2021 年第 1 期 第 16 卷

非瓣膜性心房颤动首次射频消融术后复发的影响因素及临床预测模型的建立

Influencing factors of recurrence of nonvalvular atrial fibrillation after first radiofrequency ablation and establishment of clinical prediction model 

作者：晁鹏任澎

英文作者：Chao Peng Ren Peng

单位：新疆维吾尔自治区人民医院心血管内科，乌鲁木齐830001

英文单位：Department of Cardiovascular Medicine People′s Hospital of Xinjiang Uygur Autonomous Region Urumqi 830001 China

关键词：非瓣膜性心房颤动；射频消融术；复发；临床模型

英文关键词：Nonvalvularatrialfibrillation;Radiofrequencyablation;Recurrence;Clinicalmodel

• 摘要：

• 目的 探讨非瓣膜性心房颤动首次射频消融术后复发的影响因素并建立临床预测模型。方法 收集2012年1月至2018年12月于新疆维吾尔自治区人民医院首次行射频消融术的282例非瓣膜性心房颤动患者临床资料进行回顾性分析。根据术后心房颤动复发情况分为复发组（87例）和非复发组（195例）。运用Lasso回归方法筛选影响非瓣膜性心房颤动患者首次射频消融术后复发的预测因素。应用Logistic回归方法建立合适的模型，并绘制预测首次射频消融术后复发的列线图。绘制模型的校准曲线、计算C指数、绘制受试者工作特征曲线和决策曲线进一步验证预测模型的准确性。结果 Lasso回归筛选出病史≥1年、男性、年龄≥70岁、高血压、吸烟、心房颤动类型为非阵发性、糖尿病、左心房内径≥45 mm、左心室射血分数＜50%共9个预测术后复发的相关因素。根据Lasso回归分析筛选出的9个因素构建预测模型列线图，进一步制得评分系统，得分较高的前3个危险因素分别为左心房内径≥45 mm、高血压和心房颤动类型为非阵发性，得分分别为100、66、48分，左心室射血分数＜50%、合并糖尿病、心房颤动病史≥1年、吸烟、男性和年龄≥70岁得分分别为24、22、22、12、12、8分。检测评分系统的准确性，C指数为0.749，受试者工作特征曲线下面积为0.749，临床决策曲线分析得出患者复发概率在＞8%和≤71%这个范围内，这种模型在临床上应用的准确性、净获益最高，超出该范围此模型的准确性有限，净获益将大幅度下降。结论 病史≥1年、男性、年龄≥70岁、高血压、吸烟、心房颤动类型为非阵发性、糖尿病、左心房内径≥45 mm、左心室射血分数＜50%这9种因素均会影响非瓣膜性心房颤动患者首次射频消融术后复发的概率，本研究所制得的模型准确性较高。

• Objective To discuss the influencing factors of recurrence of nonvalvular atrial fibrillation after first radiofrequency ablation and to establish clinical prediction model. Methods The clinical data of 282 patients with non-valvular atrial fibrillation underwent first radiofrequency ablation (RFCA) in People′s Hospital of Xinjiang Uygur Autonomous Region from January 2012 to December 2018 were retrospectively analyzed. According to the recurrence of atrial fibrillation after operation, they were divided into recurrence group (87 cases) and non-recurrence group (195 cases). Lasso regression method was used to screen the predictors of recurrence in patients with non-valvular atrial fibrillation after first RFCA. Logistic regression method was used to establish a suitable model, and the nomogram was drawn to predict the recurrence of atrial fibrillation in patients after first RFCA. The calibration curve, C-index, Receiver operating characteristic (ROC) curve and decision-making curve were drawn to further verify the accuracy of the prediction model. Results Lasso regression showed that 9 factors predicted the recurrence of atrial fibrillation, including medical history≥1 year, male, age≥70 years old, hypertension, smoking, non-paroxysmal atrial fibrillation, diabetes mellitus, left atrial diameter≥45 mm, left ventricular ejection fraction(LVEF)<50%. According to the 9 factors selected by Lasso regression analysis, the predictive model nomogram was constructed, and the scoring system was further developed. The top three risk factors with high score were left atrial diameter≥45 mm, hypertension and non-paroxysmal atrial fibrillation, the scores of 100, 66, 48 respectively, and the score of LVEF<50%, diabetes mellitus, history of atrial fibrillation≥1 year, smoking, male and age ≥70 years old were 24, 22, 22, 12, 12, 8, respectively. The accuracy of the scoring system, C-index was 0.749, the area under the ROC curve was 0.749. The clinical decision-making curve analysis shows that the recurrence probability of patients was in the range of>8% and≤71%. This model had the highest accuracy and net benefit in clinical application. Beyond this range, the accuracy of this model was limited, and the net benefit would be reduced greatly. Conclusions  Medical history≥1 year, male, age≥70 years old, hypertension, smoking, non-paroxysmal atrial fibrillation, diabetes mellitus, left atrial diameter≥45 mm and LVEF<50% can affect the recurrence probability of non-valvular atrial fibrillation in patients after first RFCA. The accuracy of the model established in this study is high.