Transfer RNA-derived fragment tRF-28-P4R8YP9LOND5 as a novel serum biomarker for gastric cancer: diagnostic efficacy and clinicopathological correlations

Introduction

Background

Gastric cancer (GC), ranking as the fifth most prevalent malignant tumor globally, imposes a substantial burden on society with persistently high incidence rates. In 2020, there were 1.08 million new cases and 760,000 deaths reported. The 5-year survival rate remains at approximately 20%, primarily attributed to the majority of cases being diagnosed at advanced stages, while the lack of early detection technologies remains a core bottleneck (1-5). Traditional serum biomarkers such as carcinoembryonic antigen (CEA) and cancer antigen 199 (CA199) exhibit limited clinical value: their sensitivity for early-stage GC is below 50%, and they demonstrate false-positive rates as high as 20–30% in benign conditions like gastritis or gastric ulcers (6-9). Only 50% of GC patients show elevated CEA expression (6), whereas CA199 elevation is predominantly observed in advanced GC stages (8). This underscores the urgent clinical need for novel serum biomarkers with higher sensitivity and specificity.

Transfer RNA-derived fragments (tsRNAs), a class of non-coding RNAs (18–40 nt) generated by nucleolytic cleavage of mature or precursor transfer RNAs (tRNAs) by enzymes like angiogenin (ANG) and Dicer, have garnered substantial attention in cancer precision medicine. These molecules are classified into 5’/3’tRNA halves (tiRNAs) and tRNA-derived fragments (tRFs), exerting functional roles both intracellularly—through mechanisms such as gene expression regulation [e.g., targeting messenger RNA (mRNA) 3’UTRs], protein translation interference, and metabolic reprogramming—and extracellularly via stable presence in biofluids like serum (1-3). In GC, for instance, the 3’tRF-Val promotes tumor cell proliferation by targeting EEF1A1 to facilitate MDM2-mediated p53 ubiquitination (6), while the 5’tRF-Tyr inhibits metastatic potential by binding to the RNA-binding protein hnRNPD and disrupting migration-associated mRNA interactions (7). Notably, serum tRF-29-R9J8909NF5JP has demonstrated diagnostic efficacy with an area under the curve (AUC) of 0.889 (82% sensitivity, 85% specificity), outperforming traditional markers like CEA (AUC =0.722) and CA199 (AUC =0.678) (8,9). The biogenesis of tsRNAs is tightly regulated by tissue-specific nucleases such as ANG and Dicer. ANG preferentially cleaves tRNA anticodon loops to generate 5’-tiRNAs (e.g., tRF-28-P4R8YP9LOND5), while Dicer processes precursor tRNAs into smaller 3’-tRFs (1,2). Notably, ANG overexpression in GC cells enhances the production of tsRNAs with stable stem-loop structures, which are resistant to serum RNase degradation compared to linear mRNA (10,11). Despite their stability and tumor-specific expression, systematic validation of tsRNAs as diagnostic biomarkers in GC remains scarce.

Rationale and knowledge gap

Although the advantages and prospects of tsRNAs are highly promising, the current bottleneck issues encountered cannot be overlooked: first, most studies focus on individual tsRNA molecules, lacking systematic profiling of serum tsRNA expression landscapes to decipher their synergistic regulatory networks in tumorigenesis. Second, mechanistic insights into tsRNA biogenesis (e.g., tissue-specific nuclease activity), RNA modifications [e.g., N6-methyladenosine (m6A)/5-methylcytosine (m5C) impacts on stability], and cross-talk with other non-coding RNAs [microRNAs (miRNAs), long non-coding RNAs (lncRNAs)] remain incomplete (10,12,13). Third, the development of multi-biomarker models validated in large-scale cohorts is scarce, hindering clinical translation—for example, while combining tRF-17-18VBY9M with CEA/CA199 achieved an AUC of 0.865 (8), optimizing marker combinations via machine learning to balance sensitivity and specificity remains unaddressed.

Objective

Based on this, this study aims to: (I) clarify the expression profile of serum tsRNA in GC and its association with tumor invasion and metastasis, providing new targets for mechanistic research; (II) develop and validate a multimodal biomarker combination to enhance diagnostic accuracy. We present this article in accordance with the STARD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1773/rc).

Methods

Database screening and bioinformatics analysis

Differentially expressed tRFs in GC were identified from the OncotRF (http://bioinformatics.zju.edu.cn/OncotRF) using stringent criteria P<0.05 and log₂ |fold change| >2. Three candidate molecules—tRF-28-P4R8YP9LOND5, tRF-24-2IUIX1Q7HV, and tRF-22-L7S5QKF14—were shortlisted. tRF-28-P4R8YP9LOND5 was further characterized using MINTbase v2.0 to determine its classification as a 5’-tRF, precursor tRNA origin (tRNA-Gly-GCC-1), and cleavage sites within the D-loop and anticodon loop. Genomic localization was validated via the UCSC Genome Browser (GRCh38/hg38) at chr1:161,443,304–161,443,331.

Clinical sample collection and processing

Sample cohorts

A total of 117 pretreatment GC patients (85 males, 32 females; median age 62 years, range 35–82 years) were enrolled at the Affiliated Hospital of Nantong University between January 2024 and July 2025. All cases underwent pathological confirmation, and none had received neoadjuvant therapy. Serum samples from 89 healthy donors (free of cancer and inflammation) and 51 patients with endoscopy-confirmed gastritis were included as control groups.

Paired preoperative (n=28) and postoperative (n=28) sera were collected from the same patients to monitor temporal changes. An additional 15 postoperative samples were included, yielding a total of 43 postoperative samples for comparison with healthy controls. Tumor, node, and metastasis (TNM) staging was based on a combination of preoperative imaging and postoperative histopathological assessment (n=117 for preoperative analysis; n=60 for postoperative validation). Postoperative samples were collected within 2 weeks after curative resection, excluding patients with M1 disease or incomplete follow-up (final n=43).

Sample processing

Serum was isolated by centrifugation at 3,000 ×g for 10 minutes at 4 °C, aliquoted into sterile tubes, and stored at −80 °C until use. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Affiliated Hospital of Nantong University (Ethics No. 2023-K167-01) and informed consent was obtained from all individual participants.

Cell culture

Human GC cell lines (HGC-27, NCI-N87, AGS) and normal gastric epithelial cells (GES-1) were sourced from the Chinese Academy of Sciences Cell Bank. Cells were maintained in RPMI 1640 medium (Gibco) supplemented with 10% FBS (Gibco) and 1% penicillin-streptomycin, under standard culture conditions (37 °C, 5% CO₂). Subculture or RNA extraction was performed when cells reached 80% confluence.

