High-Dimensional Single-Cell Analysis of Clinical-Grade Flow Cytometry Explains the Unpredictable Complex Shifts of Patient-Specific Immunophenotypes during B-ALL Relapse Progression

Relapsed precursor B-cell ALL (B-ALL) commonly evolves from minor diagnostic clones, but their early phenotypic characterization remains unachieved despite abundant multi-omics data. Immunophenotyping, an essential diagnostic and therapy-response monitoring tool, identifies the patient-specific expression pattern of bulk leukemia. At time of relapse, most B-ALL immunophenotypes have undergone substantial and seemingly random multi-directional modulations of their diagnostic expression pattern. For decades, these unpredictable immunophenotypic shifts at time of relapse, observed across different methods of antigen expression assessment, have prevented a deeper understanding of phenotypically-defined leukemia progression biology. Addressing this limitation, we applied unsupervised high-dimensional computational analysis of clinical-grade flow cytometry to dissect the intra-leukemic phenotypic heterogeneity at the single-cell level in longitudinally collected B-ALL patients and matched patient-derived xenografts. Our results provide AI-guided and clinically validated evidence that the observed immunophenotypic shifts during disease progression did not result from antigen expression fluctuations, but from enrichment of distinct phenotypically stable subpopulations. As our study identifies patient-specific subpopulation dynamics during disease evolution, it achieves immunophenotypically-defined leukemia progression assessment, which addresses an important unmet clinical and translational need. In each progression series, population dynamics followed a trajectory towards relapse-dominating subpopulations when selective pressures, such as xenotransplantation or in vivo chemotherapy, were applied, often from very minor abundance levels at diagnosis. Each time, the changes in relative proportions of subpopulations explained the observed immunophenotypic shift at the bulk-level. Overcoming decades-old challenges, our findings provide a new conceptual approach to investigate the role of intra-leukemic phenotypic heterogeneity in B-ALL progression to identify treatment-refractory phenotypes, which could significantly impact patient-care, inform precision-medicine options, and enhance relapse-modelling.

No relevant conflicts of interest to declare.