Introduction

Gastric cancer (GC) is a type of malignant tumour that originates from the epithelium of the gastric mucosa. Each year, over 1,000,000 new patients are diagnosed with GC every year, resulting in approximately 769,000 people deaths in 20211,2. GC has been imposing significant mental and financial burdens on individuals and society as a whole. Although there have been advancements in early diagnosis and treatment of GC, predictive biomarkers are necessary to establish precision medicine as the cornerstone for achieving efficient, reliable, and measurable outcomes in the management of GC.

Biomarkers are objectively measured and evaluated characteristics that serve as indicators of normal biologic process, pathogenic processes, or the pharmacological response to therapeutic interventions3. Current molecular markers associated with diagnosis, prognosis, and prediction of therapeutic responses in GC include tissue-based biomarkers such as metastasis-related genes [eg. human epidermal growth factor receptor 2 (Her-2) and E-cadherin], comprehensive analysis genes, epigenetic alterations, as well as genetic polymorphism4. Blood is a commonly used test material due to its easy to obtain, detectability, and minimal invasiveness. Additionally, considering the significant influence of the tumour microenvironment on carcinogenesis, invasion, and metastasis, blood components may also impact tumorigenesis. Therefore, circulating tumour cells, circulating cell-free DNA, non-coding RNAs (e.g. microRNAs, long noncoding RNAs and circular RNAs) and exosomes have been demonstrated to be specifically expressed in blood during gastric carcinogenesis5,6,7. The clinical utilization of blood antigens has facilitated the diagnosis and selection of chemical drugs, encompassing carcinoemacmbryonic antigen (CEA), alpha-fetoprotein (AFP), cancer antigen 19 − 9 (CA19-9), cancer antigen 72 − 4 (CA72-4), and cancer antigen 125 (CA125). Specific biomarkers have also been examined for diagnosis and treatment of GC, such as dysregulation of genes and non-coding RNAs found in gastric washes/gastric juice8; volatile markers detected in exhaled breath9; p-cresol and 4-hydroxybenzoic acid present in urine samples10; H. pylori key virulence factor CagA overexpressed in tissue samples11. However, most tissue-based biomarkers of GC carry the risk of a sampling error due to intratumoral heterogeneity, and adequate tissue sampling is of paramount importance. The biomarkers in blood, gastric washes/gastric juice, exhaled breath, or urine samples are still not clear and definite. The noninvasive blood biomarkers for early diagnosis and/or screening of high-risk population for GC may be a good choice.

MicroRNA(miRNA) is a class of non-coding single-stranded RNA molecules of approximately 22 nucleotides in length encoded by endogenous gene12. It plays a crucial role in the post-transcriptional regulation of gene expression in plants and animals13. Through base pairing with complementary sequences within mRNA molecules, miRNAs effectively silence their expression. This dysregulation of negative genes expression control is closely associated with tumorigenesis. In recent years, manipulation of miRNA clusters expression has been explored for cancer diagnosis, treatment, and prognosis14. MiRNAs have pivotal functions in gastric carcinogenesis, including oncogenic miRNAs (oncomiRNAs) such as miR-17, miR-130, and miR-181; tumour suppressor miRNAs like miR-27b, miR-29 family, miR-34, miR-124, miR-128, miR-218, miR-429, and miR-49715. MiRNAs can also modulate sensitivity to chemotherapeutic agents and alter the function of cancer-related signal pathways such as phosphatase and tensin homolog deleted on chromosome ten/phosphatidylinositol 3-kinase/protein kinase B (PTEN/PI3K/Akt)16,17. Notably, abnormal levels of miRNAs can discriminate cancer patients from healthy individuals, which have been detected not only in tissues but also in various body fluids. Therefore, miRNAs have emerged as promising biomarkers for cancer management including diagnosis, clinical staging, prediction of tumour behaviour, assessment of treatment response, and patient survival. For example, the significance of miR-21 and miR-222 lies in their distinct expression patterns in serum samples which emphasizes their utility as non-invasive biomarkers18. Henceforth investigating the characteristics of miRNAs in blood may hold immense promise for improving tumour prediction and diagnose19. The utilization of miRNAs as highly tissue-specific biomarkers has led to their incorporation into a diverse range of diagnostic tests, offering great potential for clinical diagnosis and determination of metastatic origin20.Therefore, investigating differentially expressed miRNAs in the blood of GC patients holds promise for identifying novel tumour markers and guiding future research directions. However, current studies on differentially expressed miRNAs in GC patients’ blood are not without limitations. In fact, comprehensive profiling using high-throughput sequencing is necessary to fully elucidate the dysregulation of blood miRNAs during gastric carcinogenesis.

