 0.0001), respectively. Genetic risk further identified heterogeneous outcomes among ICiCLe risk groups (p < 0.0001). Standard genetic risk was associated with superior OS and EFS irrespective of MRD status. We demonstrate the applicability of the WHO5 classification in routine practice and create a general framework for incorporating the WHO5 classification in risk-adapted therapy for childhood B-ALL.

Health sciences/Diseases/Cancer/Haematological cancer/Leukaemia/Acute lymphocytic leukaemia

Biological sciences/Cancer/Haematological cancer/Leukaemia/Acute lymphocytic leukaemia

B Cell Precursor ALL

RNA sequencing

MRD Directed Therapy

In the last few decades, successful adoption of risk-directed therapy has resulted in remarkably improved pediatric B-lymphoblastic leukemia/lymphoma (B-ALL) outcomes. Assessment of therapeutic response by detecting and quantitating measurable residual disease (MRD) forms the cornerstone of contemporary risk-adapted protocols.

The underlying biology of the B-ALL subtype, which includes abnormalities of chromosome number such as high hyperdiploid as well as chimeric gene fusions such as ETV6::RUNX1, are clinically relevant even in the context of MRD-directed therapies.¹ Contemporary protocols have thus adopted an integrated approach incorporating both baseline genetics and MRD as an optimal strategy for risk-directed therapy.^1,2

In the last decade, we have significantly advanced our understanding of B-ALL biology due to progress in genome sequencing technologies such as whole genome and transcriptome sequencing. These seminal studies have led to the identification of novel biological subgroups of B-ALL.^3,4 The WHO5 and ICC classifications have incorporated several of these new subtypes, such as TCF3::HLF1, PAX5 alteration (PAX5^alt), PAX5 P80R, DUX4-rearrangement (DUX4-r), MEF2D-r, ZNF384-r, NUTM1-r, ETV6::RUNX1-like, IG::MYC.^5,6 Recently, Jeha and colleagues have shown the clinical significance of the new subtypes of B-ALL in the context of the MRD-directed St. Jude Total Therapy Study 16 protocol. ⁷ It is important to validate and incorporate these genomics data into clinical practice for prognostication and targeted therapy. ^7,8

A major challenge in detecting the WHO5-recognised subtypes of B-ALL in routine practice is the heterogenous biology of B-ALL, which requires integration of copy number alterations, fusions involving multiple genes with various partners, and somatic mutations to comprehensively detect all subtypes. For diagnosis of these new entities, legacy workflows like fluorescent in situ hybridization (FISH), multiplexed PCR, and multiplex ligation-dependent probe amplification (MLPA) require multiple rounds of reflexive testing that would result in time delay. Furthermore, PCR-based tests require knowledge of both partners involved in the fusion, which is not always feasible, especially with the novel described entities.

Whole transcriptome sequencing (WTS) has been proposed as a one-stop solution for this problem; however, the clinical translation of WTS in practice is challenging due to the resource-intensive nature and turn-around time of WTS, especially if RNA-sequencing results are to impact therapeutic decisions. We have previously validated NARASIMHA, a lab-developed, cost-effective modular targeted RNA-sequencing assay to detect chimeric gene fusions with knowledge of only one partner gene involved in the fusion (for, e.g., ABL1). We hypothesized that a combination of targeted DNA and RNA sequencing, with an initial screen based on legacy techniques, could detect all the entities described in the WHO5 classification of B-ALL.

Here, we present the clinical utility of this approach in detecting WHO5 subtypes in a large, uniformly treated cohort of pediatric B-ALL. We further describe the clinical relevance of such an approach in predicting outcomes in the context of MRD-directed therapy in childhood B-ALL.

Diagnostic Evaluation of B-ALL

Diagnosis: A total of 533 newly diagnosed pediatric B-ALL, aged 1-15 years, enrolled in the ICiCLe-ALL-14 trial (Clinical Trials Registry-India number, CTRI/2015/12/006,434), from whom peripheral blood or bone marrow samples were received for molecular testing between January 2019 to March 2022, were included in this study (overview in Supplementary Figure 1). B-ALL diagnosis was made using a 10-12 color, 5-tube multi-color flow cytometry (MFC) panel as per WHO2016 criteria, as described elsewhere.⁹ Also, ploidy status was determined using the FxCycle Violet-dye-based MFC method.¹⁰ Informed consent was taken from the legal gurdians while enrolling in the ICiCLe-ALL-14 trial (Clinical Trials Registry-India number, CTRI/2015/12/006,434) for participation and data sharing including publication. Institutional Ethics Committee of ACTREC, Tata Memorial Centre gave ethical approval for this work (IEC III – 900924).

Cytogenetics Work-up: FISH was performed using locus-specific identifier (LSI) probes for BCR::ABL1 (Zytovision, Bremerhaven, Germany), TCF3::PBX1 (CytoTest, Rockville, USA), ETV6::RUNX1, and KMT2A-r (Abbott, Green Oaks, USA). Hyperdiploid B-ALL was diagnosed using centromeric FISH probes for chromosomes 4, 10, and 17 (Zytovision, Bremerhaven, Germany). SureFISH 7p12.2 IKZF1 (Agilent Dako, USA) probe was used to detect partial/whole IKZF1 gene deletions. Ploidy analysis was also performed on Giemsa-stained metaphases obtained from overnight cultured BMA/peripheral blood cells to determine the modal chromosome number. Analysis was carried out under a fluorescence microscope (Olympus BX61) using GenASIs software (Applied Spectral Imaging, Israel), and 200 intact interphase nuclei were evaluated by two independent co-observers as previously described.¹¹

Polymerase Chain Reaction based screening:Common translocations (BCR::ABL1 (E13/14A2 and E1A2), ETV6::RUNX1, TCF3::PBX1 and KMT2A::AFF1) were screened using real-time PCR. IKZF1 deletions were detected using fragment length analysis.^12,13

Next Generation Sequencing: Targeted RNA sequencing was performed in 392 patients (73.5%) who were negative for common translocations (Supplementary Figure 1, Supplementary Table 1). Patients with negative RNA sequencing (n=124) were subjected to targeted DNA sequencing. The details of these assays and bioinformatics analysis can be seen in supplementary data.

