Machine learning and experimental validation of novel biomarkers for hypertrophic cardiomyopathy and cancers

Hualei Dai,Ying Liu,Meng Zhu,Siming Tao,Chengcheng Hu,Peng Luo,Aimin Jiang,Guimin Zhang

DOI: https://doi.org/10.1111/jcmm.70034

Abstract:Hypertrophic cardiomyopathy (HCM) is a hereditary cardiac disorder marked by anomalous thickening of the myocardium, representing a significant contributor to mortality. While the involvement of immune inflammation in the development of cardiac ailments is well-documented, its specific impact on HCM pathogenesis remains uncertain. Five distinct machine learning algorithms, namely LASSO, SVM, RF, Boruta and XGBoost, were utilized to discover new biomarkers associated with HCM. A unique nomogram was developed using two newly identified biomarkers and subsequently validated. Furthermore, samples of HCM and normal heart tissues were gathered from our institution to confirm the variance in expression levels and prognostic significance of GATM and MGST1. Five novel biomarkers (DARS2, GATM, MGST1, SDSL and ARG2) associated with HCM were identified. Subsequent validation revealed that GATM and MGST1 exhibited significant diagnostic utility for HCM in both the training and test cohorts, with all AUC values exceeding 0.8. Furthermore, a novel risk assessment model for HCM patients based on the expression levels of GATM and MGST1 demonstrated favourable performance in both the training (AUC = 0.88) and test cohorts (AUC = 0.9). Furthermore, our study revealed that GATM and MGST1 exhibited elevated expression levels in HCM tissues, demonstrating strong discriminatory ability between HCM and normal cardiac tissues (AUC of GATM = 0.79; MGST1 = 0.86). Our findings suggest that two specific cell types, monocytes and multipotent progenitors (MPP), may play crucial roles in the pathogenesis of HCM. Notably, GATM and MGST1 were found to be highly expressed in various tumours and showed significant prognostic implications. Functionally, GATM and MGST1 are likely involved in xenobiotic metabolism and epithelial mesenchymal transition in a wide range of cancer types. GATM and MGST1 have been identified as novel biomarkers implicated in the progression of both HCM and cancer. Additionally, monocytes and MPP may also play a role in facilitating the progression of HCM.