Background and Purpose: Despite their use to treat cancers with specific
genetic aberrations, targeted therapies elicit heterogeneous responses.
Sources of variability are critical to targeted therapy drug
development, yet there exists no method to discern their relative
contribution to response heterogeneity. Experimental Approach: We use
HER2-amplified breast cancer and two agents, neratinib and lapatinib, to
develop a platform for dissecting sources of variability in patient
response. The platform comprises four components: pharmacokinetics,
tumor burden and growth kinetics, clonal composition, and sensitivity to
treatment. Pharmacokinetics are simulated using population models to
capture variable systemic exposure. Tumor burden and growth kinetics are
derived from clinical data comprising over 800,000 women. The fraction
of sensitive and resistant tumor cells is informed by HER2
immunohistochemistry. Growth rate-corrected drug potency is used to
predict response. We integrate these factors and simulate clinical
outcomes for virtual patients. The relative contribution of these
factors to response heterogeneity are compared. Key Results: The
platform was verified with clinical data, including response rate and
progression-free survival (PFS). For both neratinib and lapatinib, the
growth rate of resistant clones influenced PFS to a higher degree than
systemic drug exposure. Variability in exposure at labeled doses did not
significantly influence response. Drug potency strongly influenced
responses to neratinib. Variability in patient HER2 immunohistochemistry
scores influenced responses to lapatinib. Exploratory twice daily dosing
improved PFS for neratinib but not lapatinib. Conclusion and
Implications: The platform can dissect sources of variability in
response to target therapy, which may facilitate decision-making during
drug development.