Lipoprotein cholesterol

Do and colleagues¹⁶ (link) performed a two-sample MR analysis to evaluate the causal effect of low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C) and triglycerides on coronary heart disease (CHD) risk, using a total of 185 genetics variants. Summary association results were obtained from the Global Lipids Genetics Consortium¹⁷ (link) and the Coronary Artery Disease Genome-Wide Replication and Meta-Analysis Consortium,¹⁸ (link) and were downloaded from Do and colleagues’ supplementary material (standard errors were estimated based on the regression coefficients and P-values). Genetic variants were classified as instruments for each lipid fraction using a statistical criterion (P < 1 × 10⁻⁸), resulting in 73 instruments for LDL-C, 85 for HDL-C and 31 for triglycerides.
White and colleagues¹⁹ (link) performed a similar analysis, but with plasma urate levels rather than lipid fractions. 31 variants associated with urate levels (P < 5 × 10⁻⁷) were used as genetic instruments, and the required summary statistics were obtained from the GWAS catalogue [https://www.ebi.ac.uk/gwas/].

The Metabochip was designed by representatives of the Body Fat Percentage [9] (link), CARDIoGRAM (coronary artery disease and myocardial infarction) [10] (link), DIAGRAM (type 2 diabetes) [11] (link), GIANT (anthropometric traits) [3] (link), [12] (link), [13] (link), Global Lipids Genetics (lipids) [4] (link), HaemGen (hematological measures) [14] (link), ICBP (blood pressure) [15] (link), MAGIC (glucose and insulin) [16] (link)–[18] (link), and QT-IGC (QT interval) [19] (link), [20] (link) GWAS meta-analysis consortia. The array is comprised of SNPs selected across two tiers of traits (Table 1). Tier 1 is comprised of eleven traits deemed to be of primary interest: type 2 diabetes (T2D), fasting glucose, coronary artery disease and myocardial infarction (CAD/MI), low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, triglycerides, body mass index (BMI), systolic and diastolic blood pressure, QT interval, and waist-to-hip ratio adjusted for BMI (WHR). Tier 2 is comprised of twelve traits of secondary interest: fasting insulin, 2-hour glucose, glycated hemoglobin (HbA1c), T2D age of diagnosis, early onset T2D (diagnosis age<45 years), waist circumference adjusted for BMI, height, body fat percentage, total cholesterol, platelet count, mean platelet volume, and white blood cell count.
We included three design classes of SNPs on the Metabochip (Table 2):
In total, 217,695 SNPs were chosen for the array (Table 2). 20,970 SNPs (9.6%) failed during the assay manufacturing process, resulting in 196,725 SNPs available for genotyping. A summary file annotating each Metabochip SNP with ascertainment criteria, SNP assay, a list of unintended duplicate SNPs (Supplementary Table S4), and reference strand orientation for alleles is provided at http://www.sph.umich.edu/csg/kang/MetaboChip/.

