Serum uric acid levels significantly improve the accuracy of cardiovascular risk score models. (3rd October 2022)

• Record Type: Journal Article
• Authors: Moshkovits, Y; Tiosano, S; Kaplan, A; Kalstein, M; Bayshtok, G; Kivity, S; Segev, S; Grossman, E; Segev, A; Maor, E; Fardman, A
• Is Part Of: European heart journal. Volume 43(2022)Supplement 2
• Publication Date: 2022-10-03
• Subjects: Cardiology -- Periodicals; Heart -- Diseases -- Periodicals; 616.12005
• DOI: 10.1093/eurheartj/ehac544.2271
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Background: This study evaluated the impact of serum uric acid (sUA) on the accuracy of atherosclerotic cardiovascular disease (ASCVD) pooled cohort equations (PCE) model, Systematic Coronary Risk Evaluation score 2 (SCORE2) and SCORE2-Older Persons (OP). Methods: We evaluated 19, 789 asymptomatic self-referred adults aged 40–79 years who were screened annually in a preventive healthcare setting. All subjects were free of cardiovascular disease and diabetes at baseline. sUA levels were expressed as a continuous as well as dichotomous variable (categorized into sex-specific tertiles, with the upper tertiles defined as high sUA). Mortality and cancer data were available for all subjects from nationwide registries. The primary endpoint was the composite of death, acute coronary syndrome and stroke, after excluding subjects diagnosed with lymphatic spread cancer during follow up. Results: Mean age of study population was 50±8 years and 69% were men. During median follow up of 6 years [2.0–13.1], 1, 658 (8%) subjects reached the study endpoint. ASCVD, SCORE2 risk and high sUA were all independently associated with the study endpoint in the multivariable Cox regression model (p<0.001 for all). Continuous net reclassification improvement analysis showed an improvement of 13% in the accuracy of classification when high sUA was added to the PCE and SCORE2 models (p<0.001 for both). sUA remained independently associated with the study endpoint among normal-weight subjects in the SCORE 2 model (HR 1.3, 95% CI 1.1–1.6) but not among overweight individuals (p for interaction = 0.01). Addition of sUA to the models in normal-weight subgroup (N=6, 624) resulted in a significant 20% improvement in the model performance for both SCORE2 and ASCVD when sUA was incorporated as dichotomous variable (p<0.001 for ASCVD and p=0.026 for SCORE2 model). Conclusions: sUA significantly improves classification accuracy of PCE and SCORE 2 models. This effect is especially pronounced among normal weight subjects. Funding Acknowledgement: Type of funding sources: None.