μPharma: A microfluidic, AI-driven pharmacotyping platform for single-cell drug sensitivity prediction in leukemia.

BACKGROUND: Pharmacotyping, the ex vivo measurement of tumor cell responses to drugs, is particularly important for cancers lacking actionable genomic markers. However, current pharmacotyping methods are not clinically feasible due to prolonged drug incubations (days to weeks), extensive manual handling, and analytical limitations, including overlooking single-cell characteristics. Addressing these hurdles is critical for pediatric T cell acute lymphoblastic leukemia (T-ALL), an aggressive cancer with limited therapeutic options. METHODS: We developed μPharma, a pharmacotyping platform that predicts single-cell drug sensitivity without direct drug exposure by quantifying pretreatment biomarkers associated with therapeutic response. μPharma integrates an automated digital microfluidic immunofluorescence assay, optimized for suspension cells, with machine learning models trained on comprehensive single-cell features. We validated μPharma using T-ALL cell lines and patient-derived xenografts, predicting sensitivity to dasatinib and venetoclax by quantifying their target proteins, LCK and BCL2, respectively, including protein expression, phosphorylation status, spatial distribution, and cellular morphology. FINDINGS: We confirmed that phospho-LCK is predictive of dasatinib sensitivity, consistent with prior studies, and identified phospho-BCL2 as a previously unreported biomarker for venetoclax sensitivity. Integrating multiple biomarkers into machine learning models significantly enhanced predictive accuracy compared to single-marker analyses. Key informative features included spatial protein distribution and integrated protein-morphology metrics. Additionally, single-cell analysis revealed distinct cell subpopulations, suggesting intratumor heterogeneity in drug responses. CONCLUSIONS: μPharma provides rapid (4-h assay), accurate, and automated prediction of drug sensitivity at single-cell resolution using minimal clinical samples, potentially enabling same-day precision oncology decision-making. FUNDING: This work was supported by institutional start-up funds from the University of Utah, including internal supplements provided through the Immunology, Inflammation & Infectious Disease (3i) Initiative and the Diabetes & Metabolism Research Center (DMRC).