Abstract 4980: T-cell receptor clonotyping and immune infiltrate quantification with whole transcriptome RNA-seq from FFPE

Abstract

Abstract Background: Even within patients with high tumor mutational burden (TMB) and/or PD-L1 expression, objective response rates to checkpoint therapies remain below 50% (e.g. Checkmate 227). Identifying predictive biomarkers for immune checkpoint inhibition therapy is an active area of research. Tumor infiltrating lymphocytes (TILs) show potential as a predictive biomarker for immune-oncology therapies. Rearrangement of the T-cell receptor (TCR) complementarity-determining region 3 (CDR3) serves as a marker for T-cell clones, and can be detected by RNA-seq of the region. The standard mechanism of tissue storage, formalin fixation and paraffin embedding (FFPE), degrades nucleic acids and makes sequencing more challenging. We present detection of TCR CDR3 of TILs from FFPE RNA-seq as a potential biomarker. Methods: FFPE Tumors from 65 head and neck squamous cell cancer (HNSCC), 233 colorectal carcinoma (CRC), and 217 non-small cell lung cancer (NSCLC) patients were profiled using deep full-transcriptome RNA-seq. Patients were assigned a TCR score by first aligning RNA-seq reads to TCRα and TCRβ CDR3 sequences, then counting overall CDR3 read support. Gene expression profiles were also used to predict consensus molecular subtype (CMS) and estimate immune cell abundance. To assess tumor sample purity, deep whole genome (N=244) or whole exome (N=271) DNA sequencing was performed on both tumor and normal blood samples from each patient. TMB was calculated directly by counting somatic non-synonymous exonic mutations. Results: Immune populations revealed high correlation (Pearson's r = .84, p = 3.6e-169) between TCR CDR3 counts and independent gene expression-based estimates of T-cell activity. Using the CMS of colon cancer samples, we found that CMS2 patients had a significantly lower average number of reads aligned to CDR3 (46.5 vs 91.6, p = 1.5e-7) which agrees with previous reports of low immune activity in that subtype. We compared CDR3 read count against TMB and found significant correlation (Spearman's rho = 0.34, p = 0.0048) in head and neck tumors where the median TMB was 98, but no association in lung or colon tumors where TMB is higher (median TMB 169 and 132 respectively). Finally, tumor purity estimated via DNA-seq was anti-correlated with CDR3 read count (Spearman's ρ = -0.39, p = 3.3e-21). Conclusions: TCR CDR3 read support from bulk RNA-seq on 515 solid tumor FFPE samples provides a related but distinct measure of T-cell infiltration to gene expression-based estimates. TMB and TCR support are generally independent biomarkers, especially in those tissues with higher average mutation rate. TCR support has potential to complement TMB as a predictive biomarker of immune checkpoint therapy response. Citation Format: Andrew J. Sedgewick, Jacob J. Adashek, Christopher W. Szeto, Charles J. Vaske, Philippe E. Spiess, Stephen C. Benz. T-cell receptor clonotyping and immune infiltrate quantification with whole transcriptome RNA-seq from FFPE [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4980.