A Machine Learning Approach Identifies Modulators of Albiglutide on Preventing Cardiovascular and Renal Outcomes in People with Type 2 Diabetes: A Revisit to the HARMONY trial

A Machine Learning Approach Identifies Modulators of Albiglutide on Preventing Cardiovascular and Renal Outcomes in People with Type 2 Diabetes: A Revisit to the HARMONY trial

The incretin system has become an important target in the treatment of type 2 diabetes in recent years, and glucagon-like peptide 1 (GLP-1) is of particular interest for its glucose-lowering effects. The physiological response to oral ingestion of nutrients, involving the incretin system, is reduced in some patients with type 2 diabetes but may be augmented by administration of GLP-1 receptor agonists. Recent trials have demonstrated a strong cardiovascular and renal protective benefit from GLP1 Receptor agonists (GLP-1RA). And evidence shows that individuals with type 2 diabetes (T2D) are likely to benefit more from GLP-1RA compared to older drugs such as sulfonylureas, a class of oral antidiabetic medications most often used among individuals with T2D as the second-line treatment in the US. However, how the benefit of GLP-1RA varies differently across population subgroups with different characteristics is not well understood. This knowledge is of vital importance because a recent study published by our team demonstrated a cost-induced racial disparity in the access to GLP-1RA in racial minorities. Due to the high cost, racial minorities are less likely to access GLP-1RA. Understanding the heterogeneous treatment effect of GLP-1RA can help prioritize population subgroups that benefit the most from GLP-1RA (i.e., high-benefit users), so that policy-level intervention (e.g., reducing the out-of-pocket payment) can start from removing barriers to drug access in these populations first. The Harmony Outcomes trial is a randomized controlled trial, testing the cardiovascular (CV) safety of albiglutide (GLP-1RA) among individuals with type 2 diabetes, who were at high risk of CV events. The treatment efficacy of albiglutide has been reported previously at a population level and over a few pre-defined subgroups. However, the heterogeneous treatment effect has not been comprehensively examined. The recent innovation in machine learning, the causal forest algorithm specifically, enables us to study treatment heterogeneity. The traditional machine learning algorithm can build a model that directly predicts the study outcomes. For example, a traditional machine learning model can use patient characteristics to predict the future risk of cardiovascular disease. The causal forest algorithm, on the other hand, predicts the “change of outcomes” as a result of an intervention. For example, this algorithm allows the user to find individuals whose “change of outcomes” is greater through intervention. In other words, this algorithm allows the users to identify the “high-benefit receiver” of the treatment.

[{ "PostingID": 19799, "Title": "GSK-GLP116174", "Description": "A long term, randomised, double blind, placebo-controlled study to determine the effect of albiglutide, when added to standard blood glucose lowering therapies, on major cardiovascular events in patients with Type 2 diabetes mellitus" }]