Factors Associated with Graft Detachment after Primary Descemet Stripping Automated Endothelial Keratoplasty

Purpose: To identify risk factors associated with postoperative graft detachment after Descemet stripping automated endothelial keratoplasty (DSAEK). Methods: In this retrospective institutional cohort study, the study group included all consecutive eyes that underwent primary DSAEK between January 2005 and October 2015 at Villa Serena-Villa Igea private hospitals (Forlì, Italy) and developed postoperative graft detachment (n = 45). The control group included all eyes that underwent primary DSAEK during the same period and did not develop graft detachment (n = 1167). The main outcome measure was whether or not postoperative graft detachment occurred. Results: Main indications for surgery included Fuchs endothelial dystrophy (525/1212 = 41%), pseudophakic bullous keratopathy (422/1212 = 35%), and a failed penetrating keratoplasty (PK) graft (190/1212 = 16%). Postoperative graft detachment occurred in 45 of 1212 (3.7%) eyes. Medically treated glaucoma, previous trabeculectomy, previous aqueous shunt procedure, and failed PK were all associated with an increased risk for graft detachment in univariate analysis. No particular lens status at the time of graft implantation was significantly associated with graft detachment, including the presence of an anterior chamber intraocular lens [odds ratio (OR), 1.5; 95% confidence interval (CI), 0.35-6.46, P = 0.58] or aphakia at the time of graft transplantation (OR, 1.37; 95% CI, 0.41-4.57, P = 0.74). Cataract removal during surgery was not associated with an increased risk for detachment (OR, 0.77; 95% CI, 0.39-1.5, P = 0.47). Previous trabeculectomy (OR, 4.21; 95% CI, 1.53-11.56, P = 0.005) and a previous failed PK graft (OR, 3.04; 95% CI, 1.6-5.78, P = 0.001) were the only independent risk factors for graft detachment in multivariate analyses. Conclusions: Previous PK and trabeculectomy are independent risk factors for postoperative graft detachment in primary DSAEK.