方法：全面检索EMbase、The Cochrane Library、Pubmed、中国生物医学文摘数据库(CBM)、万方数据库(WanFang Database)、中国期刊全文数据库(CNKI)等数据库，检索时限均从建库至2017-01，语种不限，检索文献类型为随机对照试验； 采用Jadad量表及Cochrane协作网提供的风险偏倚评估工具评价所纳入研究的质量，采用GRADEpro软件对各评价指标的参数进行循证医学质量评价； 采用漏斗图检测所纳入文献的发表偏倚； 最后采用Review Manager 5.3软件行Meta分析。

结果：最终纳入发表时间介于2008/2016的16篇随机对照试验文献。共包含923例患眼，其中493例患眼进入PPV联合术前玻璃体腔内注射anti-VEGF组(试验组)，430例患眼进入单纯行PPV组(对照组)。Meta分析结果显示：(1)试验组术中出血的发生概率明显低于对照组\〖OR=0.06，95%CI(0.02～0.15)，P<0.01\〗，两组之间的差异具有统计学意义。(2)试验组手术持续时间明显短于对照组\〖WMD=-29.13，95%CI(-36.95～-21.30)，P<0.01\〗，两组之间的差异具有统计学意义。(3)试验组术后早期\〖OR=0.29，95%CI(0.19～0.44)，P<0.01\〗及晚期\〖OR=0.34，95%CI(0.20～0.58)，P<0. 01\〗玻璃体出血的概率均低于对照组，两组之间的差异均具有统计学意义。(4)试验组术后最佳矫正视力水平不优于对照组\〖WMD=-0.51(LogMAR)，95%CI(-1.10～0.08)，P=0.09\〗，两组之间的差异无统计学意义。(5)试验组术中医源性视网膜破裂发生概率低于对照组\〖OR=0.24，95%CI(0.14，0.40)，P<0.01\〗，两组之间的差异具有统计学意义。(6)在纳入的16篇文献共493例试验组患眼中，无1例报道出现与注射anti-VEGF类药物相关的明显眼内及全身不良反应。

结论：PDR患者行玻璃体切除术治疗前玻璃体腔内注射抗VEGF类药物是安全、有效的治疗措施，能够显著减少术中及术后并发症的发生概率，改善患者的整体治疗效果。其具体的实施方案，如注射药物的剂量、玻璃体腔内注射与手术的间隔时间等，还有待进一步的探索与完善。

METHODS: A Meta-analysis. A comprehensive retrieval was conducted using the database including EMbase, the Cochrane Library, Pubmed, CBM, WanFang Database, CNKI and so on. The retrieval time was limited from the building time of database to Jan. 2017. The randomized controlled trial was adopted with no requirements on languages. The Jadad scale and Cochrance cooperation were used as the tool of the risk and bias evaluation to analyze the literature quality. Quality estimation of evidence-based medicine on the parameters of each evaluation index was made via GRADEpro Software. The publishing biases of enclosed documents were inspected with funnel plot. At last, the Meta analysis was conducted with Review Manager 5.3.

RESULTS: Totally 16 literatures published from 2008-2016 were finally put into randomized controlled trial. A total of 923 cases were included, among which 493 cases were grouped as intravitreal injection of anti-VEGF before the combined operation of PPV group(the experimental group), and 430 cases were involved in simple PPV group(the control group). The results of Meta-analysis show:(1)The probability of intraoperative bleeding was remarkably lower than the control group \〖OR=0.06, 95%CI(0.02, 0.15), P<0.01\〗, with statistically significant differences.(2)The duration of operation was obviously shorter than that of the control group\〖WMD=-29.13, 95% CI(-36.95, -21.30), P<0.01\〗, with statistically significant differences.(3)The probabilities of both early and late postoperative vitreous hemorrhage were lower than those of the control group\〖OR=0.34, 95%CI(0.20, 0.58), P<0.01\〗, with highly statistically significant differences.(4)The best corrected visual acuity after surgery was no better than that of the control group, \〖WMD=-0.51(LogMAR), 95%CI(-1.10, 0.08), P=0.09\〗 with no statistical significance.(5)The occurrence of iatrogenic retinal rupture was lower than that of the control group\〖OR=0.24, 95%CI(0.14, 0.40), P<0.01\〗, with statistically significant differences.(6)Among the 493 patients included in the 16 literature, no case of ocular and general adverse effects due to anti-VEGF injections was reported.

CONCLUSION: It is effective and safe for the patients with proliferative diabetic retinopathy to inject anti-VEGF drugs into vitreous cavity before vitrectomy. And it can reduce the occurrence of complications during and after surgery, improving the general treatment effects. However, the specific implementation plans, such as the injection does, the interval between the injection and the operation and so on, still need further exploration and perfection.