RNA extraction and reverse transcription quantitative polymerase chain reaction (RT-qPCR)

RNA isolation

Serum RNA

Extraction was performed using the BioTeke Serum RNA Kit, following the manufacturer’s recommended protocol, with a final elution volume of 50 µL.

Cellular RNA

Isolation was carried out using Trizol reagent (Vazyme), involving chloroform extraction, isopropanol precipitation, and resuspension in RNase-free water.

cDNA synthesis

Reverse transcription of 500 ng total RNA was performed using the HiScript III 1st Strand cDNA Synthesis Kit (Vazyme) in a 10 µL reaction mixture. The reaction was incubated at 42 °C for 60 minutes, followed by heat inactivation at 70 °C for 5 minutes.

Real-time quantitative PCR

Quantitative PCR was conducted on an ABI QuantStudio 5 using ChamQ Universal SYBR qPCR Master Mix (Vazyme). Each 20 µL reaction contained 5 µL cDNA, 1 µL of each primer, 3 µL DEPC and 10 µL SYBR Mix. The thermal cycling conditions were as follows: 30 seconds at 95 °C for pre-denaturation, followed by 40 cycles of 5 seconds at 95 °C and 30 seconds at 60 °C. U6 small nuclear RNA was used as the internal control, with relative expression levels calculated using the 2⁻ᵟᵟcycle threshold (Ct) method.

Molecular validation assays

Agarose gel electrophoresis

Five microliters of RT-qPCR product were mixed with 1 µL of 6× loading buffer and separated by electrophoresis on a 2% agarose gel containing 0.5 µg/mL EB at 120 V for 20 minutes. Gel images were captured using a Bio-Rad imaging system, with U6 (100 bp) serving as a size marker to verify amplicon specificity.

Sanger sequencing

PCR products were subjected to Sanger sequencing at Sangon Biotech (Shanghai). The resulting chromatograms were aligned with the tRF-28-P4R8YP9LOND5 reference sequence in MINTbase v2.0 using BioEdit software, confirming 100% sequence identity.

Methodological evaluation

Reproducibility analysis

Intra-assay variability

Twenty mixed serum samples (GC and healthy controls) were assayed in duplicate in the same run. The coefficient of variation (CV) for Ct values was 1.60% (20.525±0.316).

Inter-assay variability

A pooled sample was tested across 10 independent runs, yielding a CV of 0.75% (20.086±0.144).

Stability tests

Room temperature stability

Mixed serum samples were stored at 25 °C for up to 24 hours. Ct values remained stable (ΔCt <1.0) throughout the period (tRF: 23.5±0.3; U6: 22.8±0.2).

Freeze-thaw stability

Samples underwent 0–10 freeze-thaw cycles between −80 °C and room temperature. Ct value fluctuations were minimal (ΔCt <0.5) after 10 cycles.

Dilution linear range

Serial dilutions (10⁰–10⁵-fold) of pooled serum cDNA were used to generate standard curves. tRF-28-P4R8YP9LOND5 exhibited a linear dynamic range with an R² of 0.9972 (Y=−3.059X + 16.26), while U6 showed an R² of 0.9956 (Y=−2.440X + 19.19).

Clinicopathological correlation analysis

Statistical analyses were conducted using SPSS Statistics 27.0 and MedCalc 20.0. Continuous variables were reported as mean ± SD and compared using independent t-tests (two groups), Wilcoxon tests (paired samples), or one-way analysis of variance (ANOVA) (multiple groups). Categorical variables were analyzed via Chi-squared tests or Fisher’s exact test (for small samples), while continuous variables like age were evaluated using Mann-Whitney U tests. Statistical significance was set at P<0.05.

Diagnostic performance assessment

Receiver operating characteristic (ROC) curve analysis was performed using GraphPad Prism 9.0 to determine AUC, sensitivity, specificity, and accuracy. Optimal cutoff values were identified using the Youden index, and AUC comparisons were carried out via the DeLong test. The diagnostic utility of tRF-28-P4R8YP9LOND5 alone and in combination with CEA, CA199, and CA724 was evaluated. For multimarker modeling, logistic regression with backward stepwise selection was used, incorporating variables based on clinical relevance and univariate significance (P<0.10).

Target gene prediction and pathway analysis

Four computational algorithms—miRanda (energy score <−20), PITA (score <−1.0), TargetScan (conserved sites with cumulative weighted context++ score >−0.3), and RNAhybrid (minimum free energy <−15 kcal/mol)—were employed to predict tRF-28-P4R8YP9LOND5 target genes. Genes commonly identified by all four algorithms (n=732) were selected as high-confidence targets. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were then performed to identify significantly enriched biological processes and signaling pathways (P<0.05).

Statistical analysis

Continuous data are presented as mean ± SD. Group comparisons were performed using independent t-tests, Wilcoxon tests, or one-way ANOVA, as appropriate. Categorical variables were analyzed with Chi-squared tests or Fisher’s exact test. ROC curve parameters were calculated using MedCalc, with optimal cutoffs determined by the Youden index. AUC comparisons were performed via DeLong tests, with P<0.05 denoting statistical significance.

Results

Database screening for tRF-28-P4R8YP9LOND5

To identify potential biomarkers for GC, three tsRNAs with significant differential expression—tRF-28-P4R8YP9LOND5, tRF-24-2IUIX1Q7HV, and tRF-22-L7S5QKF14—were retrieved from the OncotRF database (http://bioinformatics.zju.edu.cn/OncotRF) using stringent criteria: adjusted P value <0.05 and log₂ |fold change| >2. Expression profiling was conducted in serum samples from 20 GC patients, 20 healthy donors, and GC cell lines. Compared with healthy controls, tRF-28-P4R8YP9LOND5 displayed significantly elevated relative expression in GC patients (P<0.001), with median levels of 2.494 vs. 1.012, respectively. This pronounced upregulation, consistent with bioinformatics predictions, highlights its potential as a GC-specific biomarker (Figure 1A). In contrast, tRF-24-2IUIX1Q7HV showed no significant expression disparity between GC patients and healthy donors (non-significant, P=0.81), effectively excluding it from further consideration as a diagnostic or prognostic biomarker (Figure 1B). Conversely, tRF-22-L7S5QKF14 was markedly downregulated in GC patients compared to healthy donors (P<0.001), with median expression levels of 0.4239 vs. 0.9107. This reduction suggests a potential tumor-suppressive role, prompting the need for functional validation (e.g., loss-of-function assays) to elucidate its mechanistic relevance (Figure 1C). Furthermore, we collected GC and paracancerous tissues from these 24 GC patients to further validate the expression level of tRF-28-P4R8YP9LOND5 and discovered that it has a higher expression level in GC tissues than in adjacent tissues (Figure 1D). In GC cell lines (AGS, NCI-N87, HGC-27) and normal GES-1, tRF-28-P4R8YP9LOND5 exhibited the highest expression in the highly invasive HGC-27 cells (P<0.0001), with a relative expression of 5.117 vs. 1.002 in GES-1 (Figure 1E). These findings confirm the oncogenic potential of tRF-28-P4R8YP9LOND5 in cancer cells and establish HGC-27 as a robust model for downstream mechanistic investigations.