Risk stratification and treatment protocol: Patients were stratified according to the ICiCLe-ALL-14 risk stratification detailed in the supplementary data. These patients were treated according to the ICiCLe-ALL-14 risk-adapted protocol.¹⁴ Briefly, patients underwent four intensive treatment blocks (induction, consolidation, interim maintenance, delayed intensification) followed by 96 weeks of maintenance, varying treatment intensity according to risk group. The first randomization explored the toxicity impact of a shorter induction schedule of prednisolone (3 vs 5 weeks) in young, non-high-risk B-ALL patients. The second randomization assessed the survival benefits of replacing doxorubicin with mitoxantrone during delayed intensification across all patients (Supplementary Figure 2). Only fifteen patients underwent an allogeneic Hematopoietic stem cell transplant. Post-induction (PI) and post-consolidation (PC) MRD assessments were performed using our previously validated highly sensitive MFC assay.¹⁵

Classification according to WHO 2016 and WHO 2022:The pediatric B-ALL cases were assigned to defined genetic categories as per the WHO 2016 (revised 4th edition), and WHO 2022 (5th edition) Classification of Tumours of Haematopoietic and Lymphoid Tissues integrating DNA ploidy by MFC, FISH, chromosomal counting, screening PCR, NARASIMHA and targeted DNA sequencing.¹⁶

Clinical Endpoints and Statistical Analysis:Overall survival (OS) was calculated from the diagnosis date to either the last follow-up date or death with censoring at the last contact. Event-free survival (EFS) was defined from diagnosis to events like death, relapse, or refractory status, also censored at the last contact. Chi-square or Fisher’s exact test (two-sided) in MedCalc (version 14.8.1) assessed the association between categorical variables such as MRD positivity and genetic subgroups. The impact of genetic subgroups on EFS and OS was analyzed using Kaplan-Meier curves, generated on Python3 (version 3.12.0), and compared using the log-rank test. We stratified genetic risk into standard (SGR), intermediate (IGR), and high risk (HGR) based on genetic predictors of outcome and published literature for rare variants. This stratification was correlated with PI MRD and ICiCLe risk groups. Kaplan-Meier curves for these subgroups were produced using MedCalc (version 14.8.1). Multivariate analysis was performed using the Cox proportional-hazards regression analysis.

Clinical, laboratory, and risk stratification characteristics of patients enrolled: Our final analysis included 533 children with B-ALL with a male-to-female ratio of 1.99:1 and a median age of 5 years. According to the NCI risk stratification, most patients were classified as standard risk (SR, 67.2%). However, with the final ICiCLe risk stratification, the majority (57.4%) were categorized as high risk (HiR), followed by SR (22.9%) and intermediate risk (IR, 19.7%). Nearly one-third of patients (32.2%, out of 519 assessed) were detected to harbor measurable residual disease (defined as MRD ≥ 0.01%) at the PI timepoint. Similarly, 9.8% (out of 244 assessed) were positive at the PC timepoint. A summary of clinical, laboratory, and genetic characteristics, as well as MRD responses, can be seen in Table 1.

Table 1

Demographic and Laboratory Characteristics, WHO 2016 Classification, MRD characteristics and Risk Stratification of B-ALL Patients
Demographics
Total number of patients accrued	533
Age (years)	Range:- 1–15, Median:- 5
Gender	Male: Female- 1.99:1
Laboratory characteristics
TLC at presentation
1. ≥50 x 10⁹/L	81 (15.2%)
2. <50 x 10⁹/L	452 (84.8%)
WHO 2016 Genetic categories
1. Hyperdiploidy	182 (34.1%)
2. ETV6::RUNX1	68 (12.8%)
3. TCF3::PBX1	42 (7.9%)
4. BCR::ABL1	29 (5.4%)
5. BCR::ABL1 like	24 (4.5%)
6. KMT2A rearranged	19 (3.6%)
7. Hypodiploidy	6 (1.1%)
8. iAMP21	3 (0.6%)
9. NOS	160 (30.0%)
NCI Risk
SR	358 (67.2%)
HR	175 (32.8%)
ICiCLe Initial Risk (n-533)
SR	205 (38.4%)
IR	148 (27.8%)
HR	180 (33.8%)
ICiCLe Risk (n-519)
SR	119 (22.9%)
IR	102 (19.7%)
HR	298 (57.4%)
Post-induction MRD (n = 519)
Positive	167 (32.2%), Median- 0.14, Range- 0.01-88
Negative	352 (67.8%)
Post-consolidation MRD (n = 244)
Positive	24 (9.8%), Median- 0.2, Range- 0.01-86
Negative	220 (90.2%)

