Prognostic Risk Assessment of Hypertrophic Cardiomyopathy in The Context of Sarcomere Gene Variation : Insights from Multimodal Phenotypic Characterization Using Unsupervised Clustering Approaches
Year: 2025
Type: dissertation
Study design: Observational Study
Abstract: Hypertrophic Cardiomyopathy (HCM) is a disorder characterized by abnormal thickening of the heart muscle, which can impair normal cardiac function and increase the risk of severe complications, including sudden death, especially in young people and athletes.With a prevalence of approximately 0.16%-0.29% in the general adult population, HCM is primarily associated with mutations in genes encoding sarcomere proteins, most notably MYBPC3 and MYH7.Despite advances in imaging and genetics, HCM exhibits considerable genetic and phenotypic heterogeneity, complicating risk stratification, prognosis, and treatment response.To confront these challenges, this thesis presents stratification models that leverage multimodal clinical data to classify HCM patients and their relatives by risk level, with respect to their mutation profiles.Furthermore, this study proposes a methodological framework for monitoring transitions between risk clusters during follow-up periods through longitudinal dynamics assessment using the most recent clinical data, thereby contributing insights into pathological evolution and therapeutic response.This study utilizes multimodal clinical data from approximately 2600 individuals affected by HCM, including clinical health records, family history, prior cardiovascular events, imaging data, medication usage, laboratory results and other physiological metrics.From a genetic perspective, the dataset also comprises variants identified in those individuals, spanning across 8 sarcomeric genes and 40 others, along with their verified classifications.Accordingly, 3 mutation status groups were identified to categorize patients based on their variant profiles.Subsequently, quantitative features from multiple modalities were extracted and summarized using statistical metrics, focusing on temporal changes and progression of the disease.Data were standardized and transformed for compatibility, followed by dimensionality reduction using PCA and reconstruction error to determine the threshold.Three distinct clustering algorithms: K-Means, DBSCAN, and Gaussian Mixture Models were applied separately to each mutation subgroup.Optimal clustering parameters were selected using a proposed evaluation framework comprised of 3 normalized validity metrics: Davies Bouldin Index, Silhouette Score, and Calinski-Harabasz Index.The result shows that K-Means outperforms GMM and DBSCAN in cluster separation and intra-cluster compactness.Optimal clustering into 2 groups revealed variation in membership across the 3 SARC subsets.The analysis revealed that Cluster 2 was associated with exclusive stroke and higher mortality from heart failure and sudden cardiac death across all genotypes, especially in individuals with pathogenic or likely pathogenic sarcomeric gene mutations.Additionally, Cluster 2 also presented higher mortality rate, particularly when focusing on cardiomyopathy-related deaths, with fatality rates up to 11.75 times higher than those in other clusters.Moreover, individuals in Cluster 2 exhibited shorter lifespans, with forecasted survival ranging from 75.2 to 80.2 years and showed rapid disease progression during initial life stages, indicated by sooner reduction in survival probability.Interpretation of the results using factor loadings from the first three principal components revealed substantial influence from parameters such as CMR, echocardiography in determining cluster structure and promoting clearer cluster discrimination.Although the current findings provide preliminary insights, future studies should employ independent validation to evaluate the outcomes presented herein, while simultaneously reassessing the hypothesis of a novel classification model based on cluster centroids.
Author: Dang, Long Ngoc Kim
Language: English
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7132865393