Back

Pairwise genomic alterations identify prognostic tumor states in multiple cancer types

Bayati, M.; Mishra, J.; Luhari, L.; Farina-Morillas, M.; Reimand, J.

2026-04-17 cancer biology
10.64898/2026.04.14.718563 bioRxiv
Show abstract

Genomic models of cancer prognosis usually rely on individual genomic alterations, potentially overlooking clinically meaningful combinations of events. We analyzed genomic and clinical data from nearly 10,000 primary tumors across major cancer types to identify prognostic genomic interactions (PGIs), defined as pairs of genomic alterations whose joint status was associated with patient outcome beyond either alteration alone. By systematically integrating survival associations with pairwise combinations of recurrent copy-number alterations and frequently mutated driver genes, we identified 57 PGIs. These PGIs refined prognostic stratification and were linked to distinct transcriptomic programs representing immune-response, epithelial-mesenchymal transition, and proliferation-related themes. Gene-level mapping highlighted dosage-sensitive candidate genes within recurrent copy-number regions, and gene essentiality profiles supported subsets of PGI-derived gene pairs. Two PGIs were validated in independent datasets. Together, these results establish a framework for identifying prognostic combinations of genomic alterations and connecting them to pathway programs, candidate genes, and functional dependencies in human tumors.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.