To identify unique clinical phenotypes in type 2 diabetes (T2D) and investigate their treatment response to canagliflozin using latent class analysis. This was a pooled latent class analysis of the individuals in the CANVAS Program and CREDENCE trial. The co-primary endpoints were hospitalization for heart failure (HHF) and the composite of cardiovascular death (CVD) or HHF. Secondary endpoints included three-point major adverse CV events, its individual components, and all-cause mortality. We completed Cox proportional hazards models to evaluate the effect of canagliflozin across phenotypes. Four distinct phenotypes were identified: Phenotype 1 (n = 966, 6.6%), with the lowest prevalence of heart failure, kidney dysfunction and hypertension; Phenotype 2 (n = 4169, 28.7%), primarily comprising females with a high prevalence of atherosclerotic vascular disease (ASCVD); Phenotype 3 (n = 7108, 48.9%), predominately males with a high prevalence of ASCVD; and Phenotype 4 (n = 2300, 15.8%), possessing the highest prevalences of HF and renal dysfunction. A hierarchical increase in the risk of the primary endpoint was observed across the phenotypes, with the highest CV risk observed for Phenotype 4 (hazard ratio for HHF: 7.57 [95% CI: 4.19-13.69]). Canagliflozin significantly reduced HHF and the composite CVD or HHF across phenotypes (all P values for interaction > .05). We identified four clinically distinct T2D phenotypes with differential CV risks. Canagliflozin reduced the risk of CV events, irrespective of the phenotype, emphasizing its broad therapeutic acceptability.