Towards early integrated mechanism-based prediction of clinical outcome and cost-effectiveness in castration-resistant prostate cancer
J.G.C. van Hasselt (1,2)*, A. Gupta (3), Z. Hussein (3), J.H. Beijnen (1,2,4). J.H.M. Schellens (1,2,4), A.D.R. Huitema (1,2)
1. Department of Clinical Pharmacology, The Netherlands Cancer Institute, Amsterdam, the Netherlands; 2. Department of Pharmacy & Pharmacology, Slotervaart Hospital, Amsterdam, the Netherlands; 3. Eisai Limited, Hatfield, United Kingdom; 4. Department of Pharmaceutical Sciences, Division of Clinical Pharmacology and Pharmacoepidemiology, University, Utrecht, the Netherlands; *Current affiliation: Division of Pharmacology, Leiden University, Leiden, Netherlands.
Objectives: Disease progression (DP) of prostate cancer (PC) is characterized by rising levels of prostate-specific antigen (PSA) [1]. Models linking DP with clinical outcome models (CO) have been proposed for early prediction of efficacy in oncology [2]. However, predicted efficacy will also be influenced by toxicity profiles and dose reduction strategies. Cost-effectiveness is becoming increasingly important, although such analyses are mostly only performed in late-phase development. The objective of this analysis was to develop an integrated modeling framework for castration-resistant PC (CRPC) in patients treated with eribulin, which may serve as a proof-of-concept example for prediction of clinical utility and cost-effectiveness during early clinical development.
Methods and results: Based on historical data of patients receiving eribulin we developed a dynamic K-PD DP model for PSA. Model parameters were estimated with adequate precision (relative standard error, RSE <44.6%). Subsequently, a parametric Weibull model was developed which included treatment-related (time-to-PSA nadir), disease-specific (PSA growth rate), and patient-specific (ECOG score) covariates. A semi-physiological population PK-PD model for eribulin-induced neutropenia was then implemented, which included several patient-specific covariates. Markov-transition models successfully described the time-course of toxicity grades for fatigue, anemia, diarrhea and additional AEs. Another Markov-transition model was developed to describe the ECOG time-course as a surrogate for quality of life. A >50% PSA inhibition was related to a proportional decrease of increasing ECOG-transition probabilities (0.704, RSE 36%). Finally, a log-logistic survival model best described dropout.
Subsequently, an integrated simulation framework combining the different sub-models was implemented. Different simulation scenarios were defined to evaluate the impact alternative dose regimens, disease progression criteria, dose reduction protocols and specific patient-characteristics.
Conclusion: An innovative integrated simulation framework was developed that can be applied for early assessment of clinical utility and cost-effectiveness. The framework can potentially be used to support early trial design in CRPC, but also may also serve as a proof-of-concept example that can be applied for model-based drug development in oncology.
References
[1] Scher HI, Halabi S, Tannock I, et al. (2008) Design and end points of clinical trials for patients with progressive prostate cancer and castrate levels of testosterone: recommendations of the Prostate Cancer Clinical Trials Working Group. J Clin Oncol 26:1148–59.
[2] Claret L, Girard P, Hoff PM, et al. (2009) Model-based prediction of phase III overall survival in colorectal cancer on the basis of phase II tumor dynamics. J Clin Oncol 27:4103–8. doi: 10.1200/JCO.2008.21.0807