Spatial transcriptomics reveals epithelial-immune remodeling preceding malignant transformation of oral premalignant lesions

This preprint places TIGIT-high exhausted T cells inside the tissue program that precedes oral cancer, which is a descriptive finding about where the checkpoint sits, not evidence that blocking it changes outcomes. The Claidex graph holds four TIGIT programs and none of them supports the inferential step from presence to therapeutic benefit. Ociperlimab missed in first-line PD-L1-high non-small cell lung cancer (ociperlimab-tigit-advantig-302-nsclc-class-failure) and tiragolumab missed in the adjuvant setting of the same disease (tiragolumab-tigit-adjuvant-nsclc-phase3-class-efficacy-failure), both read at the time as class efficacy failures. Vibostolimab was pulled from BCG-unresponsive non-muscle-invasive bladder cancer (vibostolimab-favezelimab-tigit-bcg-unresponsive-nmibc-keynote-057-phase2-strategic-termination) and terminated in adjuvant melanoma on a safety signal (vibostolimab-pembrolizumab-tigit-adjuvant-melanoma-keyvibe-010-phase3-safety-termination). The pattern across those four is that TIGIT expression on exhausted T cells has repeatedly marked a state that antibody blockade did not usefully reverse. A premalignant setting is a genuinely different context and may behave differently, but any translational proposal arising from this work should state explicitly why interception before transformation escapes the mechanism that defeated four later-stage programs.