Huayang Gui, Yali Zhang, Xingxing Yuan
Immune checkpoint blockade produces durable benefit mainly in microsatellite instability-high/mismatch repair-deficient (MSI-H/dMMR) colorectal cancer (CRC), whereas most microsatellite-stable/mismatch repair-proficient (MSS/pMMR) tumors remain poorly inflamed and resistant to immunotherapy. This focused narrative review examines how mitochondrial DNA (mtDNA)-derived danger signals engage the cGAS-STING pathway in CRC, distinguishes canonical DNA sensing from noncanonical STING activation and STING-independent cGAS functions, and evaluates therapeutic strategies that modulate this axis. We qualitatively synthesize peer-reviewed mechanistic and translational studies addressing mtDNA release, pathway routing, cellular context, therapeutic targeting, and biomarkers in CRC, while drawing on pan-cancer and DNA-damage studies only when they clarify pathway architecture or translational constraints. No new experimental, patient-level, or case-series data are presented. Current evidence indicates that biological outcome depends on more than pathway activation alone. Signal amplitude and duration, STING trafficking and proteostasis, downstream IRF3- versus NF-κB-biased signaling, autophagy and mitophagy, metabolic state, pathway abundance, and responding cell type collectively determine whether activation supports antitumor immunity or chronic inflammation and immune suppression. Nuclear DNA damage can also activate noncanonical STING programs independently of cGAS, whereas cGAS can exert STING-independent functions. Upstream mtDNA-origin interventions may provide a more localized route to pathway activation by exploiting tumor mitochondrial stress, but cGAS is not intrinsically mtDNA-specific and any safety advantage over direct STING agonism remains unproven clinically. Clinical translation should therefore prioritize origin-resolved biomarkers, cell-specific pharmacodynamic readouts, tumor-localized and transient pathway modulation, and biomarker-stratified trials.