Machine learning-assisted lipidomics informs personalised prediabetes outcomes following lifestyle interventions

Abstract

Prediabetes is among the most rapidly growing global health challenges, affecting over 720 million individuals worldwide. Lifestyle modifications aimed at weight loss are the primary strategy for prediabetes management, offering glycaemic benefits, particularly in those with excess body weight. However, significant inter-individual variability in glycaemic responses and the risk of weight regain during long-term lifestyle interventions present major challenges, compromising effective prediabetes management. Currently, there is no established biomarkers to evaluate effectiveness of lifestyle interventions or to predict who are less likely to achieve glycaemic improvements. This critical gap restricts timely and personalised therapeutics, increasing the risk of type 2 diabetes and its complications.

Emerging lipidomics analyses revealed that shifts in circulating lipid profiles contribute to insulin resistance, β-cell dysfunction, and microbiome imbalances, all of which modulate prediabetes progression or reversion. Key lipid species, including signalling sphingolipids, pro-inflammatory lysophospholipids, nutrient-related glycerolipids, obesity-related glycerolipids, microbiome-derived short chain fatty acids, and metabolically active free fatty acids, have been widely implicated in prediabetes regulation. However, their potential as evaluative or predictive biomarkers for managing overweight/obesity-associated prediabetes through lifestyle interventions remains largely unexplored.

This thesis leveraged high coverage lipidomics to analyse serum samples from the Australian sub-cohort of the PREVention of diabetes through lifestyle Intervention and population studies in Europe and around the World study. The aim was to identify lipid biomarkers capable of evaluating or predicting glycaemic responses to an 8-week low-energy diet-induced acute weight loss (Project 1) and subsequent 3-year weight-maintenance interventions combining diet and physical activity (Project 2).