142: Population Pharmacokinetic-Pharmacodynamic (PKPD) Modeling Analyses of a Novel Anti-HER2 Antibody-Drug Conjugate (PF-06804103) in Patients With Human Epidermal Growth Factor Receptor 2 (HER2)-Expressing Solid Tumors

Statement of Purpose: PF-06804103 is an anti-HER2 immunoglobulin G1 antibody-drug conjugate comprising an anti-HER2 monoclonal antibody (trastuzumab) conjugated to the cytotoxic agent Auristatin-0101. In a multicenter, open-label, Phase 1 study (NCT03284723), PF-06804103 demonstrated antitumor activity in patients with HER2-expressing advanced/unresectable or metastatic breast cancer or gastric/esophagogastric junction cancer at doses ranging between 0.15-5 mg/kg. Grade ≥2 peripheral neuropathy (PN) was the most common adverse event leading to dose modifications and treatment discontinuation. The current analyses aimed to develop a tumor growth inhibition (TGI) model to characterize the relationship between PF-06804103 exposure and tumor dynamics, and a time-to-event (TTE) model to characterize the relationship between PF-06804103 exposure and Grade ≥2 PN risk.

Description of Methods & Materials: Post hoc empirical bayes estimates from the population pharmacokinetic model were used to predict individual patient exposures with dosing records. Radiological tumor assessments with computed tomography or magnetic resonance imaging scans, bone scan and/or bone x-rays for all known or suspected disease sites were conducted at baseline, every 6 weeks starting at Cycle 3 during treatment until disease progression and end of treatment. PN was defined and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.03). For both analyses, various structural models were explored, and potential covariates were selected and tested for significance by stepwise covariate modeling. Simulations of the final models were used to support dose selection. NONMEM v7.4.3, Perl-speaks-NONMEM v4.9.0, and R v3.6.1 were used for modeling, data processing, graphics and table generation.

Data & Results: An exposure-dependent first-order treatment effect and resistance TGI model, including the parameters for tumor cell growth rate (KG), death rate (KD) and drug-resistance (λ), characterized the tumor data well. The estimated KG, KD and λ were 0.0143 month-1, 0.0298 month-1/(µg/mL) and 1.10 month-1, respectively. Baseline body weight (BWT) and baseline tumor burden were significant covariates on KG. Simulations assuming no dose reductions or interruptions recommend a starting dose regimen no less than 3 mg/kg Q3W to provide an improved overall response rate over the standard of care treatments. A TTE model for Grade ≥2 PN was best described with a linear drug concentration-effect relationship and Weibull distribution hazard function. Patients with higher BWT and history of PN were found to have increased risk for Grade ≥2 PN. The predicted median time to first Grade ≥2 PN was 16.6 weeks, which is consistent with the observed data. Simulations project that both utilizing a lower starting dose of 3 mg/kg versus 4 mg/kg and capping the upper dose limit for heavier patients would have, at best, a moderate impact on time to first Grade ≥2 PN.

Interpretation, Conclusion or Significance: The TGI model suggests patients with higher BWT and patients with higher baseline tumor burden had higher tumor cell growth rate. The TTE model suggests the risk of PN was increased with PF-06804103 exposure, higher BWT and history of PN. A starting dose of 3 or 4 mg/kg was recommended for dose expansion. These analyses demonstrate a quantitative approach to optimize the dosing regimen of PF-06804103 for safety and efficacy.

Disclosures: All authors currently or previously were employees of Pfizer Inc and owned Pfizer stock.