Figure 1 The expression levels of tsRNAs in GC and the screening of tRF-28-P4R8YP9LOND5. (A) Scatter plot of tRF-28-P4R8YP9LOND5 expression in GC patients (red) and healthy donors (blue). (B) Scatter plot of tRF-24-2IUIX1Q7HV expression, with no significant difference (P=0.81). (C) Scatter plot of tRF-24-2IUIX1Q7HV expression in GC patients (red) and healthy donors (blue). (D) The expression level of tRF-28-P4R8YP9LOND5 in 24 pairs of GC tissues and adjacent tissues. (E) Bar graph of tRF-28-P4R8YP9LOND5 expression in three GC cell lines (AGS, NCI-N87, HGC-27) and human GES-1. ns, not significant; ***, P<0.001; ****, P<0.0001. Error bars represent mean ± SD or median. GC, gastric cancer; GES, gastric epithelial cell; SD, standard deviation; tsRNA, transfer RNA-derived fragments.

Structure and origin of tRF-28-P4R8YP9LOND5

To validate the bioinformatics predictions, we first characterized in detail the molecular structure and genomic origin of tRF-28-P4R8YP9LOND5, providing a mechanistic basis for its differential expression in GC. Multi-dimensional characterization of tRF-28-P4R8YP9LOND5 was achieved through integrative multi-database mining and experimental validation, defining its molecular architecture, biogenesis origin, chromosomal localization, and sequence fidelity to underpin subsequent functional and translational research. Analysis using MINTbase v2.0 revealed that tRF-28-P4R8YP9LOND5 (GCATGGGTGGTT CAGTGG TAGAATTCTC) is a 17-nucleotide 5’-tRF derived from the precursor tRNA-Gly-GCC-1 (tRNA35_GlyGCC_1), with genomic coordinates mapped to the human hg19/GRCh37 genome (Figure 2A). This confirms its origin from tRNA fragmentation, a hallmark feature of tsRNAs. The MINTbase v2.0 interface (Figure 2A) classifies it as an internal tRF (i tRF) generated by nucleolytic cleavage within the tRNA anticodon loop, with explicit annotation of its nucleotide sequence, precursor tRNA sources (e.g., tRNA Gly CCC), and genomic locus. The database’s nomenclature system—rooted in tRNA origin and cleavage site—designates it as an i tRF, furnishing essential sequence evidence for primer design and experimental validation.

Figure 2 Structure and origin of tRF-28-P4R8YP9LOND5. (A) MINTbase v2.0 entry: sequence, tRNA source (tRNA-Gly-GCC-1), and genomic location. (B) Secondary structure of parent tRNA, highlighting tRF-28-P4R8YP9LOND5 (red nodes) in D-loop/anticodon loop. (C) UCSC Genome Browser: chromosomal location (chr1:161,443,304–331). (D) Agarose gel: tRF-28-P4R8YP9LOND5 (75 bp) and U6 (100 bp) amplicons, confirming size. (E) Sanger sequencing: electropherogram validating tRF-28-P4R8YP9LOND5 sequence. tRNA, transfer RNA; UCSC, University of California, Santa Cruz (Genome Browser).

A molecular structure schematic (Figure 2B) illustrates that this fragment arises from cleavage within the D-loop and anticodon loop of tRNA precursors, including tRNA Gly CCC and tRNA Glu TTC. Key cleavage sites are highlighted with red nodes, revealing a biogenesis pathway involving processing of conserved domains in mature tRNAs. This structural annotation provides mechanistic clues for investigating its functional targets (e.g., mRNA binding motifs). Subsequently, chromosomal localization via the UCSC Genome Browser mapped tRF-28-P4R8YP9LOND5 to human chromosome chr1:161,443,304–161,443,331 (GRCh38/hg38), validating its nuclear genomic origin (Figure 2C). This chromosomal context is pivotal for deciphering its biogenesis pathway (e.g., ANG-mediated cleavage) and potential regulatory roles in GC. The localization data offer insights into upstream regulatory elements (e.g., promoter regions, epigenetic modifications) and genetic aberrations (e.g., copy number variations) implicated in disease pathogenesis. Agarose gel electrophoresis validated the ~75 bp amplicon of tRF-28-P4R8YP9LOND5, using U6 RNA (100 bp) as a size marker (Figure 2D). Size congruence with predicted values confirmed the specificity of RT-qPCR products, ruling out nonspecific amplification. Sanger sequencing chromatograms (Figure 2E) further verified 100% sequence identity with the MINTbase v2.0 reference sequence, eliminating experimental artefacts from primer design or amplification bias. Electrophoretic patterns matched database-derived sequences, unequivocally confirming the molecule’s identity and precluding misidentification of small RNAs.

Methodological evaluation of serum tRF-28-P4R8YP9LOND5

The RT-qPCR assay for tRF-28-P4R8YP9LOND5 demonstrated exceptional precision, with inter-assay (across runs) and intra-assay (within-run) variability systematically evaluated. Inter-assay analysis yielded a mean ± SD of 20.086±0.144 (CV =0.75%), while intra-assay measurements resulted in 20.525±0.316 (CV =1.60%). The internal control U6 RNA showed comparable precision (inter-assay CV =1.02%, intra-assay CV =2.40%), validating the consistency of methodological performance (Table 1). All CV values remained below the clinical acceptability threshold (CV <5%), ensuring robust reproducibility of results. Notably, the intra-assay and inter-assay CVs for tRF-28-P4R8YP9LOND5 (1.60% and 0.75%, respectively) were significantly lower than those reported for serum microRNAs [e.g., miR-21: CV =3.8–5.2% in GC studies (14,15)], highlighting the superior precision and stability of tsRNA-based diagnostics.

Ct values for tRF-28-P4R8YP9LOND5 and U6 remained stable over a 24-hour period at room temperature (tRF: 23.5±0.3; U6: 22.8±0.2; P>0.05) (Figure 3A), confirming a critical attribute for clinical sample handling, as this minimal temporal variation supports extended sample processing or delayed analysis in diagnostic workflows. Following up to 10 freeze-thaw cycles, Ct values for both molecules exhibited negligible fluctuations (ΔCt <0.5) (Figure 3B), demonstrating tolerance to standard biobanking protocols and ensuring assay reliability in clinically stored samples.

