SpliceMutr Enables Pan-Cancer Analysis of Splicing-Derived Neoantigen Burden in Tumors

Abstract

Abstract: Aberrant alternative splicing can generate neoantigens, which can themselves stimulate immune responses and surveillance. Previous methods for quantifying splicing-derived neoantigens are limited by independent references and potential batch effects. Here, we introduce SpliceMutr, a bioinformatics approach and pipeline for identifying splicing-derived neoantigens from tumor and normal data. SpliceMutr facilitates the identification of tumor-specific antigenic splice variants, predicts MHC-binding affinity, and estimates splicing antigenicity scores per gene. By applying this tool to transcriptomic data from The Cancer Genome Atlas, we generate splicing-derived neoantigens and neoantigenicity scores per sample and across all cancer types and find numerous correlations between splicing antigenicity and well-established biomarkers of antitumor immunity. Notably, carriers of mutations within splicing machinery genes have higher splicing antigenicity, which provides support for our approach. Further analysis of splicing antigenicity in cohorts of patients with melanoma treated with mono- or combined immune checkpoint inhibition suggests that the abundance of splicing antigens is reduced post-treatment from baseline in patients who progress. We also observe increased splicing antigenicity in responders to immunotherapy, which may relate to an increased capacity to mount an immune response to splicing-derived antigens. We find the splicing antigenicity to be higher in tumor samples when compared with normal, that mutations in the splicing machinery result in increased splicing antigenicity in some cancers, and higher splicing antigenicity is associated with positive response to immune checkpoint inhibitor therapies. Furthermore, this new computational pipeline provides novel analytical capabilities for splicing antigenicity and is openly available for further immuno-oncology analysis.

Significance: SpliceMutr shows that splicing antigenicity changes in response to ICI therapies and that native modulation of the splicing machinery through mutations increases the contribution of splicing to the neoantigen load of some The Cancer Genome Atlas cancer subtypes. Future studies of the relationship between splicing antigenicity and immune checkpoint inhibitor response pan-cancer are essential to establish the interplay between antigen heterogeneity and immunotherapy regimen on patient response.

Figure 1

SpliceMutr pipeline. The SpliceMutr pipeline uses RNA-seq data from two groups of samples and evaluates the changes in SA between them. In this example, the alternative splicing analysis compares tumor and normal RNA-seq samples. The pipeline performs splicing-aware alignment, gene expression quantification, and HLA genotyping. The splicing-aware alignment splice junction counts are then input into LeafCutter to evaluate differential splice junction usage. The tumor-specific and normal-specific splice junctions undergo transcript formation, translation, and kmerization for MHCnuggets input through SpliceMutr, then are evaluated for genotype-specific MHC binders using MHCnuggets (22). MHC binders associated with normal-specific and tumor-specific peptides are then used to calculate a per-gene and sample SA metric dependent on the type of the sample. ${\mathit{\hspace{0.17em}}\mathit{S}\mathit{A}}_{\mathit{T}}\left(\mathit{G}\right)$ is calculated for tumor samples and ${\mathit{S}\mathit{A}}_{\mathit{N}}\left(\mathit{G}\right)$ is calculated for normal samples. SA, splicing antigenicity.

Figure 2

Comparison of predicted splicing antigens from SpliceMutr to proteomics datasets. A, The log₁₀-transformed total number of MHC-binding kmers found with and without reference kmer filtering in all samples of the breast cancer TCGA cohort (BRCA). B, The percentage of IEAtlas-validated immunogenic kmers between SpliceMutr with and without reference kmer filtering in all samples of the breast cancer TCGA cohort (BRCA).

Figure 3

SA by tissue type and in relation to the tumor mutational burden. A, The SA averaged across genes per sample for tumor and normal samples for each tumor type analyzed (Wilcoxon test, Cohens d). B, The SA averaged across all genes per sample for TMB=high and TMB-low samples for all analyzed TCGA samples (linear regression, Cohens d). **, P value ≤ 0.005; ***, P value ≤ 0.0005; ****, P value ≤ 0.00005. See TCGA cancer-type abbreviations in Supplementary Table S1. SA, splicing antigenicity.

Figure 4

Pan-cancer correlation analysis of per-sample SA with nonsilent mutations in splicing machinery coding genes. Distribution of SA per sample, averaged across genes, by mutations in the splicing factor gene SF3B1 in BRCA (A) or HNSC (B). *, P value ≤ 0.05. See TCGA cancer-type abbreviations in Supplementary Table S1. SA, splicing antigenicity.

Figure 5

Per-patient SA of ICI-treated patients with melanoma per treatment arm and response type after treatment. A, The per-patient pseudo purity compared with the mean SA averaged across genes by the response for all treatment arms. Kendall Tau test. B, The mean SA averaged across genes per patient and normalized by the pseudo purity, for all treatment arms combined. Wilcoxon test and Cohens d. ****, P value ≤ 0.00005. SA, splicing antigenicity.

Figure 6

Per splice junction normalized SA for responders compared with baseline samples. The SA averaged across samples for the subset of splice junctions derived from the top 20 genes with the highest SA per sample. The black horizontal line is the median SA for baseline samples. *, P value ≤ 0.05; **, P value ≤ 0.005; ***, P value ≤ 0.0005; ****, P value ≤ 0.00005. SA, splicing antigenicity.