**Background:** Gastric cancer (GC) is the fifth most common cancer and third leading cause of cancer death globally. Tropomyosin 4 (TPM4), an actin-binding protein, has been implicated in several malignancies including breast, colon, and liver cancers, but its role in GC and across cancer types was not comprehensively characterized. This study aimed to perform a pan-cancer analysis of TPM4 and validate its function in GC.
**Methods:** The authors used multiple public databases: UCSC Xena, TCGA, GTEx, TIMER2.0, GEPIA, cBioPortal, UALCAN, and others. TPM4 mRNA expression was analyzed in 15,776 samples (TCGA+GTEx) and 15,043 paired tumor-normal samples. Prognostic value was assessed using Cox regression and Kaplan-Meier analysis. Diagnostic value was evaluated by ROC curves. Genetic alterations were examined in 2,565 pan-cancer patients via cBioPortal. DNA methylation was analyzed using UALCAN and MethSurv. mRNA modification correlations with 45 regulators (m1A, m5C, m6A) were assessed. Immune infiltration was evaluated using EPIC, ESTIMATE, and TIMER2.0. Correlations with immune checkpoint genes, TMB, MSI, and neoantigens were computed. A lncRNA-miRNA-TPM4 regulatory network was constructed using RNA22, miRWalk, miRDB, Starbase 2.0, and Cytoscape. Drug sensitivity was analyzed via GSCALite, GDSC, CTRP, and CMap. Gene Ontology and KEGG enrichment were performed on co-expressed genes. In vitro, AGS and BGC-823 GC cells were used with TPM4 knockdown via shRNA lentivirus (knockdown efficiency 71.1%). Wound healing and Matrigel transwell assays assessed migration and invasion.
**Key Results:** TPM4 was significantly upregulated (P<0.05) in 19 cancer types including STAD, and downregulated in 11 types. High TPM4 expression was associated with poor overall survival in ACC, CESC, CHOL, DLBC, ESCC, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, MESO, PAAD, STAD, UCEC, and UVM (HR>1, P<0.05). In digestive cancers, TPM4 had certain diagnostic accuracy for COAD (AUC=0.807), COADREAD (AUC=0.837), ESCA (AUC=0.725), LIHC (AUC=0.739), STAD (AUC=0.795), and high accuracy for PAAD (AUC=0.972). Genetic alterations in TPM4 occurred in 5% of pan-cancer samples (131/2565), with the highest mutation rate in ovarian cancer (>60%). TPM4 was hypomethylated in STAD, and 25 CpG sites were identified. TPM4 expression positively correlated with most m1A, m5C, and m6A methylation regulators across cancers. In STAD, top correlated regulators included METTL14 (r=0.403), YTHDF3 (r=0.421), and FTO (r=0.399). TPM4 expression correlated with immune cell infiltration in multiple cancers and was positively associated with ESTIMATE scores in STAD. TPM4 positively correlated with eight immune checkpoint genes (CD274, CTLA4, HAVCR2, LAG3, PDCD1, TIGIT, etc.) in multiple cancers. TPM4 expression was positively associated with TMB in ACC, UCSC, GBM, PAAD, and STAD; with MSI in TGCT, SARC, ACC, UVM, and STAD; and with neoantigens in ACC, TGCT, DLBC, PCPG, and THCA. A ceRNA network was constructed with four miRNAs (hsa-miR-338-3p, hsa-miR-30e-5p, hsa-miR-30b-5p, hsa-miR-206) and their target lncRNAs. Drug sensitivity analysis showed TPM4 expression correlated with sensitivity to docetaxel and 5-fluorouracil. Eight small molecule drugs were identified from CMap, with rucaparib having the highest absolute connectivity score. GO and KEGG enrichment revealed ECM-related pathways. Top hub genes included COL1A2, COL3A1, FN1, MMP2, and LUM. In vitro, TPM4 knockdown significantly inhibited migration (P<0.01) and invasion (P<0.01) in both AGS and BGC-823 cells.
**Clinical Implications:** TPM4 is a promising biomarker for diagnosis, prognosis, and immunotherapy response prediction across multiple cancers including GC. Its association with docetaxel and 5-fluorouracil sensitivity suggests potential as a chemotherapy response predictor. The identified ceRNA network provides insight into GC regulatory mechanisms. TPM4's role in promoting migration and invasion, likely through ECM remodeling, identifies it as a potential therapeutic target. However, clinical trial validation is needed before clinical application.