Characterization of Genetic Subgroups and their Correlations: Through the integration of conventional PCR, targeted RNA sequencing, cytogenetics, and immunophenotyping, we identified 13 distinct genetic subgroups within our cohort (Table 2). PAX5 ^alt (24, 4.5%), BCR::ABL1-like (23, 4.3%), DUX4-r (12, 2.3%), MEF2D-r (13, 2.4%), ZNF384-r (3, 0.6%) and TCF3::HLF (4, 0.7%) subgroups of B-ALL were exclusively identified using RNA-sequencing (Fig. 1). Among the BCR::ABL1-like category, almost equal division was seen among the BCR::ABL1-like: ABL-class (n-11) and JAK-STAT activated (n-13) classes. Among the BCR::ABL1-like: ABL-class, the 3’ partner were PDGFRB (EBF1; n = 5), ABL1 (n = 4; one each of ETV6, NUP214, RANBP2, FOXP1), ABL2 (RCSD1) and CSF1R (LBD1) and those in JAK-STAT activated class were CRLF2 (n = 9; P2RY8; n = 8 and one IGH), JAK2 (n = 4; comprising PAX5; n = 3 and one EBF1). For PAX5^alt category, the 3’ fusion partners were AUTS2 (n = 5), succeeded by MBNL1 (n = 3), ETV6 (n = 2), IGH (n = 2), ZNF521 (n = 2) and one each with CBFA2T3, FOXP1, GSE1, ANKRD12, CA10, and ESSRB. For MEF2D-r, the fusion partners were BCL9 (n = 8), HNRNPUL (n = 3), DASAP1 (n = 1) and SF1 (n = 1). For ZNF384-r cases, the fusion partners were TCF3 (n = 2) and EP300 (n = 1).

Table 2

Age and Sex Distribution, WBC Counts, and MRD characteristics across Genetic Subgroups. *p-values are not provided for genetic subgroups with fewer than 10 cases.
Genetic category	N (%)	Median age in years (range), p-value*	Male n (%), p-value*	Median WBC (range) at diagnosis (X 10⁹/L), p-value*	Percentage of PI MRD positive cases (median, range), p-value*
All categories	533	5 (1–15)	355 (66.5)	6.37 (0.23–692.3)	32.2 (0.14, 0.01- 88)
High hyperdiploidy	182 (34.1)	4 (1–13), < 0.001	125 (67.2), 1.000	4.63 (0.61 − 253.0), < 0.001	30.3 (0.01, 0.01- 11), 0.38
ETV6::RUNX1	68 (12.8)	4 (1–15), 0.003	44 (66.7), 1.000	4.85 (0.23 − 90.1), 0.064	24.2 (0.02, 0.01–5.8), 0.10
TCF3::PBX1	42 (7.9)	6(2–14), 0.154	24 (57.1), 0.177	14.185 (1.22 − 281.0), 0.003	26.2 (0.03, 0.01–3.60), 0.55
BCR::ABL1	29 (5.4)	10 (2–14), < 0.001	22 (75.9), 0.318	61.71 (1.11 − 559.4), <0.01	48.3 (0.48, 0.01–23.50), 0.01
BCR::ABL1 like	24 (4.5)	11.5 (3–15), 0.001	17 (70.8), 0.825	33.525 (0.93 − 295.1), 0.005	70.8 (1.6, 0.01-62), 0.000033
PAX ^alt	24 (4.5)	4 (1–14), 0.354	16 (69.6), 0.825	8.15 (0.59 − 208.5), 0.637	23.8 (0.04, 0.01–1.1), 0.46
KMT2A-r	19 (3.6)	3 (0–14), 0.001	12 (63.2), 0.806	19.54 (1.37 − 424.71), 0.055	26.3 (0.26, 0.01-50), 0.76
MEF2D-r	13 (2.4)	9 (3–14), 0.004	5 (38.5), 0.038	11.15 (1.31 − 119.11), 0.252	30.8 (0.05, 0.01–0.24), 1.00
DUX4-r	12 (2.3)	6 (3–14), 0.167	9 (75.0), 0.759	5.14 (1.56 − 27.73), 0.358	33.3 (0.03, 0.01–3.5), 1.00
Hypodiploidy	6 (1.1)	11 (4–13)	4 (66.7)	3.27 (1.34 − 155.1)	33.3 (0.01, 0.01–4.3)
PAX5 P80R	4 (0.7)	12 (8–14)	4 (100)	55.87 (1.99–86.8)	0
TCF3::HLF	4 (0.7)	8 (4–14)	3 (75)	6.335 (1.32 − 9.74)	75.0 (1.25, 0.01–6.5)
ZNF384-r	3 (0.6)	5 (4–16)	1 (33.3)	14.33 (4.37 − 22.32)	66.7 (6.20, 0.32–12.09)
iAMP21	3 (0.6)	6 (2–10)	2 (66.7)	3.04 (2.94–5.4)	0, 0.56
NOS	100 (18.8)	7 (1–14), < 0.001	70 (70.0), 0.481	8.07 (0.29–692.3), 0.147	33.0 (0.05, 0.01–71.7), 1.00

Targeted DNA sequencing further helped identify additional PAX5^alt (n = 4) and the PAX5 P80R group (n = 4). Using this approach, we could identify a genomic abnormality in 81.2% of all cases when compared to legacy approaches (61.7%)

Patients harboring BCR::ABL1 (median age- 10 years; p < 0.001), BCR::ABL1-like (median age- 11.5 years, p = 0.001), and MEF2D-r (median age- 9 years; p = 0.004) were significantly older compared to the entire cohort (median age- 5 years; range- 1 to 15 years). BCR::ABL1 (median WBC- 61.7 x 10⁹/L; p = 0.003) and BCR::ABL1 like (median WBC- 33.5 x 10⁹/L; p = 0.005) showed higher WBC counts as compared to the entire cohort (median WBC- 6.4 x 10⁹/L; range- 0.2 to 692.3 x 10⁹/L). Additionally, the BCR::ABL1 (48.3% MRD-positive; median MRD: 0.48; p = 0.02) and BCR::ABL1-like (70.8% MRD-positive; median MRD: 1.6; p < 0.0001) groups exhibited higher PI MRD positivity as compared to the rest (32.3% PI MRD-positive; median PI MRD: 0.14%). The MRD characteristics of each sub-group are detailed in supplementary data and highlighted in Supplementary Fig. 3.

