A Systematic Review and Meta-Analysis of Clinical Outcomes of Small Gauge Vitrectomy with or without Intravitreal Anti-Vascular Endothelial Growth Factor Agents Pretreatment for Proliferative Diabetic Retinopathy Open Access

Proliferative diabetic retinopathy (PDR) is a disease characterized by retinal ischemia, recurrent retinal neovascularization, and fibrous proliferation, which can lead to blindness without appropriate treatment [1‒3]. Patients with nonclearing vitreous hemorrhage (VH), tractional retinal detachment (RD), or extensive fibrovascular proliferation are always in need of pars plana vitrectomy [1].

Several meta-analyses have evaluated the efficacy of preoperative anti-vascular endothelial growth factor (anti-VEGF) agents application in PDR vitrectomy, in which many 20G vitrectomy trials were included [4‒6]. Due to less complications and better prognosis, 20G vitrectomy has been gradually replaced by small gauge vitrectomy in clinical practice [7‒9]. Many studies have reported the differences in surgical outcomes between small gauge and 20G vitrectomies [10‒12], and the results of studies evaluating the effectiveness and safety of preoperative anti-VEGF treatment in small gauge PDR vitrectomy are relatively different [1, 13‒21]. Therefore, it is necessary to review in a greater depth of the available studies to understand the benefits of anti-VEGF pretreatment in small gauge vitrectomy for PDR patients. Thus, we conducted this meta-analysis including newly published randomized controlled trials (RCTs) to evaluate the efficacy of anti-VEGF treatment before small gauge vitrectomy in PDR eyes.

The current study conforms to standard guidelines and was written according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [22].

This study was a meta-analysis of RCTs. Articles limited to RCTs were searched using PubMed, Embase, and the Cochrane Central Register of Controlled Trials. “VEGF”, “anti-VEGF”, “anti VEGF”, “lucentis”, “ranibizumab”, “aflibercept”, “conbercept”, “bevacizumab”, “Avastin”, “vitrectomy”, “diabetic retinopathy”, and “proliferative diabetic retinopathy” comprised the terms for sensitive search. The searches started at October 10, 2021, and ended at November 20, 2021. This study adhered to the tenets of the Declaration of Helsinki and relevant laws in China.

Inclusion criteria are as follows: (1) human study; (2) participants: patients with severe diabetic retinopathy; (3) compared the outcome of small gauge (23-G, 25-G, or 27-G) vitrectomy with preoperative anti-VEGF treatment versus small gauge vitrectomy alone; (4) at least one or more clinical outcomes representing intraoperative and/or postoperative outcome parameters must be assessed and published; (5) design: RCT. Exclusion criteria are as follows: (1) patients with other intraocular diseases that might affect the vitreoretinal surgery, such as uveitis, proliferative vitreoretinopathy, congenital vitreoretinopathies, and traumatic RD; (2) animal studies or cadaver subjects; (3) reports not published in English. For duplicated publications, only the data from the longest period of follow-up were used in the analysis.

Standardized data extraction was performed by two independent reviewers (MH-P and XY-Z). The following information was extracted: first author, publication year, design, group size, patient age, gender ratio, details of anti-VEGF agent injection, details of the surgical procedure, intraoperative and postoperative evaluating parameters, and follow-up periods. The methodological quality of each included studies was assessed by two researchers independently according to the 12-item scale [4, 23]: a trial with a score of >7 was considered high quality, >4 but no more than 7 was considered moderate quality, and no more than 4 was considered low quality. Disagreements were evaluated by kappa test, and agreement was achieved by a discussion with the corresponding author (GM-W).

Statistical analyses were performed using Stata SE v.12.0 software. RR and its 95% CI were calculated for dichotomous data; WMD and its 95% CI were calculated for continuous data. χ² test and I² were used to assess the statistical heterogeneity. Fixed-effect model would be used if heterogeneity was insignificant (p > 0.1, I² <50%). If heterogeneity was substantial (p < 0.1, I² >50%), subgroup analysis (different anti-VEGF agent) and sensitivity analysis were conducted to identify the source of the heterogeneity. If the heterogeneity could not be eliminated, a random-effects model would be used when the result of meta-analysis had clinical homogeneity. Begg’s funnel plot and Egger’s linear regression test were used to evaluate the publication bias. p < 0.05 was considered to indicate statistical significance.