In recent years, the utilization of tumour databases, sequencing technology, and bioinformatics analysis has become prevalent in cancer research at the molecular level21,22. These advancements provide novel ideas and tools to identify differentially expressed miRNAs from patients’ blood samples and explore their functional pathways that potentially impact gastric carcinoma. Therefore, this study searched for the relevant datasets from the GEO database23. Subsequently, the bioinformatics tools were employed to analyze the acquired data to identify key miRNAs that were implicated in biological mechanisms through functional enrichment and protein network construction. The results were detected in human blood samples and further analyzed for their prognosis and diagnostic value in patients with GC. These identified crucial blood miRNAs and their corresponding target genes have the potential to serve as biomarkers for early screening and diagnosis of GC. CagA has been identified to be involved in tumorigenesis-correlated signal pathways. Therefore, the significance and effectiveness of the key miRNAs was also assessed through relevance analysis with different Pathological Tumour-Node-Metastasis (pTNM) stages, H. pylori infection and CagA expression by comparison them with clinical markers. The flow chart illustrating the methods employed in this study can be seen in Fig. 1.

Fig. 1
figure 1

Flow chart of the methods utilized in the present study.

Full size image

Materials and methods

Data selection

GEO (https://www.ncbi.nlm.nih.gov/geo/) is developed by the National Center for Biotechnology Information (NCBI) and serves as a repository for high-throughput gene expression data submitted by scientific researchers worldwide. The database was queried using the keywords “gastric cancer”, “blood”, “human”, and “miRNA”. Subsequently, GSE113486, GSE112264, and GSE113740 were identified as the most appropriate datasets containing 115 cases of gastric cancer samples and 151 cases of normal samples. Additionally, GSE108307 was utilized to elucidate miRNAs influenced by H. pylori infection.

The initial screening of differentially expressed MiRNAs and their predicted target genes

SangerBox, an R language-based biometric analysis platform (http://sangerbox.com), was utilized for the differential analysis of sample files using the GEO convenience converter and DECenter. Subsequently, the network tool library in SangerBox was employed to perform robust rank aggregation (RRA) analysis. As a result, 20 miRNAs (10 up-regulated and 10 down-regulated) exhibiting significant changes in the GC group compared to normal samples were identified as research targets. Targetscan (http://www.targetscan.org/), miRDB (http://mirdb.org/), and miRtarBase (https://mirtarbase.cuhk.edu.cn) were utilized as online prediction websites for identifying potential target genes of these miRNAs.

Functional enrichment of differentially expressed genes

The DAVID 6.8 database (https://david.ncifcrf.gov/) is widely utilized in the field of bioinformatics to provide comprehensive and systematic information for large-scale gene or protein lists. In this study, it was employed to perform GO and KEGG24,25,26 enrichment analyses on differentially expressed genes and their associated signalling pathways, aiming to gain insights into the underlying molecular mechanisms involved in gastric carcinoma. Statistical significance was determined based on a P-value < 0.05 and a minimum gene count of ≥ 2.

PPI network

PPI network is composed of individual proteins connected through their interactions. Studying PPI network can facilitate understanding of functional links among proteins and identification of core regulators. STRING (https://string-db.org) is a protein interaction database that offers exhaustive specimens and comprehensive interaction information. After obtaining the PPI network consisting of 20 differentially expressed miRNAs and their target genes, the significant sub-network and its corresponding genes were analyzed using Cystoscape’s MCODE plug-in clustering function module (version 3.6.1). The minimum confidence score required for an interaction was set at medium level (0.900).

