Abstract
Purpose:
A number of new treatments have recently been approved for the treatment of diabetic macular edema (DME). These studies have been designed as phase 3 clinical trials to meet regulatory requirements and so they have a number of similarities. Indeed, the principal criteria of these trials have been to define the proportion of patients with a BCVA gain of ≥3-lines in an ITT population with the last observation carried forward. The objective here is to explore the methodological differences between DME studies including RISE & RIDE (ranibizumab; RANI); VIVID & VISTA (aflibercept; AFLI); FAME (fluocinolone acetonide implant; FAc); and, MEAD (dexamethasone implant; DEX).
Methods:
This is discussed in terms of HbA1C values, prior therapies, planned analysis and the use of rescue therapies.
Results:
HbA1C values: The spread of HbA1C values was not equivalent, which is important in the management of DME. Prior therapies: In FAME, all patients were treated initially with laser. In all other trials, patients were treatment naïve (MEAD, 28%; RANI, 70%; AFLI, 55 to 95%). Planned analysis: Some trials had a planned stratification (RANI and FAc) that optimize the comparison between the arms, the others (AFLI, DEX) had chosen planned subgroup analysis. In FAME alpha risk weighting was used to stratify patients and to define a chronic DME population. Use of rescue therapies: Exclusion criteria have a bearing on the final analysis and study outcomes. Indeed, in FAME, patients were not excluded based on rescue therapy. In MEAD, however, it was prohibited and was used to exclude patients from subsequent analysis. For AFLI, rescue therapy was allowed, but it was controlled, which may explain the differences between the responses seen in the placebo arms (i.e., 12 to 21%). The exclusion of patients in DEX and AFLI mean it is unclear how many patients received rescue therapy and there is a high chance of truncated visits. For FAc, 15% of patients in the treated-arm received off-protocol therapies and patients were included in follow-up visits and analysis.
Conclusions:
Trial design is a very important issue that is commonly overlooked to compare trials. Good safety results could reflect the trial design and rate of drop-outs. This emphasises the importance of understanding trial design and its bearing on subsequent analysis.