Data were obtained from the third National Health and Nutrition Examination Survey (NHANES III), a probability sample of the US civilian, noninstitutionalized population that included an oversample of non-Hispanic blacks and Mexican Americans [23 ]. NHANES III, conducted over the period 1988–1994, was unique among large US surveys because it incorporated assays of serum apolipoproteins B and A1 (ApoB, ApoA1; restricted to survey years 1988–1991). The analytic population from the seven-year survey contained 4,447 male and 4,733 female participants who were aged 18+ years, not pregnant, had fasted 8–19 hours before their laboratory examination, and had data available on basic anthropometry and fasting serum TGs (excluding three persons with a TG concentration >15 mmol/L). Participants were asked to complete a household interview and a standardized examination, including measurement of the standing waist circumference (in the horizontal plane at the level just above the right iliac crest, at minimal respiration) [24 ,25 (link)]. Serum TGs were measured enzymatically after hydrolysis to glycerol (Hitachi 704 Analyzer; Boehringer Mannheim, Indianapolis, Indiana); the coefficient of variation was 3–5 percent over the study and across the clinical range. Additional details of all laboratory procedures are available elsewhere [26 ]. Calculations of low-density lipoprotein (LDL) cholesterol concentration were limited to participants with TG concentrations below 4.5 mmol/L (a requirement of the Friedewald equation [27 (link)]) who had fasted at least nine hours.
Sampling weights from NHANES III were used with the software programs SAS, SAS/Graph (Release 8.2, SAS Institute, Cary, NC), and SUDAAN (Release 8.0, Research Triangle Institute, Research Triangle, NC), to estimate the sizes of the represented adult populations, to describe the distributions in the population of risk factors associated with LAP and BMI, and to perform analyses using multivariable linear regression. The analyses thus incorporated sampling weights that accounted for unequal selection probabilities (clustered design, planned oversampling, and differential nonresponse) [28 ]. Based on the sampling weights assigned, the analytic cohort represented an estimated total of 100,048,439 US adults aged 18+ years, 50.5 percent (SE 0.7) of them women, with a distribution of race-ethnicity that was 76.0 (1.5) percent non-Hispanic white, 10.5 (0.6) percent non-Hispanic black, 5.3 (0.5) percent Mexican American, and 8.1 (1.1) percent other.
Sex-specific bubble plots of population density by the values for WC (to the nearest cm) and TG concentration (to the nearest 0.1 mmol/L) were prepared to represent US adults in three age ranges (figure 1). The area of each bubble on these plots is proportional to the estimated number of men or women represented by those intersections. A sex-specific hypothetical minimum value for WC (that is, the waist size that theoretically contained only abdominal muscle, viscera, and vertebral bone) was estimated by calculating the mean minus two standard deviations of the log-transformed WC value among the estimated 15.00 million persons aged 18–24 years. These estimated minimum WC values (65 cm for men and 58 cm for women) were very similar to minimum values reported in a survey of 18 year-old Canadians in 1981 [29 (link)].
The minimum WC values were used to define sex-specific origin points (near the left-lower corner of each panel in figure 1) that represented a hypothetical state in which TG concentrations were arbitrarily set to zero and the waist size (greater for men than women) comprised primarily lean truncal tissue. A comparison of sex-specific bubble plots by age group confirmed that increasing age was accompanied by a shift of the population density upward (increasing waist size) and to the right (increasing circulating TG). Consistent with this empirical observation, the LAP was defined to describe the extent to which an individual had travelled the route – in theory – of both increasing waist and increasing TG:
LAP for men = (WC [cm] - 65) × (TG concentration [mmol/L])
LAP for women = (WC [cm] - 58) × (TG concentration [mmol/L])
In order to avoid having nonpositive values for LAP, any waist values for men that were 65 cm or less (five men in the NHANES III sample, all aged 18–22 years) were revised upward to 66.0 cm. No women in the entire NHANES III sample had a waist circumference less than 58.4 cm.

The TG/HDL-C ratio was considered a continuous variable and calculated as follows: TG/HDL-C ratio = triglycerides divided by high-density lipoprotein cholesterol.

Plasma of 14-week-old mice were used to measure total cholesterol in plasma lipoprotein fractions. Lipoproteins were separated from plasma by size exclusion chromatography fast protein LC (FPLC) using a Superose 6 column on an FPLC system with a Model 500 pump from Waters (Milford, MA). In short, and as described in a previous study^{83 (link)}, a 100 μL aliquot of mouse plasma pooled equally from four different mice per group was injected into a 1.0-ml sample loop and separated with a buffer (0.15 M NaCl, 0.01 M Na₂HPO₄, and 0.1 mM EDTA) at a flow rate of 0.5 mL/min. Sixty fractions of 300 µL were collected, each containing its correspondent lipoproteins. Batch analysis was performed to measure circulating total cholesterol (Fujifilm Medical Systems U.S.A., Inc. cat. 999-02601) in plasma lipoprotein fractions according to the manufacturer’s protocol.