IKZF1 deletion was identified in 69 (15.1%) of the 458 patients tested and showed a high prevalence in the BCR::ABL1 (56.5%) and BCR::ABL1-like (55.6%) genetic subgroups. Additionally, IKZF1 deletions were detected across various other subgroups, including NOS, high hyperdiploidy, PAX5^alt, DUX4-r, KMT2A-r, hypodiploidy, ETV6::RUNX1, and iAMP21 (Supplementary Fig. 4).

Re-stratification from the WHO 2016 to WHO 2022 classification: There was no change in the classification of 373 patients (ETV6::RUNX1, High hyperdiploidy, BCR::ABL1, TCF3::PBX1, BCR::ABL1 like KMT2A-r, hypodiploidy, iAMP21) when they were classified as per the WHO 2022 compared to the WHO 2016. Among WHO 2016-defined NOS subgroup (n = 160, 30% of the entire cohort), 60 (37.5%) children were reclassified into newly recognized subtypes such as PAX5^alt (n = 24, 15%), DUX4-r (n = 12, 7.5%), MEF2D-r (n = 13, 8.1%), PAX5 P80R (n = 4, 2.5%), TCF3::HLF (n = 4, 2.5%), and ZNF384-r (n = 3, 1.9%) as seen in Fig. 1A.

Patient Outcomes

The 2-year EFS and OS of the entire cohort were 75.8% and 85.1%, respectively (median EFS and OS not reached). The median follow-up of the cohort was 26.8 months. The presence of PI MRD predicted a significantly inferior EFS [HR- 2.0; 95% CI- 1.42 to 2.89; (p < 0.0001)] and OS [HR- 1.7; 95% CI- 1.08 to 2.64; (p = 0.01)]. Similarly, PC MRD positivity was associated with markedly worse EFS [HR- 22.0; 95% CI- 4.37 to 110.93; (p < 0.0001)] and OS [HR- 2.4; 95% CI- 0.78 to 7.70; (p = 0.02)] (Supplementary Fig. 5). The impact of ICiCLe risk stratification on patient outcomes is detailed in the supplementary data (Supplementary Fig. 6).

Outcomes for individual genetic subgroups: On evaluating the impact of genetic subgroups, the following were associated with inferior EFS: PAX5^alt (HR- 2.8; 95% CI- 1.18 to 6.68; p = 0.0001), BCR::ABL1 (HR- 2.5; 95% CI- 1.17 to 5.51; p = 0.0002), MEF2D-r (HR- 2.3; 95% CI- 0.75 to 7.20; p = 0.024), and BCR::ABL1-like with ABL-class fusion (HR-2.3; 95% CI- 0.68 to 7.55; p = 0.043) (Table 3). Subgroups associated with inferior OS were PAX5^alt (HR-4.3; 95% CI- 1.46 to 12.96; p < 0.0001) and BCR::ABL1 (HR-2.4; 95% CI- 0.98 to 6.11; p = 0.004). Similarly, IKZF1 deletion was associated with inferior EFS (HR-3.0; 95% CI- 1.72 to 5.07; p < 0.0001) and OS (HR-1.8; 95% CI- 0.95 to 3.51; p = 0.022). TCF3::HLF subgroup also exhibited poor EFS, with three out of four affected children having refractory disease and one experiencing an early relapse; however, statistical analysis was not performed due to the small sample size.

Table 3

Event-free and overall survival rates for genetic categories of B-ALL, as classified by WHO 2022, including hazard ratios and p-values. The genetic categories with fewer than 10 cases are not included in the table.
	Overall survival			Event free survival
	2 years OS	P-value	Hazard ratio (95% Confidence Interval)	2 years EFS	P-value	Hazard ratio (95% Confidence Interval)
IKZF1 del	77.5	0.022	1.8 (0.95 to 3.51)	46.2	< 0.0001	3.0 (1.72 to 5.07)
PAX5^alt	46.6	< 0.0001	4.3 (1.46 to 12.96)	39.1	0.0001	2.8 (1.18 to 6.68)
BCR::ABL1	67.6	0.004	2.4 (0.98 to 6.11)	50.6	0.0002	2.5 (1.17 to 5.51)
MEF2D-r	66.1	0.114	2.2 (0.51 to 9.33)	53.8	0.024	2.3 (0.75 to 7.20)
BCR::ABL1 like, ABL-Class	90.9	0.821	0.9 (0.23 to 3.11)	60.6	0.043	2.3 (0.68 to 7.55)
NOS	84.2	0.714	1.1 (0.67 to 1.79)	65.9	0.015	1.6 (1.03 to 2.33)
TCF3::PBX1	77.3	0.133	1.6 (0.77 to 3.26)	67.4	0.253	1.3 (0.76 to 2.36)
Hypodiploidy	83.3	0.961	1.1 (0.14 to 7.90)	66.7	0.684	1.3 (0.27 to 6.64)
KMT2A-r	72.0	0.214	1.8 (0.5469 to 5.6249)	67.5	0.645	1.2 (0.50 to 2.95)
BCR::ABL1 like, JAK-STAT	84.6	0.730	1.2 (0.35 to 4.33)	63.5	0.98	1.0 (0.37 to 2.65)
DUX4-r	82.5	0.791	0.8 (0.23 to 2.97)	73.3	0.58	0.7 (0.27 to 1.94)
ETV6::RUNX1	100	0.027	0.4 (0.23 to 0.72)	98.0	0.006	0.4 (0.27 to 0.65)
High Hyperdiploidy	92.1	0.0006	0.4 (0.29 to 0.65)	88.9	< 0.0001	0.4 (0.26 to 0.49)

