AUTHOR=Zhang Bo , Zhou Zhulin , Zhang Bingjie , Wang Dan TITLE=Efficacy and Safety of Various Treatments for Proliferative Diabetic Retinopathy: A Systematic Review and Network Meta-Analysis JOURNAL=Frontiers in Pharmacology VOLUME=12 YEAR=2021 URL=https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2021.709501 DOI=10.3389/fphar.2021.709501 ISSN=1663-9812 ABSTRACT=
Diabetic retinopathy is the main cause of visual impairment and blindness. The proliferative diabetic retinopathy at the severe stage of diabetic retinopathy is more harmful to vision and even leads to total blindness. To evaluate the visual acuity, central retinal thickness, and adverse reactions of various treatments for proliferative diabetic retinopathy through a systematic network meta-analysis. The relevant research published in English or Chinese from January 1, 2011, to February 1, 2021, was systematically searched by using PubMed, science network, EMBASE, MEDLINE, Cochrane Library, China National Knowledge Infrastructure, Wanfang, and other electronic databases. A total of 15 studies were selected, including 3,222 eyes of PDR patients. Our results show that in terms of visual score improvement, ranibizumab alone (69.90%) and laser + ranibizumab (67.90%) are the best. However, if the groups were grouped again according to the dose and times of ranibizumab injection, the results showed that 0.5 mg ranibizumab injection per month (58.0%) had the best effect on vision improvement. For the change of central retinal thickness, the thickness decreased the most after the laser combined with ranibizumab (96.5%). After the same subgroup analysis, the results were further refined into the best effect of laser combined with 0.3 mg ranibizumab per quarter (72.7%). In addition, our analysis of complications also showed that the overall incidence of adverse reactions of PRP (11.1 ± 12.4, %) was greater than that of ranibizumab (10.6 ± 13.0, %). However, more high-quality randomized controlled trials with longer follow-up using standard methods are still needed to verify the correlation.