Figure 3 Methodological validation of tRF-28-P4R8YP9LOND5 assay. (A) Time stability (0–24 hours)—Ct values for tRF-28-P4R8YP9LOND5 and U6 remain unchanged (P>0.05), confirming room temperature stability. (B) Freeze-thaw stability (0–10 cycles)—minimal Ct variation (ΔCt <0.5), validating robustness under repeated freezing. (C,D) Linear regression of dilution factor vs. Ct (tRF-28-P4R8YP9LOND5 and U6)—high R² (>0.995) ensures accurate quantification across wide concentration ranges. Ct, cycle threshold.

Linear regression of log-transformed dilution factors against Ct values for tRF-28-P4R8YP9LOND5 yielded an R² of 0.9972 (Y=−3.059X + 16.26) (Figure 3C), spanning six orders of magnitude (10⁻⁶ to 10⁰ dilutions), indicative of high sensitivity and precise quantification across broad concentration ranges. A comparable linear relationship for U6 (R²=0.9956, Y =−2.440X + 19.19) (Figure 3D) enabled reliable normalization in both low-expression (healthy donors) and high-expression (GC patients) contexts (Figure 1A). Notably, tRF-28-P4R8YP9LOND5 exhibited a steeper slope, reflecting enhanced sensitivity to template dilution compared with U6.

Analysis of serum tRF-28-P4R8YP9LOND5 expression and its correlation with clinicopathological parameters

With the assay validated for high precision and stability, we investigated the clinical relevance of tRF-28-P4R8YP9LOND5 expression in relation to tumor progression and patient characteristics. Serum tRF-28-P4R8YP9LOND5 levels were significantly elevated in GC patients compared to healthy donors (P<0.001) and gastritis patients (P<0.001), with no difference between healthy donors and gastritis patients (P=0.31; Figure 4A). This highlights its specificity for differentiating GC from benign conditions and indicates tumor-specific overexpression in GC. Preoperative tRF-28-P4R8YP9LOND5 levels were significantly higher than postoperative levels (P<0.001), with individual line graphs showing a postoperative downward trend, suggesting its utility as a postoperative monitoring marker. Surgical resection reduced tRF-28-P4R8YP9LOND5 expression, as postoperative levels in 43 patients (including 28 paired samples) were comparable to healthy donors (P=0.14), supporting a direct correlation with tumor burden (Figure 4B,4C). Expression was significantly associated with tumor invasion depth (T3–T4 > T1–T2, P<0.001), advanced TNM stage (III–IV > I–II, P<0.001), lymph node metastasis (P=0.03), and neurovascular invasion (P=0.002) (Figure 4D-4G), reflecting its link to aggressive tumor biology. Clinicopathologic analysis (Table 2) showed higher expression in patients with tumor diameter ≥5 cm (21/32, P=0.03), neurovascular invasion (21/56, P=0.01), and advanced TNM stages (32/47, P=0.02), consistent with tumor invasiveness. No correlations were found with sex, age, histologic differentiation, or Lauren classification (all P>0.10), indicating its focus on tumor biology rather than demographics. The strong association with advanced disease (Figure 4E, Table 2) suggests tRF-28-P4R8YP9LOND5 may serve as a prognostic marker for poor outcome. Its correlations with tumor size, staging, and invasive features, combined with postoperative downregulation, highlight its potential for predicting aggressive phenotypes and monitoring treatment response.

Figure 4 Serum tRF-28-P4R8YP9LOND5 expression and clinicopathological correlations. (A) tRF-28-P4R8YP9LOND5 levels in GC (n=117, red), healthy donors (n=89, blue), and gastritis patients (n=51, green). (B) Paired preoperative vs. postoperative tRF-28-P4R8YP9LOND5 levels in GC patients (n=28). (C) Postoperative tRF-28-P4R8YP9LOND5 levels in GC patients (n=43, orange) compared with healthy donors (n=86, blue). (D) tRF-28-P4R8YP9LOND5 expression in early-stage (T1–T2, n=44, pink) and advanced-stage (T3–T4, n=70, orange) GC, alongside healthy donors (n=89, blue). (E) tRF-28-P4R8YP9LOND5 levels in TNM stage I–II (n=39, pink) vs. III–IV (n=78, orange) GC patients and healthy donors (n=89, blue). (F) tRF-28-P4R8YP9LOND5 expression in lymph node metastasis-positive (n=19, orange) and -negative (n=96, pink) GC patients. (G) tRF-28-P4R8YP9LOND5 levels in nerve/vascular invasion-positive (n=44, orange) and -negative (n=46, pink) GC patients. Error bars represent median with IQR in (A-C) (E-G), and mean ± SD in (D). Significance is denoted as: ns, not significant; *, P<0.05; **, P<0.01; ***P<0.001; ****P<0.0001. GC, gastric cancer; IQR, interquartile range; SD, standard deviation; TNM, tumor-node-metastasis.

Diagnostic value of tRF-28-P4R8YP9LOND5 in GC serum

Building on the strong correlation between tRF-28-P4R8YP9LOND5 expression and aggressive tumor traits, we evaluated its diagnostic performance as a standalone marker and in combination with conventional biomarkers. Using serum data from 117 GC patients, 89 healthy controls, and 51 gastritis patients, we constructed ROC curves to assess single-marker and multi-marker panel performance. Although we utilized the entire sample set for this analysis, we acknowledge that a more robust approach involving random division into training and validation sets or implementing cross-validation would enhance the reliability of our model. This is a limitation of our current study and will be addressed in future work with larger, multicenter datasets (Figure 5A-5F).

Figure 5 ROC curve analyses for single and combined biomarkers in GC. (A-C) GC versus healthy controls; (D-F) GC versus gastritis patients. Cohort: 117 GC patients, 89 healthy controls, and 51 gastritis patients. Methods: ROC curves were generated using MedCalc (version 20.0; MedCalc Software). AUC comparisons were performed using the DeLong test, with P<0.05 considered statistically significant. AUC, area under the curve; CA, carbohydrate antigen; CEA, carcinoembryonic antigen; GC, gastric cancer; ROC, receiver operating characteristic; tRFs, tRNA-derived fragments.

