Characterizing Variants of Uncertain Drug Resistance (VUDRs) Using Quantitative Measurements at Clinical Exposures

Somatic missense mutations in oncogenes drive resistance to anticancer drugs, yet many variants remain clinically uncharacterized. Analogous to Variants of Uncertain Significance (VUS) in genetic disorders, these Variants of Uncertain Drug Resistance (VUDRs) in cancer lack the functional annotation needed to guide clinical management. Here, we applied a standards-driven deep mutational scanning platform that connects quantitative concentration-response measurements to human pharmacokinetics across 4922 missense variants (>96% coverage) at approved (400 mg QD) and investigational (400 mg and 500 mg BID) human doses of imatinib. Resistance phenotypes for 18 standards spanning 2 orders of magnitude of drug sensitivity showed strong quantitative performance and clinical concordance. Analyzing 257 clinical VUDRs, >10% conferred modest levels of resistance that might be overcome by dose escalation with generic imatinib instead of a branded alternative, potentially alleviating financial toxicity. Integration with global germline data also revealed ancestry-specific variants with the potential to create private VUDRs. These preclinical data establish the first generalizable framework for high throughput resistance variant classification directly tied to known human doses.