Temporal changes in tongue color during immune checkpoint inhibitor therapy in patients with non-small-cell lung cancer: a prospective observational study using digital tongue diagnosis

Background: Tongue diagnosis (TD), a key component of traditional East Asian medicine, employs a unique pattern-based diagnostic system. Digital TD enables quantitative assessment of tongue characteristics, like body and coating color, enhancing objectivity and reproducibility. While abnormal tongue features (including dark red, bluish, or pale appearance) have been documented in patients with cancer, the relationship between longitudinal changes in tongue characteristics and immune checkpoint inhibitor (ICI) treatment response or survival outcomes in non-small-cell lung cancer (NSCLC) remains underexplored. This multicenter, prospective, observational study investigated whether longitudinal tongue changes differ by ICI response and predict survival in patients with NSCLC. Methods: We enrolled patients with stage IIIB, IIIC, or IV NSCLC scheduled to receive second-line or subsequent pembrolizumab or atezolizumab following first-line platinum-based therapy failure. Digital tongue images were collected every 9 weeks from baseline to week 45. Linear mixed models evaluated temporal parameter changes and compared responders (durable clinical benefits ≥6 months) versus nonresponders. Multivariate Cox models adjusted for sex and age assessed tongue lightness changes as a prognostic value for progression-free survival (PFS) and overall survival (OS). Survival distributions were compared using Kaplan–Meier curves. Results: Of 170 enrolled participants, 140 were included in the analysis. Early in treatment, tongue lightness decreased in the body, fur, root, and center areas in both responders and nonresponders; however, the darkening was more pronounced in nonresponders, with significant visit-by-response interaction effects. In multivariate Cox analysis, lightness changes of the tongue body were significantly associated with PFS (hazard ratio [HR] = 0.93; 95% confidence interval [CI], 0.88–0.99; p = 0.019) and showed a trend for OS (HR = 0.93; 95% CI, 0.86–1.00; p = 0.062). Lightness changes of the tongue center were also significantly associated with PFS (HR = 0.95; 95% CI, 0.90–0.99; p = 0.027). Kaplan–Meier analysis confirmed that patients with a greater decrease in tongue body lightness had significantly shorter OS (p = 0.049). Conclusion: Digital TD diagnosis, particularly monitoring tongue lightness changes, may provide a valuable noninvasive prognostic tool for patients with NSCLC undergoing ICI therapy. It offers information for both PFS and OS, potentially complementing current biomarkers for cancer immunotherapy.