The participants in this study were recruited from Kaiser Permanente Northern California, a large integrated healthcare system that provides comprehensive care to more than 3 million people in San Francisco and the greater Bay Area. Details of participant recruitment have been described previously[12, 13]. Briefly, we enrolled adults between 60 and 72 years old who lived within 50 miles of the research facility, and who did not have a diagnosis of cardiovascular disease, cancer, end-stage renal disease, liver failure, dementia, or human immunodeficiency virus infection. By January 2001, 3,054 apparently eligible subjects were randomly chosen to be sent letters inviting participation. From these invitations, a total of 1,024 subjects were recruited, consisting of 639 men and 385 women. In the final year of recruitment, men, Hispanics, and African-Americans were selectively oversampled in order to better match the cases enrolled in the Atherosclerotic Disease, Vascular Function, and Genetic Epidemiology study. After excluding people who were taking dyslipidemia medications from this analysis, 714 subjects were included in the study. The study protocol was reviewed and approved annually by the institutional review boards at both Stanford University and the Division of Research at Kaiser Permanente Northern California. Written informed consent was obtained from all participants.
Risk Factor Measurement
All the study subjects attended a research clinic visit at Stanford University and completed a health survey about medical diagnoses, medication use, smoking history, family history, race and ethnicity, dietary intake, and physical activity. Subjects were asked to bring their medications, which were independently reviewed and recorded by the interviewer. Resting blood pressure was determined using standard sphygmomanometers. Weight, height, and waist circumference were measured by trained and certified staff using a stadiometer, balance scale, and tape. Blood tests were drawn for biomarker analysis, which included fasting lipid levels, glucose, C-reactive protein, and Lp-PLA2 mass and activity levels[16, 17].
Stanford Seven-Day Physical Activity Recall
The Stanford Seven-Day Physical Activity Recall, a semi-structured interview, was used to estimate the amount of time that a person engaged in moderate-, hard-, and very hard-intensity activities during the previous seven days[18, 19]. A trained interviewer guided the subject through the recall process, day-by-day, to determine the duration and intensity of physical activities performed, as well as time spent sleeping. Time spent in light activity was estimated by subtracting the time included in sleep and moderate-, hard-, and very hard-intensity activities from the total hours in the recall period. Total energy expenditure was estimated by multiplying the hours spent in each level of activity with the estimated energy expenditure value for each intensity category.
Block Food Frequency Questionnaire
The Block Food Frequency Questionnaire was used to characterize dietary intake. It queries average consumption of 106 foods by portion size and includes 13 questions on dietary supplementation, six questions on restaurant eating, five summary questions, eight questions on fat use or low-fat foods, and seven demographic and health-related questions.
Self-reported race was collected during the eligibility screening survey, on the baseline health survey and from the Kaiser Permanente data sources. In addition, birthplace, grandparents' race, and grandparents' country of origin were collected. The algorithm for assigning race was as follows: if self-reported race and grandparents' race were concordant, then subjects were coded to that race category (> 80% of subjects). In discordant cases, race was assigned by a hierarchy of grandparents' race, grandparents' country of origin, self reported race on the baseline health survey, and self-reported race on the screening survey. Only the major racial groups of White, African-American, Hispanic and Asian are used for this analysis. This method of classifying race has been validated in this cohort by high-throughput genotyping of more than 450,000 single nucleotide polymorphisms (SNPs) among African-Americans and Whites.
Blood samples were collected at the baseline visit after an overnight fast and stored in aliquots frozen at -80 °C. Lp-PLA2 mass (ng/ml) was measured using a dual enzyme linked immunoassay (PLAC test, diaDexus Inc., South San Francisco, CA). Intra- and inter-assay coefficients of variation were < 5% and < 8%, respectively, and sensitivity across the assay range was < 0.5 ng/mL. Lp-PLA2 activity (nmol/ml-min) was measured by a colorimetric activity method (CAM test). Intra- and inter-assay coefficients of variation were < 4% and < 6%, respectively, and sensitivity across the assay range was < 5 nmol/ml-min.
Single Nucleotide Polymorphism Selection
A literature search for all of the documented SNPs related to Lp-PLA2 mass and activity levels was conducted. Only those SNPs that were both documented in the literature to be related to Lp-PLA2 and existed in our database were included in the analysis. The SNPs included in our database were selected for likely association with atherosclerosis as determined a priori from literature review and by comparing up and down-regulated genes in diseased and non-diseased vascular endothelium. The genetic covariates included three SNPs in the PLA2G7 locus (Lp-PLA2 gene) that have been documented to be associated with Lp-PLA2 activity: Ala379Val (rs1051931), Arg92His (rs1805017), and Ile198Thr (rs1805017). The SNPs were coded as 0 for homozygous major, 1 for heterozygous, and 2 for homozygous minor.
Baseline characteristics were presented as means and standard deviations for normally distributed continuous variables and as counts and proportions for categorical variables. Medians and interquartile ranges were used for non-normally distributed continuous variables. Differences in the baseline variables between racial groups were compared using a general linear model analysis of variance (ANOVA).
Linear regression was used to examine the association between race and Lp-PLA2 mass and activity in separate models. The covariates for multivariable analysis were selected a priori and included biological, lifestyle, demographic, and genetic characteristics. The biological covariates included age, race, history of diabetes mellitus, history of hypertension, quantitative systolic blood pressure measurement, body mass index (BMI), waist circumference, LDL cholesterol level, high density lipoprotein (HDL) cholesterol level, asymmetric dimethylarginine (ADMA), and C-reactive protein (CRP); the lifestyle and demographic covariates were current smoking status, physical activity, percent of calories from saturated fat, percent of calories from carbohydrates, education level, and income.
A simple linear regression model was first used to analyze the relation between race and Lp-PLA2 mass and activity adjusted for only sex. The biological, lifestyle, and genetic covariates were each included in separate multivariable models in addition to a single multivariable model with all the covariates. Data were analyzed using SAS, version 9.1 (SAS Institute, Cary, North Carolina).