Selection and survival curve analysis of the key MiRNAs and relative target genes

The sub-network nodes of the PPI network were clustered to sort the most important genes that were negatively regulated by key miRNAs. The frequency of a miRNA’s occurrence was used to determine its significance in the study. ONCOLNC and GEPIA tools were utilized for interactive exploration of survival dependencies, and correlation analysis between key nodal genes and the of GC patient survival rate. The survival curves of the key miRNAs and the target genes were examined by the Kapler-Meier Plotter online tools (https://kmplot.com/analysis/) to determine the influences of their expression levels on GC patient prognosis with statistical significance was defined as P<0.05.

The human blood specimens and qRT-PCR determination

The human peripheral blood specimens were collected from Xiajin County People’s Hospital, Dezhou, and Shandong University Affiliated Hospitals, Jinan, Shandong Province. The participants were allocated into two groups: an experimental group (n = 49) and a control group (n = 37). The experimental group comprised newly diagnosed GC patients with definitive histopathological confirmation who had no prior antitumor therapy and absence of severe comorbidities or complications. Demographic characteristics (age and gender), clinicopathological parameters (pTNM stage), and immunohistochemical marker profiles including Ki67, E-cadherin, CK19, Her-2, AFP, CEA, CA19-9, CA125, and CA72-4 were extracted from standardized histopathology reports. The control group consisted of age- and gender-matched healthy volunteers recruited through randomized selection. Exclusion criteria for controls included personal history of malignancy or precancerous lesions, chronic systemic diseases affecting physiological function or metabolic homeostasis, and occupational exposure to known carcinogens. This research has been approved by the Ethics Committee of Shandong University School of Basic Medical Sciences (No.: ECSBMSSDU2021-1–097).

Total RNA was extracted using Invitrogen TRIzol (Thermo Fisher SCIENTIFIC, Code: 15596018). For miRNA quantification, the synthesis of cDNA was performed using Evo M-MLV RT Kit for qPCR (Accurate Biology, Code: AG11707). The qPCR primers of miR-575, miR-125a-3p, miR-124-3p, miR-4276 and miR-29b-3p were designed and synthesized with the Bulge-Loop miRNA qRT-PCR Primer sets (RIBOBIO, Code: MQPSCM001). Real-time PCR was carried on with SYBR Green Premix Pro Taq HS qPCR Kit (Accurate Biology, Code: AG11701) on a Bio-Rad CFX-96 real-time system. U6 was detected as the internal control. CagA expression was determined with the primers forward: 5’-CAAGTCCGTGGGCATCATGT-3’ and reverse: 5’-GAGGACACTCGGTCTCTAGC-3’. All qRT-PCR reactions were conducted in triplicates, and relative quantification was calculated by the 2-ΔΔCt method (95% confidence interval) with calibration to the corresponding control. The expression data were analyzed with a Student’s t-test. The P < 0.05 was considered significant difference in statistics.

ROC analysis

The ROC curves were constructed using the expression data of key miRNAs through SPSS 26.0 software, with area under the ROC curve (AUC) values serving as diagnostic indicators. The P < 0.05 was considered to be statistically significant.

Statistical analysis

All experiments were performed in triplicate. Quantitative data were analyzed using GraphPad Prism 8.0.2 (GraphPad Software) and SPSS 26 (IBM) for statistical computations and graphical representations, with final figure preparation conducted in Adobe Photoshop 2023. Statistical comparisons were carried out using Student’s t-test and one-way ANOVA, with results expressed as mean ± standard deviation (SD). Statistical significance was defined as P < 0.05 for all analyses.

Results

The blood MiRNAs differentially expressed in gastric carcinoma

The SangerBox was utilized for the analysis of GSE113486, GSE112264, and GSE113740 datasets to screen and identify aberrantly expressed miRNAs in the blood samples of GC patients. The results revealed that in GSE113486, there were 2086 miRNAs exhibiting statistically significant differences between the GC group and normal controls. The top 10 significantly up-regulated and down-regulated miRNAs in GSE113486 were presented in Fig. 2A. The top 10 most significantly up-regulated and down-regulated miRNAs in GSE112264 were shown in Fig. 2B. The top 10 differentially expressed miRNAs in GSE113740 were displayed in Fig. 2C. A more detailed analysis of the GEO database was shown in Figure S1. The RRA algorithm was utilized to identify the top 10 up-regulated miRNAs and the top 10 down-regulated miRNAs from the three datasets (Fig. 2D).