For the ROC curve analysis of GC versus healthy controls (Figure 5A-5C), single-marker performance (Figure 5A) revealed that tRF-28-P4R8YP9LOND5 exhibited a moderate AUC of 0.737 (55.6% sensitivity, 94.4% specificity; Table 3), indicating diagnostic utility for GC but potential underdiagnosis due to suboptimal sensitivity. CEA demonstrated a moderate AUC of 0.724 with 97.8% specificity but only 23.1% sensitivity, effectively excluding healthy individuals but prone to missing GC cases. In contrast, CA199 (AUC =0.596) and CA724 (AUC =0.628) showed near-random performance with minimal sensitivity, underscoring their lack of standalone diagnostic utility. For dual-marker combinations (Figure 5B), the tRF-28-P4R8YP9LOND5 + CEA panel achieved the highest AUC of 0.808, improving sensitivity to 70.1% while maintaining 91.0% specificity—significantly outperforming single markers—whereas combinations with CA199 (AUC =0.759) or CA724 (AUC =0.788) showed minimal or no improvement. In multi-marker panels (Figure 5C), tRF-28-P4R8YP9LOND5 + CEA + CA199 reached the peak AUC of 0.821 (76.9% sensitivity, 88.8% specificity); adding CA724 to the panel (tRF-28-P4R8YP9LOND5 + CEA + CA724) resulted in a higher AUC of 0.846 with balanced sensitivity and specificity, while tRF-28-P4R8YP9LOND5 + CA199 + CA724 performed poorly (AUC =0.802). The four-marker panel (all biomarkers) achieved an AUC of 0.86. The four-marker panel (tRF-28-P4R8YP9LOND5 + CEA + CA199 + CA724) matched the three-marker AUC of 0.821 but with decreased specificity, suggesting redundancy of CA724. Overall, CEA provided high specificity but low sensitivity, tRF-28-P4R8YP9LOND5 balanced moderate sensitivity with high specificity, and CA199/CA724 were ineffective alone; the tRF-28-P4R8YP9LOND5+CEA pair emerged as the optimal dual combination, while the three-marker tRF-28-P4R8YP9LOND5 + CEA + CA199 panel maximized diagnostic efficiency by balancing sensitivity and specificity, recommending its prioritization to reduce costs by excluding non-contributory markers like CA724. Clinically, standalone use of CA199/CA724 is not advised, and multi-marker panels with tRF-28-P4R8YP9LOND5 and CEA as core markers, supplemented by CA199, are preferred to optimize diagnostic accuracy.

For the ROC curve analysis of GC versus gastritis patients (Figure 5D-5F), single-marker analysis (Figure 5D) revealed that tRF-28-P4R8YP9LOND5 exhibited an AUC of 0.849 [95% confidence interval (CI): 0.782–0.916], with 72.6% sensitivity and 86.3% specificity for differentiating GC from gastritis (Table 4); while these findings indicate diagnostic utility, the relatively small gastritis cohort (n=51) may compromise specificity estimates, necessitating multicenter validation to stabilize results. CEA showed a lower AUC of 0.602 with 94.1% specificity but only 28.2% sensitivity, while CA199 (AUC =0.442) and CA724 (AUC =0.625) exhibited suboptimal performance, reinforcing their limited standalone value. In dual-marker combinations (Figure 5E), the tRF-28-P4R8YP9LOND5 + CA724 panel achieved the highest AUC of 0.872, improving sensitivity to 80.3% while maintaining 82.4%, surpassing tRF-28-P4R8YP9LOND5 + CEA (AUC =0.856) and tRF-28-P4R8YP9LOND5 + CA199 (AUC =0.859). For multi-marker panels (Figure 5F), three-marker tRF-28-P4R8YP9LOND5 + CEA + CA199 (AUC =0.865) and tRF-28-P4R8YP9LOND5 + CEA + CA724 (AUC =0.877) were outperformed by tRF-28-P4R8YP9LOND5 + CA724 + CA199 (AUC =0.881), with the full panel (all biomarkers) reaching 0.883, while the four-marker panel (all biomarkers) reached 0.883 with balanced sensitivity (91.5%) and specificity (70.6%). Overall, tRF-28-P4R8YP9LOND5 demonstrated superior single-marker efficacy, and the tRF-28-P4R8YP9LOND5 + CA724 dual panel emerged as the optimal balance for differentiating GC from gastritis, while multi-marker panels offered limited incremental benefit.

Prediction of downstream target genes of tRF-28-P4R8YP9LOND5

To decipher the biological mechanisms and signaling cascades governed by tRF-28-P4R8YP9LOND5, bioinformatics methodologies were applied to predict target genes and conduct functional enrichment analysis. Four computational tools—miRanda (2,678 predicted genes), PITA [1,725], TargetScan [2,637], and RNAhybrid [8,560]—were leveraged to identify potential target genes. A Venn diagram (Figure 6A) illustrated the inter-database overlap and unique sequences, uncovering a conserved set of 732 genes consistently predicted across all four algorithms. This cross-database consensus approach enhances prediction robustness by filtering for genes repeatedly identified, thereby refining the candidate list for subsequent experimental validation.

Figure 6 Downstream target gene prediction and functional pathway enrichment analysis of tRF-28-P4R8YP9LOND5. (A) Multi-database potential target gene prediction: four bioinformatics databases—miRanda [2,678], PITA [1,725], TargetScan [2,637], and RNAhybrid [8,560], were used to predict target genes of tRF-28-P4R8YP9LOND5. (A) Venn diagram illustrates the overlaps and unique sequences of prediction results across databases. (B) GO enrichment analysis of target genes: covers three major categories—biological processes, cellular components, and molecular functions. (C) KEGG pathway enrichment analysis of target genes: bar chart shows significantly enriched signaling pathways of target genes. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Functional annotation via GO revealed that the shared target genes were enriched across three primary categories, each offering distinct insights into tRF-28-P4R8YP9LOND5’s potential roles. In biological processes, terms such as “cardiac muscle cell proliferation” and “heart growth” emerged, though these likely reflect metabolic adaptation in GC cells or stromal cell activation within the tumor microenvironment rather than direct involvement in myocardial physiology; concurrently, “cell motility” and “cellular response to chemical stimulus” highlighted potential involvement in GC cell migration and invasion, critical processes in metastatic dissemination. For cellular components, enrichment of “TORC2 complex” (implicated in cell growth regulation) and “signal recognition particle” (involved in protein trafficking) suggested tRF-28-P4R8YP9LOND5 may modulate intracellular signaling assemblies or secretory pathways, while “lipoprotein particle” enrichment correlated with dysregulated lipid metabolism in GC cells, a hallmark of tumor nutrient acquisition. In molecular functions, “cytokine binding” and “inositol 1,4,5-trisphosphate binding” pointed to roles in intercellular communication and calcium signaling networks that could influence GC cell proliferation, whereas “serine hydrolase activity” enrichment was tied to extracellular matrix degradation and inflammatory signaling, mechanisms critical for tumor invasion (Figure 6B).

