Abstract 4282: Total RNA sequencing of frozen biopsies combined to k-mer sequence analysis identify new hallmarks of immune checkpoint therapy response in melanoma

Abstract Introduction: Predicting response to immune checkpoint inhibitors (ICI) is key for stratifying patients and identifying factors of resistance. The main recognized ICI response biomarkers in melanoma are T-cell gene expression signatures, PD-L1 expression and tumor clonal mutational burden, all of which are specific of responding tumors but nevertheless not robust enough to guide therapeutic strategies. Biomarkers of progression (eg. TGF-B) are scarce and inconsistent across studies. All current gene expression-based biomarkers of ICT-response are based on formalin-Fixed paraffin-embedded (FFPE) samples, sequenced using RNA capture or polyA+ libraries. This technique is known to incompletely represent RNA populations. Patients, Material, Methods: Here we introduce the first total RNA-seq dataset from fresh frozen biopsies collected from 80 advanced melanoma patients, prior to treatment with anti PD-1 or combined anti-PD-1/anti-CTA4 therapy. The cohort included 43 responders and 37 non-responders based on RECIST best objective response. To ensure all possible RNA biomarkers are captured, we applied a k-mer based computational protocol that extracts all subsequences of fixed size (31 nt) from the raw RNA-seq reads and performs differential expression statistics on k-mers. This identifies response-specific RNA sequences independently of annotated genes, enabling identification of novel RNAs and RNA isoforms. Results: In responding patients, this analysis identified a large number of immunoglobulin and HLA mRNA fragments, consistent with an immune-active microenvironment. However, the most remarkable finding was the specific expression in non-responding tumors of centromeric satellite repeats (HSAT-II and ALR elements) and of a set of intergenic loci which independently described the progressor population. Altogether, total RNA-seq biomarkers were found in most non-responding patients, providing the basis for a strong predictor of resistance. This contrasted with results in FFPE/capture seq data that showed few notable k-mers in the progressor group, due to the incomplete RNA sequencing. Our approach revealed a number of other unexpected markers, including a set of circular RNAs in responders and a mutant of the PRAME tumor-specific antigen in progressors. Our main finding is that the combination of frozen sample conservation and total (ribodepleted) RNA sequencing reveals a unique source of non-coding RNAs produced from pervasive transcription of intergenic and repeated regions, likely associated to chromatin derepression, that are not detected by the usual FFPE/capture seq methods. Conclusion: We argue that these novel RNA sequences will be instrumental in the development of improved ICI-response predictors and a better understanding of resistance to ICI. Citation Format: Hugues Hermann, Séverine Roy, Sandrine Agoussi, Emilie Routier, Céline Boutros, Djaouida Belkadi, Haoliang Xue, Antoine Lainé, Stephan Vagner, Daniel Gautheret, Caroline Robert. Total RNA sequencing of frozen biopsies combined to k-mer sequence analysis identify new hallmarks of immune checkpoint therapy response in melanoma. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4282.