Fig. 2
figure 2

The differentially expressed blood miRNAs in gastric carcinoma. (A) The top 10 up-regulated and down-regulated miRNAs in GSE113486. (B) The top 10 up-regulated and down-regulated miRNAs in GSE112264. (C) The top 10 up-regulated and down-regulated miRNAs in GSE113740. (D) The highest-scoring up-regulated and down-regulated miRNAs identified using RRA algorithm. The selected key miRNAs were shown in red frames.

Full size image

The identification of target genes directly silenced by the most differentially expressed miRNAs and the subsequent enrichment analysis of GO and KEG

By integrating and comparing predictions from Targetscan, Mirtarbase and miRDB, a total of 691 over-expressed genes and 1074 down-regulated ones were determined as potential targets of the selected 20 miRNAs. GO and KEGG enrichment analyses using DAVID revealed the molecular functions (MFs), cellular components (CCs), biological processes (BPs), and signalling pathways associated with these target genes. GO enrichment results for Up-regulated genes suggested that key BPs included autophagosomes maturation, regulation of mitochondrial membrane potential, and folic acid metabolic process. The main MFs identified were protein binding, metal ion binding, and RNA polymerase II core promoter proximal region sequence-specific DNA binding. The CCs observed included cytosol, cytoplasm, and nucleus (Fig. 3 A). On the contrary, the down-regulated genes (targets of up-regulated miRNAs) were primarily involved in BPs such as regulation of transcription from RNA polymerase II promoter, platelet-derived growth factor receptor signalling pathway, and DNA-templated positive regulation of transcription. The main MFs encompassed protein binding, chromatin binding, transcriptional activator activity, and RNA polymerase II core promoter proximal region sequence-specific binding. The CCs comprised nucleoplasm, nucleus and focal adhesion (Fig. 3B). The signalling pathways implicated in the regulation of down-expressed genes included focal adhesion, prolactin signalling pathway, PI3K-Akt signalling pathway, and focal adhesion (Fig. 3 C). KEGG analysis indicated that endocytosis, proteoglycans in cancer, and glycosylphosphatidylinositol-anchor biosynthesis were among the main signalling pathways involved in overexpressed genes (targets of down-regulated miRNAs)-correlated mechanism (Fig. 3D).

The analysis of proteins interaction network and identification of key blood MiRNAs

Furthermore, the 20 most differently expressed blood miRNAs and their corresponding target genes were analyzed in the STRING database (Fig. 3E). The results of PPI network analysis were imported into Cytoscape software for further investigation. A total of 38 nodes were identified, indicating a close relationship among the top 20 differently expressed blood miRNAs. These miRNAs were found to be involved in multiple signalling pathways, suggesting their potential roles in tumorigenesis. Additionally, the up-regulated gene-encoded proteins network consisted of 413 nodes with 143 edges, an average node degree of 0.692, and an average local clustering coefficient of 0.253. The enrichment P value was less than 0.0287 (Figure S2A). Similarly, the down-regulated gene-encoded proteins network comprised of 1323 nodes with 1021 edges, an average node degree of 1.54, and an average local clustering coefficient of 0.319. The enrichment P value was less than 2.62e-10 (Figure S2B). These findings indicate the data is reliable and can be utilized for further analysis.

Fig. 3
figure 3

The identification of target genes directly regulated by the key blood miRNAs from the three datasets by bioinformatics analysis. (A) The principal BPs, MFs, and CCs of down-regulated miRNAs and corresponding up-regulated target genes. (B) The principal BPs, MFs, and CCs of up-regulated miRNAs and corresponding down-regulated target genes. (C) Signalling pathway enrichment results of up-regulated miRNAs and corresponding down-regulated target genes in KEGG (The permission to use KEGG software has been obtained from Kanehisa Laboratories). (D) Signalling pathway enrichment results of down-regulated miRNAs and the corresponding up-regulated target genes in KEGG. (E) Network of target gene-encoded proteins corresponding to all 20 most significantly differentially expressed blood miRNAs.

