Search for a command to run...
Immune checkpoint inhibitors (ICIs) have transformed oncology for microsatellite instability-high (MSI-H) colorectal cancer, yet 85–95% of colorectal cancer patients carry microsatellite-stable (MSS) tumors and derive no benefit from current ICI regimens. We propose that VDAC1-mediated mitochondrial DNA (mtDNA) gate-jamming — suppression of VDAC1 oligomerization by HK-II docking, Bcl-xL binding, and outer mitochondrial membrane cholesterol loading — explains this selectivity by silencing the cGAS–STING innate immune signal required for spontaneous T cell priming. To test this hypothesis at scale, we computed a transcriptomic Gate-Jamming Score (tGJS = 0.4 × HK2 + 0.3 × BCL2L1 + 0.3 × TSPO, rank-normalized) and conducted three sequential analyses: (S1) pan-cancer TCGA (n = 10,071, 33 cancer types) — null result (ρ = +0.38 vs ICI response rate, p = 0.14); (S2) COADREAD MSS/TP53-wildtype clean room (n = 209) — five Bonferroni-significant inverse correlations between tGJS and immune markers including HAVCR2 (ρ = −0.349, p_bonf = 5 × 10⁻⁶), CXCL10 (ρ = −0.231, p_bonf = 0.015), and cGAS (ρ = −0.208, p_bonf = 0.049); (S3) IMvigor210 urothelial carcinoma atezolizumab cohort (n = 348) — null result (Wilcoxon p = 0.965, Cox HR = 0.898, p = 0.455). The flanking nulls (S1, S3) define the framework’s domain: the gate-jamming signal is detectable only when VDAC1-mediated mtDNA release is the dominant cytosolic DNA source and innate priming is the rate-limiting step. The S2 clean room results, combined with the three-layer therapeutic hypothesis (VDAC1 gate-opener + cGAMP/DNA eraser inhibitor + checkpoint blockade) independently derived from the same data by three AI analytical systems, motivates protein-level validation in MSS colorectal cancer and combination ICI trials in this specific population.