Objective To explore the longitudinal associations of body roundness index (BRI), visceral adiposity index (VAI), and metabolic score for visceral fat (METS-VF) with the risk of new-onset atrial fibrillation (AF).

Methods This study included participants from the UK Biobank who were free of AF or pregnancy at baseline and completed the first and second assessments of BRI, VAI, and METS-VF. The changes in BRI, VAI, and METS-VF were classified using K-means clustering analyses, and the cumulative adiposity indices were also calculated. The primary outcome was new-onset AF. Three Cox regression models were employed to investigate the longitudinal associations of the BRI, VAI, and METS-VF changes with the risk of incident new-onset AF. The results were presented as hazard ratios (HRs) and the corresponding 95% confidence intervals (CIs). Restricted cubic spline analyses were performed to explore potential non-linear associations between baseline or cumulative adiposity indices and the risk of new-onset AF. C-index analyses were conducted to evaluate the predictive value of BRI, VAI, and METS-VF for new-onset AF. Subgroup analyses were performed according to age, gender, race, smoking status, alcohol consumption, and physical activity. Polygenic risk scores were applied to account for genetic susceptibility and investigate potential interactions between adiposity indices and genetic risk. Univariate linear regression analyses were performed to evaluate the relationships of cumulative adiposity indices and magnetic resonance imaging and dual X-ray absorptiometry parameters, including visceral adipose tissue (VAT) volume, VAT mass, trunk fat volume, and trunk fat mass. We further applied the eXtreme Gradient Boosting (XGBoost) algorithm, with the feature importance being measured to evaluate the predictive value of each adiposity index for imaging parameters. Mendelian randomization analysis was further conducted to investigate the potential causal relationship between trunk fat mass and AF.

Results A total of 12 776 participants were included. Over a median follow-up of 9.60 years, 761 (5.96%) new-onset AF events were recorded. Participants were divided into four classes based on the changes in adiposity indices. In the fully adjusted model, compared to participants in Class 1 of BRI, those in Class 3 (HR＝1.30, 95%CI 1.04-1.63, P＝0.023) and Class 4 (HR＝2.17, 95%CI 1.61-2.93, P＜0.001) were associated with significantly higher risks of new-onset AF. Regarding METS-VF, participants in Class 4 of METS-VF also demonstrated a significantly higher risk of new-onset AF compared to those in Class 1 (HR＝1.66, 95%CI 1.15-2.39, P＝0.007). However, no significant association was observed between different classes of VAI and the risk of new-onset AF. For every 1 standard deviation increase in cumulative BRI, VAI, and METS-VF, the fully adjusted HRs of new-onset AF were 1.23 (95%CI 1.13-1.35), 1.02 (95%CI 0.94-1.10), and 1.23 (95%CI 1.12-1.35), respectively. Cumulative adiposity indices (BRI, VAI, and METS-VF) were divided into quartiles. Using the first quartile as reference, participants in the highest quartiles of BRI (HR＝1.40, 95%CI 1.10-1.79, P＝0.007) and METS-VF (HR＝1.44, 95%CI 1.13-1.83, P＝0.003) both exerted a significantly higher risk of new-onset AF. Regarding VAI, no significant association was observed (HR＝1.00, 95%CI 0.81-1.23, P＝0.988). Restricted cubic spline analyses revealed non-linear relationships between cumulative BRI, baseline/cumulative VAI, and baseline/cumulative METS-VF with new-onset AF risk (all P_overall＜0.05, P_non-linear＜0.05). In the C-index analysis, BRI demonstrated the highest predictive performance for new-onset AF, followed by METS-VF and VAI. Subgroup analysis indicated a stronger association between METS-VF and the risk of new-onset AF amongst participants younger than 60 years (P_interaction＝0.008). Polygenic risk score analysis stratified by genetic risk demonstrated a synergistic effect between BRI and genetic risk with new-onset AF, with the overall risk of new-onset AF increasing as both BRI and genetic risk increased. Linear regression analysis revealed a positive correlation between cumulative BRI with VAT volume, VAT mass, trunk fat volume, and trunk fat mass. The feature importance plot derived from the XGBoost algorithm indicated that cumulative BRI had the greatest predictive value on VAT volume, VAT mass, trunk fat volume, and trunk fat mass. Mendelian randomization analysis confirmed a significant causal relationship between trunk fat mass and AF.

Conclusions There are significant non-linear associations between BRI, METS-VF, and VAI with new-onset AF. Higher BRI and METS-VF are significantly associated with a higher risk of new-onset AF, whereas no significant association is observed for the VAI. BRI exhibits a positive correlation with VAT and trunk fat, and demonstrates superior performance in predicting new-onset AF compared to VAI and METS-VF. Monitoring and managing BRI may be important in the early detection and intervention of AF.