Prediction and Risk Stratification for 30-Day Mortality in Stable Acute Pulmonary Embolism: Recalibration of the Multimarker Calculator and Evaluation of the AHA/ACC Classification
Yuzhi Tao, Xianglong Meng, Yu Zhang, Qin Luo, Chunling Dong, Hong Chen, Jie Zhang, Yiwei Shi, Xiaomao Xu, Maoyun Wang, Ling Zhu, Juhong Shi, Yingqun Ji, Hong Chen, Zhe Cheng, Yongjun Tang, Yanxia Li, Chaosheng Deng, Pinyao Lu, Linfeng Xi, Xincheng Li, Rui Liang, Yuhao Li, Zhushan Fu, Zhaofei Chen, Shuai Zhang, Jixiang Liu, Ran Miao, Changming Xiong, Behnood Bikdeli, Manuel Monreal, Shengfeng Wang, Zhenguo Zhai, Chen WangAbstract
Background
Risk stratification in haemodynamically stable acute pulmonary embolism (PE) remains challenging. The multimarker calculator (MMC) provides individualized 30-day mortality estimates, whereas the 2026 AHA/ACC Acute PE Clinical Categories (AHA/ACC classification) provide a phenotype-based framework incorporating clinical severity, right ventricular dysfunction, cardiac biomarker elevation, and respiratory compromise. We evaluated these approaches and recalibrated the original MMC in a Chinese multicentre cohort.
Methods
We included adults with haemodynamically stable acute PE from the prospective, multicentre China pUlmonary thromboembolism REgistry Study (2012–2021). The primary outcome was 30-day all-cause mortality. The MMC, 2019 European Society of Cardiology (ESC) model, and AHA/ACC classification were assessed using discrimination, decision-curve analysis, risk classification, and subgroup analyses. Logistic recalibration updated the original MMC linear predictor. The AHA/ACC classification was evaluated using 7-, 4-, and simplified phenotype-based 3-category schemes.
Results
Among 11,568 patients, 331 died within 30 days (2.9%). The MMC and 7-category AHA/ACC classification showed better discrimination than the ESC model (AUC, 0.704 and 0.680 vs. 0.606). The original MMC overestimated intermediate-high-risk mortality (mean predicted risk, 16.2%; observed risk, 9.4%). Recalibration improved calibration and identified a smaller intermediate-high-risk subgroup comprising 2.2% of patients, with 13.2% observed mortality. In the 3-category AHA/ACC classification, mortality increased stepwise across category B, partial phenotype involvement, and dual cardiopulmonary phenotype involvement (0.7%, 2.7%, and 6.7%).
Conclusions
The recalibrated MMC may be useful for individualized 30-day mortality risk estimation, whereas the AHA/ACC classification may offer a simple framework for phenotype-based risk grouping.