Full size image

The Cytoscape software was utilized to determine the number of occurrences of the blood miRNAs on the key nodes, thereby assessing the significance of these miRNAs (Table S1). The findings indicated that hsa-miR-124-3p and hsa-miR-29b-3p were involved in up-regulated cases, while hsa-miR-125a-3p, hsa-miR-4276, and hsa-miR-575 were associated with down-regulated cases (Fig. 2, in red frames). These miRNAs could potentially serve as crucial biomarkers for early diagnosis and treatment of gastric carcinoma. Furthermore, the sub-network genes obtained from the PPI network analysis represented potential targets for these key miRNAs and were subjected to survival analysis. The results revealed 27 blood miRNA-targeted genes that exhibited an association with GC (Table S2). According to online data, similar specific miRNA expression patterns were not found in colon adenocarcinoma (COAD), liver hepatocellular carcinoma (LIHC) and lung adenocarcinoma (LUAD), which were also derived from epithelial cells, confirming the specificity of this key miRNAs expression model in gastric carcinogenesis (Figure S3).

The key blood MiRNAs were identified through their expression levels detection in human blood samples

The demographic information (including age, gender, and pTNM) of the patients who provided blood samples and their correlation with the expression of key miRNAs was presented in Table 1.

Table 1 The correlation between key MiRNAs expression and the age, gender, and pTNM stages of gastric cancer patients providing blood specimens.
Full size table

The results of qRT-PCR, which determined the expression levels of selected key miRNAs in human blood samples, were depicted in Fig. 4A-E. Furthermore, the expression levels of tissue-based markers and blood antigens to help diagnosis and treatment of GC and five key miRNAs across different pTNM stages from GC patients’ samples and online databases were also analyzed, which were shown in Fig. 4 F, S4, S5, and Tables 2 and 3. From these findings, it was preliminarily observed that there was a significant up-regulation of hsa-miR-124-3p and hsa-miR-29b-3p, as well as a down-regulation of hsa-miR-575, hsa-miR-125a-3p and hsa-miR-4276, whose expression levels in human blood specimens were consistent with bioinformatics predictions and online data. Moreover, it was determined that the patients’ age and gender did not exert a significant influence on the expression of key miRNAs (P > 0.05). However, there were significant differences in the expression levels of key miRNAs among different stages of pTNM (*P < 0.05) individually, and these expression trends were partially dependent on the pathological stages (from I to III); whereas the expression trends of each clinical marker did not exhibit substantial variation or correlation with the stages. Further testing and verification using larger cohorts is necessary to confirm these results.

Fig. 4
figure 4

The expression of key miRNAs in human blood specimens by RT-qPCR detection and the correlation with the stages. (A) The expression of hsa-miR-124-3p increased in the blood samples of GC patients (P < 0.01 vs. normal). (B) The blood expression of hsa-miR-125a-3p decreased in the blood samples of GC patients (P < 0.01 vs. normal). (C) The expression of hsa-miR-29b-3p increased in the blood samples of GC patients (P < 0.05 vs. normal). (D) The expression of hsa-miR-4276 decreased in the blood samples of GC patients (P < 0.01 vs. normal). (E) The expression of hsa-miR-575 decreased in the blood samples of GC patients (P < 0.01 vs. normal). (F) The expression levels of key miRNAs in different stages of pTNM (P < 0.05 vs. Stage I).

Full size image
Table 2 The expression levels of clinical markers and key MiRNAs in human blood species of GC patients.
Full size table
Table 3 The correlation between clinical markers expression levels and different clinical stages The correlation between different clinical stages and markers expression levels.
Full size table

The diagnostic value of key blood MiRNAs in patients with GC

The survival curves indicating the correlation between the expression of key miRNAs and patient prognosis could be found in Figure S6. Additionally, Figure S7 displayed the survival curves for key miRNAs’ target genes and corresponding patient prognosis. The expression levels of hsa-miR-124-3p, hsa-miR-29b-3p, and hsa-miR-575 were closely associated with GC patient prognosis (P < 0.05). However, regarding hsa-miR-125a-3p and hsa-miR-4276, their expressions in human blood specimens were consistent with bioinformatics predictions but inconsistent with GC patient prognosis. Furthermore, although the prognosis analysis indicated that there was no significant association between hsa-miR-4276 expression and patient survival rate (P > 0.05), lower expression levels of hsa-miR-4276 tended to decreased patient survival rate. The prognosis analysis also revealed that most target genes’ expressions were related to patients’ likelihoods for survival.

