Passenger mutations link cellular origin and transcriptional identity in human lung adenocarcinomas.

DNA damage is preferentially repaired in expressed genes; thus, genome-wide correlations between somatic mutation patterns and normal cell transcription may reflect tumor cell origins. Accordingly, we found that aggregate lung adenocarcinoma (LUAD) and squamous cancer (LUSC) somatic mutation density associated most strongly with distal (alveolar) and proximal (basal) lung cell-type-specific gene expression, respectively, consistent with presumed LUAD and LUSC cell origins. Analyzing individual genomes, 21% of LUADs bore mutational footprints of proximal airway origins, with 38% classified as ambiguous. Distal origin LUADs, enriched for KRAS and STK11 drivers, occurred mainly in smokers; proximal origin LUADs, enriched for EGFR drivers, were more common in never-smokers. Ambiguous origin LUADs showed APOBEC signatures and SMARCA4 alterations. TP53 mutant LUADs with non-distal cell origins preferentially exhibited non-distal transcriptional identity. Our study reveals a complex interplay between lineage and identity in LUAD evolution and offers a scalable strategy to infer tumor origins in human cancers.