In this systematic review, we analyzed the effect of preoperative IVR compared with preoperative no IVR and preoperative IVC on the intraoperative and postoperative outcomes of vitrectomy.
Our study compared preoperative IVR versus no IVR for intraoperative outcomes. Our findings indicate that the IVR group had significantly lower mean surgery time and rates of intraoperative electrocoagulation, silicone oil tamponade, intraoperative bleeding, and iatrogenic retinal breaks. There was no significant difference in the rate of relaxing retinotomy between the two groups. Preoperative IVR can potentially reduce atrophy of most retinal neovascularization and neovascularization membrane fibrosis in the fundus retina and front of the optic disc. This reduction effectively decreases the likelihood of intraoperative bleeding and electrocoagulation events. Additionally, the proliferative membrane may become thinner, layered, and loose, resulting in poor adhesion to the retina, which can reduce the difficulty of dissociation of the proliferative membrane and ultimately reduce the incidence of iatrogenic retinal breaks. Due to fewer retinal breaks, air and gases become the main internal tamponading agents instead of silicone oil [15].
At the same time, preoperative IVR reduces intraoperative bleeding and surgical instrument replacement, provides clear intraoperative vision, increases retinal visibility, simplifies operations, reduces the surgical difficulty, and dramatically shortens surgery time. Our results showed high heterogeneity in the mean surgery time. The duration of surgery is influenced by the complexity of the surgical case and the occurrence of intraoperative complications. In Comyn et al.'s [9], there was no significant difference in surgery time between the two groups, but the slightly higher median surgeon-defined complexity score in the IVR group may have affected surgery time, resulting in heterogeneity. Furthermore, the heterogeneity may also be due to surgeons' different learning curves for vitrectomy. Due to the limited studies, we could not do a subgroup analysis to assess the exact source of heterogeneity, but we conducted a sensitivity analysis, and the result was stable.
In comparing postoperative BCVA in the IVR group and the no IVR group, our results showed that the mean change in BCVA was greater in the IVR group at one month and three months after surgery; there was no significant difference in the mean change in BCVA between the two groups at 6 months and at least 12 months after surgery. Our results showed high heterogeneity in mean change in BCVA one month after surgery. In Guan et al.'s study [14], it was found that inflammation was more severe in the preoperative group that did not receive intravitreal injection (IVR). This inflammation affected the absorption of macular edema, resulting in little mean change in best-corrected visual acuity (BCVA) one month after surgery, which may have contributed to the observed heterogeneity. We performed a sensitivity analysis and found the result to be stable.
Several perspectives explain the postoperative BCVA changes. On the one hand, Guan et al.'s study [14] found that the improvement in BCVA and the reduction in macular edema were consistent at one and three months after surgery in the IVR group. Compared with the preoperative no IVR group, the center macular thickness (CMT) of the IVR group was significantly different at one and three months after surgery and then decreased slowly. The changing trend is consistent with the changing trend of BCVA after surgery. At the last 6-month follow-up, the CMT and BCVA in the IVR group were not significantly different from those in the no IVR group. On the other hand, preoperative IVR can reduce the rate of early postoperative VH, resulting in faster vision recovery, while preoperative IVR cannot reduce the rate of late postoperative VH, resulting in no significant difference in final BCVA between the two groups [8]. According to our results, preoperative IVR therapy did not improve long-term visual acuity in patients. The primary factors influencing postoperative long-term visual acuity are lens transparency, macular edema, and retinal condition.
In the comparison of postoperative complications in the IVR group and the no IVR group, our results showed that the preoperative IVR group could reduce the incidences of early postoperative VH and postoperative NVG, and the incidences of late postoperative VH, postoperative INV, and recurrent retinal detachment had no significant difference between the two groups. Wakabayashi et al. [39] reported that high intraocular VEGF levels in PDR patients during primary vitrectomy were identified as an important risk factor for early postoperative VH. Bevacizumab, one of the anti-VEGF drugs, has a half-life of 6.7 days in the vitreous, and it can still be detected in the retina 14 days after intravitreal injection [40]. Although most of the vitreous was removed during the vitrectomy, bevacizumab in the retina may still be effective [41]. The half-life of ranibizumab in the vitreous is approximately 9 days [42], and similarly, residual ranibizumab in the retina after vitrectomy may remain effective. Therefore, based on the above studies and our results, preoperative IVR can reduce the incidence of early postoperative VH. Postoperative NVG is a devastating complication and a significant cause of vitrectomy failure in patients with severe PDR. Our results showed that preoperative IVR could reduce the incidence of postoperative NVG. Still, this result should be treated with caution because we found a risk of publication bias through the Harbord test.
Our study suggests that preoperative IVR may not be a significant factor in determining the rate of relaxing retinotomy and the occurrence of postoperative INV and recurrent retinal detachment. However, it is essential to note that our findings were based on a limited sample size of only two studies. Further research is required to determine whether preoperative IVR has an impact on these factors.
In the comparison of preoperative IVR versus preoperative IVC, our results showed that the intraoperative outcomes (mean surgery time, the rates of intraoperative electrocoagulation, silicone oil tamponade, relaxing retinotomy, intraoperative bleeding, and iatrogenic retinal breaks) and postoperative outcomes (mean postoperative BCVA, the incidence of postoperative NVG, recurrent retinal detachment) were not significantly different between the two groups. The mechanism of action of ranibizumab is binding to VEGF-A, while conbercept binds not only to VEGF-A but also to VEGF-B and PIGF [43]. Compared with ranibizumab, conbercept binds to VEGF with a higher affinity and has a longer half-life in the vitreous humor [44, 45]. However, our results suggested that both preoperative IVR and preoperative IVC improve surgical outcomes with no significant difference. This may be because the smaller molecular weight of ranibizumab is more permeable in the retina, thus offsetting its lower binding affinity [46]. It is worth noting that we found that postoperative BCVA may be at risk of publication bias through the Egger test. And based on the current study, we pooled and analyzed only three studies showing BCVA at the final follow-up. In future studies, it is necessary to pay attention to the changes in visual acuity during different follow-up periods.
Our systematic review showed that preoperative IVR was significantly better than preoperative no IVR for PDR patients and had the same effect as preoperative IVC. Preoperative IVR can not only reduce surgery time, intraoperative electrocoagulation, silicone oil tamponade, intraoperative bleeding, and iatrogenic breaks but also significantly improve early postoperative BCVA and reduce the incidences of early postoperative VH and postoperative NVG. However, there are several unavoidable limitations to our meta-analysis. First, just 4 RCTs were included, whereas most studies were retrospective. Second, differences in patient characteristics and criteria defining the complexity of patients' conditions in each study may represent an important source of bias. Finally, computer-based literature searches might not contain all relevant research. Gray literature was likewise ineligible for inclusion.
In conclusion, in the context of the available studies, preoperative IVR was associated with more significant early visual improvement, better intraoperative outcomes, and fewer postoperative complications than preoperative no IVR. There was no significant difference in outcome between preoperative IVR and preoperative IVC. Therefore, preoperative IVR is an effective adjunct to vitrectomy for PDR, and there is no inferiority in surgical outcomes compared to preoperative IVC. However, due to the low quality of the evidence, further high-quality studies are needed to assess its effectiveness.