Treating chronic atrophic gastritis: identifying sub-population based on real-world TCM electronic medical records

Yu-man Wang, Jianhui Sun, Run-xue Sun, Xiaoyu Liu, J Y Li, Runze Li, Yan-Ru Du, Xuezhong ZhouFrontiers in Pharmacology 2024 · DOI

Abstract

Based on Patient Similarity Network (PSN) identification and community detection analysis, Chronic Atrophic Gastritis (CAG) population division can provide useful recommendations for clinical treatment. This method is useful for CAG illness classification and genotyping investigations and can be used for other complicated chronic diseases.

Introduction

Chronic Atrophic Gastritis (CAG) is a high-prevalence condition characterized by the loss of gastric glandular cells, often serving as a precursor to gastric cancer. In clinical practice, CAG is notoriously heterogeneous; patients present with a wide array of symptom phenotypes, making standardized treatment difficult.

Traditional Chinese Medicine (TCM) is inherently designed for individualized "syndrome differentiation," but its logic is often qualitative and difficult to scale. This study aims to bridge the gap between empirical TCM success and the need for a standardized, data-driven understanding of disease subtypes by using real-world data to identify hidden patient subgroups and their underlying biological mechanisms.

Methods

The researchers developed a multi-stage computational pipeline to analyze over 5,000 real-world TCM electronic medical records (EMRs).

Workflow of the study, from clinical data extraction to network pharmacology analysis

The process involved:

  1. Data Curation: Extracting and annotating 6,253 EMRs from the Hebei Hospital of TCM (2013-2023), standardizing terms via ICD-11 and the Chinese Pharmacopoeia.
  2. Patient Similarity Network (PSN): Building a network where patients are nodes and edges represent symptom similarity calculated via the Jaccard coefficient.
  3. Community Detection: Applying the BGLL and BigClam algorithms to partition the network into 176 distinct patient subgroups.
  4. Phenotype Enrichment: Identifying significant symptoms and pulse patterns for major modules using Relative Risk (RR) and chi-square tests.
  5. Network Pharmacology: Mapping effective herbs for each subgroup to molecular targets and performing Gene Ontology (GO) and KEGG pathway enrichment.

Key Insights and Results

The analysis revealed that while the population was divided into 176 modules, nearly half the cohort (47.5%) clustered into three primary subtypes: M29, M3, and M0.

Community structure of the patient similarity network showing distinct subgroups

Subtype Characteristics

  • M29 (n=860): Associated with "Turbidity Syndrome" and gastric fullness. Treatments like Citrus medica were linked to NF-κB and apoptosis signaling.
  • M3 (n=818): Linked to "Liver-Stomach Disharmony," characterized by bitter taste and yellow urine. Treatments like Angelica sinensis were associated with JAK-STAT signaling.
  • M0 (n=419): Specific to "Dampness-Heat" syndromes, presenting with nausea and vomiting. Scutellaria baicalensis was a key herb, linked to the PI3K-Akt pathway.

The study population was balanced (52.6% male) and primarily aged 40–79. Notably, 65.8% of hospitalized patients showed clinical improvement, validating the efficacy of the treatments analyzed.

Conclusion and Future Directions

This research demonstrates that traditional TCM "dialectical classification" has an objective basis in patient symptom clusters and molecular biology. While limited by its single-center retrospective nature, the study provides a robust computational framework for precision medicine in complex chronic diseases. Future work will focus on multi-center validation and the integration of multi-omics data to further refine these subtypes.

Summary updated

Abstract

Based on PSN identification and community detection analysis, CAG population division can provide useful recommendations for clinical CAG treatment. This method is useful for CAG illness classification and genotyping investigations and can be used for other complicated chronic diseases.
BibTeX
@inproceedings{wang2024treating,
  author = {Yu-man Wang and Jianhui Sun and Run-xue Sun and Xiaoyu Liu and J Y Li and Runze Li and Yan-Ru Du and Xuezhong Zhou},
  title = {{Treating chronic atrophic gastritis: identifying sub-population based on real-world TCM electronic medical records}},
  booktitle = {Frontiers in Pharmacology},
  year = {2024},
  doi = {10.3389/fphar.2024.1444733},
}
Figure