Cell cycle-related genetic variants are frequently acquired at relapse of acute lymphoblastic leukemia and predict worse post-relapse prognosis

Abstract Introduction:While the genetic landscape of acute lymphoblastic leukemia (ALL) at diagnosis has been deeply elucidated, its genetic evolution at relapse remains poorly understood. In this study, we compared the targeted next-generation sequencing (NGS) results of ALL before and after relapse and investigated their clinical significance. Methods:This was a retrospective study conducted at a single center, Severance hospital, in South Korea. Patients diagnosed with ALL who relapsed between January 2, 2013, and June 30, 2025, and had available NGS results from bone marrow or whole blood samples both before and after relapse were included in this study. The NGS results were either retrieved from previous medical records or newly generated through additional analyses if frozen-stored samples were available. To minimize false-negative results, only the NGS results performed on samples with a blast proportion of ≥10% were used in this study. Although the targeted gene panels used for NGS were not completely uniform, all shared a common set of 124 genes. Therefore, non-synonymous variants within these 124 genes were used for analysis. Acquired variants were defined as those not detected before relapse but detected after relapse. Enrichment of each KEGG pathway among acquired variants was assessed using a one-sided binomial test, with the null hypothesis that each acquired variant was randomly drawn from the 124 genes; for a pathway with x genes overlapping with the background 124 genes, the expected probability of each acquired variant falling within the pathway would be x/124. To account for multiple comparisons, the Benjamini-Hochberg procedure was applied for correction. Survival analysis was conducted using multivariate Cox regression, adjusting for age group, sex, and disease subtype as covariates. Results: A total of 41 patients were included in this study: 21 pediatric, 15 adolescent and young adult, and 5 adult; 21 male, and 20 female. Disease subtypes comprised 12 Philadelphia-positive B-lymphoblastic leukemia (Ph+ B-ALL), 21 Philadelphia-negative B-ALL, and 8 T-lymphoblastic leukemia. A total of 166 acquired variants were identified, with 33 (80.5%) patients having at least 1 acquired variant. ABL1 was the most frequently acquired variant. It was acquired in 8 (19.5%) patients, all of whom had Ph+B-ALL. Other known ALL-related variants, including CDKN2A (5 patients, 12.2%), CDKN2B (4, 9.8%), IKZF1 (2, 4.9%), and TP53 (2, 4.9%), were also acquired with various frequencies. The cell cycle pathway was the most significantly enriched among acquired variants, with 32 acquired variants identified in 17 patients (P = 3.46 × 10-6, false discovery rate = 7.33 × 10-4). Genes overlapping between this pathway and the acquired variants were ABL1, ATM, ATRX, CDKN2A, CDKN2B, EP300, and TP53. Notably, even when ABL1 was excluded, the cell cycle pathway remained the most significantly enriched, with 21 acquired variants in 12 patients (P = 4.69 × 10-3). Survival analysis showed that the mere presence of any acquired variant (irrespective of gene or pathway) did not have a significant impact on post-relapse survival (PRS) (hazard ratio [HR] = 0.642, P = 0.384). Acquired variants in ABL1 were also not significantly associated with PRS, possibly due to the availability of resistance-circumventing drugs (HR = 1.74, P = 0.416). In contrast, acquired variants in the cell cycle pathway were significantly associated with shorter PRS (HR = 2.72, P = 0.033). Conclusions: Cell cycle-related genetic variants, not limited to ABL1, are frequently acquired at relapse of ALL and are associated with worse post-relapse prognosis. Our findings suggest that genetic evolution is common at relapse and that genetic re-testing after relapse can be helpful for stratifying post-relapse prognosis.