To examine whether the circulating substrate mix may be related to the incidence of heart failure (HF) and cardiovascular (CV) mortality and how it is altered by canagliflozin treatment.


We measured fasting glucose, free fatty acids (FFA), glycerol, β-hydroxybutyrate, acetoacetate, lactate, and pyruvate concentrations in 3,581 samples from the CANagliflozin cardioVascular Assessment Study (CANVAS) trial at baseline and at 1 and 2 years after randomization. Results were analyzed by univariate and multivariate Cox proportional hazards models.


Patients in the lowest baseline FFA tertile were more often men with a longer duration of type 2 diabetes (T2D), higher urinary albumin excretion, lower HDL-cholesterol levels, higher history of CV disease (CVD), and higher use of statins and insulin. When all seven metabolites were used as predictors, FFA were inversely associated with incident hospitalized HF (hazard ratio [HR] 0.33 [95% CI 0.21–0.55]), while glycerol was a positive predictor (2.21 [1.45–3.35]). In a model further adjusted for 16 potential confounders, including prior HF and CVD and pharmacologic therapies, FFA remained a significant negative predictor. FFA and glycerol also predicted CV mortality (HR 0.53 [95% CI 0.35–0.81] and 1.81 [1.26–2.58], respectively) and all-cause death (0.50 [0.36–0.70] and 1.64 [1.22–2.18]). When added to these models, background insulin therapy was an independent positive predictor of risk of death. Canagliflozin treatment significantly increased plasma FFA and β-hydroxybutyrate regardless of background antihyperglycemic therapy.


A constitutive metabolic setup consisting of higher lipolysis may be beneficial in delaying or preventing hospitalized HF; a further stimulation of lipolysis by canagliflozin may reinforce this influence.

Clinical trial reg. no. NCT01032629, clinicaltrials.gov

This article contains supplementary material online at https://doi.org/10.2337/figshare.19747645.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.
You do not currently have access to this content.