Development and validation of clinical prediction models for cardiorespiratory fitness in atrial fibrillation patients following radiofrequency catheter ablation

Guiling Zhao^1,2,3Jian Sun²Qianji Che²Wenqing Xu⁴Mengmeng Song²Doa El-Ansary^5,6Roger David Adams⁷Jia Han⁸Shu Meng^2,4*Yi-Gang Li^2*

• ¹Department of Cardiology, Shanghai Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, Shanghai, China
• ²Department of Cardiology, Shanghai Jiaotong University School of Medicine Xinhua Hospital, Shanghai, China
• ³Department of Sport Rehabilitation, Shanghai University of Sport, Shanghai, China
• ⁴Department of Rehabilitation, Shanghai Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, Shanghai, China
• ⁵School of Health and Biomedical Sciences, RMIT University, Melbourne, Australia
• ⁶Department of Surgery, The University of Melbourne, Melbourne, Australia
• ⁷UC Research Institute for Sport and Exercise, University of Canberra, Canberra, Australia
• ⁸College of Rehabilitation Sciences, Shanghai University of Medicine and Health Sciences, Shanghai, China

Background: Assessment of cardiorespiratory fitness (CRF) is imperative in patients with atrial fibrillation (AF) who have had radiofrequency catheter ablation (RFCA). This study aimed to develop and validate CRF prediction models in this population. Methods: 141 AF patients with RFCA were recruited. The cardiopulmonary exercise test was used to assess CRF with VO₂peak and METsₘₐₓ. Multidimensional predictors (demographics, serum biomarkers, cardiovascular parameters, and motor function parameters) were analyzed through Spearman correlation analysis and stepwise multivariate linear regression analysis. The internal validity of the prediction equation was tested by paired Student’s t-test, Pearson correlation analysis and Bland-Altman analysis. Results: Sex, BMI, ln NT-proBNP, glucose (GLU), 6-minute walking distance (6MWD), and systolic blood pressure (SBP) were found to be significantly associated with CRF in this population. Multivariate linear regression generated the equations: VO₂peak = 35.080 − 0.286 * BMI − 1.927 * Sex − 1.090 * ln NT-proBNP + 0.011 * 6MWD − 0.039 * SBP − 0.512 * GLU, and METsₘₐₓ = 9.646 − 0.447 * Sex − 0.260 * ln NT-proBNP − 0.140 * GLU − 0.078 * BMI − 0.016 * SBP + 0.004 * 6MWD, (VO₂peak: adjusted R2 = 0.506, and METsmax: adjusted R2 = 0.469, both P ＜ 0.01). Pearson correlations between the predicted values and the measured values showed good validity (VO₂peak: r = 0.616, and METsmax: r = 0.581, both P ＜ 0.01). The Bland-Altman analysis showed that the predicted VO₂peak values were slightly lower than the measured values (mean difference = -0.13; 95% limits of agreement: -5.20 to 4.93), while the predicted METsₘₐₓ values were in close agreement with the measured values (mean difference = -0.00; 95% limits of agreement: -1.59 to 1.59). Conclusion: Sex, BMI, NT-proBNP, glucose, 6MWD, and SBP are robust predictors of VO₂peak and METsmax in AF population after RFCA. This study generates and internal validates the first multivariable CRF prediction models with easy-to use clinical paraments in AF patients after RFCA, thereby providing safe and effective alternatives to conventional CPX, which may help to optimize personalized patient management.