Conversely, high hyperdiploidy was associated with superior EFS (HR-0.4; 95% CI- 0.26 to 0.49; p < 0.0001) and OS (HR-0.4; 95% CI- 0.29 to 0.65; p = 0.0006). ETV6::RUNX1 also predicted superior EFS (HR-0.4; 95% CI- 0.27 to 0.65; p = 0.006) and OS (HR- 0.4; 95% CI- 0.23 to 0.72; p = 0.027) as seen in Table 3.

Genetic risk assignment: We devised a 3-tier genetic risk model (SGR, IGR, and HGR) incorporating variables found to be significantly predictive of EFS on univariate analysis in the present study. Furthermore, if the number of available cases was < 10 in our study, available literature for those categories was considered (for, e.g., TCF3::HLF). Genetic categories with a hazard ratio (HR) < 0.75 were classified as SGR, those with an HR between 0.75 and 2 as IGR, and those with an HR > 2 as HGR. The details of this classification can be seen in Fig. 3. The 2-year EFS for SGR, IGR, and HGR was 92.6%, 71.2%, and 50.3%, respectively (p < 0.0001), and the 2-year OS for SGR, IGR, and HGR were 94.3%, 81.9%, and 71.6%, respectively (p < 0.0001; Table 4).

Table 4

Genetic risk-based survival outcomes and the impact of post-induction measurable residual disease positivity.
Impact of Genetic Risk on Patient Outcomes (n-533)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
Standard genomic risk (SGR)	1	92.6%	< 0.0001	1	94.3%	< 0.0001
Intermediate genomic risk (IGR)	2.77 (1.91 to 4.02)	71.2%		2.37 (1.48 to 3.79)	81.9%
High genomic risk (IGR)	5.34 (3.61 to 7.89)	50.3%		3.81 (2.34 to 6.19)	71.6%
Survival analyses in genetic risk categories in ICiCLe initial SR/IR risk categories (n-352)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
ICiCLe SR/IR & SGR	1	93.8%	< 0.0001	1	94.2%	< 0.0001
ICiCLe SR/IR & IGR	2.81 (1.76 to 4.48)	74.0%		2.34 (1.34 to 4.08)	81.4%
ICiCLe SR/IR & HGR	5.70 (3.18 to 10.20)	50.0%		4.38 (2.21 to 8.68)	70.6%
Survival analyses in genetic risk categories in ICiCLe initial HiR risk categories (n-181)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
ICiCLe HiR & SGR	1	89.0%	< 0.0001	1	94.5%	0.026
ICiCLe HiR & IGR	2.67 (1.39 to 5.10)	64.4%		2.35 (0.96 to 5.74)	83.8%
ICiCLe HiR & HGR	4.36 (2.53 to 7.51)	51.1%		3.28 (1.55 to 6.93)	73.1%
Impact of PI MRD in standard genomic risk (n-248)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
SGR with PI MRD Negative	1	93.9%	0.47	1	95.9%	0.96
SGR with PI MRD positive	0.75 (0.36 to 2.80)	90.6%	0.47	1.02 (0.42 to 2.48)	91.7%	0.96
Impact of PI MRD in intermediate genomic risk (n-130)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
IGR with PI MRD Negative	1	77.4%	0.017	1	82.8%	0.239
IGR with PI MRD positive	2.03 (1.03 to 4.00)	58.7%	0.017	1.56 (0.69 to 3.50)	79.9%	0.239
Impact of PI MRD in high genomic risk (n-141)
	Event free survival			Overall survival
	Hazard ratio (95% confidence interval)	2 year EFS	p-value	Hazard ratio (95% confidence interval)	2 year EFS	p-value
HGR with PI MRD Negative	1	67.4%	< 0.0001	1	79.6%	0.054
HGR with PI MRD positive	2.71 (1.65 to 4.43)	34.2%	< 0.0001	1.86 (0.97 to 3.55)	68.3%	0.054

Genetic Risk Identifies Heterogeneous Outcomes within ICiCLe Risk Groups

Since patients in this cohort were treated with risk-adapted protocol, we further evaluated the effect of genetic risk stratification in risk groups treated with similar intensities. The ICiCLe initial SR and IR were combined for this analysis, while the HiR group was analyzed separately. Genetic risk stratification retained its prognostic significance within both ICiCLe initial SR/IR and HiR groups. The 2-year EFS of SGR, IGR, and HGR within ICiCLe initial SR/IR were 93.8%, 74.0%, and 50.0%, respectively (p < 0.0001), and that within ICiCLe initial HiR was 89.2%, 64.4% and 51.1% respectively (p < 0.0001) (Fig. 4). Similarly, the 2-year OS of SGR, IGR, and HGR within ICiCLe initial SR/IR were 94.2%, 81.4%, and 70.6%, respectively (p < 0.0001), and that within ICiCLe initial HiR was 94.5%, 83.8% and 73.1% respectively (p = 0.026) (Table 4).

