Methods
Study
The study cohort consisted of 917 women presenting with symptoms and/or signs of ischemic heart disease undergoing coronary angiography enrolled in the National Heart, Lung, and Blood Institute–sponsored WISE and followed for adverse outcomes. Exclusion criteria for WISE included any contraindication to provocative stress testing; thus, enrolled subjects almost exclusively had stable symptoms including typical, atypical, and nonanginal chest pain. Myocardial ischemia was defined as an ischemic response on any noninvasive tests, the details of which are reported elsewhere.
Baseline Evaluations
Initial evaluation in addition to coronary angiography included collection of demographic, medical history, symptom data, physical examination, and blood sampling for lipids, reproductive hormones, and inflammatory markers. The design and methodology are described in detail elsewhere.
Coronary Angiography Core Laboratory
All angiograms were quantitatively evaluated (WISE angiographic core laboratory) masked to all clinical findings including symptoms and results of noninvasive testing. All coronary segments with any visual abnormality (even luminal irregularities) were measured quantitatively. Cine film (rare) was measured with electronic calipers, and digital recordings on CD-ROM (most) were analyzed using a computer-based edge detection algorithm described previously. An angiographic severity score was prospectively developed that assigned points according to the category of severity of percent diameter stenosis (0–19, 20–49, 50–69, 70–89, 90–98, and 99–100) adjusted for the presence of partial or complete collaterals. Scores were weighted by lesion location, with more proximal lesions receiving a higher weighting factor. Individual lesion point totals were then added together to yield a final score. A severity score of 5 was assigned to those women with no CAD. A severity score ≥50 represents severe, diffuse 3-vessel obstructive CAD.
Definitions
No CAD was defined as normal-appearing coronary arteries and, when measured, no diameter stenosis ≥20% diameter narrowing in any epicardial coronary artery because in our experience, (1) any luminal irregularity when measured quantitatively will yield at least a 20% diameter reduction versus a completely normal reference and (2) it is common to obtain a 0 to 19% diameter reduction when measuring normal segments as a result of vessel tapering. Nonobstructive CAD was defined as at least 1 diameter stenosis ≥20 but <50%. Obstructive (single, double, or triple vessel) CAD was defined as at least 1 diameter stenosis ≥50%.
Follow-up Procedures
Each patient gave informed consent, and all centers had institutional review board approval for the inclusion of patients in this targeted cohort study as well as for collection of follow-up data.
Outcome data used in this report were collected in 2 consecutive collection phases. During the first phase, patients were contacted at 6 weeks and at 1-year intervals after enrollment. During telephone contact, a scripted interview was completed by an experienced nurse or physician at the respective center: each patient or family member was queried for occurrence of major adverse cardiac events or hospitalizations. In the event of death, a death certificate and/or physician narrative was obtained. During the second phase, for those who were alive at last contact and had not withdrawn consent, this was later followed by a National Death Index search. All deaths were adjudicated as cardiovascular (CV) or non-CV by a committee of senior WISE investigators blinded to angiographic findings. Major adverse events included CV mortality and nonfatal myocardial infarction (MI). A secondary outcome was hospitalization for angina.
Statistical Analysis
Continuous variables were expressed as means ± SD. The primary measure of interest for the prognosis analysis was time to the first occurrence of either death or nonfatal MI. Cox proportional hazards modeling was used to test the associations between angiographic findings and adverse outcomes. Because of the skewed distribution of the CAD severity score, it was log transformed for this analysis. Multivariate modeling was conducted in subsequent steps: (a) unadjusted associations, (b) adjusting for age, and (c) adding to the models known cardiac risk factors that were also strong predictors in this cohort, including diabetes, ever smoking, and hypertension. From the Cox proportional hazard model, the relative risk ratio and 95% CIs were calculated. Unadjusted Kaplan-Meier survival curves were also calculated. Receiver operator characteristic (ROC) analysis was used to estimate the optimal threshold among the WISE angiographic severity score classes for predicting CV mortality as well as estimating sensitivity and specificity. All tests were 2 sided, and a P value <.05 was considered statistically significant. Data were analyzed using SAS version 9.3 (Cary, N.C).