In order to further elucidate the diagnostic value of key blood miRNAs in gastric carcinoma, ROC analysis was performed based on their expression levels. The results demonstrated that miR-124-3p, miR-125-3p and miR-29b-3p held potential for diagnosis, with area under the ROC curve (AUC) values of 0.952, 0.812 and 0.824 respectively. Combining miR-124-3p, miR-125-3p, and miR-29b-3p yielded an improved ROC value of 0.975. Despite relatively lower ROC values (0.556 and 0.41) for hsa-miR-4276 and hsa-miR-575 respectively, the model composed of these key miRNAs exhibited excellent discriminatory ability between patients with and without GC (Fig. 5A-F).

Fig. 5
figure 5

The key miRNAs could have diagnostic value in GC determined by ROC analysis (A) The value of hsa-miR-124-3p AUC was 0.952, P < 0.001. (B) The value of hsa-miR-125a-3p AUC was 0.812, P < 0.001. (C) The value of miR-29b-3p AUC was 0.824, P < 0.001. (D) The value of hsa-miR-4276 AUC was 0.556, P > 0.05. (E) The value of hsa-miR-575 AUC was 0.41, P >0.05. (F) The value of (hsa-miR-124-3p + hsa-miR-125a-3p + miR-29b-3p) AUC was 0.975, P < 0.001.

Full size image

The model was further validated in GSE108307, which involved the miRNAs induced by H. pylori infection. Cross-validation demonstrated the presence of up-regulated hsa-miR-29b-3p and down-regulated hsa-miR-4276 in the dataset (Figure S8). Furthermore, CagA expression was assessed in human blood specimens using qRT-PCR. The results indicated a significant overexpression of CagA in GC patients compared to normal individuals (Fig. 6 A). The expression levels of five key miRNAs and clinical markers were detected in CagA-positive samples from GC patients respectively. In comparison with the clinical markers, a diagnostic model composed of these key miRNAs provided accurate diagnoses, demonstrating a stronger correlation between the CagA-positive group and expression levels of these key miRNAs, while the expression patterns of clinical markers were inconsistent with that of CagA-positive group in GC (Fig. 6B-D). In conclusion, the identified key miRNAs exhibited significant diagnostic value for detecting GC through analysis of blood samples. These findings also suggested that while bioinformatics predictions may be valuable for diagnosis purposes, results obtained through different analytical approaches may not always align completely with each other.

Fig. 6
figure 6

The expression levels of key miRNAs and clinical markers in CagA-positive patients’ blood specimens by RT-qPCR detection. (A) CagA was remarkably overexpressed in human GC blood specimens compared to the normal controls (P < 0.05). (B) The expression pattern of five key miRNAs in CagA-positive blood samples from GC patients, which showed up-regulation of hsa-miR-124-3p and hsa-miR-29b-3p, as well as down-regulation of hsa-miR-125-3p, hsa-miR-4276, and hsa-miR-575. (C) The clinical markers expression levels including Ki67, E-Cadherin, CK19, and Her-2 in CagA-positive blood samples, while the expression of E-Cadherin and Her-2 showed down-regulated. (D) The clinical markers expression levels including AFP, CEA, CA19-9, CA125 and CA72-4 in CagA-positive blood samples. None of them determined overexpression.

Full size image

Discussion

In the research, GSE113486, GSE112264 and GSE113740 datasets were initially analyzed using bioinformatics tools to identify key blood miRNAs in gastric carcinoma. RRA analysis were performed to rank the corresponding miRNAs. The top 20 differentially expressed blood miRNAs (10 up-regulated and 10 down-regulated) were selected as the aims for predicting target genes and their relative biological activities. Among these, there were 691 up-regulated differentially expressed genes and 1074 down-regulated genes that might be directly silenced by these 20 blood miRNAs. GO and KEGG enrichment analyses of the differentially expressed genes were conducted using DAVID to explore the biological processes and signalling pathways primarily involved in this system. Based on this analysis, a PPI network of target genes suggested five crucial blood miRNAs: hsa-miR-124-3p, hsa-miR-125a-3p, hsa-miR-29b-3p, hsa-miR-4276, and hsa-miR-575. The subsequent qRT-PCR results obtained from human blood samples validated the expression levels of these pivotal miRNAs in relation to pTNM stages, which exhibited a consistent pattern with that of clinical markers in GC. Survival curve analysis further supported the significant roles of these key blood miRNAs in gastric carcinoma. ROC analysis along with cross-validation using H. pylori-induced miRNAs as well as comparative study with clinical markers all indicated that these key miRNAs could have valuable implications for diagnosing GC.