Genetic Risk Identifies Heterogeneity in Standard Risk B-ALL Irrespective of MRD Status

Patients in the SGR group exhibited superior outcomes regardless of PI MRD status, with 2-year EFS rates of 93.9% for MRD-positive and 90.6% for MRD-negative patients (p = 0.469), and 2-year OS rates of 95.9% and 93.5%, respectively (p = 0.954). In the IGR group, OS was similar between MRD-negative (84.2%) and MRD-positive (79.9%) patients (p = 0.239). In the HGR group, OS was 79.6% for MRD-negative and 68.3% for MRD-positive patients (p = 0.054). However, MRD status remained a significant predictor of EFS in the IGR and HGR groups, with 2-year EFS rates of 58.7% for MRD-positive versus 77.4% for MRD-negative in the IGR group (p = 0.017), and 34.2% for MRD-positive versus 67.4% for MRD-negative in the HGR group (p < 0.0001) (Table 4, Fig. 4).

Multivariate analysis

On multivariate analysis, genetic risk stratification (HR-2.0; 95% CI- 1.62 to 2.45; p < 0.0001), initial ICiCLe risk (HR-1.3; 95% CI- 1.05 to 1.57; p = 0.015), and PI MRD (HR-2.0; 95% CI- 1.42 to 2.74; p < 0.0001) independently predicted EFS, with genetic risk stratification (HR-1.7; 95% CI- 1.29 to 2.17; p = 0.0001) and PI MRD positivity (HR-1.6; 95% CI- 1.03 to 2.38; p < 0.0001) also independently impacting OS (Supplementary Table 2).

Pioneering recent studies have utilized unbiased broad-spectrum approaches such as WTS/WGS to uncover novel genetic subtypes^4,17, which have been recognized as distinct entities in the latest WHO5/ICC classification of B-ALL.¹⁶ Routine clinical application of WTS/WGS assays remains challenging. Herein, we demonstrate that targeted RNA and DNA sequencing, in combination with standard FISH/cytogenetics, is routinely implementable in clinical practice and can detect all of the WHO5-recognized subtypes of B-ALL. We demonstrate that a targeted sequencing approach identifies prognostically relevant subgroups even in the era of contemporary MRD-directed therapies.

Among the recently defined genetic categories, we found frequencies of 4.5% for BCR::ABL1-like, 4.5% for PAX5^alt, 2.44% for MEF2D-r, 2.25% for DUX4-r, and 0.75% each for PAX5 P80R, and TCF3::HLF, and 0.56% for ZNF384-r. The proportion of PAX5^alt (4.5% vs. 7.5%)⁴, and DUX4-r (2.25% vs. 4%)³ cases in our cohort are lower than that reported in the discovery cohorts. The NARASIMHA assay used in this study focuses only on detecting chimeric gene fusions. Unbiased sequencing approaches such as WTS that include gene expression profiling might have uncovered further cases of BCR::ABL1-like and DUX4-r. Additionally, we could not detect other specific categories, such as ETV6::RUNX1-like, due to a lack of gene expression profiling. Another factor to consider is that we report results only on a large cohort of children uniformly recruited in the ICiCLe-ALL-14 trial. Interestingly, transcriptomic sequencing findings from the St. Jude Total Therapy Study 16 trial reported similar BCR::ABL1-like (3%) and PAX5^alt (4.85%) cases.⁷ In total, 81.2% of the children analyzed could be assigned to a defined genetic subgroup using our approach of targeted RNA/DNA sequencing. In comparison, 18.7% of cases remained uncharacterized (B other) in contrast to 12.1% B-other cases reported in the St. Jude Total Therapy Study 16.⁷ While this gap highlights the need for broader spectrum genomic profiling in future studies, our approach provides a viable framework that detects most of the WHO5-recognised genetic abnormalities in children with B-ALL in a trial setting.

