Abstract CT112: Exposure-response analysis of pembrolizumab in patients with advanced melanoma and non-small cell lung cancer enrolled in KEYNOTE-001, -002, and -006

• Published in: Cancer research (Chicago, Ill.) Ill.), 2016-07, Vol.76 (14_Supplement), p.CT112-CT112
• Main Authors: Chatterjee, Manash S., Turner, David C., Ahamadi, Malidi, de Greef, Rik, Freshwater, Tomoko, Mayawala, Kapil, Dong, David, Stone, Julie, de Alwis, Dinesh, Kondic, Anna
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2016-CT112

Background: The anti-PD-1 monoclonal antibody pembrolizumab has demonstrated durable antitumor activity against several advanced malignancies and is generally well tolerated. In the KEYNOTE-001, -002, and -006 trials, patients with advanced/metastatic melanoma and non-small cell lung cancer (NSCLC) were given pembrolizumab doses ranging from 2 mg/kg Q3W to 10 mg/kg Q2W. The objective of this current analysis was to explore the relationship between exposure to pembrolizumab and safety/efficacy in patients with advanced melanoma or NSCLC to help select an appropriate therapeutic dose. Methods: Exposure-response analyses (using graphical exploration and nonlinear mixed effects modeling) of tumor size data were used to characterize the relationship between pembrolizumab exposure and tumor size reduction. Data for these analyses were derived from 1366 patients with melanoma and 496 with NSCLC; exposure-response analyses were performed separately for each indication. Tumor size, defined as the sum of longest dimensions (SLD) of target lesions, was the response readout and individual steady-state area under the curve (AUC) estimates from a population PK model were used as an integrated exposure measure across all concentration data for each patient. Additionally, exposure-adverse event (AE) logistic regression was performed on the integrated dataset, focusing on AEs of special interest (AEOSIs), defined as a broad category of potentially immune-related AEs. Results: The graphical analysis for both melanoma (stratified by IPI pretreatment status) and NSCLC (stratified by PD-L1 expression status) identified an almost flat relationship between exposure and change in tumor size from baseline at 24 wk (melanoma) or 18 wk (NSCLC), with substantial overlap in confidence intervals. In agreement with the exploratory graphical analyses, individual pembrolizumab exposures showed a small and statistically insignificant influence on the final model estimated tumor decay parameter for both melanoma (slope = 0.131, P = 0.20 for IPI-naive; and slope = 0.1, P = 0.25 for IPI-experienced), and NSCLC (slope = 0.196, P = 0.54) patients. Clinical trial simulations, using the tumor size model to normalize for covariates, predicted little variation and considerable overlap in the confidence intervals for response across the dose regimens studied for both populations, confirming lack of clinically meaningful differences between 2 mg/kg and 10 mg/kg. The exposure-AEOSI analysis did not id