Pharmacostatistical models can quantify different relationships and improve decision making in personalized medicine and drug development. Our objectives were to develop models describing non–small-cell lung cancer (NSCLC) dynamics during first-line treatment with erlotinib, and survival of the cohort.Methods:
Data from patients with advanced NSCLC (n = 39) treated first-line with erlotinib (150 mg/day) were analyzed using nonlinear mixed effects modeling. Exposure-driven disease-drug models were built to describe tumor metabolic and proliferative dynamics evaluated by positron emission tomography (PET) using 2′-deoxy-2′-[18F]fluoro-D-glucose (FDG) and 3′-[18F]fluoro-3′-deoxy-L-thymidine (FLT), respectively, at baseline, weeks 1 and 6 after starting erlotinib treatment. A parametric time-to-event model was built to describe overall survival (OS). Demographics, histology, mutational, smoking, and baseline performance statuses were tested for their effects on models developed, in addition to tumor dynamics on survival.Results:
An exponential relationship described progression, and a concentration-driven drug effect model described erlotinib effect. An activating epidermal growth factor receptor (EGFR) mutation increased the drug effect as assessed using FDG-PET by 2.19-fold (95% confidence interval [CI]:1.35–4.44). An exponential distribution described the times-to-death distribution. Baseline FDG uptake (p=0.0005; hazard ratio [HR] =1.26 for every unit increase, 95%CI: 1.13–1.42) and relative change in FDG uptake after 1 week of treatment (p=0.0073; HR=0.84 for every 10% drop, 95%CI: 0.71–0.91) were significant OS predictors irrespective of the EGFR mutational status. FLT-PET was statistically less significant than FDG-PET for OS prediction.Conclusion:
Models describing tumor dynamics and survival of advanced NSCLC patients first-treated with erlotinib were developed. The impacts of different covariates were quantified.