Non-ischemic Primary and Secondary Cardiomyopathy
Jinquan Zhang, BMSc
undergraduate
West China School of Public Health, Sichuan University
Chengdu, Sichuan, China (People's Republic)
Jinquan Zhang, BMSc
undergraduate
West China School of Public Health, Sichuan University
Chengdu, Sichuan, China (People's Republic)
Hongyu Chen, BMSc
undergraduate
West China Medical Center of Sichuan University, Sichuan, China (People's Republic)
Lutong Pu, BMSc
Master
West China Hospital, Sichuan, China (People's Republic)
Zihuan Tang, BMSc
undergraduate
West China Medical Center of Sichuan University, Sichuan, China (People's Republic)
Weitang Qi, BMSc
Master
West China Hospital, Sichuan, China (People's Republic)
Jie Wang, PhD
Dr
West China Hospital
Chengdu, Sichuan, China (People's Republic)
.jpeg.jpg "Yuchi Han, MD photo")
Yuchi Han, MD
Professor, Medicine
The Ohio State University, Ohio, United States
Yucheng Chen, MD
Doctorate
West China Hospital, Sichuan, China (People's Republic)
Hypertrophic cardiomyopathy (HCM) is the leading cause of sudden cardiac death (SCD) in young adults. Although the 2014 ECS HCM-SCD risk prediction model has been widely used in clinical practice, its performance is variable and limited. Furthermore, a previous study has reported that radiomics features derived from late gadolinium enhancement (LGE) - cardiovascular magnetic resonance (CMR) could improve SCD risk prediction. Therefore, we aim to develop and validate a new SCD risk prediction model by combining existing individual clinical risk factors with radiomics features.
Methods:
326 radiomic features of 545 sequential participants with HCM who underwent native T1 and LGE imaging were collected. The endpoint of SCD or aborted SCD events was defined as SCD, resuscitating after cardiac arrest, and appropriate implantable cardiac defibrillator (ICD) discharge due to ventricular tachycardia or fibrillation. Feature selection was performed using the logistic regression via Least Absolute Systolic, Selective Operator (LRLasso) algorithm, and the Naive Bayes (NB) algorithm. Internal validation was conducted by the 1000 bootstrap method.
Results:
During a median follow-up of 52 months (interquartile range, 32 months), 44 participants (8%) with HCM experienced SCD or aborted SCD events. The novel HCM-SCD model for predicting individual risk of SCD at 5 years includes clinical predictors [age (hazard ratio (HR): 1.02, 95% CI: 1.00-1.04; P = 0.08), unexplained syncope (HR: 2.06, 95% CI: 0.93-4.59; P = 0.08) and left ventricular outflow tract gradient (HR: 1.00, 95% CI: 0.99-1.01; P = 0.47) ], LGE-radiomics features [SumOfSquares(3) (HR: 2.65, 95% CI: 0.43-16.4; P = 0.29) and Homogeneity(3) (HR: 0.58, 95% CI: 0.07-4.69; P = 0.61)], and T1 mapping-radiomics features [entropy(HR: 1.77, 95% CI: 1.24-2.55; P < 0.05)]. In the internal validation, the area under the curve of the novel model was 0.77 (95% CI: 0.69-0.86). The distribution of all points on the calibration curve is fairly close to the diagonal line predicting the endpoint after 5 years, indicating that the new prediction model performed well in predicting risk in participants with high-risk HCM.
Conclusion:
The novel SCD risk model combining clinical and radiomic features in HCM provides robust discrimination in identifying HCM patients at high risk of SCD.