KEGG pathway mapping highlighted the broad oncogenic relevance of tRF-28-P4R8YP9LOND5, with target genes significantly enriched in “human diseases”, “immune system”, and “signal transduction” categories. Bar chart visualization (Figure 6C) emphasized robust enrichment in “pathways in cancer”, “pancreatic cancer”, and “bladder cancer” pathways, underscoring its particular relevance to digestive system malignancies. The “p53 signaling pathway” enrichment suggested direct modulation of tumor suppressor-mediated apoptosis and DNA repair, potentially contributing to uncontrolled GC cell proliferation, while activation of “NOD-like receptor signaling” and “Th17 cell differentiation” pathways implied indirect promotion of GC progression through modulation of innate immunity and intestinal inflammatory niches. Involvement in the “Hippo signaling pathway”—key to balancing cell proliferation and apoptosis—indicated tRF-28-P4R8YP9LOND5 may disrupt tumor suppressive mechanisms normally governing organ development, thereby influencing GC cell growth.

Collectively, integrative multi-database prediction and GO/KEGG analyses have illuminated the functional repertoire and signaling networks of tRF-28-P4R8YP9LOND5, providing mechanistic clues for understanding its role in GC pathogenesis and prioritizing candidates for downstream validation.

Discussion

Key findings

GC remains a profound global health burden, with survival rates stagnating due to the dominance of late-stage diagnosis (16,17). Traditional serum biomarkers such as CEA and CA199 are inadequate for clinical needs, underscoring the urgent imperative for innovative diagnostic strategies (18,19). tsRNAs, characterized by their tumor-specific expression patterns and stability in biofluids, offer a compelling avenue for noninvasive early detection (1-3,20). In this study, we conducted a comprehensive validation of tRF-28-P4R8YP9LOND5 as a novel serum biomarker, demonstrating its utility as both a standalone marker and in combination with conventional markers to enhance early diagnostic precision.

Using the OncotRF database, three candidate tsRNAs were initially evaluated, with only tRF-28-P4R8YP9LOND5 demonstrating significant overexpression in GC serum (n=117) and the highly invasive HGC-27 cell line (P<0.001). Originating from the D-loop and anticodon-loop cleavage of tRNA-Gly-GCC, this 17 nt 5’-iRF maps to chr1:161,443,304–161,443,331, suggesting a potential ANG-mediated specific cleavage mechanism (1,2). Methodologically, RT-qPCR assays exhibited exceptional precision (intra-assay/inter-assay CV =1.60%/0.75%), stability (Ct value fluctuations <0.5 after 24-hour room temperature storage or 10 freeze-thaw cycles), and a broad six-log dynamic range (R²>0.99), outperforming serum miRNAs in terms of reproducibility (21,22). These technical attributes solidify tRF-28-P4R8YP9LOND5’s candidacy for clinical application.

Clinicopathologic analyses revealed robust positive associations between tRF-28-P4R8YP9LOND5 expression and aggressive tumor characteristics, including deep tumor invasion (T3–T4 stages, P<0.001), advanced TNM staging (III–IV, P=0.02), and neurovascular invasion (P=0.002). Notably, postoperative tRF-28-P4R8YP9LOND5 levels rapidly normalized to healthy control ranges (P<0.001), closely aligning with tumor burden dynamics (8,18,23), which suggests its potential as a real-time marker of disease progression. For early-stage GC (T1–T2), the AUC of 0.793 implies that integrating tRF-28-P4R8YP9LOND5 with other early-specific tsRNAs could further optimize diagnostic performance (9,24,25). As a single marker, tRF-28-P4R8YP9LOND5 achieved an AUC of 0.737 (55.6% sensitivity, 94.4% specificity). Combinations with CEA/CA199/CA724 improved diagnostic performance, with the highest AUC reaching 0.883 (91.5% sensitivity) in differentiating GC from gastritis, and 0.821 (76.9% sensitivity) in distinguishing GC from healthy controls. This synergistic effect likely arises from the complementary roles of tsRNAs and conventional markers: while tsRNAs like tRF-28-P4R8YP9LOND5 regulate post-transcriptional processes such as mRNA stability and translation (1,2), CEA and CA199 reflect tumor-associated antigen expression. By integrating these orthogonal molecular signatures, the panel captures both transcriptional dysregulation and antigenic alterations, surpassing the diagnostic accuracy of single-modality approaches (19,23,26). In differentiating GC from gastritis, a four-marker panel achieved high sensitivity (91.5%) but with compromised specificity (70.6%), emphasizing the importance of context-dependent marker selection in clinical practice (8,23).

Strengths and limitations

This study is not without limitations. First, the single-center design and limited number of early-stage GC samples (n=47) may compromise the generalizability of our findings. Moreover, the failure to isolate exosomal tsRNA could have introduced non-tumor-derived signals, as serum tsRNAs may originate from various cellular sources, including stromal, immune, and blood cells. Exosomal tsRNAs, which are actively secreted by cancer cells and exhibit higher tumor specificity (27,28), were not fractionated in this study, potentially leading to an underestimation of diagnostic specificity (11,15,17,28). Future studies employing ultracentrifugation or commercial exosome isolation kits will be essential to address this methodological gap. Second, functional validation of tRF-28-P4R8YP9LOND5’s mechanism, such as clustered regularly interspaced short palindromic repeats-Cas9 (CRISPR-Cas9) knockout experiments, RNA immunoprecipitation (RIP), and m6A methylation analysis, was not performed in this work (6,7,10,12). While our study presents strong evidence for the diagnostic potential of tRF-28-P4R8YP9LOND5, further experimental validation, such as dual-luciferase reporter assays, is needed to confirm its direct interaction with the predicted target genes. Our current findings are based on bioinformatics predictions and clinical data, and we recognize that functional validation will be essential for fully understanding the biological mechanisms underlying tRF-28-P4R8YP9LOND5’s role in GC. CA724 demonstrated variable utility across cohorts, with notable contributions in differentiating GC from gastritis (e.g., tRF + CA724 + CA199, AUC =0.881), though its role in GC vs. healthy controls remained limited. Larger datasets are needed to validate its consistency (3,23,26,29). The small gastritis cohort (n=51) also introduced variability in specificity estimates, necessitating validation in larger, multicenter cohorts.

Comparison with similar research

When contextualized within the broader landscape of tsRNA research, while tRF-28-P4R8YP9LOND5 exhibited a moderate single-marker AUC of 0.737, its combination with conventional markers (AUC =0.821–0.883) approached the diagnostic accuracy of tRF-29 (AUC =0.889) while offering broader serum detectability across tumor stages (9). This discrepancy may be attributed to variations in sample composition, such as the proportion of early- vs. advanced-stage tumors, or differences in tsRNA tissue specificity. Notably, tRF-29 is predominantly expressed in tumor tissues, whereas tRF-28-P4R8YP9LOND5 exhibits broader serum detectability, which may explain its higher sensitivity in mixed-stage cohorts. In GC vs. gastritis patients, the tRF-28-P4R8YP9LOND5 + CA724 dual-marker panel achieved an AUC of 0.872, surpassing tRF-28-P4R8YP9LOND5 + CEA (AUC =0.856) and highlighting CA724’s potential to improve specificity in differentiating GC from benign inflammation. Notably, the tRF-28-P4R8YP9LOND5, CEA, CA199, and CA724 multi-marker panel achieved an AUC of 0.883 in differentiating GC from gastritis, surpassing single tsRNA markers like tRF-29 (AUC =0.889) and demonstrating synergistic value through complementary molecular mechanisms.

