humanSTK11/LKB1 co-mutation with KRAS in lung adenocarcinoma drives primary PD-1/PD-L1 resistance despite TMB-intermediate/high status (Skoulidis 2018)1. The "KL" subgroup had ORR 7.4% vs 35.7% for KRAS/TP53-comutated ("KP") tumors; replicated in multiple subsequent NSCLC cohorts and used in trial stratification.
humanPTEN loss drives ICI resistance in melanoma via PI3K-AKT activation, VEGF/CCL2 secretion, reduced autophagy-dependent killing, and reduced CD8 infiltration (Peng 2016)2. Replicated in melanoma and pan-cancer.
mouseWNT/β-catenin activation — strong mouse biology, weaker clinical signal. Spranger 2015 established that melanoma β-catenin activation excludes CD103+ Batf3+ DCs via reduced CCL43. Clinical cohorts show weaker, more context-dependent correlations. WNT-inhibitor + ICI combinations have been limited by toxicity; β-catenin status is not a clinical biomarker.
humanPBRM1 LOF as ICI sensitizer in ccRCC has not replicated. Miao 20184 found PBRM1 LOF enriched in responders in two cohorts. Subsequent CheckMate-009/010/025 (Braun 2020 Nat Med) and IMmotion analyses did not confirm the predictive value.
humanIPRES signature underperformed as an independent predictor. Hugo 20165 defined an EMT/wound-healing/ECM/angiogenesis transcriptional signature in innately resistant melanoma. Subsequent validation has been mixed — IPRES overlaps heavily with TGF-β and stromal signatures and adds little above CD8, PD-L1, TMB, or pan-fibroblast measures.
humanBRCA2 association with ICI response did not replicate (from the same Hugo paper).
mouseLactate → AARS1 → PD-L1 K280 lactylation stabilizes PD-L1 against HUWE1-mediated degradation6 — but exogenous lactate paradoxically enhances anti-PD-L1 efficacy preclinically. The paradox must be reconciled before clinical translation. Directly implicates metabolic inputs in checkpoint protein turnover.
mouseHILPDA-sustained KLF5-driven fatty acid synthesis palmitoylates PD-L1 at Cys272, stabilizing membrane PD-L17. TRIM21 engagement by fenretinide degrades HILPDA and restores anti-PD-1 efficacy in breast cancer mouse models. A post-translational PD-L1 regulatory layer distinct from transcriptional control. Preclinical; single tumor type; fenretinide has prior clinical exposure which helps translation.
humanNOTCH3 → RBPJ → PVR → TIGIT in CRC8: tumor-intrinsic NOTCH3 transcriptionally upregulates PVR, engaging TIGIT on CD8 cells. Low/mutant NOTCH3 predicts better ICB survival in 102-patient + MSKCC cohorts. No pharmacologic NOTCH3 inhibitor tested in vivo.
mousePKMYT1 inhibition with clinical-grade RP-6306 activates cGAS-STING in castration-resistant prostate cancer9. Type I/II IFN signaling upregulates CCL5/CXCL10, enhances CD8 infiltration, and potentiates anti-PD-L1 in syngeneic models. Prostate cancer has been stubbornly ICI-refractory; a druggable cell-cycle-kinase–innate-sensing connection is clinically provocative.
The emerging literature in this period is heavily focused on post-translational regulation of PD-L1 (palmitoylation, lactylation) as a distinct layer from transcriptional control. If these preclinical mechanisms yield clinical PTM-modulating strategies, they would complement rather than replace PD-L1-targeting antibodies. None is clinical yet.