Mehaffy, N. S. | Search bioRxiv

bioRxiv2026

Inference of cancer driver mutations from tumor microenvironmentcomposition: a pan-cancer study with cross-platform external validation

Cancer driver mutations shape the tumor microenvironment (TME), yet whether TME composition alone can predict genotype has not been systematically evaluated across cancers with external validation. We trained machine learning models to predict driver mutation status from TME cell-type composition signatures derived from bulk transcriptomes. Tissue-specific TME signatures (22-28 programs per cancer) were scored from RNA-seq data in TCGA for glioblastoma (GBM, n=157 total; n=90 EGFR-amplification evaluable), breast cancer (BRCA, n=1,082 total; n=994 evaluable), lung adenocarcinoma (LUAD, n=510 total; n=502 evaluable), and colorectal cancer (CRC, n=592 total; n=524 evaluable), then externally validated on independent cohorts spanning different platforms: CPTAC (GBM, n=65), METABRIC (BRCA, n=1,859), GSE72094 (LUAD, n=442), and GSE39582 (CRC, n=585). Of 15 driver-cancer pairs tested, 14 achieved external AUC [≥]0.65, with top performance for ERBB2 amplification in BRCA (AUC=0.980), BRAF mutation in CRC (0.899), and TP53 mutation in BRCA (0.871). TME-predicted ERBB2 status stratified overall survival in METABRIC (Cox HR=1.73, p=7.95x10-8). Marginal KRAS performance in LUAD (AUC=0.615) reflected opposing TME profiles in KRAS+STK11 versus KRAS+TP53 co-mutant tumors. These results demonstrate that TME composition encodes sufficient information to infer driver mutations across cancers.

bioinformatics↗

bioRxiv2025

Microenvironment-Inferred Genotyping: An Exclusionary Classifier for EGFR Amplification When DNA Testing Fails

EGFR amplification occurs in approximately 40-50% of glioblastoma (GBM) cases and is critical for treatment selection [1]. However, GBM tissue samples frequently yield insufficient material for comprehensive molecular testing due to extensive necrosis and tissue quality limitations [2]. This affects thousands of patients annually in the United States [4]. We developed a microenvironment-inferred genotyping approach, enabling molecular classification by measuring the "oligodendrocyte desert" effect when direct genetic testing is impossible. Using single-cell RNA-seq data from 102 GBM patients (1.47M cells) [19], we identified oligodendrocyte exclusion patterns associated with EGFR amplification. We developed a 13-feature machine learning classifier and computationally validated it across an independent external cohort (CPTAC, n=96) and cross-pipeline technical validation using TCGA data processed through three distinct bioinformatic pipelines (n=148 patients) [20,21,22,23]. EGFR-amplified tumors created detectable oligodendrocyte deserts (60-70% depletion, p<0.001). Our exclusionary classifier achieved AUC 0.845 in discovery cohorts and 0.756 average across validation analyses (n=244 unique patients). With a positive predictive value of 94%, this tool identifies high-confidence candidates for EGFR-targeted therapies who would otherwise be excluded from treatment. To our knowledge, this is the first algorithm enabling microenvironment-inferred genotyping from routine RNA-seq data, providing rescue diagnosis for EGFR classification when DNA testing fails.

cancer biology↗