An equation for estimating low-density lipoprotein-triglyceride content and its use for cardiovascular disease risk stratification

Figures

• 

  Figure 1. Correlation of BQLDL-TG with lipid parameters and validation of eLDL-TG equation. Relationship between LDL-TG measured by the BQ reference method (BQLDL-TG) (N = 40,202) and nonHDL-C (Panel A), TG (Panel B), and HDL-C (Panel C). Development of the equation for estimated LDL-TG (eLDL-TG) (Panel D). Equation parameters are from the training dataset (N = 20,191) and graphed results are from the validation dataset (N = 20,011). Results are colored by TG (Panels A, C, and D) and nonHDL-C (B).

• 

  Figure 2. Kaplan-Meier survival curves and lipid test evaluation for ASCVD events in the UKB dataset. Kaplan-Meier survival curves in the UKB dataset for people without any lipid medications or incomplete follow-up data (N = 271,760) for all ASCVD events were calculated for LDL-C quintiles (Panel A) and eLDL-TG quintiles (Panel B). Other lipid tests were divided into quintiles for survival curves and the Wilcoxon Chi-Square is ranked (Panel C). The same set of lipid tests was evaluated with logistic regression for ASCVD and the AUC is ranked (Panel D).

• 

  Figure 3. Discordance analysis and survival curves in the UKB dataset. Discordance analysis in UKB dataset for people without any lipid medications or missing follow-up data (N = 271,760). The percentile of LDL-C is plotted against the percentile of eLDL-TG and divided into quadrants (1-green, 2-blue, 3-purple, 4-red (Panel A) and Kaplan-Meier survival curves between these four quadrants are plotted in Panel B. Panel C shows unadjusted model for the ASCVD incidents per quadrants. Panel D shows model for the ASCVD incidents adjusted for nonlipid variables: age, sex, race, SPB, antihypertensive medication, diabetes, and smoking. * (Panel A) is concentration in mg/dl of the 50th percentile.

• 

  Figure 4. Percentile relationships between eLDL-TG and LDL-C in the NIH and NHANES datasets. Data is plotted as percentiles for eLDL-TG and LDL-C in an NIH dataset of NMR lipoprotein results (N = 13,793, Panels A, B). Results are colored by LDL particle number as a percentile (Panel A) and average LDL size as a percentile (Panel B). Data is plotted as percentiles for eLDL-TG and S-LDL-C in NHANES (Panels C–F). Results are colored by percentiles of nonHDL-C (Panel C), percentiles of TG (Panel D), lipoprotein phenotypes, according to the Sampson classification system (Panel E), and the pooled cohort equations (PCE) calculated as if all patients were 55-year old White males and expressed in percentiles (Panel F).

• 

  Figure 5. Evaluation of eLDL-TG as a risk-enhancer test in NHANES. The percentile of LDL-C is plotted against the percentile of eLDL-TG (Panel A) and shown in a Venn diagram (Panel B). Points are colored by which risk-enhancer tests are present. Group A (light grey, N = 28,079) are normolipidemic patients below the cut-off for LDL-C (160 mg/dl; 4.14 mmol/L), TG (175 mg/dl; 1.97 mmol/L) and eLDL-TG (80th percentile, 44.6 mg/dl; 0.50 mmol/L). Group B (green, N = 1,109) are only positive for the LDL-C cut-off. Group C (blue, N = 457) are only positive for the TG cut-off. Group D (red, N = 752) are only positive for the eLDL-TG cut-off and represent a potentially unrecognized high-risk group. Group E (yellow, N = 6,656) is positive for both eLDL-TG and either LDL-C or TG, or all three tests.