MRD Xenotransplantation Prospectively Identifies Treatment-Selected Acute Lymphoblastic Leukemia Subpopulations with Relapse-Initiating Potential

Despite complete remissions being achieved in most newly diagnosed acute lymphoblastic leukemia (ALL) patients, relapse remains a significant clinical challenge. Interrogation of matched ALL samples has provided insights into the cellular characteristics and evolutionary trajectory of relapse-fated clones, but requires blasts obtained at time of relapse to provide a reference point for the retrospective analysis. Despite the strong prognostic value of minimal residual disease (MRD), MRD-positive samples have been under-utilized for investigating disease progression as the low leukemia burden presents significant challenges to the evaluation of the residual blasts. To overcome this limitation, we performed patient-derived xenografting (PDX) with MRD samples to enable deeper characterization of residual blasts, including functional assessment of their leukemia propagating ability.

In marked contrast to a previous study in adult ALL, we achieved engraftment of MRD blasts from 5 of 7 pediatric ALL patient samples, obtained at the end of induction (EOI) therapy, by intravenous injection into non-conditioned NSG mice. MRD blasts expanded with patient-specific kinetics, reaching overt leukemia in recipients of three of the samples and remaining at MRD levels in recipients of the remaining two. The absence of normal human hematopoiesis in the MRD-engrafted mice (termed EOI-PDX), combined with the expansion of residual blast numbers, enabled clear phenotypic definition of therapy-selected, leukemia-propagating subpopulations, including those present at <1%. Subsequent reanalysis of diagnosis (DX) and EOI clinical samples for the identified subpopulations exposed a treatment response-mediated disruption of the established population hierarchy that explained the phenotypic shifts observed during disease progression. The relevance of the subpopulations identified as potential leukemia drivers was confirmed by their dominance in subsequent clinical relapses. Furthermore, phenotype-based sorting of viable blast subpopulations enabled the identification and comparison of metabolic characteristics of treatment sensitive and resistant blasts.

This immunophenotypic study recapitulates the enrichment of minor clones during ALL progression described by retrospective sequencing studies, but it does so prospectively. By expanding on the quantitative and qualitative information that can be gathered from MRD samples, EOI-PDX identify the blast populations that are most likely to drive relapse, prior to clinical disease progression. By providing a platform to achieve early identification and ongoing monitoring of therapy-resistant subpopulations with strong leukemia-propagating capacity, this approach could refine treatment-response assessment and relapse-risk stratification. Further, by expanding the numbers of viable treatment-selected MRD blasts available for interrogation, EOI-PDX could enable early identification of vulnerabilities in chemotherapy-resistant blasts.