Predictors of response to glucagon-like peptide-1 receptor agonists: a meta-analysis and systematic review of randomized controlled trials

The aim of the present meta-analysis is the identification of the characteristics of patients, which predict the efficacy on HbA1c of glucagon-like peptide-1 receptor agonists (GLP-1 RA). A Medline and Embase search for exenatide OR liraglutide OR albiglutide OR dulaglutide OR lixisenatide was performed, collecting randomized clinical trials (duration > 12 weeks) up to September 2016, comparing GLP-1 RA at the maximal approved dose with placebo or active drugs. Furthermore, unpublished studies were searched in the www.clinicaltrials.gov register. For meta-analyses, the outcome considered were 24- and 52-week HbA1c. Separate analyses were performed, whenever possible, for subgroups of trials based on several inclusion criteria. In addition, meta-regression analyses were performed for comparisons for which 10 or more trails were available. A total of 92 trials fulfilling the inclusion criteria were identified. In placebo-controlled trials (n = 41), the 24-week mean reduction of HbA1c with GLP-1 RA was - 0.75 [- 0.87; - 0.63]%. Shorter-acting molecules appear to be more effective in patients with lower fasting glucose, whereas longer-acting agents in patients with higher fasting hyperglycaemia. Obesity and duration of diabetes do not seem to moderate the efficacy of GLP-1 RA, whereas in non-Caucasians and older patients liraglutide could be less effective. At 52 weeks, only 9 placebo-controlled trials were available for preventing any reliable analyses. Using a variety of approaches (meta-analyses of subgroup of trials, meta-regression, systematic review of subgroup analyses in individual trials, and meta-analyses of subgroups of patients), we identified some putative predictors of efficacy of GLP-1 RA, which deserve further investigation.