Different forms of evaluation of atherosclerotic load and its different predicted powers compared with framingham risk score

Abstract Introduction The screening of carotid and/or ileo-femoral subclinical atherosclerosis using vascular ultrasound could be a cost-effective tool in the reclassification of cardiovascular risk in primary cardiovascular prevention. But until today we don't have evidence indicating which atherosclerotic load indicator variable (such as plaque area, number of atherosclerotic plaques or affected territories) has the most predictive power for future cardiovascular events. Objectives a. Evaluate the prevalence of carotid and ileo-femoral subclinical atherosclerosis (SubAth). b. Evaluate the predictive power of each atherosclerotic load indicator variable for cardiovascular events (CVE). c. to compare those methods with Framingham Score for prediction of CVE. Material and methods We retrospectively evaluated patients from our registry called CARFARE (CARDIOMETABOLIC RISK FACTORS REGISTRY) conducted in the context of a cardiovascular primary prevention program (n=6108). In this analysis we included only those patients who completed a follow up ≥24 months from the first vascular evaluation. Exclusion criteria: <40 or >65 years old, BMI>40 kg/m2, previous cardiovascular and/or cerebrovascular event, chronic stable angina. We used Framingham score for the cardiovascular risk assessment. We evaluated the following atherosclerotic load variables using high resolution vascular ultrasonography (Mannheim Consensus): carotid plaques area (AR-C), ileo-femoral plaques area (AR-IF), and total (carotid+ileo-femoral, AR-TOT), number of affected territories (N-TER). Primary composite endpoints (P-CVE): myocardial infarction, stroke, hospitalization due to angina or acute heart failure. For prediction power analysis we used ROC curves and logistic regressions (unadjusted and adjusted by age, sex, BMI, major cardiovascular risk factors and treatment). Results We included 1431 patients (55.0±9.78 y.o., 38% women, follow-up period of 789±23 days). The total P-CVE rate was 3.77% during this period (54 events). The overall prevalence of atherosclerosis in this population was 60.7%. The AUC of the ROC curves for P-CVE prediction was 0.648, for Framingham 0.706, for AR-C 0.726, for AR-IF 0.746, for AR-TOT and 0.79 for N-TER (graph). The AUC of AR-TOT and N-TER were significantly higher than the AUC of the Framingham score (p=0.017 and p=0.0004, respectively) for prediction of P-CVE. The dichotomization criteria according to ROC were: AR-C>5.80 mm2, AR-IF>23.0 mm2, AR-TOT>43.9 mm2, and N-TER>1. In the logistic regressions adjusted for prediction of P-CVE the OR were: Framingham 1.05 (95% CI: 1.02–1.07), AR-C 4.00 (95% CI: 1.89–8.47), AR-IF 4.01 (95% CI: 2.02–8.32), AR-TOT 4.35 (95% CI: 2.13- 8.91) and N-TER 6.95 (95% CI: 3.05–15.8). Comparison of ROC curves for CVE Conclusions The carotid/ileo-femoral SubAth scan was a more potent predictor of cardiovascular events than the Framingham score, particularly those variables that indicate extension of multiterritorial affection like AR-TOT or N-TER.