In this meta-analysis, in which only patients without metformin at baseline were included, GLP-1RAs (compared with placebo) were associated with a significant reduction in MACE incidence with a trending forward reduction in all-cause mortality. Metformin has been first line therapy recommended for treatment of patients with T2D for a long time. Results from CVOTs where GLP-1RAs an SGLT2 inhibitors added to standard therapy all have found non-inferiority with many finding superiority respect to placebo [6–14] led to a rethinking of the paradigm of metformin as the first drug choice, and GLP-1RAs arise in some guidelines as the first line therapy in patients with cardiovascular disease o high cardiovascular risk [15, 25].
This discussion is not focused on the glucose lowering effect of metformin or GLP-1RAs but on the impact in terms on reduction in cardiovascular events. Regarding this issue, the supporting data for the cardiovascular benefit of metformin is drawn for the most part from a sub study of the UKPDS trial; all patients included were overweight and control group was usual care (diet, with sulfonylurea or insulin). This study had a small sample size and was not powered to prove cardiovascular benefit . Currently there is no published data from placebo controlled cardiovascular outcomes trials with metformin in patients with T2D. There have been several meta-analyses to investigate the effects of metformin on cardiovascular events, but data have not been particularly conclusive [28–31].
Cardiovascular benefit observed with GLP-1RAs use in CVOTs is overwhelming. However, most of the patients evaluated in these trials received metformin at baseline. An exploratory subgroup analysis was recently performed in some of these trials, reporting that the reduction in MACE was independent of background metformin therapy. Two different meta-analysis including data from 3 GLP-1RAs CVOTs evidenced that GLP1-RAs were associated with a significant reduction in the risk of MACE of 20 % and 23 %, respectively [19–20]. None of the mentioned meta-analyses reported other outcomes than MACE. In this context, the results of this meta-analysis become relevant, since it includes a great number of patients from six different trials.
GLP1-RAs promotes glycemic reduction for the treatment of T2D by glucose-dependent control of insulin and glucagon secretion which results in a low risk of hypoglycemia. GLP1-RAs also decelerate gastric emptying, lower circulating lipoproteins, inflammation, increased satiety and decreased body weight . Furthermore, GLP1-RAs are associated with a variety of positive cardiovascular and renal effects beyond glycemic control. GLP-1RAs improves blood pressure, vascular tone and cardioprotection against myocardial ischemia/stunning . Although improving glycemic control is associated with a reduction of microvascular complications in people with T2D,  its impact in reduction of macrovascular complications is not so categorical. Despite a reduction in the incidence of myocardial infarction and death due to an early glycemic control reported in UKPDS, trials like VADT, ADVANCE and ACCORD failed to demonstrate cardiovascular benefit with an intensive glucose control strategy. [34–36]. From the above it follows that cardiovascular benefit exceeds glycemic control, the use of GLP-1RAs could maintain the positive cardiovascular profile regardless use of metformin.
The reduction in MACE observed in this meta-analysis in patients receiving GLP-1RAs without metformin follows the same line as CVOTs. Metformin increases glucagon-like peptide-1 secretion . In patients receiving GLP1-RAs this effect is avoided. This meta-analysis cannot answer the question about what is the mechanism of GLP-1AR that leads to a reduction in MACE, but it may suggest that this benefit does not depend on the use of metformin at baseline.
A meta-analysis of CVOTs, comparing GLP-1RAs and placebo, comprising 56004 patients showed not only a reduction in MACE but also a significant reduction in all-cause mortality and in cardiovascular death . In the present meta-analysis, which only includes patients without baseline metformin, no significant reduction in the mentioned results has been observed. The cause could be the small number of patients included in our study. The total of patients included in our meta-analysis represents 20% of the one mentioned above. The fact that only four studies reported all-cause of death and three trials cardiovascular death, could reduce even more the power to detect a difference between these two groups of patients. Since metformin trials didn't show a reduction in these endpoints, we don't think that this is the reason why we didn't observe the benefit described before.
This meta-analysis presents several limitations. First, limitations related with clinical heterogeneity (popular characteristics, different schemes of antihyperglycemic drugs, different follow-up). However, the sensitivity analysis showed the same directionality of the results. Second, the analysis included only trial-level data without having the individual data. Consequently, exploratory analysis of certain subgroups according to baseline characteristics could not be performed. Third, the characteristics of patients who were not treated with metformin at baseline may be not necessarily similar to those of the total populations of the included studies. Unfortunately, the characteristics of the subgroups of patients without metformin at baseline have not been published in all included studies. However, the characteristics of the total populations were similar. The reported body mass index ranged between 30.6 and 32.5 kg/m2 and the baseline HbA1c level ranged between 7.3% and 8.7%. Similarly, the use of statins ranged between 72.2% and 86.4% and the use of insulin ranged between 44.7% and 60.4%. Finally, the number of studies reporting all cause mortality and cardiovascular mortality was small. Therefore, it is necessary to have larger studies to confirm or refute our findings.
In conclusion, in the present meta-analysis GLP-1RAs reduced the incidence of MACE in patients with T2D regardless use of metformin. These results support the fact that metformin would not be needed to obtain positive cardiovascular effects when this class of drugs are administered.