Background: 
Dyslipidemia is a key risk factor for type 2 diabetes and cardiovascular disease. Novel lipidomics methods are providing new insights into the pathophysiology of these diseases. However, human studies are limited and no previous studies have examined relationships between the plasma lipidome and insulin secretion in humans.

Objective: 
We examined whether dysregulation of certain lipid species or classes was associated with reduced insulin sensitivity and/or impaired insulin secretion measured by gold-standard methods, as well as other cardiometabolic risk factors in overweight or obese, non-diabetic adults.

Methods: 
In 65 overweight or obese (mean BMI=31.5±5.2 kg/m2) non-diabetic adults (35M/19F; mean age=31.3±8.5 years), we performed lipid profiling of 459 lipid species across 26 lipid classes (liquid chromatography mass spectrometry). Gold-standard methods were used to assess body composition (% body fat, dual X-ray absorptiometry), insulin sensitivity (hyperinsulinemic-euglycemic clamps) and total, first-phase, and second-phase insulin secretion (intravenous glucose tolerance tests). Additional measures included anthropometry (BMI, waist-to-hip ratio), oral glucose tolerance tests (OGTT) for fasting and 2-hour post-OGTT blood glucose concentrations, and measurement of serum inflammatory markers: high-sensitivity C-reactive protein (hsCRP by immunoturbidimetry), tumor necrosis factor (TNF), interleukins (IL)-1β, -6, and -10, and adipokines: leptin, adiponectin, resistin, and adipsin (multiplex assay; flow cytometry). Multivariable regression analyses were performed with adjustment for age, sex, and % body fat, and all analyses were adjusted for multiple testing using Benjamini-Hochberg correction.

Results: 
On univariable analyses, fasting glucose, 2-hour post-OGTT glucose, fasting insulin, and insulin sensitivity were not associated with lipid species or classes (all p>0.05). However, total and second-phase insulin secretion were positively associated with the lysophosphatidylinositol (LPI) lipid class (β=781.9 mU/L, p=0.01 and β=521.3 mU/L, p=0.01, respectively). IL-1β was positively associated with the cholesterol ester (CE) lipid class (β=11.9 pg/ml, p=0.04) and adiponectin was inversely associated with several diacylglycerol (DG) and triacylglycerol (TG) lipid species (all p<0.05). After adjustment for age, sex, and % body fat, LPI remained associated with total (p=0.02) and second-phase (p=0.01) insulin secretion, and insulin sensitivity was negatively associated with the dihydroceramide (dhCer) lipid class (β= -1.91 mg/kg/min, p=0.01) and several DG and TG lipid species (all p<0.05). IL1β and adiponectin were no longer associated with any lipid classes or species after adjustment for age, sex, and % body fat (all p>0.05).

Conclusions: 
In a well-characterised cohort of overweight or obese non-diabetic adults, we found that increased dhCer and LPI were associated with reduced insulin sensitivity and impaired insulin secretory response, respectively, as measured by gold-standard methods. Our findings suggest that these lipid classes may be involved in the pathophysiology of type 2 diabetes.

Plasma lipidomics are associated with insulin sensitivity and secretion in overweight and obese non-diabetic adults