Abstract 4137833: The plasma cell-free transcriptome non-invasively profiles the immune response and myocardial damage in immune checkpoint inhibitor-induced myocarditis.

Introduction: Immune checkpoint inhibitor-induced myocarditis (ICI-m) is a severe complication of cancer immunotherapy, marked by cardiac inflammation. Despite its high mortality rate, diagnosing ICI-m remains challenging due to the lack of sensitive and non-invasive diagnostic modalities. Plasma cell-free mRNA (cf-mRNA) has recently been highlighted as a promising tool for non-invasive tissue profiling. Hypothesis: The plasma cf-mRNA profile contains disease-specific immune and tissue transcriptomic signatures of ICI-m that can be leveraged for disease-specific precision diagnostics. Aims: To establish and validate a novel cf-mRNA bench and bioinformatic pipeline for noninvasive profiling of the immune and tissue landscape of patients with ICI-m. Methods: We isolated, DNase-treated, and sequenced bulk cfRNA from frozen plasma of patients with ICI-treated cancer patients with no autoimmunity (group A, n=5), extracardiac autoimmunity (group B, n=7), and ICI-m with or without extracardiac autoimmunity (group C, n=10). Publicly available healthy control cfRNA was downloaded (n=30) as external controls. We used cell type-specific gene panels curated from single-cell RNA-sequencing datasets of circulating immune cells and cardiac cells for cellular deconvolution of the cf-mRNA. Signature scores of the sum of the counts of each cell type-specific gene were used to compare cell type contributions between patient groups. Results: Cancer and inflammatory gene pathways were significantly enriched in the cf-mRNA of all ICI-treated cancer patients. Cardiac pathways and conduction genes were significantly enriched in group C vs. A and B patients. Signature scores of Temra CD8+ T cells and cardiomyocytes, and gene expression of CCL5 and MYH6 were elevated in group C patients. Our custom ICI-m classifier containing 3 cardiac- and 3 immune cell-specific genes differentiated group C from A (AUC 0.903, 95% CI 0.858-0.903) and B patients (AUC 0.983, 95% CI 0.970-0.983) effectively, outperforming a diagnostic classifier obtained from unsupervised feature selection from all differentially expressed genes. Conclusions: Using our novel plasma cf-mRNA platform, we developed a diagnostic classifier that captures the disease-specific immune and tissue transcriptomic signatures of ICI-m, highlighting the advantages of this disease-specific approach over conventional biomarker discovery, and demonstrating the broader implications of the platform for the field of precision diagnostics.