Identification of Potential Predictive Biomarker Candidates Through Strategic Analysis of Cytokine Profiles Across Multiple Anti-PD-1 Clinical Trials

• Activity Number: 170 - SPEED: Biopharmaceutical Methods and Application I, Part 1
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Jeea Choi* and Ying Amanda Wang and John Millholland and Albert Reising and Jan Christoph Brase and Xiaoshan Wang and Connie Wong and Kitty Wan and Yiqun Yang and Gullu Gorgun and Parul Patel and Hemant Patel
• Companies: Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis and Novartis
• Keywords: Biomarkers; anti-PD-1; Cytokines; Immuno-Oncology therapy; Melanoma; neuroendocrine tumor

Abstract

Cytokines are essential signaling molecules secreted by immune cells to modulate immune responses. Monitoring cytokine changes in patients treated with IO therapies may enable a better understanding of the drug’s immunomodulation mechanisms and identifying potential biomarkers. We have developed a statistical strategy to enable systematic analysis of cytokine profiling data across multiple anti-PD1 clinical trials with different study designs and sample size. This strategy includes a linear mixed model with an effort to mitigate challenges, such as repeated measurement data structure, imbalance due to missingness, small sample size, and heterogeneity. We performed univariate as well as multiple marker analysis to evaluate associations with tumor reduction. In addition, we provide shiny applications for within and across study data visualization to maximize the value of the data. Cytokine data were collected in multiple clinical trials including Melanoma cancer patients and NET cancer patients. From the analysis, we observed that early increase of interferon gamma and decrease of C-reactive protein appear to trend with tumor shrinkage in the Melanoma study.