The currently utilized tumour markers exhibit limited sensitivity and specificity, while the absence of a universally accepted “gold standard” further compounds this issue. Simultaneously, relying on a single tumour marker proves challenging in accurately capturing the intricate nature of tumours. Consequently, identifying effective markers and employing combined detection methods will serve as an efficacious approach to enhance the diagnostic value of tumour markers. MiRNAs have emerged as pivotal regulators in intricate biological processes, encompassing the pathogenesis of cancer27. Human malignancies exhibit distinctive characteristics such as sustained proliferation signals, activation of invasion and metastasis, angiogenesis, and evasion from immune destruction; miRNAs can actively participate in each of these features28,29. During carcinogenesis, miRNAs are involved in various signal pathways, including but not limited to miR-135a, miR-21, miR-218, miR-221/222, miR-375, and miR-451.They function as modulators (e.g. miR-148a, miR-155, miR-181b, miR-218, miR-374b-5p, and miR-499), epigenetic regulators (e.g. miR-34b/c, miR-129, and miR-212), and influencers of drug resistance (e.g. let-7a, miR-106a, miR-148a-3p, miR-495-3p, and miR-508-5p)30. The expression profiles of miRNAs in different cancer cells and tissues can be detected in the circulation, including blood, plasma and other body fluids. Mature miRNAs demonstrate remarkable stability in body fluids and possess high specificity for different cancer states, establishing them as potential non-invasive tumour marker. Consequently, blood miRNAs are increasingly recognized as invaluable biomarkers for the diagnosis and treatment of cancers31,32,33,34. There have been advances in the investigation of blood miRNAs in gastric carcinoma. For instance, the study conducted by Tasuku Matsuoka et al. in 2018 unveiled the up-regulation of several miRNAs, including miR-101, miR-106b, miR-125a, miR-129, miR-130b, miR-148b, miR-181c, miR-199a, miR-21, miR-23a, miR-27a, miR-29a, miR-212, miR-215, miR-218, miR-222-221, miR-331, miR-335, miR-370, miR-375, miR-449, miR-486, and miR-512, in blood samples as potential specific markers for diagnostic/prognostic/therapeutic of GC4. Similarly, in 2020, Soudeh Ghafouri-Fard et al. provided a comprehensively summary of the oncogenic roles of miR-17, miR-130, and miR-181 as well as the tumor suppressor functions of miR-124, miR-128, miR-27b, miR-29 family, miR-218, miR-34, miR-429, and miR-497 in blood samples from patients with gastric carcinoma15. In addition, a 2021 study from Mona Noohi et al. demonstrated an elevation in blood miR-21 level among the gastritis patients infected by H. pylori35. To summarized, despite the crucial blood miRNAs profiles may vary in gastric carcinoma and require identification, all aforementioned studies have consistently demonstrated the significant potential of circulating miRNAs as prognostic and diagnostic biomarkers in GC, thereby positioning them as promising therapeutic targets36,37.

