Transcriptomics-based liquid biopsy panel for early non-invasive identification of peritoneal recurrence and micrometastasis in locally advanced gastric cancer

Ping’an Ding, Haotian Wu, Jiaxiang Wu, Tongkun Li, Renjun Gu, Lilong Zhang, Peigang Yang, Honghai Guo, Yuan Tian, Jinchen He, Jiaxuan Yang, Ning Meng, Xiaolong Li, Lingjiao Meng, Qun ZhaoJournal of Experimental & Clinical Cancer Research 2024 · DOI

TL;DR

Researchers have developed a novel six-gene expression panel that can accurately detect peritoneal recurrence and micrometastasis in locally advanced gastric cancer (LAGC) using a non-invasive liquid biopsy. This panel, integrated into a risk stratification model (RSA), significantly improves early detection and risk assessment, offering a promising tool for personalized treatment strategies.

What Problem Does This Paper Address?

Gastric cancer is a leading cause of cancer-related mortality, with peritoneal dissemination being a primary driver of recurrence and poor prognosis. Early detection of peritoneal recurrence and micrometastasis is crucial for effective treatment, but current diagnostic methods like CT and PET-CT scans have limitations, and invasive procedures like staging laparoscopy are costly and not always feasible. The absence of reliable, non-invasive biomarkers for early detection poses a significant challenge in managing LAGC patients and improving their outcomes.

Methods

The study employed a multi-stage approach:

  • Biomarker Discovery: Genome-wide transcriptome profiling was performed on publicly available datasets (TCGA, GEO) and paired gastric cancer tissue samples with peritoneal metastasis. Differential gene expression analysis identified candidate genes.
  • RT-qPCR Validation: The expression of candidate genes was validated using RT-qPCR in pilot cohorts of gastric cancer and adjacent non-malignant tissues, as well as peripheral blood samples.
  • Risk Stratification Model (RSA) Development: A six-gene expression panel was developed. This panel, combined with clinical features, was used to build a risk stratification model (RSA) via logistic regression.
  • Multi-specimen Validation: The RSA model was validated across multiple independent cohorts using various sample types:
  • Surgical Resection Specimens: Training and validation cohorts of fresh frozen tissue samples.
  • Gastroscopic Biopsy Specimens: To assess the feasibility of using less invasive biopsy samples.
  • Peripheral Blood Specimens (Liquid Biopsy): To evaluate the non-invasive potential of the panel.
  • Statistical Analysis: Univariate and multivariate logistic regression, ROC curve analysis (AUC), DeLong's test for AUC comparison, confusion matrices, calibration curves, and Kaplan-Meier survival analysis were used to assess predictive accuracy and clinical utility.

Datasets

The study utilized the following datasets:

  • Publicly Available Datasets:
    • Gene Expression Omnibus (GEO) database (GSE15081)
    • The Cancer Genome Atlas (TCGA) database
  • Clinical Cohorts:
    • Training Cohort (Surgical Specimens): 196 samples from the Fourth Hospital of Hebei Medical University (FHHMU) and Shijiazhuang People's Hospital (SJZPH).
  • Validation Cohort (Surgical Specimens): 133 samples from Baoding Central Hospital (BDCH), Nanjing University Jinling Hospital (JLNJ), and Wuhan University People's Hospital (WHPH).
  • Gastroscopic Biopsy Cohort: 103 matched specimens from LAGC patients across five institutions.
  • Peripheral Blood (Serum) Training Cohort: 120 LAGC patients from FHHMU.
  • Peripheral Blood (Serum) Validation Cohort: 123 LAGC patients from four institutions.
  • Tumor Marker-Negative Cohort: 56 patients.
  • POCY1 Cohort: 66 patients with confirmed peritoneal lavage cytology positivity.
  • Peritoneal Metastasis (PM) Cohort: 95 patients with confirmed PM by laparoscopic exploration.
  • Other Gastrointestinal Cancers Cohort: Colorectal (n=35), hepatocellular carcinoma (n=38), and pancreatic adenocarcinoma (n=29) for specificity testing.

