Tumor NK ligand profiles as predictors of allogenic NK cell therapy responsiveness.

80

Background: Although NK cells offer a versatile and safe therapeutic platform, the understanding of natural killer (NK) cell therapy remains limited, particularly regarding mechanisms underlying effective tumor targeting. For clinical use, target recognition and anti-tumor efficacy need to be improved. In this study, we analyzed the activating and inhibitory signal integration to predict tumor responsiveness to NK cell therapy. Methods: Gene expression data for 13 activating and 10 inhibitory NK ligands were obtained from 6,892 solid tumors, including lung, breast, colorectal, and stomach cancers from the TCGA TARGET cohorts, and normalized against non-cancerous tissues from the GTEx database. From these data, NK ligand combination that were specifically enriched in particular cancer types were identified. The tumor NK ligand expression profiles were further used to compute a NK sensitivity prognostic index by subtracting inhibitory from activating signals. This index was validated in vitro using cytotoxicity assays of 50 NK donors against 5 cancer cell lines with different prognostic indexes. Results: We identified distinct NK ligand combinations that were exclusively enriched in each cancer type: TIGIT ^high NKG2D ^neg PD1 ^neg (62.6% in lung adenocarcinoma), NKG2D ^high TACTILE ^high (79.3% in lung squamous cell carcinoma), CD59 ^high 2B4 ^high (58.8% in breast cancer), NKG2D ^high TIGIT ^low NKp44 ^neg (51.0% in colorectal cancer) and NKp30 ^high TIM3 ^high (69.1% in stomach cancer). Factors such as NK cell donor identity and single ligand expression levels did not associate with NK sensitivity. In contrast, the NK prognostic index derived from composite ligand expression strongly correlated with in vitro NK cell-mediated cytotoxicity (R ² =0.86, p=0.02). SW620 cells, which exhibited the highest prognostic index, were the most sensitive to NK cell killing, whereas MDA-MB-231 cells, which had the lowest prognostic index, were the most resistant. Conclusions: NK ligand signatures and the derived tumor prognostic index may serve as biomarkers for predicting NK cell therapy response and help physicians better stratify patients for NK-based therapeutic approaches.