ACCORD: Premixed insulin use, renal events predict CV outcomes in patients with diabetes

CHICAGO — A new analysis of data from the ACCORD trial showed that use of premixed insulin and renal microvascular events predicted the occurrence of MI, stroke and MACE in patients with type 2 diabetes.

During a presentation at the American Heart Association Scientific Sessions, Md Shamsuzzama, PhD, data scientist in the FDA’s Office of Computational Science and research scientist at Philips Research North America, said the findings “provide important clinical insights for reducing CV events in type 2 diabetic patients.”

Previously, the multicenter ACCORD trial assessed three potential strategies to lower the risk for major CV events in 10,251 patients with type 2 diabetes: intensive glucose control (HbA1c < 7%), intensive BP control and the treatment of multiple blood lipids with combination therapy of a fibrate and statin. Results of the trial showed that none of these strategies reduced the combined risk for fatal or nonfatal CV events. In fact, lower HbA1c levels had a positive effect on mortality.

“ACCORD, a large trial with a mortality safety signal, provided an opportunity to identify important predictors of clinical value,” Shamsuzzama said. “Despite ACCORD’s landmark findings, a data-driven, systematic evaluation of predictors of mortality and MACE among hundreds of variables has not been conducted.”

Therefore, Shamsuzzama and colleagues performed a sub-analysis of the trial to identify predictors of total mortality, CV death, hospitalization or death due to HF, stroke, MI and MACE in patients with type 2 diabetes. After identifying 378 risk factors — including demographic data, baseline history, HbA1c, BP, lipids, concomitant medications and other laboratory results — the researchers used Random Forest Survival (RFS), a machine-learning analytic strategy, to narrow down the top predictors.

Shamsuzzama reported that premixed insulin use and predefined composite renal microvascular events were important predictors of stroke, nonfatal MI and MACE. The variables predicted these events with 74% to 79% accuracy, according to Shamsuzzaman. Meanwhile, variables associated with changes in renal function predicted total mortality, CV mortality and hospitalization or death due to HF with about 90% accuracy. Changes in urinary renal function biomarkers also were predictive of fatal events.

Additional analyses with Cox regression models yielded similar results. However, Shamsuzzaman said the associations still warrant further investigation.

“From this study, we can conclude that RSF is a rapid and flexible data-driven approach to identify potential outcome predictors among a large number of variables,” he added. “With RSF, we can generate prediction models with low error rates and select predictors for outcomes of interest.” – by Stephanie Viguers

References:

Shamsuzzaman M, et al. Abstract 165. Presented at: American Heart Association Scientific Sessions; Nov. 10-12, 2018; Chicago.

NIH. Landmark ACCORD Trial Finds Intensive Blood Pressure and Combination Lipid Therapies do not Reduce Combined Cardiovascular Events in Adults with Diabetes. https://www.nih.gov/news-events/news-releases/landmark-accord-trial-finds-intensive-blood-pressure-combination-lipid-therapies-do-not-reduce-combined-cardiovascular-events-adults-diabetes. Accessed Dec. 6, 2018.

Disclosure: Shamsuzzaman reports no relevant financial disclosures.