Here, we developed a three-tier genetic risk model incorporating variables predictive of EFS in a large cohort of children treated uniformly with MRD-directed risk-adapted ICiCLe-ALL-14 protocol. In addition to the high hyperdiploidy and ETV6::RUNX1, the SGR group also included DUX4-r. Children with DUX4-r had a 2-year EFS of 73.3 months (HR- 0.73; 95% CI- 0.27 to 1.94) in our cohort, and although the presence of DUX4-r was not associated with statistically significant better EFS, possibly due to only 12 cases, DUX4-r has been reported to be associated with superior outcome in earlier reports.⁷ KMT2A-r are often linked with adverse outcomes in both infants and adults with B-ALL. Our study included a cohort of 1–15-year-olds, and similar to the study where non-infant children with KMT2A-r were evaluated,¹⁸ our cohort of KMT2A-r B-ALL demonstrated an intermediate 2-year EFS of 67.5%, thus being grouped into IGR. Interestingly, only the BCR::ABL1-like with ABL class-r was associated with inferior outcome (median EFS 31.3 months; HR-2.3; 95% CI- 0.68 to 7.55; p = 0.043) in our cohort and included in HGR, while the JAK-STAT perturbed BCR::ABL1-like group (median not reached; HR- 1.00; 95% CI- 0.37 to 2.65; p = 0.98) was not associated with significant outcome difference and was included in the IGR category. Although BCR::ABL1-like B-ALL is generally associated with inferior prognosis irrespective of age, previous reports have suggested that MRD-directed therapy may be beneficial in abrogating the prognostic impact.¹⁹ In a retrospective multi-centric study of the pre-tyrosine kinase inhibitor (TKI) era, BCR::ABL1-like ABL-class-r had inferior outcomes.²⁰ Our cohort of BCR::ABL1-like B-ALL has a relatively larger proportion of ABL-class-r cases, which might have allowed the identification of distinct inferior outcomes of this subgroup. Among the patients of BCR::ABL1-like ABL-class-r, nine received TKI, despite which their outcome remained poor. The PAX5^alt group had the worst 2-year EFS (39.1%, HR- 2.8; 95% CI- 1.18 to 6.68, p = 0.0001) in our cohort and is included in the HiR category. Initial studies from the SJRCH/COG group have reported an intermediate prognosis of PAX5^alt with MRD-directed therapy,^4,7 however, based on the dismal EFS in our cohort, we have included the PAX5^alt in the HGR category. This finding aligns with the recently published study by Chang and colleagues, where approximately 50% of children with PAX5^alt ALL relapsed, highlighting the poor prognosis of this subtype in standard-risk pediatric patients.²¹ We included IKZF1 deletion, MEF2D-r and TCF3::HLF in the HGR category, which is in line with previous findings by the SJRCH/COG group.^4,7

The genetic risk groups proposed here were associated with distinct EFS across ICiCLe risk groups and had an independent association with EFS on multivariate analysis, in addition to ICiCLe-ALL-14 initial risk and PI MRD status. On combining genomic data with PI MRD assessments, we found out that the SGR group was associated with uniformly superior prognosis irrespective of MRD status. This, however, should not be interpreted as a rationale to deescalate therapy as the patients with MRD positive in the SGR category were all treated with high-risk consolidation in this cohort. MRD positive status was associated with inferior outcomes within the IGR and HGR categories, suggesting that a combination of genomic risk group and MRD status might form the most optimum strategy for risk-adapted protocols in children with B-ALL.

We acknowledge the limitation of our study of relatively shorter follow-up and lack of gene expression profiling. Incorporating unbiased sequencing approaches will help further delineate the genetic makeup of the NOS cases and may identify missed cases. However, the stepwise approach, targeted RNA/DNA sequencing strategy, and the genomic risk score described in this manuscript are easily implementable in routine practice and, together with MRD status, may help in refining risk-stratification approaches in children with B-ALL.

In conclusion, we demonstrate the applicability of the WHO5 classification in routine practice and create a general framework for the incorporation of the WHO5 classification in risk-adapted therapy for childhood B-ALL.

Acknowledgements:

We would like to express our sincere gratitude to Rakhi Salve, Bhagyashree Satam, Tuhina Srivastava, Harsh Chindarkar and Dr. Shivangi Harankhedkar for their invaluable assistance in entering and managing the patient data for this study. Their contributions were crucial to the successful completion of this work. We also acknowledge Mr Sitaram Ghogale and Mr Nilesh Deshpande for processing samples for flow cytometry testing.

Authorship and conflict-of-interest statements:

SR, NP and GC designed the study, conducted research, analyzed, interpreted the data, and wrote the manuscript. PB, SJ, SC, PT, DS, SG and PS conducted research conducted research and analyzed data. SS, AC, CD, NM, SB and GN recruited patients and analyzed data. Dr Nikhil Patkar is supported by an India Alliance (Wellcome-DBT) Senior Fellowship for Clinicians and Public Health Researchers. Dr Gaurav Chatterjee is supported by an India Alliance (Wellcome-DBT) Intermediate Fellowship for Clinicians and Public Health Researchers. Dr Nikhil Patkar has received research funding from Illumina Inc and ThermoFisher Scientific.

The author(s) declare no other competing financial interests.

Agreement to Share Publication-Related Data and Data Sharing Statement:

Dr Nikhil Patkar, Department of Haematopathology, CCE Building, Advanced Centre for Treatment Research and Education in Cancer, Tata Memorial Centre, Kharghar, Maharashtra, India, Pin:410210. Email: [email protected]; [email protected]. Phone: +91-22-27405000

