Peripheral blood lymphocytes constitute a central arm of the immune system, and their role in the initiation and progression of glaucoma is attracting increasing attention. To investigate whether shifts in peripheral lymphocyte-subset profiles are associated with primary open-angle glaucoma susceptibility and to evaluate the predictive value of lymphocyte-related indices for glaucoma onset. This retrospective case-control study enrolled 428 consecutive primary open-angle glaucoma patients attending the Department of Ophthalmology between January 2023 and January 2024 (glaucoma group), and 475 age- and sex-matched healthy subjects undergoing routine health screening during the same period (HC group). Demographic data, laboratory results, and quantified peripheral lymphocyte subsets were extracted from electronic medical records. Logistic regression was used to identify glaucoma-related lymphocyte variables, and evaluated their discriminative performance using ROC curves, calculating the AUC, optimal cut-off, sensitivity, and specificity. Compared with controls, glaucoma patients exhibited significantly higher absolute counts and percentages of CD4⁺ T cells, lower absolute counts and percentages of CD8⁺ T cells and NK cells, and an elevated CD4⁺/CD8⁺ ratio (all P 1.57; sensitivity 57.94%, specificity 82.32%) and 0.657 (95% CI 0.625–0.688, P < 0.001) for NK-cell count (cut-off ≤ 525 cells/µL; sensitivity 79.67%, specificity 47.58%). Combining the two indices increased the AUC to 0.787 (95% CI 0.759–0.813, P < 0.001), with sensitivity 67.76% and specificity 75.16%. The AUC of the combined model was significantly superior to that of either index alone (all P < 0.001). In this retrospective analysis, peripheral CD4⁺ and CD8⁺ T-cell imbalances were associated with glaucoma susceptibility and demonstrated moderate predictive accuracy for disease onset. The CD4⁺/CD8⁺ ratio, especially when combined with NK-cell count, shows promise as a potential candidate biomarker for early glaucoma risk assessment, pending prospective validation.
Yu et al. (Fri,) studied this question.