Abstract
Importance
Predicting risk of Type 2 Diabetes Mellitus (T2DM) accurately allows allocation of resources to prevent its development. Fasting blood triglycerides are highly predictive for T2DM. Triglyceride remnant lipoproteins (TRL) more accurately reflect pathophysiological changes that underlie progression to T2DM, such as pancreatic steatosis and inflammation. We hypothesized TRL-related factors could improve risk prediction for development of T2DM.
Methods
We included individuals aged 35–74 years from the ELSA-Brasil cohort who had HbA1c and an oral glucose tolerance test at baseline. Regression models were used to predict incident T2DM, starting with medical history, metabolic syndrome traits (age, sex, hypertension, waist circumference [WC], HbA1c, triglycerides) and hsCRP, adding TRL-related measurements, including plasma concentration, particle size, as well as cholesterol and triglyceride content. TRL features were measured by NMR spectroscopy. Discrimination was assessed with area under receiver operator curves (AUROCs).
Results
Among 4,466 individuals at-risk, there were 353 new cases of T2DM after 3.7 (SD=0.6) years follow-up. We derived an 8-variable model with AUROC 0.891 (95% CI: 0.870–0.913). Overall TRL-related markers did not improve predictive capacity for T2DM. However, TRL particle diameter (TRLZ) increased AUROC, particularly in individuals without glucose abnormalities at baseline (Hba1c <5.7%). In this subgroup, AUROC increased from 0.761 (95% CI: 0.739–0.798 – Baseline-model) to 0.823 (95% CI: 0.783–0.862 – TRLZ model) (p-value=0.00006). Consistently, the net reclassification improvent (NRI) of the model with TRLZ improved by 10.27% (95% CI: 1.0–24.1, p=0.00041). In subjects with pre-diabetes at baseline, TRLZ is highly correlated with obesity, insulin resistance and inflammation (sum of z-scores for WC + HOMA-IR + hsPCR: R2=0.25), but this was less important in individuals with Hba1c <5.7% (R2=0.15). Sensitivity analyzes confirmed the results with different inclusion criteria (pre-diabetes defined with TTGO only) and excluding patients diagnosed with DM2 incident based only on fasting blood glucose.
Conclusions
TRL particle diameter improves prediction of T2DM, particularly in subjects with no glycemic abnormalities at baseline.
Funding Acknowledgement
Type of funding source: Public Institution(s). Main funding source(s): Brazilian Ministry of Health (Department of Science and Technology), Ministry of Science, Technology and Innovation, and the National Council for Scientific and Technological Development (CNPq)