Risk factors and survival analysis in high-risk penetrating keratoplasty: a retrospective cohort study in 220 patients

High-risk penetrating keratoplasty (HRPK) is associated with significantly reduced graft survival compared to low-risk cases, with 5-year survival below 35% in developed countries. However, risk factors and outcomes may differ in developing regions where infectious and traumatic indications are more common. This study evaluates the 1- to 5-year graft survival rates and identified independent predictors of graft failure in HRPK from a tertiary center in China. We performed a retrospective cohort study of 220 HRPK adult patients between December 2019 and December 2023, meeting standardized high-risk criteria including corneal neovascularization in two or more quadrants, re-transplantation, active inflammation, or combined procedures. Preoperative, intraoperative, and postoperative data were collected. Graft failure was defined as central corneal opacity of ≥ 4 mm or persistent corneal edema. Multivariable Cox regression analysis identified independent risk factors, and a prediction nomogram was created and validated via bootstrapping. Graft survival rates at 1, 3, and 5 years were 82.7%, 46.3%, and 34.7%, respectively. Graft failure occurred in 113 cases (51.36%) over the total follow-up period. Compared to non-failure cases, the failure group had higher rates of re-transplantation (53.1% vs. 38.3%, p = 0.039), peripheral anterior synechiae (63.7% vs. 47.7%, p = 0.024), and 4-quadrant neovascularization (34.5% vs. 17.8%, p = 0.036). Cox proportional hazards regression identified advancing age (HR 1.02 per year, 95% CI 1.01–1.04, p = 0.023), graft diameter ≥ 9 mm (HR 1.74, 95% CI 1.12–2.71, p = 0.02), and 4 quadrants of neovascularization (HR = 2.53, P < 0.01) as independent predictors of failure. Peripheral anterior synechia and re-transplantation were also associated with increased risk. The five-variable predictive nomogram showed moderate predictive performance for 3-year (AUC = 0.715) and 5-year (AUC = 0.770) survival rate, with strong calibration. Subgroup analysis indicated that mycophenolate mofetil was mostly administered to higher-risk patients; however, propensity-matched analysis demonstrated no significant survival advantage (50.0% vs. 50.0% failure rate, p = 0.856). This study identified advanced age, a large graft diameter (≥ 9 mm), and 4 quadrants of neovascularization as significant predictors of HRPK failure, with an especially elevated risk during the initial three postoperative years. Risk-based immunosuppressive strategies and close monitoring are essential to improving long-term outcomes in this high-risk population.