This investigation followed the ethical standards set forth by the World Medical Association throughout the sample collection period from January 2021 to February 2024. Blood samples were collected from 60 patients diagnosed with primary hepatocellular cancer and 60 medical examiners at Affiliated Hospital of Nantong University. Three milliliters of blood were collected from each participant through venipuncture, and serum was isolated via centrifugation. Additionally, 36 pairs of GC tissues and the non-cancerous tissues that surrounded them were taken from our hospital’s operating room. All of the aforementioned patients received a pathology-confirmed diagnosis, gave their informed permission in accordance with ethical standards. Ethics committee approval was secured from the Affiliated Hospital of Nantong University (ethics review report number: 2018-L055).

Total RNA isolation was performed with Trizol, followed by assessment of RNA quality and quantity. The Multi-sample library preparation kits was utilized to prepare a small RNA library using 1 μg total RNA from each sample. The library was sequenced, and the clean reads obtained were analyzed in comparison to MINTbase utilizing MINTmap software, leading to the generation of new tsRNA predictions for reads absent in MINTbase. The analysis of changes in tsRNAs across various groups was conducted using EdgeR software.

The qRT-PCR reaction was conducted utilizing a Bio-Rad CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories, Inc., United States). U6 was utilized as an internal control, while RNA extracted from a pooled serum of 10 healthy donors acted as an external control. The sequences of the primers can be found in Supplementary Table S1. The 2^−ΔΔCT method was used to analyze the data using the formula: ΔΔCt = ΔCt_target (Ct_target−Ct_reference) - ΔCt_{external control} (Ct_{external control}−Ct_reference) during the reaction.

The amplification products of each tsRNA underwent verification through Sanger sequencing. The purified amplicons underwent sequencing with Applied Biosystems technology. Sequences underwent analysis with the ABI3730XL DNA Analyzer and were compared to tsRNA on a sequence-by-sequence basis.

A set of human GC cell lines (MGC-803, AGS, MKN-45, SGC-7901, BGC-823) along with human gastric mucosal epithelial cell lines (GES-1) was sourced from the Chinese Academy of Sciences (Shanghai, China). All cells were cultured in RPMI 1640 medium (Corning, Manassas, VA, United States) enriched with 10% fetal bovine serum (Gibco, Waltham, MA, United States) and 1% penicillin and streptomycin (HyClone, Logan, UT, United States) under standard conditions of 37°C and 5% CO₂ in a humidified incubator.

Total RNA was extracted from serum samples of GC patients utilizing a total RNA purification and isolation kit (BioTeke, Wuxi, Jiangsu, China). In contrast, total RNA from tissue and cell samples was obtained using TRIzol reagent (Invitrogen, Carlsbad, CA, United States). The process of reverse transcription of total RNA to cDNA was executed with a revert aid RT reverse transcription kit (Thermo Fisher Scientific, Waltham, MA, United States). This was carried out in a 10 µL reaction volume, incubated at 42°C for 60 min, and subsequently inactivated at 70°C for 5 min. Prior to the reaction, quantification was necessary.

The nomenclature and structures of tsRNAs: MINTbase version 2.0, ¹ UCSC Genome Browser database, ² and Transfer RNA Database. ³

Identification of tsRNAs and Target Genes: Miranda, ⁴ and Pictar. ⁵

Screening for Differentially Expressed Genes: Kyoto Encyclopedia of Genes and Genomes (KEGG), ⁶ Gene Ontology enrichment analysis ⁷ and Reactome Pathway Database. ⁸

The analysis of high-throughput sequencing results was conducted utilizing the R package edgeR, with significance determined at p < 0.05 and |Log₂FC|>2. The analysis of data was conducted using SPSS Statistics Version 20.0 (IBM SPSS Statistics, Chicago, IL, United States) and GraphPad Prism 9.0 (GraphPad Software, San Jose, California, United States). Prior to conducting the analysis, the normality of the data was evaluated through normality and lognormality tests, which confirmed that all data adhered to a normal distribution. Group differences were assessed through the application of t-tests. The ROC curve and the area under the curve (AUC) were utilized to assess the diagnostic efficacy of tsRNAs in GC. Prior to plotting the ROC curve, binomial logistic regression was conducted. Cox’s proportional hazards model was utilized for the multivariate analysis. The risk ratio along with its 95% confidence interval (CI) was documented for each marker. The determination of tsRNA cut-off values utilized the Youden index, whereas the cut-off values for CEA, CA199, and CA724 were established based on the reference range from the Affiliated Hospital of Nantong University, specifically 5 ng/mL, 37 U/mL, and 10 U/mL, respectively. Independent experiments were conducted a minimum of three times, and statistical significance was established when the P-value was less than 0.05.

The deregulation of ncRNA expression is closely associated with the initiation and advancement of cancers. Studies have focused on ncRNAs as potential diagnostic biomarkers because of their tissue specificity and convenient and stable detection in human tissues and body fluids. tsRNAs have been shown to circulate in a very stable, cell-free state inside the serum. The study design is presented in Figure 1. We initially utilized a high-throughput sequencing technique to identify the differentially expressed tsRNAs in GC tissues and adjacent non-cancerous specimens. The identified tsRNAs with p < 0.05 and |Log₂FC|>2 were differentially expressed.

