PersonALL: a genetic scoring guide for personalized risk assessment in pediatric B-cell precursor acute lymphoblastic leukemia

Abstract

Background: Recurrent genetic lesions provide basis for risk assessment in pediatric acute lymphoblastic leukemia (ALL). However, current prognostic classifiers rely on a limited number of predefined sets of alterations.

Methods: Disease-relevant copy number aberrations (CNAs) were screened genome-wide in 260 children with B-cell precursor ALL. Results were integrated with cytogenetic data to improve risk assessment.

Results: CNAs were detected in 93.8% (n = 244) of the patients. First, cytogenetic profiles were combined with IKZF1 status (IKZF1^normal, IKZF1^del and IKZF1^plus) and three prognostic subgroups were distinguished with significantly different 5-year event-free survival (EFS) rates, IKAROS-low (n = 215): 86.3%, IKAROS-medium (n = 27): 57.4% and IKAROS-high (n = 18): 37.5%. Second, contribution of genetic aberrations to the clinical outcome was assessed and an aberration-specific score was assigned to each prognostically relevant alteration. By aggregating the scores of aberrations emerging in individual patients, personalized cumulative values were calculated and used for defining four prognostic subgroups with distinct clinical outcomes. Two favorable subgroups included 60% of patients (n = 157) with a 5-year EFS of 96.3% (excellent risk, n = 105) and 87.2% (good risk, n = 52), respectively; while 40% of patients (n = 103) showed high (n = 74) or ultra-poor (n = 29) risk profile (5-year EFS: 67.4% and 39.0%, respectively).

Conclusions: PersonALL, our conceptually novel prognostic classifier considers all combinations of co-segregating genetic alterations, providing a highly personalized patient stratification.

Fig. 1. Co-segregation analysis of genetic subgroup defining alterations and copy number aberrations affecting genes recurrently altered in pediatric B-ALL.

Co-occurrence and mutual exclusivity between various B-ALL subgroup defining alterations and/or detected copy number aberrations are labeled with green and brown colors, respectively. Significant (p < 0.05) associations revealed by pair-wise Fisher’s exact test are marked with asterisks.

Fig. 2. Event-free survival of high-hyperdiploid patients with presence or absence of double trisomy affecting chromosomes 4 and 6.

High-hyperdiploid patients with extra chromosomes of 4 and 6 showed higher estimated 5-year EFS rate than other high-hyperdiploid patients in our cohort (93.2% vs. 60.4%).

Fig. 3. Event-free survival of 260 patients classified by combined cytogenetic and IKZF1 copy number status.

Risk groups were determined as outlined in Fig. S6. Patients in the IKAROS-low, IKAROS-medium and IKAROS-high groups showed an estimated 5-year EFS rate of 86.3%, 57.4% and 37.5%, respectively.

Fig. 4. Event-free survival of 260 patients with B-cell precursor ALL, classified according to patient-specific composition of all disease-relevant aberrations associated with prognostic significance in our patient cohort.

Excellent, good, high and ultra-poor risk groups showed an estimated 5-year EFS rate of 96.3%, 87.2%, 67.4% and 39.0%, respectively. Patient-specific scores were calculated as outlined in Table S6.

Fig. 5. Distribution of cytogenetic subtypes across four risk groups determined based on non-overlapping ranges of patient specific cumulative scores generated from the prognostic value of single genetic lesions.

a The excellent risk group almost exclusively comprised patients with ETV6/RUNX1 gene fusion or high-hyperdiploid (HHD) karyotype. b, c The good and the high-risk groups were dominated by patients who classified as B-other in the current study. d Two-thirds of the patients in the ultra-poor risk group carried BCR/ABL1 fusion or alterations characteristic of the Ph-like subtype, with an additional one-fifth of the children in this group harboring KMT2A rearrangement or iAMP21 genotype.

Fig. 6. Validation of PersonALL, a novel classification approach assigning patients to four prognostic subgroups based on highly individualized cumulative scores reflecting the weighted impact of all detected and prognostically relevant genetic aberrations.

a Event-free survival of 606 patients with B-cell precursor ALL included in the TARGET ALL Phase 1 Pilot or Phase 2 Expansion studies. Excellent, good, high and ultra-poor risk groups showed an estimated 5-year EFS rate of 93.8%, 87.2%, 70.4% and 52.5%, respectively. b Event-free survival of 866 (260 + 606) patients with B-cell precursor ALL included in the merged discovery (in-house) and validation cohort. Excellent, good, high and ultra-poor risk groups showed an estimated 5-year EFS rate of 94.7%, 87.2%, 69.8% and 46.8%, respectively.