The flowchart of studies from initial results to final inclusions is shown in Figure 1. Of the 169 potentially relevant studies yielded by electronic searches, ten RCTs published between 2008 and 2020, which met the predefined inclusion criteria, were included in this meta-analysis [1, 13‒21].

A total of 699 eyes in the 10 included trials [1, 13‒21] were enrolled in this meta-analysis. Table 1 shows the main characteristics of the included trials. Six [1, 13‒15, 18, 19] out of the 10 trials matched groups for preoperative visual acuity. Six [1, 15‒17, 20, 21] out of the 10 trials matched groups for sex. Eight [1, 13‒17, 20, 21] out of the 10 trials matched groups for age. Four out of the 10 trials matched groups for HbA1C [1, 13, 15, 17]. Four out of the 10 trials matched groups for type of diabetes [17, 18, 20, 21]. Three out of the 10 trials matched groups for duration of diabetes [17, 18, 20]. One out of the 10 trials matched groups for hyperlipidemia [13]. Three out of the 10 trials matched groups for hypertension [13, 15, 17]. Five out of the 10 trials [1, 13, 17‒19] reported masking surgeons. Two out of the 10 trials [1, 13] reported masking patients. Regarding attrition rate, seven trials had 100% completeness of follow-up, one trial [15] lost 5 eyes (6%) to follow-up, one trial [1] lost 10 eyes (4%) to follow-up, and the other trial [13] lost 2 eyes (3%) to follow-up. Overall, six out of the 10 trials were deemed to be high quality (score >7) (Table 2).

Five studies reported both pre- and postoperative best-corrected visual acuity (BCVA) in logarithm of the minimal angle of resolution format. Pooled results of these data revealed no statistically significant differences in preoperative baseline BCVA (WMD = 0.003; 95% CI: −0.151–0.158; p = 0.966, I² = 0.0%) and postoperative BCVA (WMD = −0.066; 95% CI: −0.157–0.024; p = 0.151, I² = 0.0%) (Fig. 2) at the last follow-up between the anti-VEGF treated group and the vitrectomy-alone group.

The included trials applied differential scales for the evaluation of intraoperative bleeding. Clinically significant intraoperative bleeding (defined as moderate to severe bleeding which cannot stop spontaneously or with transient bottle/pressure elevation) reported by six studies was included in this meta-analysis. Analysis of these data revealed that the incidence of clinically significant intraoperative bleeding was significantly less in the anti-VEGF treated group than in the vitrectomy-alone group (RR = 0.258, 95% CI 0.135–0.491, p < 0.001, I² = 68.1%). Because heterogeneity was evident for this outcome, a random-effects model was used to combine the data. As subgroup analysis failed to eliminate the heterogeneity, sensitivity analysis was performed. Results pooled from the largest group of homogeneous studies (five trials) showed that the difference between anti-VEGF-treated and control groups was still significant (RR = 0.218, 95% CI 0.146–0.327, p < 0.001, I² = 45.3%) (Fig. 3).

Four trials reported the frequency of intraoperative endodiathermy application. Analysis of these data showed that the frequency of intraoperative endodiathermy application was statistically significantly less in the study group than in the control group (WMD = −2.882; 95% CI −4.826 to −0.938; p = 0.004, I² = 95.8%) (Fig. 4). As sensitivity analysis failed to eliminate the heterogeneity, a random-effects model was used to combine the data. Owing to the significant heterogeneity for this outcome, the pooled result should be interpreted with caution.

The frequency of intraoperative iatrogenic retinal breaks was reported by 5 trials. Analysis of these data revealed that the anti-VEGF group had a lower proportion of eyes suffering from intraoperative iatrogenic retinal breaks than the control group (RR = 0.555, 95% CI 0.417–0.738, p = 0.001, I² = 0.0%) (Fig. 5).

Seven trials reported data of intraoperative silicone oil tamponade. The forest plot showed that the incidence of intraoperative silicone oil tamponade in the anti-VEGF treated group was significantly less than that in the vitrectomy-alone group (RR = 0.532, 95% CI 0.391–0.725, p < 0.001, I² = 7.1%) (Fig. 6).

