Session Title: Vascular Diseases and Diabetic Retinopathy I
Session Date/Time: Thursday 11/09/2014 | 08:00-10:00
Paper Time: 08:16
Venue: Boulevard C
First Author: : B.Kuppermann USA
Co Author(s): : B. Ding I. Stoilov
Head to head studies comparing VEGF inhibitors for diabetic macular edema are not currently available. Comparison of results across clinical studies is limited by differences in the patient populations and baseline patient characteristics. In the absence of head to head data, we sought to facilitate comparison between studies using subanalysis to generate patient populations with similar baseline characteristics.
Data from phase 3, randomized, controlled clinical trials of ranibizumab (RESTORE, DRCR.net Protocol I, RISE, and RIDE), bevacizumab (BOLT), and aflibercept (VIVID and VISTA) were compared.
Study designs, primary outcome measures, inclusion and exclusion criteria, baseline characteristics, and visual acuity (VA) outcomes were compared side-by-side. Inclusion criteria were similar with the exception that RESTORE and Protocol I allowed patients with baseline VA up to 20/32 Snellen equivalent whereas the other studies capped vision at 20/40. In order to produce a patient population with comparable inclusion/exclusion criteria and baseline VA, a subgroup analysis was conducted including all patients in DRCR.net Protocol I who had a baseline VA of 20/40 Snellen equivalent or worse. In this exploratory, hypothesis-generating analysis, no formal statistical comparisons were conducted across studies and meta-analysis was not attempted.
Baseline VA varied according to inclusion/exclusion criteria and was highest in Protocol I (mean EDTRS letter score=66). Duration of diabetes at baseline ranged from 14.1 years (VIVID) to 17.6 years (VISTA). The VIVID trial had the highest percentage of Asians (20%). Only 1.5% of patients in VIVID/VISTA were level 61 (mild PDR) compared to 25-30% in RIDE/RISE. Across studies, most visual and anatomic improvement occurred during the first year of treatment, with generally stable vision in the second and third years despite reduction in injection frequency in ranibizumab PRN dosing regimen arms (mean annual injections as low as 2-4 in the second and third years). Subgroup analysis of Protocol I year 1 data (VA 20/40 or worse, n=581, 68% of study subjects) yielded a population with baseline EDTRS letter score of ~57 (comparable to RIDE/RISE, VIVID/VISTA and BOLT). The average number of injections was 8.1 for ranibizumab+prompt laser and 8.9 for ranibizumab+deferred laser. Mean BCVA gains from baseline, +11.7 letters (prompt laser) and +11.5 letters (deferred laser) were greater than in the all-comers population (9 letters). Among subgroup subjects, 43% (prompt) and 34% (deferred) gained ≥15 letters from baseline, which was similar to patients in RISE, RIDE, VIVID, and VISTA.
Although cross-trial comparisons have known limitations, when adjusted for baseline VA, comparison across DME trials showed generally comparable improvements in VA measures among patients receiving ranibizumab or aflibercept on monthly or less than monthly basis. VA gains with bevacizumab were a bit more modest. The comparatively smaller VA gains in Protocol I where patients with better VA were allowed into the trial compared with RISE, RIDE, VIVID, and VISTA are potentially related to ceiling effects owing to the VA inclusion levels, per study protocol.