文献速递瓣膜外科

AI心电图年龄与再次主动脉瓣置换术后死亡率的关联

Transcatheter valve-in-valve versus surgical aortic valve replacement for failed bioprostheses: Artificial-intelligence electrocardiogram age as a marker of mortality.
PubMed 文献 · J Thorac Cardiovasc Surg · 2026-09-25

本研究比较了失败生物瓣行再次外科主动脉瓣置换(redo SAVR)与瓣中瓣经导管主动脉瓣置换(ViV-TAVR)的结局,并评估术前AI心电图年龄对全因死亡率的影响。经倾向性评分匹配后167对患者分析显示,ViV-TAVR术后跨瓣压差更高且严重瓣膜-患者不匹配更常见,但手术死亡率无显著差异;AI-ECG年龄差阳性是全因死亡的独立预测因子。

要点

  • AI-ECG年龄差阳性是再次主动脉瓣置换术后全因死亡的独立预测因子(HR 1.20)。
  • ViV-TAVR较redo SAVR有更高的出院前跨瓣压差和严重瓣膜-患者不匹配。
  • AI-ECG年龄差增大与术后房颤、输血、ICU停留和住院延长相关。

核心结论

  • AI-ECG年龄差与再次主动脉瓣置换术后死亡率独立相关,提供超越实际年龄的预后信息。
  • 年龄差增大还与多项术后结局指标相关,支持其作为无创生理储备指标的潜力。

研究设计与样本

  • 回顾性队列研究,纳入2010至2025年因生物瓣失败行再次AVR的1558例患者。
  • 排除合并手术者后剩244例redo SAVR和369例ViV-TAVR,经1:1倾向性评分匹配得167对。
  • 术前心电图用卷积神经网络估算AI-ECG年龄,年龄差定义为AI-ECG年龄减实际年龄。

主要发现

  • 匹配队列平均实际年龄72±11岁,AI-ECG年龄68±10岁,平均年龄差-3.6±9岁(范围-22至+15岁)。
  • ViV-TAVR较redo SAVR有更高的中位出院前跨瓣压差(13 [9-19] mmHg vs 9 [7-13] mmHg,P < .001)。
  • ViV-TAVR严重瓣膜-患者不匹配更常见(25% vs 11%,P < .001)。
  • 手术死亡率redo SAVR为2.4%,ViV-TAVR为0.6%(P = .37)。
  • 年龄差阳性独立预测全因死亡(HR 1.20,95%CI 1.16-1.24,P < .001),治疗方式与年龄差无显著交互作用(HR 1.02,95%CI 0.96-1.07,P = .571)。

临床怎么用

  • 对于生物瓣失败需再次AVR的患者,术前AI-ECG年龄差可辅助识别生理储备较差、死亡风险较高者。
  • ViV-TAVR虽手术死亡率低,但需关注更高的跨瓣压差和严重瓣膜-患者不匹配风险。

局限性

  • 回顾性设计,排除合并手术患者可能限制外推性。
  • AI-ECG年龄估算依赖特定卷积神经网络模型,未报告外部验证。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVE: To analyze outcomes of patients with failed stented bioprostheses undergoing redo surgical aortic valve replacement (redo SAVR) versus valve-in-valve transcatheter AVR (ViV-TAVR) and evaluate the impact of preoperative artificial-intelligence electrocardiogram (AI-ECG) age on all-cause mortality. METHODS: Between 2010 and 2025, 1558 patients underwent repeat AVR for a failed bioprosthesis. To obtain a more homogenous cohort, patients with concomitant procedures were excluded, leaving 244 redo SAVR and 369 ViV-TAVR cases. Propensity-score matching (1:1) was then applied, yielding 167 balanced pairs. Preoperative ECGs were analyzed using a convolutional neural network to estimate AI-ECG age. The age gap was defined as AI-ECG age minus chronological age, with positive values indicating a physiologically older age. RESULTS: The matched cohort had a mean chronological age of 72 ± 11 years and AI-ECG age of 68 ± 10 years. Mean age gap was -3.6 ± 9 years (range, -22 to +15 years). Compared with redo SAVR, ViV-TAVR was associated with greater median predismissal gradients (13 [9-19] mm Hg vs 9 [7-13] mm Hg, P < .001), and more frequent severe prosthesis-patient mismatch (25% vs 11%, P < .001). Operative mortality was 2.4% for redo-SAVR and 0.6% for ViV (P = .37). A positive age gap was an independent predictor of all-cause mortality (hazard ratio, 1.20; 95% CI, 1.16-1.24; P < .001). No significant interaction between treatment modality and age gap was observed (hazard ratio, 1.02; 95% CI, 0.96-1.07; P = .571). An increasing age gap was associated with measures of postoperative outcomes, including atrial fibrillation, blood product transfusion, prolonged intensive care unit stay, and prolonged hospitalization. CONCLUSIONS: AI-ECG age gap is independently associated with mortality after repeat aortic valve replacement and provides prognostic information beyond chronological age. Increasing age gap is also associated with several markers of postoperative outcomes, supporting its role as a noninvasive measure of physiologic reserve.

来源:J Thorac Cardiovasc Surg,查看原文。版权归原作者所有。

引用

  1. Danesh S, Schaff HV, Arghami A, et al. Transcatheter valve-in-valve versus surgical aortic valve replacement for failed bioprostheses: Artificial-intelligence electrocardiogram age as a marker of mortality. J Thorac Cardiovasc Surg. 2026. PMID: 42790622. doi:10.1016/j.jtcvs.2026.09.010

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。