In this study, utilizing bioinformatics tools, five crucial blood miRNAs (hsa-miR-124-3p, hsa-miR-125a-3p, hsa-miR-29b-3p, hsa-miR-4276, and hsa-miR-575) were identified as potential targets for diagnosis and treatment of gastric carcinoma. These key miRNAs and their respective target genes were involved in the biological processes closely correlant to tumorigenesis. Notably, platelet-derived growth factor receptor signalling pathway and transcription RNA polymerase II have been implicated in gastric carcinoma and development38,39.Additionally, focal adhesion and the PI3K-Akt signalling pathway are intricately linked to carcinogenesis16,40. Furthermore, through the identification of upstream miRNAs regulating target genes, the results obtained from Cytoscape analysis revealed that hsa-miR-124-3p exhibited the highest frequency among key miRNAs in the PPI network. Both examination of human blood specimens and survival curve analysis provided evidence supporting a correlation between hsa-miR-124-3p and gastric carcinogenesis. Additionally, ROC analysis further emphasized the significance of hsa-miR-124-3p in diagnosing gastric cancer. Up-regulation of hsa-miR-124-3p was observed in blood samples from CagA-positive GC patients. Moreover, several genes among the predicted targets of hsa-miR-124-3p exhibit a robust association with GC prognosis. For instance, ANXA5, a member of the Annexin family implicated in tumorigenesis and development across various cancers including GC41,42,43,44, and CAV1 to significantly regulate E-cadherin expression as well as alterations in cell morphology and migration ability of GC cells45,46. The impact of hsa-miR-125a-3p on immunity and carcinogenesis has been investigated through its regulation of tumour-associated signal pathways, including the Hippo pathway47. In this study, aberrant expression of hsa-miR-125a-3p was detected in blood specimens from patients with GC. Importantly, CagA also induced down-regulation of hsa-miR-125a-3p, consistent with its inhibition observed in blood samples from GC patients. Our findings also revealed a potential downstream target of hsa-miR-125a-3p as PRDM1, which exhibited a correlation between its expression and survival probability among GC patients. The involvement of PRDM1 in the regulation of B cell and T cell differentiation, as well as its crucial role in immunosuppression, has been demonstrated to be associated with various types of cancers48,49. Hsa-miR-29b-3p plays a pivotal role in tumorigenesis and metastasis50,51,52,53. Research has demonstrated that hsa-miR-29b-3p exhibits potential as a crucial circulating miRNA with clinical significance in endometrial cancer54. Additionally, activation of PER1, one of the predicted key target genes of hsa-miR-29b-3p in our findings, can effectively impede the progression of pancreatic cancer55. The correlation between expression levels of circulating hsa-miR-29b-3p and gastric cancer prognosis and diagnosis was revealed through GC blood samples analysis in this study. The Up-regulation of hsa-miR-29b-3p was also observed in qRT-PCR results of CagA-positive blood samples from GC patients. Hsa-miR-4276 has been implicated in conferring resistance to influenza A infection in lung epithelial cells56. However, limited research exists regarding the involvement of hsa-miR-4276 in tumorigenesis, as well as its target genes RNF217 and IP6K1. Conversely, among the key nodes identified through Cytoscape analysis, there was a significant enrichment of downstream genes regulated by hsa-miR-4276, suggesting a strong correlation between hsa-miR-4276 and its target genes with GC. This possibility is further supported by patient blood specimens’ check and predictions based on H. pylori-induced miRNAs, as well as the observed down-expression level in CagA-positive blood samples. The protein MSRB3, which has been identified as a crucial regulator of proliferation and migration in GC cells, making it a potential marker for predicting peritoneal metastasis and poor prognosis57,58, was predicted as the target of hsa-miR-575 in this study. Hsa-miR-575 has also been identified to regulate development of gastric cancer by targeting PTEN59. Differential expression of hsa-miR-575 was observed in GC through blood checks and its down-regulation by CagA was also confirmed. The patterns reflecting the correlation with pTNM stages and H. pylori infection in GC patients of these blood-based key miRNAs were also found to be consistent with their expression levels, which seemed to be better than those of clinical markers. Considering the feasibility of combining miRNA markers with clinical blood markers, we compared the diagnostic accuracy for gastric cancer detection using clinical blood markers alone, miRNA markers alone, and the combination of both approaches. The results demonstrated that the highest positive detection rate was achieved when using miRNA markers alone (Figure S9). Therefore the present findings highlight the robustness and significance of the identified blood miRNAs in gastric carcinogenesis and diagnosis, emphasizing their potential as promising biomarkers.

However, further in-depth exploration is warranted. In addition to miRNAs, it has also been reported that eleven cytokines exhibit significant increases in blood samples from patients with GC. To precisely determine the expression levels of key miRNAs and their target genes at different stages of GC, a larger number of blood samples from GC patients are required. The identified blood miRNAs such as miR-21 not included in this model also can serve as controls to assess the significance of the five key miRNAs. Furthermore, it is imperative to investigate the correlation between five key miRNAs and pTNM stages as well as H. pylori infection in gastric carcinoma, as well as elucidate the underlying mechanisms involved.