Key Insights and Ideas

  • A Six-Gene Signature for Peritoneal Recurrence: The study successfully identified a specific panel of six genes (BUB1, CKS2, PCNA, CHEK1, NEK2, and NCAPG2) that are significantly upregulated in patients with peritoneal recurrence. This provides a molecular basis for predicting this critical complication.
  • Liquid Biopsy Efficacy: A major breakthrough is the demonstration that this gene panel retains high predictive accuracy even when analyzed in peripheral blood (serum) samples. This transitions the diagnostic approach from invasive tissue biopsies to a non-invasive liquid biopsy, significantly improving patient comfort and accessibility.
  • Integrated Risk Stratification Model (RSA): Combining the six-gene panel with clinical parameters into a single RSA model yielded superior predictive performance compared to either component alone. This highlights the power of integrating molecular and clinical data for more robust prognostication.
  • Early Detection of Micrometastases: The RSA model shows promise in identifying peritoneal micrometastases (POCY1), a challenging subset of patients often missed by current diagnostics, enabling earlier intervention.
  • Clinical Utility Beyond Gastric Cancer: While primarily focused on gastric cancer, the panel showed higher diagnostic accuracy for PM in LAGC compared to other gastrointestinal cancers, suggesting potential for broader applications or at least a high specificity for gastric cancer.

Key Results

  • The developed six-gene panel demonstrated high predictive accuracy for peritoneal recurrence.
  • The RSA model, integrating the 6-mRNA panel with clinical features, achieved an AUC of 0.966 in the training cohort and 0.961 in the validation cohort for predicting peritoneal recurrence in surgical specimens.
  • In liquid biopsy samples, the RSA model maintained high performance with an AUC of 0.963 for predicting peritoneal recurrence.
  • The RSA model showed excellent performance in predicting peritoneal micrometastasis (POCY1) with an AUC of 0.941.
  • The RSA model also demonstrated high accuracy in predicting peritoneal metastasis (PM) with an AUC of 0.970.
  • Survival analysis revealed significantly worse 5-year recurrence-free survival (RFS) in high-risk groups identified by the RSA model compared to low-risk groups across multiple cohorts.

Limitations and Future Work

  • Retrospective Design and Selection Bias: The study's retrospective nature may have introduced selection bias, and the modest sample size, particularly for patients with peritoneal recurrence, could impact the robustness of the model.
  • Limited Cohort Diversity: The study primarily focused on Chinese patients, limiting the generalizability of findings to diverse populations. Future international, multicenter studies with larger sample sizes are recommended.
  • Integration with Other Biomarkers: The authors suggest future research should investigate the integration of their panel with other known clinical markers (e.g., HER2, PDL1) and DNA mutation status to further enhance diagnostic accuracy.
  • Pan-Cancer Applications: While the panel showed specificity for gastric cancer, further expansion of sample size and prospective studies are needed to analyze the role of these genes in other gastrointestinal cancers.
  • Comparison with Tissue vs. Blood: While no significant difference was found in diagnostic performance between tissue and blood samples, the authors acknowledge potential statistical bias due to small sample sizes and suggest further investigation.

The study concludes that the developed 6-mRNA panel and RSA model offer a significant advancement in the early detection and risk stratification of peritoneal recurrence and metastasis in LAGC, paving the way for more personalized and effective patient management.

Summary updated

Abstract

Our study provides a novel gene expression biomarker panel that significantly enhances early detection of peritoneal recurrence and micrometastasis in patients with LAGC. The RSA model's predictive capability offers a promising tool for tailored treatment strategies, underscoring the importance of integrating molecular biomarkers with clinical parameters in precision oncology.
BibTeX
@article{ding2024transcriptomics,
  author = {Ping’an Ding and Haotian Wu and Jiaxiang Wu and Tongkun Li and Renjun Gu and Lilong Zhang and Peigang Yang and Honghai Guo and Yuan Tian and Jinchen He and Jiaxuan Yang and Ning Meng and Xiaolong Li and Lingjiao Meng and Qun Zhao},
  title = {{Transcriptomics-based liquid biopsy panel for early non-invasive identification of peritoneal recurrence and micrometastasis in locally advanced gastric cancer}},
  journal = {Journal of Experimental \& Clinical Cancer Research},
  year = {2024},
  doi = {10.1186/s13046-024-03098-5},
}
Figure