Explanations of findings

The clinical relevance of tRF-28-P4R8YP9LOND5 is likely rooted in its modulation of key signaling pathways. Bioinformatics-driven target gene predictions identified enrichment in the p53 signaling pathway, Hippo pathway, and cell migration-associated genes, such as claudin-1 (30). Claudin-1, a tight junction protein overexpressed in GC and strongly linked to epithelial-mesenchymal transition (EMT) and lymph node metastasis (22), may be post-transcriptionally regulated by tRF-28-P4R8YP9LOND5 to disrupt cell-cell adhesion, thereby facilitating tumor invasion—a mechanism previously documented in colorectal cancer (22). Additionally, enrichment in NOD-like receptor signaling and Th17 cell differentiation pathways suggests a potential crosstalk with Helicobacter pylori infection to promote tumor microenvironment inflammation (14,27,31). Metabolic gene enrichment, particularly for ACADSB—a key enzyme in fatty acid β-oxidation—implicates tRF-28-P4R8YP9LOND5 in tumor metabolic reprogramming. ACADSB promotes GC cell survival under nutrient stress by reallocating lipid metabolism to energy production (32), a mechanism consistent with a recent study demonstrating that tRF-23-Q99P9P9NDD enhances GC progression through ACADSB targeting (32). These convergent findings suggest conserved metabolic regulatory mechanisms for tsRNAs in gastric carcinogenesis.

Implications and actions needed

Looking ahead, three strategic research directions are proposed. First, mechanistic investigations using CRISPR-Cas9, luciferase reporter assays, and exosomal tsRNA sequencing will help elucidate tRF-28-P4R8YP9LOND5’s regulatory networks and distinguish tumor-derived signals from non-tumor components (6,10,12,30). Second, multi-center clinical validation with cohorts exceeding 1,000 participants will be critical to validate diagnostic models, develop portable POCT devices based on recombinase polymerase amplification (RPA), and integrate artificial intelligence (AI) algorithms for intelligent diagnostic systems (9,26,33-35). Lastly, the development of antisense oligonucleotides targeting tRF-28-P4R8YP9LOND5 could foster the creation of “theranostic” strategies that integrate diagnosis and treatment (6,36).

In conclusion, tRF-28-P4R8YP9LOND5 serves as a valuable serum biomarker for GC, with multi-marker panels (tRF + CEA + CA199 + CA724) providing optimized diagnostic accuracy (AUC =0.821–0.883) for noninvasive screening, particularly in differentiating GC from gastritis and monitoring tumor burden. Despite its limitations, the marker’s stability, tumor specificity, and involvement in multiple oncogenic pathways provide a strong foundation for tsRNA-based precision medicine. While awaiting multicenter validation, these findings highlight the transformative potential of tsRNAs in GC management, bringing the field closer to implementing personalized oncology approaches (1,3,33).

Conclusions

This study represents the first systematic validation of tRF-28-P4R8YP9LOND5 as a serum biomarker for GC, demonstrating that its integration with conventional markers significantly enhances diagnostic accuracy and establishes a cost-effective and highly compatible novel testing strategy for clinical practice. Specifically, the multimodal panel incorporating tRF-28-P4R8YP9LOND5, CEA, and CA199 offers a viable approach to implement noninvasive GC screening in high-incidence regions (e.g., East Asia, Eastern Europe), where current early detection rates remain suboptimal. This strategy capitalizes on tRF-28-P4R8YP9LOND5’s tumor-specific overexpression in serum and the established clinical utility of conventional markers, thereby addressing the unmet need for accurate, accessible diagnostic tools in resource-constrained settings. While mechanistic studies and translational applications require further exploration, the tumor-specific expression pattern of tRF-28-P4R8YP9LOND5 and its enrichment in key oncogenic pathways highlight the potential of tsRNAs as pivotal molecules for precision diagnosis and treatment of GC. Looking forward, advancements in single-cell sequencing and portable detection technologies (3,35) position tsRNAs to drive a paradigm shift in GC screening—evolving from “opportunistic testing” to “precision-targeted screening”—thereby improving early detection rates and patient outcomes on a global scale.

Acknowledgments

We would like to express our sincere gratitude to all participants who contributed serum samples to this study, as well as the medical staff at the Affiliated Hospital of Nantong University and Affiliated Dongtai Hospital of Nantong University for their assistance in sample collection and clinical data management.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1773/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1773/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1773/prf

