Abstract 5550: Translational biomarkers for SAR439459, an anti-TGFβ antibody for cancer immunotherapy

Abstract

Abstract TGFβ is a potent immunosuppressive cytokine, acting on multiple cell types of both the innate and adaptive immune systems. Emerging preclinical and clinical data implicate TGFβ in tumor immune evasion, poor prognosis and resistance to PD1/PDL1 directed checkpoint therapy. Thus, neutralization of TGFβ is suggested to relieve immunosuppression via mechanisms that are distinct yet complementary to checkpoint inhibitors such as PD1. SAR439459 is a potent human pan-neutralizing anti-TGFβ antibody that has entered first-in-human studies in advanced solid tumors (NCT03192345*). I. Biomarkers for indication selection: A gene expression signature of TGFβ pathway activation was derived from analysis of TGFβ-stimulated versus naïve cancer cell lines. Queries of the signature against The Cancer Genome Atlas (TCGA) revealed that primary head and neck, ovarian, and colorectal cancers are enriched for activation. II. Biomarkers for patient selection: An analysis of the patients enriched for activation revealed that mesenchymal tumors predominate. Machine learning methods were applied to panels of gene expression data for colorectal (CRC), ovarian serous and head and neck squamous cell carcinoma to derive a two-gene biomarker for selection of patients with mesenchymal tumors. Implementation in a logistic regression model achieves good for prediction in CRC, and similar performance in other indications. Patient cohorts are enriched over three-fold for the mesenchymal subtype and false-negative rates are acceptable. Implementation of the two-gene biomarker on a low complexity platform and its validation on an independent panel of samples is described. III. Biomarkers for assessment of the tumor microenvironment: In partnership with NeoGenomics, the immune contexture of patient tumors was evaluated using MultiOmyx, a multiplex IHC assay, on CRC and melanoma. Multiplexing was conducted with 12 biomarkers (jointly describing 22 immune cell types) on single FFPE section from each tumor sample. Pilot studies included a range of inflammation to evaluate how well the analytics assess each tumor type and correlate to possible treatment effects. Statistical methods were developed to assess differences at the cell population level, including replicate concordance, volcano plots for analyses of variance, and correlation matrices. The MultiOmyx assay demonstrated excellent technical reproducibility and precision, a favorable dynamic range, inflammation status differences in select immune cells and regions of interest, and included both positive and negative correlations between cell populations. * https://clinicaltrials.gov/ct2/show/NCT03192345 Citation Format: Joachim Theilhaber, Jean Cavallo, Stephen L. Madden, Charlene Manning, Sherry Cao, Parminder Mankoo, Robert Pomponio, Hongjing Qu, Natalia Malkova, Gary Shapiro, Christopher Winter, Dmitri Wiederschain, Michele Sanicola-Nadel, Fangxian Sun, Tun Tun Lin, Richard C. Gregory, Jack Pollard. Translational biomarkers for SAR439459, an anti-TGFβ antibody for cancer immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5550.