Moorman A V., Enshaei A, Murdy D, Joy M, Boer JM, den Boer ML et al. Integration of genetics and MRD to define low risk patients with B-cell precursor acute lymphoblastic leukaemia with intermediate MRD levels at the end of induction. Leukemia 2024; 38. doi:10.1038/S41375-024-02329-0.
Moorman A V., Enshaei A, Schwab C, Wade R, Chilton L, Elliott A et al. A novel integrated cytogenetic and genomic classification refines risk stratification in pediatric acute lymphoblastic leukemia. Blood 2014; 124: 1434–1444.
Lilljebjörn H, Henningsson R, Hyrenius-Wittsten A, Olsson L, Orsmark-Pietras C, Von Palffy S et al. Identification of ETV6-RUNX1-like and DUX4-rearranged subtypes in paediatric B-cell precursor acute lymphoblastic leukaemia. Nat Commun 2016; 7. doi:10.1038/NCOMMS11790.
Gu Z, Churchman ML, Roberts KG, Moore I, Zhou X, Nakitandwe J et al. PAX5-driven Subtypes of B-progenitor Acute Lymphoblastic Leukemia. Nat Genet 2019; 51: 296.
Alaggio R, Amador C, Anagnostopoulos I, Attygalle AD, Araujo IB de O, Berti E et al. The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Lymphoid Neoplasms. Leukemia. 2022; 36. doi:10.1038/s41375-022-01620-2.
Duffield AS, Mullighan CG, Borowitz MJ. International Consensus Classification of acute lymphoblastic leukemia/lymphoma. Virchows Arch 2023; 482: 11–26.
Jeha S, Choi J, Roberts KG, Pei D, Coustan-Smith E, Inaba H et al. Clinical Significance of Novel Subtypes of Acute Lymphoblastic Leukemia in the Context of Minimal Residual Disease–Directed Therapy. Blood Cancer Discov 2021; 2: 326–337.
Tanasi I, Ba I, Sirvent N, Braun T, Cuccuini W, Ballerini P et al. Efficacy of tyrosine kinase inhibitors in Ph-like acute lymphoblastic leukemia harboring ABL-class rearrangements. Blood 2019; 134: 1351–1355.
Gudapati P, Khanka T, Chatterjee G, Ghogale S, Badrinath Y, Deshpande N et al. CD304/neuropilin-1 is a very useful and dependable marker for the measurable residual disease assessment of B-cell precursor acute lymphoblastic leukemia. Cytometry B Clin Cytom 2020; 98: 328–335.
Tembhare P, Badrinath Y, Ghogale S, Patkar N, Dhole N, Dalavi P et al. A novel and easy FxCycle™ violet based flow cytometric method for simultaneous assessment of DNA ploidy and six-color immunophenotyping. Cytometry Part A 2016; 89: 281–291.
et al. Cytogenetic Profile in 7209 Indian Patients with de novo Acute Leukemia: A Single Centre Study from India. J Cancer Ther 2016; 07: 530–544.
van Dongen JJ, Macintyre EA, Gabert JA, Delabesse E, Rossi V, Saglio G et al. Standardized RT-PCR analysis of fusion gene transcripts from chromosome aberrations in acute leukemia for detection of minimal residual disease. Report of the BIOMED-1 Concerted Action: investigation of minimal residual disease in acute leukemia. Leukemia 1999; 13: 1901–28.
Caye A, Beldjord K, Mass-Malo K, Drunat S, Soulier J, Gandemer V et al. Breakpoint-specific multiplex polymerase chain reaction allows the detection of IKZF1 intragenic deletions and minimal residual disease monitoring in B-cell precursor acute lymphoblastic leukemia. Haematologica 2013; 98: 597–601.
Das N, Banavali S, Bakhshi S, Trehan A, Radhakrishnan V, Seth R et al. Protocol for ICiCLe-ALL-14 (InPOG-ALL-15-01): a prospective, risk stratified, randomised, multicentre, open label, controlled therapeutic trial for newly diagnosed childhood acute lymphoblastic leukaemia in India. Trials 2022; 23: 1–20.
Tembhare PR, Subramanian PG PG, Ghogale S, Chatterjee G, Patkar N V., Gupta A et al. A High-Sensitivity 10-Color Flow Cytometric Minimal Residual Disease Assay in B-Lymphoblastic Leukemia/Lymphoma Can Easily Achieve the Sensitivity of 2-in-106 and Is Superior to Standard Minimal Residual Disease Assay: A Study of 622 Patients. Cytometry B Clin Cytom 2020; 98: 57–67.
Alaggio R, Amador C, Anagnostopoulos I, Attygalle AD, Araujo IB de O, Berti E et al. The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Lymphoid Neoplasms. Leukemia 2022; 36: 1720–1748.
Li JF, Dai YT, Lilljebjörn H, Shen SH, Cui BW, Bai L et al. Transcriptional landscape of B cell precursor acute lymphoblastic leukemia based on an international study of 1,223 cases. Proc Natl Acad Sci U S A 2018; 115: E11711–E11720.
Attarbaschi A, Möricke A, Harrison CJ, Mann G, Baruchel A, De Moerloose B et al. Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/ KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study. J Clin Oncol 2023; 41: 1404–1422.
Roberts KG, Pei D, Campana D, Payne-Turner D, Li Y, Cheng C et al. Outcomes of children with BCR-ABL1–like acute lymphoblastic leukemia treated with risk-directed therapy based on the levels of minimal residual disease. J Clin Oncol 2014; 32: 3012–3020.
den Boer ML, Cario G, Moorman A V., Boer JM, de Groot-Kruseman HA, Fiocco M et al. Outcomes of paediatric patients with B-cell acute lymphocytic leukaemia with ABL-class fusion in the pre-tyrosine-kinase inhibitor era: a multicentre, retrospective, cohort study. Lancet Haematol 2021; 8: e55–e66.
Chang TC, Chen W, Qu C, Cheng Z, Hedges D, Elsayed A et al. Genomic Determinants of Outcome in Acute Lymphoblastic Leukemia. J Clin Oncol 2024. doi:10.1200/JCO.23.02238.

There is NO conflict of interest to disclose.

Supplementarydraft5.docx

Download PDF

Reviewer #2 agreed at journal
31 Oct, 2024
Reviewer #1 agreed at journal
25 Oct, 2024
Reviewers invited by journal
24 Oct, 2024
Editor assigned by journal
23 Oct, 2024
Submission checks completed at journal
23 Oct, 2024
First submitted to journal
22 Oct, 2024
Unknown event
22 Oct, 2024

You are reading this latest preprint version

Clinical Relevance and Applicability of the 2022 World Health Organization Classification of Childhood B Lymphoblastic Leukemia in the Context of MRD-Directed Therapy

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1