Funding: This study was supported by the Yancheng Science and Technology Bureau project (No. YCBE202238) and the Yancheng Health Commission project (No. YK2021084), the Project of Jiangsu Provincial Health Commission (No. H2019068), and the Jiangsu Pharmaceutical Vocational College University-Local Collaborative Innovation Research Project (No. 202590508).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1773/coif). All authors report that this study was supported by Yancheng Science and Technology Bureau project (No. YCBE202238) and the Yancheng Health Commission project (No. YK2021084), the Project of Jiangsu Provincial Health Commission (No. H2019068), and the Jiangsu Pharmaceutical Vocational College University-Local Collaborative Innovation Research Project (No. 202590508). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Affiliated Hospital of Nantong University (Ethics No. 2023-K167-01) and informed consent was obtained from all individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Shen Y, Yu X, Zhu L, et al. Transfer RNA-derived fragments and tRNA halves: biogenesis, biological functions and their roles in diseases. J Mol Med (Berl) 2018;96:1167-76. [Crossref] [PubMed]
2. Liu B, Cao J, Wang X, et al. Deciphering the tRNA-derived small RNAs: origin, development, and future. Cell Death Dis 2021;13:24. [Crossref] [PubMed]
3. Sun C, Fu Z, Wang S, et al. Roles of tRNA-derived fragments in human cancers. Cancer Lett 2018;414:16-25. [Crossref] [PubMed]
4. Geng J, Sun Z, Li H. The Roles of tRNA-Derived Fragments in Cancer: Updates and Perspectives. Int J Mol Sci 2025;26:5822. [Crossref] [PubMed]
5. Li B, Li H, Li Y, et al. Genetic Control of tRNA-Derived Fragments Contributes to Cancer Risk. Cancer Res 2025;85:3855-74. [Crossref] [PubMed]
6. Cui H, Li H, Wu H, et al. A novel 3’tRNA-derived fragment tRF-Val promotes proliferation and inhibits apoptosis by targeting EEF1A1 in gastric cancer. Cell Death Dis 2022;13:471. [Crossref] [PubMed]
7. Cui H, Liu Z, Peng L, et al. A novel 5’tRNA-derived fragment tRF-Tyr inhibits tumor progression by targeting hnRNPD in gastric cancer. Cell Commun Signal 2025;23:88. [Crossref] [PubMed]
8. Mao C, Zhang Z, Fang R, et al. A novel tRNA-derived fragment tRF-17-18VBY9M works as a potential diagnostic biomarker for gastric cancer. J Cancer Res Clin Oncol 2024;150:263. [Crossref] [PubMed]
9. Li X, Zhang Y, Li Y, et al. A comprehensive evaluation of serum tRF-29-R9J8909NF5JP as a novel diagnostic and prognostic biomarker for gastric cancer. Mol Carcinog 2023;62:1504-17. [Crossref] [PubMed]
10. Liang Y, Ji D, Ying X, et al. tsRNA modifications: An emerging layer of biological regulation in disease. J Adv Res 2025;74:403-14. [Crossref] [PubMed]
11. Yang P, Li Z, Chen X, et al. Non-canonical small noncoding RNAs in the plasma extracellular vesicles as novel biomarkers in gastric cancer. J Hematol Oncol 2025;18:39. [Crossref] [PubMed]
12. Wu Z, Zhou R, Li B, et al. Methylation modifications in tRNA and associated disorders: Current research and potential therapeutic targets. Cell Prolif 2024;57:e13692. [Crossref] [PubMed]
13. Saad AAA, Zhang K, Deng Q, et al. The functions and modifications of tRNA-derived small RNAs in cancer biology. Cancer Metastasis Rev 2025;44:38. [Crossref] [PubMed]
14. Sousa C, Ferreira R, Azevedo NF, et al. Helicobacter pylori infection: from standard to alternative treatment strategies. Crit Rev Microbiol 2022;48:376-96. [Crossref] [PubMed]
15. Allemani C, Matsuda T, Di Carlo V, et al. Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37 513 025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet 2018;391:1023-75. [Crossref] [PubMed]
16. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
17. Ilic M, Ilic I. Epidemiology of stomach cancer. World J Gastroenterol 2022;28:1187-203. [Crossref] [PubMed]
18. Zhang S, Li JZ, Du T, et al. The Modified Glasgow Prognostic Score Predicts Survival in Gastric Cancer Patients with Normal CEA and CA19-9. Can J Gastroenterol Hepatol 2022;2022:3953004. [Crossref] [PubMed]
19. Tong W, Ye F, He L, et al. Serum biomarker panels for diagnosis of gastric cancer. Onco Targets Ther 2016;9:2455-63. [Crossref] [PubMed]
20. Zhao Y, Li X, Ye C, et al. The biogenesis, mechanism and function of the tRNA-derived small RNA (tsRNA): a review compared with microRNA. Am J Cancer Res 2023;13:1656-66.
21. Jia Y, Tan W, Zhou Y. Transfer RNA-derived small RNAs: potential applications as novel biomarkers for disease diagnosis and prognosis. Ann Transl Med 2020;8:1092. [Crossref] [PubMed]
22. Ling H, Fabbri M, Calin GA. MicroRNAs and other non-coding RNAs as targets for anticancer drug development. Nat Rev Drug Discov 2013;12:847-65. [Crossref] [PubMed]
23. Xie Y, Zhang S, Yu X, et al. Transfer RNA-derived fragments as novel biomarkers of the onset and progression of gastric cancer. Exp Biol Med (Maywood) 2023;248:1095-102. [Crossref] [PubMed]
24. Liu X, Zhang G, Zhao L. Detection of transmembrane protein 100 in breast cancer: Correlation with malignant progression and chemosensitivity. Cytojournal 2024;21:65. [Crossref] [PubMed]
25. Murugan R, Manivannan P, Gochhait D, et al. Utility of fine-needle aspiration cytology combined with flow cytometry in extramedullary hematolymphoid lesions - A cross-sectional study. Cytojournal 2024;21:79. [Crossref] [PubMed]
26. Gao H, Zhang Q, Wu W, et al. The diagnostic and prognostic value of tsRNAs in gastric cancers: a systematic review and meta-analysis. Expert Rev Mol Diagn 2023;23:985-97. [Crossref] [PubMed]
27. Jia H, Zhang L. tRNA-derived small RNAs in disease immunity. Theranostics 2025;15:245-57. [Crossref] [PubMed]
28. Zhang S, Gu Y, Ge J, et al. tRF-33-P4R8YP9LON4VDP inhibits gastric cancer progression via modulating STAT3 signaling pathway in an AGO2-dependent manner. Oncogene 2024;43:2160-71. [Crossref] [PubMed]
29. Wang Y, Weng Q, Ge J, et al. tRNA-derived small RNAs: Mechanisms and potential roles in cancers. Genes Dis 2022;9:1431-42. [Crossref] [PubMed]
30. Luan N, Chen Y, Li Q, et al. TRF-20-M0NK5Y93 suppresses the metastasis of colon cancer cells by impairing the epithelial-to-mesenchymal transition through targeting Claudin-1. Am J Transl Res 2021;13:124-42.
31. Wang XY, Zhou YJ, Chen HY, et al. 5’tiRNA-Pro-TGG, a novel tRNA halve, promotes oncogenesis in sessile serrated lesions and serrated pathway of colorectal cancer. World J Gastrointest Oncol 2023;15:1005-18. [Crossref] [PubMed]
32. Zhang Y, Gu X, Li Y, et al. Transfer RNA-derived fragment tRF-23-Q99P9P9NDD promotes progression of gastric cancer by targeting ACADSB. J Zhejiang Univ Sci B 2024;25:438-50. [Crossref] [PubMed]
33. Zong T, Yang Y, Zhao H, et al. tsRNAs: Novel small molecules from cell function and regulatory mechanism to therapeutic targets. Cell Prolif 2021;54:e12977. [Crossref] [PubMed]
34. Yuan J, Gu W, Xu T, et al. Dysregulated transfer RNA-derived small RNAs as potential gastric cancer biomarkers. Exp Biol Med (Maywood) 2024;249:10170. [Crossref] [PubMed]
35. Kim HK, Yeom JH, Kay MA. Transfer RNA-Derived Small RNAs: Another Layer of Gene Regulation and Novel Targets for Disease Therapeutics. Mol Ther 2020;28:2340-57. [Crossref] [PubMed]
36. Slack FJ, Chinnaiyan AM. The Role of Non-coding RNAs in Oncology. Cell 2019;179:1033-55. [Crossref] [PubMed]