ObjectiveTo assess the clinical efficacy of vitrectomy with intravitreal ranibizumab (IVR) at different injection time for proliferative diabeticretinopathy (PDR). MethodsThis was a prospective, comparative, and randomized study. Ninety-seven eyes of 97 patients were enrolled and randomly assigned to three different treatment groups: 30 eyes (30 patients) in the preoperative IVR group, 32 eyes (32 patients) in the intraoperative IVR group and 35 eyes (35 patients) in the no IVR injection group. The best corrected visual acuity (BCVA) (F=0.18) and the grading of vitreous hemorrhage (χ2=1.39) before surgery did not differ significantly among the 3 groups, respectively (P > 0.05). All eyes enrolled underwent conventional 23-gauge pars plana vitrectomy (PPV). The preoperative IVR group received intravitreal 0.5 mg/0.05 ml ranibizumab injection 3 to 7 days before PPV, intraoperative IVR group received intravitreal 0.5 mg/0.05 ml ranibizumab injection at the end of PPV and non-drug injection group received PPV only. Postoperative BCVA, fundus color photography, optical coherence tomography examination was performed in all eyes at 1 week and 1, 3, 6, 9, 12 months after surgery. Early RVH was defined as RVH occurred within 1 week to 1 month postoperatively; while late RVH was defined as RVH occurred 1 month later after the operation. ResultsThe mean BCVA were all improved among the 3 groups compared with the preoperative vision at 1 month after operation. At the beginning of 3 months after surgery, the average BCVA of the preoperative injection group and the intraoperative injection group tended to stable; while 3 eyes in the non-drug injection group began to decreased. There was no significant difference in average BCVA at 1, 3 and 12 months of follow-up periods among the 3 groups (F=1.42, 1.17, 0.26; P > 0.05). The incidences of early RVH were 16.7%, 9.4%, 28.6% in the preoperative injection group, intraoperative injection group, and non-drug injection group, respectively (χ2=5.12, P < 0.05). The incidence of early RVH in the intraoperative injection group reduced compared to preoperative injection group and non-drug injection group (χ2=4.04, 4.93; P < 0.05). The incidences of late RVH were 13.3%, 9.4%, 14.3% in preoperative injection group, intraoperative injection group, and non-drug injection group, respectively (χ2=0.47, P > 0.05). The average centeral foveal thickness (CFT) decreased among the 3 groups in different degrees at 1 month when compared with that of 1 week after operation and the decreasing was statistically significant (F=59.50, P < 0.05). A subgroup pairwise analysis showed no significant difference of decreasing CFT in preoperative injection group compared with that of intraoperative injection group (t=0.23, P > 0.05). The average CFT of the 3 groups had different degrees of thickening at 3, 6, 9, 12 months after surgery, and the increasingof CFT among the 3 groups were not differ significantly (F=2.92, 2.86, 3.07, 3.12; P > 0.05). ConclusionsThe adjunctive use of IVR can reduce the incidence of early postoperative RVH in vitrectomy for PDR, decrease in macular thickness and obtain favorable visual recovery. The effect of preoperative IVR injection was slightly better than that of the intraoperative IVR injection.