Oral Presentation The Annual Scientific Meeting of the Endocrine Society of Australia and the Society for Reproductive Biology 2014

Lipid profiling: Applications in diagnosis and disease risk stratification  (#97)

Peter Meikle 1
  1. Baker IDI Heart & Diabetes Institute, Melbourne, VIC, Australia

The metabolic syndrome incorporating obesity, hypertension, dyslipidemia and elevated plasma glucose has reached epidemic proportions in many countries leading to an increased prevalence of type 2 diabetes (T2D) and cardiovascular disease (CVD).  Dyslipidemia, as assessed by standard measures (raised plasma triglycerides and LDL-cholesterol, and decreased HDL-cholesterol) is an independent risk factor for T2D and CVD.  However, current risk prediction algorithms have limited accuracy.   Further to this, the mechanistic links between dyslipidemia, T2D and CVD are complex and not well understood.  Lipidomics presents a new set of tools to address these issues.

We have developed a targeted lipidomics platform using liquid chromatography electrospray ionization-tandem mass spectrometry to profile 300-400 lipids from 10 mL plasma.  We have applied this technology to multiple clinical and population based cohorts to define the plasma lipid profiles associated with T2D and CVD and evaluate the potential application of these profiles to diagnose, assess disease risk and inform on disease biology. 

Regression analysis adjusting for covariates (age, sex, systolic blood pressure and obesity) identified multiple lipid species that were significantly associated with T2D or CVD.  Many of these lipids also displayed an association with disease severity suggesting that they are altered prior to the onset of acute stage disease.  Multivariate analysis incorporating unsupervised feature correlation minimization and reliefF feature selection was employed to create and test multivariate classification models incorporating different numbers of lipids and other risk factors.  Relatively few lipids (3-20) were required to achieve maximum classification and prediction accuracy.  Models based on lipids generally performed better than models based solely on traditional risk factors.  Validation of these models on independent cohorts and application to prospective studies is providing additional evidence of these findings.    

Plasma lipid profiling can provide insight into disease pathogenesis and may contribute to a new approach to risk stratification for T2D and CVD.