Comparative study of disease progression for heart failure with different etiologies via time-ordered network analysis.

Haoran Sun, Xiuhong Li, Hao Yuan, Chengyi Wang, Guangde Zhang, Hongbo ShiPubMed 2022

TL;DR

This study introduces a novel time-ordered network analysis approach to unravel the distinct molecular progression pathways of heart failure (HF) across different etiologies. By identifying dysregulated long non-coding RNA (lncRNA)-mRNA regulatory pairs (LMRPs) over time, the researchers uncovered unique functional characteristics for each HF cause and developed promising diagnostic biomarkers.

What Problem Does This Paper Address?

Heart failure (HF) is a critical endpoint for numerous cardiovascular diseases, leading to significant morbidity and mortality globally. While HF patients often present with similar symptoms, the underlying molecular mechanisms can vary considerably depending on the initial cause (etiology). Current clinical approaches tend to treat HF uniformly, potentially overlooking these crucial differences. This paper aims to bridge this gap by providing a deeper understanding of how HF progresses differently based on its origin, moving beyond static analyses to capture dynamic molecular changes. This knowledge is vital for developing more precise and effective treatments.

Methods

The core of this research lies in a novel computational framework designed to analyze the dynamic interplay between lncRNAs and mRNAs in HF. Key methodological components include:

  • LncRNA-mRNA Regulation Pair (LMRP) Identification: The study developed a method to identify dysregulated LMRPs by considering three factors: (1) differential gene expression, (2) altered gene interaction (edge dysregulation), and (3) the influence of LMRP dysregulation on interacting genes. This goes beyond simple correlation to capture functional impact.
  • Time-Ordered Network Construction: A "time-ordered background IncRNA-mRNA regulation network" (TO-BLMRN) was constructed using a breadth-first search algorithm. This network represents the progression of LMRP dysregulation across different "time-ordered levels," allowing for the analysis of dynamic changes.
  • Biomarker Identification: Both single-molecule (mRNA/lncRNA) and LMRP-based biomarkers were identified using random forest and support vector machine (SVM) classification algorithms. These biomarkers were then evaluated for their ability to distinguish HF patients from healthy controls.

Datasets

The study utilized publicly available RNA-sequencing (RNA-seq) and microarray data from the Gene Expression Omnibus (GEO) database.

  • Training Sets: Datasets GSE141910, GSE135055, and GSE116250 (RNA-seq) were used for training. These included samples from patients with dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM), ischemic cardiomyopathy (ICM), peripartum cardiomyopathy (PPCM), and healthy non-heart failure (NF) controls.
  • Validation Sets: RNA-seq data from GSE133054 and GSE46224, along with microarray data from GSE1145, were used for validation.

Key Insights and Ideas

  • Dynamic vs. Static HF Progression: The study highlights that while late stages of HF might share common functional characteristics (e.g., myocardial remodeling), the progression pathways and early functional disturbances are distinct for different etiologies.
  • LMRP as a Key Regulator: The research emphasizes the importance of dysregulated LMRPs in driving HF progression, suggesting that these regulatory pairs are not just markers but active participants in disease development.
  • Etiology-Specific Functional Signatures: Different HF causes exhibit unique functional enrichments at various stages of progression. For instance, myocardial energy metabolism appears crucial in early DCM and PPCM, while immune responses are prominent earlier in HCM.
  • Novel Biomarker Discovery: The developed method successfully identified panels of diagnostic biomarkers that can differentiate HF patients from controls, and importantly, can distinguish between different HF etiologies.

Key Results

  • Distinct Functional Pathways: The study revealed that myocardial energy metabolism is significantly enriched in the initial stages of DCM and PPCM. Cardiomyocyte morphology alterations were observed early in DCM and HCM. Immune responses appeared earlier in HCM compared to other etiologies.
  • Biomarker Performance: The identified single-molecule and LMRP-based biomarkers demonstrated high classification efficiency, with accuracies and AUC values often exceeding 0.90 in both training and validation sets, effectively distinguishing HF patients from NF controls across different etiologies. For example, specific mRNA panels for DCM-NF achieved accuracies and AUCs of 0.956/0.988 and 1.000/1.000.
  • LMRP Identification: The study identified thousands of dysregulated LMRPs across different HF etiologies, with a significant proportion of these LMRPs containing disease-related genes.

Limitations and Future Work

  • Data Availability: The authors note that time-series gene expression data for human cardiomyopathy is not readily available, which limits the direct application of time-series analysis. Their TO-BLMRN approach aims to mitigate this by inferring temporal order.
  • Validation Scope: While multiple test sets were used for validation, further experimental validation of the identified biomarkers and functional insights in independent cohorts and preclinical models would strengthen the findings.
  • Personalized Treatment: The study's findings provide a foundation for personalized treatment strategies by elucidating etiology-specific mechanisms, but translating these insights into clinical practice will require further research and development.

Summary updated

Abstract

This study revealed distinct functional characteristics during the progression of HF from different causes and facilitated the discovery of candidate diagnostic biomarkers for HF.
BibTeX
@inproceedings{sun2022comparative,
  author = {Haoran Sun and Xiuhong Li and Hao Yuan and Chengyi Wang and Guangde Zhang and Hongbo Shi},
  title = {{Comparative study of disease progression for heart failure with different etiologies via time-ordered network analysis.}},
  booktitle = {PubMed},
  year = {2022},
}
Figure