Abstract 5904: Intra-tumor microbes identified by RNAseq associated with response to immune checkpoint blockade in metastatic melanoma

Abstract

Abstract The tumor microbiome has recently been shown to play a key role in the context of oncogenesis, cancer immune phenotype, cancer progression and treatment outcomes in a variety of cancers. We investigated the possible associations between tumor microbiome and successful treatment outcomes with immune checkpoint blockade (ICB) in patients with metastatic melanoma. We evaluated RNAseq from tumor samples, collected prior to the start of treatment with ICB, from 71 patients with metastatic melanoma. Samples were provided by eight members of the Oncology Research Information Exchange Network (ORIEN). Non-response was determined as change in treatment after less than 12 months. Patients maintaining the same treatment regimen for greater than 12 months were classified as responders. We applied our custom tool, {exotic} (Exogenous sequences in Tumor and Immune Cells), to carefully identify non-human sequences within the RNAseq data. After filtering reads aligning to the human reference genome, reads were further filtered of common laboratory contaminants, taxa inversely correlated with input RNA quantity, and taxa frequently found in the negative controls of microbiome experiments. A differential abundance analysis was performed on the response groups at every taxonomic level utilizing DESeq2. We calculated expression signatures using {tmesig}, and related them to ICB response using {IOSig}. We observed significantly enriched taxa (p-value < 0.05) with a high (>1.00) fold-difference in abundance between responders and non-responders found within the tumor RNAseq data, including Fusobacterium nucleatum and several viruses in responders, and Delftia lacustris and Fungi in non-responders. These microbes were associated with immune cell expression signatures, including Th17 cells and CD8+ T-cells. We calculated the gene expression scores of 30 signatures with literature precedence for the ability to predict ICB treatment outcomes in melanoma. The receiver operator characteristic (ROC) curve of the random forest classification model for prediction of response to ICB using the combined expression signature scores resulted in an AUC of 0.8750. Combining expression signature scores with microbe relative abundances at the genus level improved the ability to predict ICB response (AUC = 0.8958). Combining tumor expression signatures with curated tumor microbiome relative abundances improves the performance of predictive models for treatment outcomes with ICB in melanoma. Citation Format: Caroline E. Wheeler, Samuel Coleman, Rebecca Hoyd, Afaf Osman, Louis Denko, Aik Choon Tan, Daniel Spakowicz, Ahmad Tarhini. Intra-tumor microbes identified by RNAseq associated with response to immune checkpoint blockade in metastatic 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 5904.