Abstract
IMPORTANCE: Previous studies have reported an association between hypoglycemia and cardiovascular (CV) events in people with type 2 diabetes (T2D), but it is unclear if this association is causal or identifies a high-risk patient phenotype. OBJECTIVE: To evaluate the associations between hypoglycemia and CV outcomes. DESIGN, SETTING, AND PARTICIPANTS: This secondary analysis was a post hoc assessment of the multinational, double-blind CARMELINA (Cardiovascular and Renal Microvascular Outcome Study With Linagliptin; 2013-2016) and CAROLINA (Cardiovascular Outcome Trial of Linagliptin vs Glimepiride in Type 2 Diabetes; 2010-2018) randomized clinical trials of the antihyperglycemic drug, linagliptin, a dipeptidyl peptidase 4 inhibitor. Participants were adults with T2D at high CV risk with or without high kidney risk. By design, participants in the CARMELINA trial had longer duration of T2D and had a higher CV risk than participants in the CAROLINA trial. Data analyses were conducted between June 2021 and June 2023. INTERVENTION: Linagliptin or placebo in the CARMELINA trial, and linagliptin or glimepiride in the CAROLINA trial. MAIN OUTCOMES AND MEASURES: The primary outcome for both trials was CV death, myocardial infarction (MI), or stroke (3-point major adverse CV events [3P-MACE]). For the present analyses, hospitalization for heart failure (HF) was added. Hypoglycemia was defined as plasma glucose less than 54 mg/dL or severe hypoglycemia (episodes requiring the assistance of another person). Associations between the first hypoglycemic episode and subsequent CV events and between nonfatal CV events (MI, stroke, hospitalization for HF) and subsequent hypoglycemic episodes were assessed using multivariable Cox proportional hazards regression models. Sensitivity analyses explored the risk of CV events within 60 days after each hypoglycemic episode. RESULTS: In the CARMELINA trial (6979 patients; 4390 males [62.9%]; mean [SD] age, 65.9 [9.1] years), there was an association between hypoglycemia and subsequent 3P-MACE plus hospitalization for HF (hazard ratio [HR], 1.23; 95% CI, 1.04-1.46) as well as between nonfatal CV events and subsequent hypoglycemia (HR, 1.39; 95% CI, 1.06-1.83). In the CAROLINA trial (6033 patients; 3619 males (60.0%); mean [SD] age, 64.0 [9.5] years), there was no association between hypoglycemia and subsequent 3P-MACE plus hospitalization for HF (HR, 1.00; 95% CI, 0.76-1.32) and between nonfatal CV events and subsequent hypoglycemia (HR, 1.44; 95% CI, 0.96-2.16). In analyses of CV events occurring within 60 days after hypoglycemia, there was either no association or too few events to analyze. CONCLUSIONS AND RELEVANCE: This study found bidirectional associations between hypoglycemia and CV outcomes in the CARMELINA trial but no associations in either direction in the CAROLINA trial, challenging the notion that hypoglycemia causes adverse CV events. The findings from the CARMELINA trial suggest that both hypoglycemia and CV events more likely identify patients at high risk for both. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT01897532 (CARMELINA) and NCT01243424 (CAROLINA).