Subgroup analyses in GLP1-RA trial meta-analysis to understand the varying outcomes seen with different GLP-1RAs
Cardiovascular, mortality, and kidney outcomes with GLP-1 receptor agonists in patients with type 2 diabetes: a systematic review and meta-analysis of cardiovascular outcome trials
Introduction and methods
GLP-1 receptor agonists (GLP-1RAs) have been demonstrated to reduce CV risk in patients with type 2 diabetes (T2DM) . This class of drugs lowers HbA1c by stimulating glucose-dependent insulin secretion and by reducing glucagon secretion, gastric emptying and appetite. Moreover, treatment with a GLP-1RA is also associated with modest improvements in lipid levels and lower blood pressure and bodyweight.
Treatment is associated with a low risk of hypoglycemia [2,3]. Various GLP-1RAs have been developed, with different structure and duration of action. The trials in which their effect has been evaluated had varying sizes and were in different patient populations. Effects of these drugs on CV outcomes are not entirely consistent [4-14].
This is a systematic review and meta-analysis of all large, placebo-controlled, CV outcome trials of GLP-1RAs, with the aim to obtain robust estimates of the effects of this class on CV outcomes overall and in specific patients subgroups. Data of ELIXA (lixisenatide), LEADER (liraglutide), SUSTAIN-6 (semaglutide), EXSCEL (exenatide), Harmony Outcomes (albiglutide), REWIND (dulaglutide), and PIONEER 6 (oral semaglutide) were included (in total 56004 patients).
CV outcomes of interest were major adverse CV events (MACE), a composite of CV death, myocardial infarction (MI) and stroke. Moreover, the effects on hospitalization for heart failure (HHF) and death from any cause were studied. Regarding kidney outcomes, worsening of kidney function and a broader composite outcome (new-onset macroalbuminuria, decline in eGFR, progression to end-stage kidney disease, or death attributable to kidney causes) were considered. Severe hypoglycemia, retinopathy, pancreatitis and pancreatic cancer were monitored as key safety outcomes. By lack of individual level data, summary statistics from the included individuals trials were used. Estimated median follow-up was 3.2 years (range: 1.3- 5.4 year).
- In the pooled analysis, GLP-1RA treatment was associated with 12% lower risk of MACE (HR: 0.88, 95%CI: 0.82-0.94, P<0.0001). Number needed to treat (NNT) to prevent 1 MACE event was 75 (95%CI: 50-151) during a median follow-up of 3.2 years.
- When assessing the components of MACE separately, GLP-1RA therapy resulted in a lower risk of CV death (HR: 0.88, 95%CI: 0.81-0.96, p<0.0001), fatal or non-fatal stroke (HR: 0.84, 95%CI: 0.76-0.93, P<0.0001) and fatal or non-fatal MI (HR: 0.91, 0.84-1.00, P=0.043).
- Subgroup analyses were performed based on primary vs. secondary prevention, high vs. low baseline HbA1c (cutoff varied by trial), short vs. long duration of follow-up (<3 vs. ≥3 years) and daily vs. weekly drug dosing. No significant heterogeneity was seen for the effect of GLP-1RA between those groups.
- Possible heterogeneity of the effect of GLP-1RA was noted for homology to human GLP-1, as the exendin 4-based drugs (lixisenatide and exenatide) showed an HR of 0.95 (95%CI: 0.85-1.06) compared with HR: 0.84 (95%CI: 0. 79-0.90) for the other drugs, although the P-interaction was not significant (P=0.06).
- GLP-1RA treated patients showed a lower risk of HHF (HR: 0.91, 95%CI: 0.83-0.99, P=0.02).
- The broader composite kidney outcome was evaluated in 5 out of 7 trials, as PIONEER 6 and Harmony Outcomes did not provide data on kidney events. Treatment with a GLP-1RA lowered the risk of this outcome by 17% (HR: 0.83, 95%CI: 0.78-0.89).
- Worsening kidney function was not significantly reduced by GLP-1RA treatment (HR: 0.87, 95%CI: 0.73-1.03).
- Incidences of the safety outcomes did not differ significantly between GLP-1RA- and placebo groups.
This meta-analysis shows a 12% reduction in the risk of MACE, and mostly reflected a lower risk of death from CV causes and of fatal or non-fatal stroke. Several subgroup analyses were performed to address the proposed explanations for the different effects on CV outcomes observed in different GLP-1RA trials. A hint of an interaction between the treatment effect and homology to human GLP-1 was noted, but other pharmacological characteristics or differences in patient populations between the trials did not lead to significant heterogeneity between subgroups.