Development of a nomogram to predict graft survival after penetrating keratoplasty.

Abstract

PURPOSE\nTo develop a nomogram to predict the 3- and 5-year likelihood of graft survival after penetrating keratoplasty (PK) based on preoperative assessment and intraoperative plan.\n\n\nDESIGN\nRetrospective clinical case-control study.\n\n\nMETHODS\nData from 1029 consecutive PKs in 903 eyes of 835 patients performed at a single tertiary center from May 2007 to September 2018 were extracted from electronic medical records and evaluated for corneal graft failure, defined as irreversible and visually-significant graft edema, haze, or scarring. Thirty-seven variables were assessed by multivariable Cox models. A nomogram to predict the probability of graft survival was created.\n\n\nRESULTS\nMean recipient age was 57.1±22.0 years and mean follow-up was 4.22±3.05 years. Overall, 37.4% of grafts failed during follow-up. Eleven variables were significantly associated with graft failure, including active microbial infection at the time of PK (hazard ratio [HR]=5.10, 95% confidence interval [CI] 3.53-7.37), intraocular silicone oil at the conclusion of the PK (HR=4.28, 95% CI 2.38-7.71), history of systemic autoimmune disease (HR=2.83, 95% CI 1.63-4.90), four quadrants of corneal neovascularization (HR=2.76, 95% CI 1.56-4.86), any prior anterior segment surgery (HR=2.41, 95% CI 1.55-3.75), and lens status as anterior chamber intraocular lens (ACIOL) at the conclusion of surgery (HR=2.35, 95% CI 1.30-4.26). The nomogram exhibited a concordance index of 0.76 (95% CI 0.74-0.78); internal calibration plots depicted strong correlation between prediction and observation of graft survival.\n\n\nCONCLUSIONS\nPK graft prognosis may be predicted relatively accurately based on 11 variables. Although established from retrospective data, this nomogram would be valuable for data-driven patient counseling prior to corneal transplantation.