Tong Lu, Dan Lu, Bert L. Lum, Mark Stroh, Priya Agarwal , Nalin Tikoo , Jin Y. Jin, Amita Joshi
Genentech, Inc.
Objectives: Exposure-response (E-R) analysis is extensively used in drug development and regulatory decision-making, and is still a developing field in oncology given many unique challenges. The survey will provide an overview of methodologies for E-R analysis in FDA oncology review and other E-R methodologies, and discuss pros and cons.
Methods: 28 New Molecular Entity (NME) in oncology approved by FDA from 2010 to 2013 were surveyed based on “clinical pharmacology and biopharmaceutics review” that publicly available [1]. Both exposure-efficacy (E-E) and exposure-safety (E-S) analyses were reviewed. FDA’s approach was chosen when different from sponsor. Results are summarized as pie charts and table with pros and cons as well as case examples.
Results: In total, there were 37 E-E and 32 E-S analyses performed for the 28 NMEs. Typical E-E analysis included Kaplan-Meier (K-M) plots & Cox proportional hazard (Cox PH) model of survival data (19 cases), and logistic plot & regression model for binary data (11 cases). Compared to K-M plots stratified by exposure, which can be influenced by confounding factors, Cox PH model assess the E-E relationship by adjusting for those factors, and has been combined with K-M plots in majority of cases (15/19). For E-S analysis, logistic plot & regression model played a central role (21 cases) given most safety data were treated as binary endpoints. Box plots can be used to visualize E-R relationship for categorical data; but it might fail to identify one that is not sufficiently steep; logistic plot is more sensitive to detect E-R relationship, but both plots can be influenced by confounding factors, which has been considered by logistic regression model. Additional methodologies that were not in the review include longitudinal PK/PD model for biomarker or tumor response data, longitudinal and repeat time-to-event model for categorical data, parametric model and case-matching analysis for survival data, etc. These methodologies may be more robust to refine E-R relationship by integrating data in longitudinal fashion; however, comparisons of these methodologies for E-R assessment in oncology need further evaluation.
Conclusions: The survey reviewed and summarized the E-R analysis for recently approved oncology drugs by FDA, together with other E-R methodologies. It provides a framework for appropriate application and further advancement of E-R methodology to support dose justification and optimization especially in oncology.
References:
[1] http://www.accessdata.fda.gov/scripts/cder/drugsatfda/. Clinical Pharmacology & Biopharmaceutics Reviews (Summary Basis of Approval)
[2] Mehrotra S, Florian J Jr, Gobburu J. Don’t get boxed in: commentary on the visual inspection practices to assess exposure-response relationships from binary clinical variables. J Clin Pharmacol. 2012;52(12):1912-7
[3] Paule I, Girard P, Freyer G, Tod M. Pharmacodynamic models for discrete data. Clin Pharmacokinet. 2012 ;51(12):767-86
[4] Mould DR, Walz A-C, Lave T, Gibbs JP and Frame B. Developing Exposure/Response Models for Anticancer Drug Treatment: Special Considerations. CPT: PSP. 2015; 4(1): 1-16
Reference: PAGE 24 (2015) Abstr 3609 [www.page-meeting.org/?abstract=3609]
Poster: Methodology - Other topics