Oncolytic viruses in combination cancer therapy: clinical lessons, biomarker-guided strategies, and future treatment directions

Oncolytic viruses (OVs) preferentially infect and replicate within tumors, induce immunogenic cell death, and remodel immune-poor microenvironments to support T-cell infiltration. These features provide a rationale for combining OVs with immune checkpoint blockade, adoptive cellular therapies, bispecific T-cell engagers, cancer vaccines, and innate immune agonists, and for engineering viruses to express immunostimulatory cytokines. Clinical outcomes, however, have been inconsistent. In MASTERKEY-265, adding intratumoral talimogene laherparepvec to pembrolizumab did not significantly improve progression-free or overall survival in advanced melanoma, despite a 62% objective response rate in the preceding 21-patient phase Ib cohort and evidence of increased intratumoral CD8+ T cells and PD-L1 expression. In contrast, single-arm clinical data for RP1 plus nivolumab in anti-PD-1-refractory melanoma showed objective responses, including durable responses in a subset; the U.S. Food and Drug Administration granted accelerated approval for the regimen, but randomized confirmation remains required. Together with the failure of PHOCUS, these findings indicate that benefit is determined by viral platform, tumor immune state, disease and prior-treatment setting, delivery feasibility, and resistance biology rather than immune activation alone. We critically synthesize positive and negative evidence across checkpoint and non-checkpoint strategies and distinguish biomarkers derived from OV studies from those extrapolated from checkpoint-inhibitor trials. Candidate biomarkers and pharmacodynamic readouts are organized into a three-tier research framework covering pretreatment characterization, platform–partner hypotheses, and on-treatment monitoring; their utility for selecting OV-based combinations still requires validation. We further examine delivery barriers, antiviral immunity, tumor heterogeneity and resistance, next-generation vector engineering, and computational approaches to molecularly guided patient selection. Prospective validation in biomarker-stratified, adaptive trials is needed to improve patient selection, match viral platforms with rational partners, and translate biological activity into durable clinical benefit.