Pharmacotypes across the Genomic Landscape of Pediatric Acute Lymphoblastic Leukemia and Impact on Treatment Response

Abstract

INTRODUCTION The current paradigm of risk stratification in childhood acute lymphoblastic leukemia (ALL) utilizes a combination of clinical features, leukemia somatic genomics and early treatment response as measured by minimal residual disease (MRD). Although ALL subtypes and MRD are strongly related to treatment outcomes, the pharmacological basis of their prognostic impact remains incompletely understood. Therefore, we sought to characterize pharmacotypes (drug sensitivities) across the complete spectrum of ALL molecular subtypes and determine their association with in vivo treatment response and survival outcomes. METHODS This was a prospective study of 805 children with newly-diagnosed ALL from Total Therapy XV, XVI, and XVII trials at St Jude Children's Research Hospital. We performed ex vivo pharmacotyping of 18 drugs on primary ALL diagnostic blasts to determine drug sensitivities (LC50 - dose required to kill 50% of leukemia cells). RNA-sequencing was performed to derive molecular subtype. We then evaluated the associations of drug LC50 with clinical features, ALL subtype, MRD, and event-free survival (Figure 1). RESULTS We analyzed 5,447 drug measurements across 23 ALL subtypes. LC50s of prednisolone, asparaginase, and mercaptopurine were associated with age, and/or with leukocyte count at diagnosis. Sensitivities to 14 of 18 drugs varied across ALL molecular subtypes. For cytotoxic agents, B-ALL subtypes with favorable prognosis (ETV6-RUNX1 and hyperdiploidy) exhibited greatest sensitivity to asparaginase and glucocorticoids, whereas high-risk subtypes (KMT2A, BCR-ABL1, and BCR-ABL1-like) were resistant. Novel subtypes DUX4 and ETV6-RUNX1-like ALL showed significant resistance to cytotoxic drugs. For targeted agents, hyperdiploid ALL was most sensitive to venetoclax, whereas TCF3-PBX1, BCR-ABL1, and T-ALL were dasatinib-sensitive. The poor-prognosis ETP-ALL was resistant to most cytotoxic drugs compared to T-ALL. Although BCR-ABL1, BCR-ABL1-like, and CRLF2 are closely related transcriptionally, these three subtypes had distinctive drug sensitivity profiles. In B-ALL, sensitivities of prednisolone, asparaginase, dexamethasone, cytarabine and thiopurines were positively correlated with MRD. By contrast, these drugs were not correlated with MRD in T-ALL. Instead, sensitivities of dasatinib and nelarabine correlated positively, and venetoclax negatively, with MRD in T-ALL. Finally, we classified patients into six functional clusters based on ALL pharmacotypes, each with distinct patterns of somatic genomic aberrations and MRD response. Drug sensitivity cluster was significantly associated with event-free survival, even after adjusting for MRD. Importantly, pharmacotyping identified a T-ALL subset with poor prognosis under conventional chemotherapy but also with unique sensitivity to targeted agents, pointing to therapeutic intervention to improve survival in this group. CONCLUSIONEx vivo drug sensitivities vary widely across ALL subtypes and strongly influence in vivo MRD. Our study comprehensively described pharmacological heterogeneity of ALL and the association of drug sensitivities with survival outcomes, providing insight into the pharmacological basis of inter-patient variability in ALL treatment outcomes and highlighting opportunities for individualizing therapy for this blood cancer. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal