Vitreoretinal surgery (VRS) is the main treatment of severe proliferative diabetic retinopathy (PDR), intravitreal injection of anti-VEGF drugs before VRS on severe PDR has been widely used in clinic, but the auxiliary effect of different anti-VEGF drugs on VRS is unclear.

This study was to observe the efficacy of preoperative intravitreal injection of anti-VEGF drugs for the treatment of severe PDR, and to compare the efficacy of intravitreal conbercept (IVC) versus intravitreal ranibizumab (IVR) before VRS.

A retrospective, comparative study was performed.Niety-six severe PDR patients (98 eyes) who experienced VRS were collected in Tianjin Eye Institute from October 2014 to January 2016.There were 32 eyes in the preoperative IVC group, 31 eyes in the preoperative IVR group and 35 eyes in the simple VRS group.The intravitreal injection was performed 3 days prior to VRS in the preoperative IVC group and preoperative IVR group.All operations were conducted by the same physician.The duration of VRS, number of endodiathermy, incidence of severe hemorrhage and iatrogenic retinal breaks were recorded.The follow-up visit ranged from 3 to 6 months with an average of (3.51±0.59) months, the changes of best corrected visual acuity (BCVA) and incidence of postoperative complications were compared among the three groups.

The average operation duration was (84.84±11.03) minutes and (83.45±14.24) minutes in the preoperative IVC group and preoperative IVR group respectively, which were significantly shorter than (104.57±13.42) minutes in the simple VRS group, with significant differences between them (t=6.622, 6.604, both at P<0.01). The incidences of endodiathermy were lower in the preoperative IVC group (31.25%) and preoperative IVR group (29.03%) than that in the simple VRS group (60.00%), with significant differences between them (χ²=5.558, P=0.018; χ²=6.359, P=0.012). The incidence of severe hemorrhage in the simple VRS group was 42.86%, which was significantly higher than that in the preoperative IVC group (15.63%) and preoperative IVR group (12.90%), with significant differences between them (χ²=5.920, P=0.015; χ²=7.195, P=0.007). There were no statistical differences in the incidence of iatrogenic retinal breaks and postoperative vitreous hemorrhage, temporary increase of intraocular pressure, recurrent retinal detachment, glaucoma among the three groups (all at P>0.05). The mean BCVAs were significantly different among the 3 groups of different timepoints (F_group=4.077, P=0.020; F_time=100.937, P=0.000). The postoperative mean BCVA at 3 months after surgery was (0.80±0.37), (0.82±0.32) and (1.03±0.52) in the preoperative IVC group, preoperative IVR group and simple VRS group respectively, which was significantly improved compared with preoperative results (all at P<0.01). The postoperative BCVAs were improved in preoperative IVC group and preoperative IVR group compared with the simple VRS group, with significant differences between them(both at P<0.05).

The preoperative intravitreal injection of anti-VEGF drugs combined with VRS is conductive to shorten the average operation time, reduce the rate of intraoperative hemorrhage, endodiathermy as well as improve the BCVA.Conbercept and ranibizumab had the similar auxiliary effect for VRS for the treatment of severe PDR.

Proliferative diabetic retinopathy; Vascular endothelial growth factor; Conbercept; Ranibizumab; Vitreoretinal surgery