Early Versus Delayed Vitrectomy for Vitreous Hemorrhage Secondary to Proliferative Diabetic Retinopathy
Rodrigo Anguita, Lorenzo Ferro Desideri, Philipp Schwember, Neil Shah, Syed Naqib Ahmed, Antony Raharja, Janice Roth, Sobha Sivaprasad, Louisa Wickham
Abstract
OBJECTIVE: To compare the clinical outcomes of early pars plana vitrectomy (PPV) versus delayed PPV in patients with first episode of vitreous hemorrhage (VH) secondary to proliferative diabetic retinopathy (PDR). DESIGN: Retrospective, comparative, interventional study. SUBJECTS, PARTICIPANTS, AND/OR CONTROLS: Consecutive patients with type 1 or II diabetes diagnosed with new onset VH secondary to PDR who underwent PPV at Moorfields Eye Hospital between December 2014 and December 2016. Exclusions were prior vitrectomy, iris neovascularization, neovascular glaucoma, macular edema, or presence of tractional/rhegmatogenous retinal detachment. METHODS, INTERVENTION, OR TESTING: Patients were divided into two groups based on the timing of their surgery: early PPV (≤6 weeks) and delayed PPV (>6 weeks). Demographic and clinical features, including best-corrected visual acuity (BCVA), expressed in logMAR at baseline and 12 months were collected. Statistical analyses, including propensity score matching, were performed using Python 3.10, Scikit-learn, Pandas, and GraphPad Prism 10. MAIN OUTCOME MEASURES: BCVA at 12 months postoperatively, reoperation rates, and severity of complications. RESULTS: A total of 178 eyes were analyzed (48 early PPV, 130 delayed PPV). The mean (SD) number of weeks before surgery was 3.36 (SD 1.6) for the early PPV group and 22.56 (SD 17.23) for the delayed PPV group (P < .0001). Baseline BCVA prior to PPV was similar between groups (P = .08). At 12 months, the early PPV group had significantly better BCVA (0.40 logMAR vs 0.67 logMAR; P = .02). Patients without evidence of posterior vitreous detachment on ultrasound or OCT showed more pronounced differences (0.3 logMAR vs 0.7 logMAR; P = .001). The early PPV group had fewer sight-threatening complications (P = .005). Multivariable logistic regression identified initial BCVA, early PPV, and absence of preoperative panretinal photocoagulation as significant predictors of better visual outcomes. CONCLUSIONS: Early PPV significantly improves visual outcomes and reduces severe complications in patients with VH secondary to PDR. These findings support the benefits of early surgical intervention to enhance long-term visual prognosis in these patients. However, a randomized clinical trial is warranted.