AS332 - Utility of the Aviir Risk Score in Predicting Incident Coronary Heart Disease in the Women's Health Initiative


PI: Tim Assimes - Stanford School of Medicine

WHI Study Overview
The objective of this study is to robustly test the performance of the Aviir algorithm to predict the risk of CHD in the short term in a subset of women enrolled in the WHI Observational Study (WHI OS) or the placebo arm of the WHI clinical trials (WHI CT). This study will complement the studies of cardiovascular disease in the WHI cohort by providing both the raw data of biomarker concentrations and the algorithmically calculated Aviir Risk Score, which will be valuable for future scientific research within the WHI study group.  Accordingly, the specific aims of this study are to:
1.      To measure 7 biomarkers in ~700 cases with an incident AMI, revascularization, or cardiac death within five years of baseline evaluation, and ~700 random baseline controls (case-cohort sampling design) from the WHI OS.  Where necessary, measure baseline serum lipids in cases and controls using standardized assays.
2.      To assess Aviir’s algorithm for its ability to predict near-term events (0-5 years from baseline evaluation) relative to the FRS among cases and controls at intermediate risk for clinical CHD at baseline.
3.      To assess the predictive ability of Aviir’s algorithm against additional more comprehensive reference predictive models that incorporate covariates not included in the FRS such as the use of antihypertensive agents, family history of CAD, and hs-CRP. Furthermore, assess the predictive value of the Aviir score in additional subgroups including subjects at either low or high baseline risk as per FRS, and subjects with no history of diabetes at baseline.
Synopsis of Participant Selection Criteria
              All centrally and locally confirmed cases of MI, CHD death, or self-reported cases of revascularization within 5 years of baseline blood draw were identified in the OS and CT control arms as potential cases from the August 14, 2009 database. Controls were OS and CT control participants free of all 3 case conditions within 5 years of their baseline blood draw. Cases and controls were subjected to the following exclusion criteria: (1) had history of cardiovascular disease, MI, stroke or TIA; (2) used Statin at baseline; (3) had Type 1 diabetes at baseline (being diagnosed as Diabetics before age 29 and on insulin at baseline); (4) did not have enough serum sample (<250ul); (5) did not sign supplemental consent except for those who were decease prior to supplemental consent; (6) had missing values on age, family history of MI, history of diabetes, systolic BP, diastolic BP, smoking status, use of antihypertensive or lipid altering drugs at baseline. A total of 829 cases and 55,582 controls met the inclusion criteria    
Synopsis of Laboratory Methods
MIRISK VP Assay Testing
1544 WHI serum samples were tested on the seven MIRISK VP immunoassays developed and validated on Luminex xMAP® (Luminex Corp. Austin, TX) and Meso Scale Discovery MULTI-SPOT® (MSD, Gaithersburg, MD) multiplex technologies.
Eotaxin, FasLigand, sFas were tested in a 3-plex assay on Luminex technology while CTACK, IL-16, HGF, MCP-3 were tested in a 4-plex assay on the MesoScale Discovery platform.
Samples were batched in 32 samples per plate and run in duplicate per plate. Each plate also had a calibration curve and four samples designated as process controls (also run in duplicate) and 8 replicates of background wells (sample diluent only).
The four samples were run on each plate to measure plate-to-plate reproducibility. They were sourced from Bioreclamation and are well-characterized pooled human sera prepared from clinically asymptomatic or “Normal” individuals, as well as individuals previously diagnosed with Rheumatoid Arthritis, Systemic Lupus Erythrematosus, and Cardiovascular Disease. A total of 50 plates were run and the average concentration of each sample plus 3 standard deviations was used to set an acceptable range. An assay on any given plate with process control values outside the acceptable range was removed as an outlier. Any individual sample with an intra-assay %CV > 30% was removed as an outlier. If a sample was detected below the Limit of Detection (LOD) of any assay, 0.5*LOD was substituted.
MIRISK VP Lipid Panel Testing
Serum samples were also tested on the following lipid panel: Cholesterol, HDL and Triglyceride.
Abbott Architect Plus ci4100 technology was utilized and methodologies were enzymatic, accelerator selective detergent and glycerol phosphate oxidase respectively.
LDL-C, VLDL-C, Chol/ HDL ratio and Non-HDL-C were subsequently calculated.