Statistical results showed 435 tsRNAs were dysregulated (249 were up-regulated, and 186 were down-regulated). The expression heatmap and volcano plot are presented in Figures 2A, B. Ten candidates were selected from the top five up-regulated and top five down-regulated tsRNAs, based on the average expression level and the uniformity of fluorescence signals within each group (Figure 2C; Supplementary Table S2). To confirm the results of RNA sequencing, we employed the same three pairs of GC tissues for qRT-PCR. The results demonstrated that eight tsRNAs (the up-regulated tsRNAs include: tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, tRF-30-IK9NJ4S2I7L7, and the down-regulated tsRNAs include: tRF-38-W6RM7KYUPRENRHD2, tRF-37-LBRY73W0K5KKOV2, tRF-36-JB59V3WD8YQ84VD, tRF-25-MBQ4NKKQBR, tRF-36-0KFMNKYUHRF867D, tsRNAs named followed version 2.0 of MINTbase) expression levels were consistent with the high-throughput sequencing (Figure 2D).

The structure and cleavage site of the eight tsRNAs were visualized in Figure 3A. To further validate the eight tsRNAs expression levels, we obtained 36 pairs of tissue samples from GC patients and assessed the expression levels of the eight tsRNAs. Results showed that the three up-regulated tsRNAs expressed higher levels in carcinomas than paracancerous specimens, and the five down-regulated tsRNAs were lowly expressed (Figure 3B). The results of unpaired comparisons were consistent with the results of paired comparisons within 36 pairs of tissue samples (Figure 3C). Simultaneously, we observed the expression levels of eight tsRNAs across several GC cell lines and discovered that the three up-regulated tsRNAs were markedly elevated in GC cells relative to GES-1 cell line, whereas the five down-regulated tsRNAs exhibited the inverse pattern (Figure 3D). This section assessed the expression of eight tsRNAs in GC tissue samples and cell lines, revealing that these tsRNAs were dysregulated in GC.

The collection of peripheral blood samples from patients for testing is the most convenient diagnostic method; hence, we sought to determine if these tsRNAs were differentially expressed in the serum of 60 GC patients compared to 60 healthy controls. The serum concentration of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7 showed a statistically significant increase in GC patients, whereas tRF-38-W6RM7KYUPRENRHD2, tRF-37-LBRY73W0K5KKOV2, tRF-36-JB59V3WD8YQ84VD, tRF-25-MBQ4NKKQBR, and tRF-36-0KFMNKYUHRF867D showed an opposite trend (Figure 4A). The expression levels of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7 in tissues were correlated with the expression levels in the same patient’s serum (Figure 4B).

For each tsRNA group, we split 60 GC patients in half, with 30 patients in the high expression group and 30 in the low expression group, according to the median serum expression of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7, respectively. Clinicopathological parameters were evaluated the correlations with the three up-regulated tsRNAs expression levels in serum using the χ² test. The results demonstrated a strong correlation between gender (p = 0.0321), differentiation grade (p = 0.0062), and TNM stage (p = 0.035) and tRF-31-PNR8YP9LON4VD expression. There was a significant correlation between the expression of tRF-30-MIF91SS2P4FI and age (p = 0.0242), TNM stage (p = <0.0001), tumor depth (p = 0.0149), and lymph node metastasis (p = 0.0242). A statistically significant connection (p = 0.0016) was found between the expression of tRF-30-IK9NJ4S2I7L7 and TNM stage (Table 1). The elevated expression of tsRNAs may serve as a possible indicator for predicting tumor malignant development. Subsequently, we categorized the substantially varied clinicopathological features into groups to further evaluate the disparities in tsRNA expression levels within each group. The results demonstrated that the expression of three upregulated tsRNAs increased with the early TNM stages (I/II) and T1/T2 tumor depth of gastric cancer (Figures 4C, D).

We further assessed the diagnostic effectiveness of three highly expressed tsRNAs for GC based on their expression levels in patients with GC using ROC analysis. The diagnostic AUC values of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7 were 0.808 (95% confidence interval, 0.694–0.924), 0.774 (95% CI, 0.657–0.892), and 0.843 (95% CI, 0.747–0.940), respectively (Figure 4E). We compared the diagnostic efficiency of the three up-regulated tsRNAs with GC clinical biomarkers CEA, CA199, and CA724. ROC analysis indicated that the three tsRNAs had better diagnostic efficacy than CEA, CA199, and CA724. AUC increased to 0.956 (95% CI, 0.906–1) when all biomarkers were combined for diagnosis. Additionally, we analysed the diagnostic efficacy of the three tsRNAs in TNM Ⅰ/Ⅱ stage of GC based on the clinicopathological parameters of the patients. We discovered that their diagnostic value increases even further in the early stages of GC. The diagnostic AUC values of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7 were 0.849, 0.851 and 0.847 (Figure 4E). In contrast, CEA and CA199 have limited diagnostic value (AUC values of 0.564 and 0.578) in the early stages of GC.

Consequently, we aimed to further examine and forecast the involvement of these tsRNAs in the advancement of gastric cancer. We predicted downstream target genes of tRF-31-PNR8YP9LON4VD, tRF-30-MIF91SS2P4FI, and tRF-30-IK9NJ4S2I7L7 using miRanda and PicTar databases. The prediction results of the two databases were intersected, and the results showed 1,187 target genes that were most likely bound to tRF-30-IK9NJ4S2I7L7, 445 to tRF-30-MIF91SS2P4FI, and 2,114 to tRF-31-PNR8YP9LON4VD (Figure 5A). Integrating all 3,435 predicted target genes, the KEGG signaling network was shown to be considerably enriched in pathways related to cancer, the MAPK signaling pathway, and the control of the actin cytoskeleton, according to an enrichment analysis (Figure 5B). Gene Ontology functional enrichment study suggested that the up-regulated tsRNAs of GC may play a significant role in histone modification, signal release, and cell growth in biological processes (Figure 5C). The phrase having the highest statistical significance inside a cluster was selected to represent that cluster. The enrichment analysis is depicted in Figure 5D. It is evident that certain genes play a role in multiple functions. Nonetheless, the processes behind tsRNA expression in the control of biological activity in GC cells require additional investigation.