Integrated Molecular and Phenotypic Profiling of Circulating Tumor Cells in Gastric Cancer: A Translational Perspective

ABSTRACT

Gastric cancer often presents at advanced stages, limiting treatment outcomes. Circulating tumor cells (CTCs), particularly those undergoing epithelial–mesenchymal transition (EMT), have emerged as promising biomarkers for real‐time disease monitoring. Using magnetic‐activated cell sorting (MACS), we enriched both epithelial (EpCAM ⁺ ) and mesenchymal (CD90 ⁺ ) CTCs and evaluated the expression of the tumor‐suppressive microRNA miR‐29b in CTCs of gastric cancer patients. This study aimed to explore the clinical relevance of perioperative CTC phenotypes and miR‐29b as potential diagnostic and prognostic markers. We conducted a multi‐phase study integrating systematic review, bioinformatics, and experimental analyses to investigate CTC phenotypes and ECM‐associated gene expression in GC. Peripheral blood samples were collected from 30 treatment‐naïve gastric cancer patients at two time points: before initiation of therapy and 1 month after curative‐intent gastrectomy. Circulating tumor cells (CTCs) were isolated using CD45‐negative magnetic‐activated cell sorting (MACS). Immunofluorescence staining was performed to phenotype CTCs based on EpCAM (epithelial) and CD90 (mesenchymal) expression. Total RNA was extracted from tumor tissues and isolated CTCs to evaluate ECM‐related genes and miR‐29b expression using quantitative real‐time PCR. Bioinformatics‐guided gene selection and statistical analysis were used to assess diagnostic and prognostic relevance. CTCs were detected in 83.3% (25/30) of gastric cancer patients prior to treatment. Mesenchymal CTCs (CD90 ⁺ ) were markedly more frequent than epithelial CTCs (EpCAM ⁺ ), with mean baseline counts of 14.07 versus 0.5 cells per 10 mL of blood, respectively. Following gastrectomy and chemotherapy, mesenchymal CTCs significantly declined (mean: 1.18; p  = 0.0064), while epithelial CTCs were nearly undetectable ( p  = 0.0246). Based on perioperative CTC dynamics, patients were stratified into three risk groups: high‐risk (16%), intermediate‐risk (68%), and low‐risk (16%). All observed deaths occurred in the high‐risk group during 14 months of follow‐up, with Kaplan–Meier analysis showing significantly lower survival in this group ( p  = 0.0107). Expression analysis revealed significant downregulation of miR‐29b in tumor tissues (fold change = 0.5163; p  = 0.0136) and in CTCs (fold change = 0.2049; p  < 0.0001) compared to healthy controls. Although miR‐29b levels were broadly downregulated in gastric cancer patients, no significant differences were observed among clinicopathological subgroups within the patient cohort. ROC analysis showed strong diagnostic performance of miR‐29b in tissue (AUC = 0.883, 95% CI: 0.798–0.969), with 96.7% sensitivity and 70% specificity. CD90 serves as a sensitive mesenchymal marker for detecting CTCs that lack epithelial features, enabling more comprehensive identification of aggressive CTC phenotypes in gastric cancer. Moreover, the perioperative dynamics of CTCs—especially postoperative increases—provide a valuable tool for patient risk stratification and survival prediction. Importantly, our findings suggest that miR‐29b expression in CTCs, together with perioperative CTC dynamics, may serve as a non‐invasive biomarker for monitoring disease progression and recurrence in gastric cancer.