Seven trials reported data of surgical time. Analysis of these data showed that surgical time in the anti-VEGF treated group was significantly less than that in the vitrectomy-alone group (WMD = −18.286; 95% CI −24.452 to −12.119; p < 0.001, I² = 70.5%) (Fig. 7). As subgroup analysis and sensitivity analysis failed to eliminate the heterogeneity, a random-effects model was used to combine the data. Owing to the significant heterogeneity for this outcome, the pooled result should be interpreted with caution.

Six trials reported data of postoperative VH in the first month after vitrectomy (early postoperative VH). Analysis of these data showed that the proportion of eyes suffering from early postoperative VH was significantly less in the anti-VEGF treated group than in the vitrectomy-alone group (RR = 0.513, 95% CI 0.386–0.680; p < 0.001, I² = 21.1%) (Fig. 8).

The data of late postoperative VH (after 1 month) after vitrectomy was reported by 2 trials. Analysis of these data revealed no significant differences in late postoperative VH development between the anti-VEGF treated group and the vitrectomy-alone group (RR = 0.744, 95% CI 0.344–1.606; p = 0.451, I² = 0.0%) (Fig. 9).

Five trials reported data of postoperative RD. Analysis of these data showed that the proportion of eyes suffering from postoperative RD was significantly less in the anti-VEGF treated group than in the vitrectomy-alone group (RR = 0.420, 95% CI 0.196–0.899; p = 0.026, I² = 0.0%) (Fig. 10).

Six trials reported data of postoperative rubeosis iridis or neovascular glaucoma (NVG). Analysis of these data revealed a borderline p value (RR = 0.437, 95% CI 0.177–1.078; p = 0.072, I² = 0.0%) in the proportion of eyes with postoperative progression or persistence of rubeosis iridis/NVG between the anti-VEGF treated group and the vitrectomy-alone group (Fig. 11).

Using Begg’s test (p = 0.452, continuity corrected) and Egger’s test (p = 0.101), publication bias was not found to be significant.

The present meta-analysis revealed that anti-VEGF treatment before small gauge vitrectomy for PDR could reduce the incidence of clinically significant intraoperative bleeding, intraoperative iatrogenic retinal breaks, intraoperative endodiathermy, intraoperative silicone oil tamponade, and surgical time. Also, it can reduce the incidence of early postoperative VH and postoperative RD.

Treatment of advanced PDR has always been a challenge [6]. Surgical outcomes may limit visual prognosis of the disease and it could be compromised by intraoperative and postoperative complications. Perioperative VH resulting in the formation of adherent fibrinous clots may necessitate frequent application of diathermy and exchange of instruments [1]. Intravitreal anti-angiogenic therapy is being widely used for neovascular complications of PDR. Results of fluorescein angiography revealed a reduction in leakage from the foci of neovascularization and regression of the neovascular component of fibrovascular tissue in eyes with PDR within 1 week after anti-VEGF injection [5]. With the regression of neovascularization, this may lead to reduction of intraoperative bleeding, provide better surgical view, and reduce the need for endodiathermy, thereby resulting in less frequent instrument changes and shorter operating time [24]. In this study, the forest plots of clinically significant intraoperative bleeding, frequency of endodiathermy, and surgical time indicated that preoperative anti-VEGF treatment could significantly benefit the surgical process of small gauge diabetic vitrectomy.

Iatrogenic retinal break is one of the major intraoperative complications of PDR patients, it can develop via different mechanisms, such as during proliferative fibrovascular membrane dissection or the induction of posterior vitreous detachment [10, 25]. In this meta-analysis, we compared the incidence of iatrogenic retinal breaks in small gauge vitrectomy with anti-VEGF pretreatment versus in small gauge vitrectomy alone for PDR. The pooled results indicated that preoperative anti-VEGF agents application was associated with fewer patients suffering from intraoperative iatrogenic retinal breaks during PDR vitrectomy. It was presumed that after preoperative anti-VEGF treatment, vascular leakage, as well as the retinal thickness and congestion, was reduced, and the tissue was more resistant to traction [5]. Less intraoperative bleeding also provide a clear retinal visualization, which may be the other reason for the decrease in iatrogenic retinal breaks [5].

