Physicians' Academy for Cardiovascular Education

Baseline risk of outcomes does not predict the treatment effect of GLP-1 receptor agonist

Effect of Exenatide Once-Weekly on Clinical Outcomes in Patients with Type 2 Diabetes Mellitus and Cardiovascular Disease: Insights from the EXSCEL Trial

Presented at AHA 2017 Scientific Sessions in Anaheim, CA, USA by Robert J. Mentz (Duke University, Durham, NC, USA)

Nov. 15, 2017 - news


The double-blind, placebo-controlled EXSCEL trial was conducted at 687 sites in 35 countries enrolling 14,752 patients. The trial investigated the effects of the GLP-1 receptor agonist exenatide 2 mg once weekly on CV-related outcomes in patients with type 2 diabetes (T2DM). The primary endpoint was MACE, a composite of CV death, non-fatal MI and non-fatal stroke.

After a median follow-up of 3.2 years (2.2-4.4), treatment with exenatide showed a reduction in primary endpoint compared to placebo (HR: 0.91, 95%CI: 0.83-1.00, P<0.001 for non-inferiority and P=0.061 for superiority) and all-cause mortality was reduced in the exenatide group compared to placebo (HR: 0.86, 95%CI: 0.77-0.97, nominal P=0.016).

This raised the question whether the magnitude of the treatment effect depended on a patient’s baseline risk. Therefore, a risk score for the endpoints of all-cause mortality and MACE based on baseline characteristics was created. The aim of this analysis was to evaluate whether the magnitude of the treatment effect depended on a patient’s risk profile.

A predictive model was used with stepwise selection of baseline characteristics and risk analysis was done by calculating a risk score for each patient. A new time-to-event model for each endpoint was developed including calculated risk score, treatment assignment and their interaction.

Main results


Baseline characteristics including age, prior CV events, comorbidity burden and laboratory parameters provided prognostic value for mortality and MACE in the EXSCEL trial population. The effects of exenatide on mortality and MACE were consistent across the spectrum of baseline risk.


Discussant Eldrin Lewis (Brigham and Women’s Hospital, Boston, MA, USA) said that in the past, diabetes trials have focused on surrogate outcomes, such as HbA1c, BP, LDL levels, eGRF and others. However, studies evaluating new diabetes drugs have shown an increased CV risk. Therefore, guidelines now recommend to evaluate the CV risk during all phase 2 and 3 trials testing new antidiabetic therapies. CV events should include CV mortality, MI, and stroke and hospitalization for ACS. The strategy in diabetes treatment is changing and drugs with non-glucose mechanisms are considered such as empaglifozin, canagliflozin, liraglutide, and semaglutide. Taking a closer look at patients enrolled in this trial showed that 73% had prior CVD with a diabetes duration of 12 years, 9.8% was Asian, 7.7% Hispanic, 6.0% Black, 70% had CAD, 22% CVD, 24% PAD, 16% HF and 3969 had no CVD. This trial showed ‘noninferior to placebo with respect to cardiovascular safety but was not superior to placebo with respect to efficacy’ and ‘difference in all-cause death was not considered to be statistically significant on the basis of the hierarchical testing plan’. Although this trial consisted of sound statistical methods, a large sample size and patients were well stratified, there were also some drawbacks: the primary endpoint was not met, there was no validation cohort, the clinical utility of the model can be discussed, there were potentially competing risks and there was no information on safety. Future steps that can be taken are examination of safety vs. efficacy across quintiles of risk, addressing disparate risk for all-cause mortality and MACE, creating a risk profile that can easily be administered by researchers and clinicians and evaluating the patients with pre-existing HF to get a better understanding of their risk.


- Our reporting is based on the information provided at the AHA 2017 congress -