PREVENT vs PCE thresholds for coronary artery calcium referral: Detection-optimized strategies in ELSA-Brasil.

BACKGROUND

Risk equations estimate 10-year ASCVD risk but were not designed to guide coronary artery calcium (CAC) testing, although guidelines consider CAC for risk refinement in borderline or intermediate-risk adults.

OBJECTIVES

To compare the Predicting Risk of Cardiovascular Disease EVENTs (PREVENT) with Pooled Cohort Equations (PCE) for detecting CAC, identifying optimized CAC referral thresholds.

METHODS

In a cross-sectional analysis of 3477 ASCVD-free ELSA-Brasil participants, 10-year ASCVD risk was estimated using both tools, and reclassification assessed. Discrimination for CAC >0 and CAC ≥100 Agatston units was compared using area under receiver operating characteristic curve (AUC-ROC). Prespecified thresholds (PCE 7.5%; PREVENT 3.7% in men, 4.9% in women) and Youden-derived cutpoints were evaluated for sensitivity, specificity, scanning proportion, and number needed to scan (NNS).

RESULTS

Discrimination for CAC ≥100 was similar (AUC 0.81 vs 0.80 for PCE and PREVENT, respectively; p > 0.05). Using standard thresholds (7.5% for PCE; 3.7%/4.9% for PREVENT in men/women), 36.5% vs 38.8% of with CAC ≥100 were classified below cutpoints, more often in women (70.8% vs 67.4%). Youden-derived thresholds (5.6% vs 2.7%) increased sensitivity (73.3% vs 78.4%) but reduced specificity (74.1% vs 66.7%) and expanded referral eligibility (30.6% vs 38.2%), respectively.

CONCLUSIONS

Both tools showed comparable discrimination for CAC ≥100, but standard cutpoints miss individuals, especially women. Lower, optimized thresholds (PCE 5.6%; PREVENT 2.7%) may improve detection but warrant further evaluation.