Silicone oil is always used in cases with severe intraoperative complications, where a complete removal of the fibrovascular tissue was not possible, or if the surgery had been particularly long and complex [19]. Although silicone oil facilitates retinal reattachment and reduces the chances of intraocular hemorrhage by producing extended intraocular tamponade, it may lead to significant postoperative complications such as cataract, corneal decompensation, and intraocular pressure elevation [26]. In this study, the pooled result indicated a decreased incidence of silicone oil tamponade in the anti-VEGF group than in the control group. The reduced use of silicone oil may benefit PDR patients’ outcome, and the consequent decreased silicone oil-related complications may also improve the patients’ satisfaction.

Early postoperative VH in patients with PDR has always been a major concern for both surgeons and patients [27]. On the patient’s side, expectation regarding visual improvement is not met [27]. On the surgeon’s side, this postoperative complication interferes with fundus examination, detection of iatrogenic retinal breaks, and performing laser therapy [27]. In this meta-analysis, we analyzed the incidences of early postoperative VH and late postoperative VH separately. The pooled results revealed lower incidence of early postoperative VH in the anti-VEGF treated group but not late postoperative VH. The sources of early and late postoperative VH were reported to be different by previous studies [4, 17]. Early postoperative VH was associated with dissection of fibrovascular membranes, recurrent bleeding from initial bleeding site, surgically injured retinal tissue, and increased vascular permeability, while late recurrent VH was attributed mainly to the recurrent growth of neovascularization [4, 6, 15, 17, 28, 29]. Effects of anti-VEGF agents are known to be transient, and as intravitreal injected drug clearance is enhanced in vitrectomized eyes, the anti-VEGF effect would have lasted even shorter [15, 30]. Anti-VEGF therapy before surgery may have exerted anti-VEGF effects during the early postoperative period; it would diminish quickly, which results in similar late postoperative VH incidence in the case group and control group [15].

Rubeosis iridis and NVG occur frequently after vitrectomy for diabetic eye disease [31]. Although retinal ischemia is presumed to be the stimulus for iris neovascularization, the free diffusion of vasoproliferative factor induced by intraocular manipulation may also play a decisive role in the development/persistent of NVG after vitrectomy [31]. In this study, we achieved a borderline result [32] in the pooled analysis of postoperative rubeosis iridis/NVG. Considering the clinical significance of anti-VEGF therapy and the recommendations for borderline result reporting [32], we interpret this result as borderline significant. This result suggests that preoperative anti-VEGF treatment may lower the risk of postoperative persistence or deterioration of rubeosis iridis/NVG in a certain extent.

In this study, we also compared pre- and postoperative BCVA between the anti-VEGF treated group and the vitrectomy-alone group. The pooled results revealed no significant differences in both pre- and postoperative BCVA between the study group and controls. This result is consistent with the results of several previous studies [1, 13, 15] and the meta-analysis performed by Zhang et al. [6] but different from the meta-analyses conducted by Zhao et al. [5] and Zhao et al. [4]. In these former meta-analyses, 20G vitrectomy trials were included. The different technical characteristics and advantages in small gauge vitrectomy as compared with 20G vitrectomy may partially contribute to this difference. Further studies with larger sample sizes are needed to verify our result.

This meta-analysis has several limitations: First, the inconsistent dosage and interval between injection and diabetic vitrectomy, the different follow-up times of different trials, and the unreported baseline systemic data in some included studies might potentially influence the results of our study. Second, this meta-analysis was restricted to data from published articles. Third, as the number of small gauge vitrectomy trials is small, the optimal interval of preoperative anti-VEGF injection in small gauge vitrectomies was not evaluated.

Based on the available evidence, the use of anti-VEGF agents as a preoperative adjuvant treatment for small gauge vitrectomy in PDR patients could achieve easier surgery and fewer intra- and postoperative complications. Better-designed studies with larger sample sizes are needed to confirm our findings and further evaluate the efficacy of different anti-VEGF agents and the optimal interval and dosage for preoperative anti-VEGF therapy in PDR patients.

All analyses were based on published studies, and thus, no ethical approval and informed consent are required.

The authors have no conflicts of interest to declare.

No funds were used in the conduct and completion of this study.

Minghang Pei performed the literature search, data extraction, statistical analysis, drafted the manuscript, and revised the submitted manuscript. Guangming Wan conceived the study, coordinated and participated in the entire process of drafting, and revised the manuscript. Xinyu Zhao contributed to the literature search, data extraction, statistical analysis, and revision of the manuscript. All authors read and approved the final manuscript.

All data in this article were based on studies that have already been published. Further inquiries can be directed to the corresponding author.