Lysine crotonylation-related long non-coding RNAs: a novel prognostic framework for gastric carcinoma

Introduction

Ranked fifth in occurrence and fourth in lethality among all cancers, gastric carcinoma (GC) is a critical issue for global health monitoring (1-3). While we have come a long way with improved diagnostic methods and treatment approaches ranging from surgical interventions to chemotherapy and novel immunotherapies, patients with advanced-stage gastric cancer continue to face an uphill battle with disappointing clinical outcomes (4). Metastatic cases, in particular, boast alarmingly low 5-year survival rates. This grim picture is largely due to the molecular diversity of gastric cancer, which causes inconsistent responses to conventional treatments (5). Despite programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) blockade inhibitors revealing novel possibilities, landmark studies such as CheckMate-649 reveal that only a fraction of patients experience lasting benefits, underscoring the intricate nature of the tumor immune ecosystem (6,7). Consequently, there is a pressing demand to uncover innovative prognostic biomarkers and therapeutic targets that can enhance risk assessment and establish the groundwork for customized therapeutic approaches in General Clinical settings.

Spanning over 200 nucleotides without coding for proteins, long non-coding RNAs (lncRNAs) have become a hotspot in oncological studies (8). These molecular players wear many hats, controlling mechanisms across multiple stages, encompassing epigenetic, transcriptional, and post-transcriptional, and getting their hands dirty in crucial cellular activities like cell division, programmed cell death, cancer spread, and metabolic makeovers (9-11). The writing’s on the wall: when these lncRNAs go haywire, they can either fuel cancer growth or slam the brakes on it, depending on the cancer type, with gastric cancer being no exception (12). In addition, cutting-edge research has pulled back the curtain on how lncRNAs are key players in reshaping the TME and helping cancers evade the immune system—qualities that make them promising candidates for predicting outcomes and opening new treatment avenues (13,14).

Evolutionarily conserved across species, lysine crotonylation (Kcr) is defined by the conjugation of a crotonyl group to lysine residues found in both nuclear histones and non-histone proteins (15). Unlike its cousin acetylation, crotonylation brings a hydrophobic and structurally rigid element to the party, which can fundamentally transform chromatin architecture and give active gene transcription a significant boost (16). Fresh differentiation and each have put Kcr front and center in regulating cellular metabolism, steering stem cell differentiation, and managing stress responses (17). In addition, when crotonylation goes off the rails, it has been fingered as playing a part in oncogenesis, tweaking the expression of major oncogenic drivers through a metabolic-epigenetic tag team (18,19). That said, the exact dance between Kcr and lncRNAs within the GC arena remains something of a black box. Whether lncRNAs linked to crotonylation can serve as prognostic yardsticks or help shape the immune scenery in GC is still up in the air.

Through our current investigation, we have identified a collection of lncRNAs associated with crotonylation and developed a reliable predictive model for GC patients. Beyond simply forecasting survival outcomes, we thoroughly examined how this risk score correlates with the immune tumor landscape, mutation load, and responsiveness to treatments. Our findings introduce an innovative categorization framework that could refine prognostic evaluations and guide immunotherapy approaches for GC. Additionally, we highlighted UBOX5-AS1 as a standout lncRNA among our findings and confirmed its role in GC cellular models. We present this article in accordance with the MDAR and TRIPOD reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2752/rc).

Methods

Data acquisition and preprocessing

To gather RNA sequencing expression data, specifically fragments per kilobase of transcript per million mapped reads (FPKM) values were normalized via log₂ (x + 1) transformation, and pertinent clinical details, we delved into The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/). For our analysis, we excluded patients who were missing key clinical details or had follow-up periods of less than 30 days—without exception. Using standard annotation tools, we converted all Ensembl IDs to their official gene symbols. Additionally, we compiled a carefully curated list of genes involved in crotonylation regulation by referencing existing scientific literature and reputable public databases (20). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Screening of crotonylation-related lncRNAs

To identify lncRNAs involved in Kcr, we performed a comprehensive co-expression analysis. We investigated connections between known crotonylation gene expressions and the complete lncRNA expression profile through the Pearson correlation coefficient. LncRNAs were classified as KcrlncRNAs if they showed a significant correlation with crotonylation regulators (|R| >0.4 and P<0.001). These selected KcrlncRNAs were subsequently used to develop the predictive model.

Development and validation of the prognostic risk model

We divided the complete TCGA-stomach adenocarcinoma (STAD) cohort—comprising 407 gastric cancer cases—into training and testing datasets with a 50:50 split, leveraging the “caret” R package to randomize the allocation. Prognostic candidates were first identified via univariate Cox analysis (P<0.05). The ‘glmnet’ package was then utilized to execute least absolute shrinkage and selection operator (LASSO) regression, thereby preventing overfitting and narrowing down the feature set. Subsequent multivariate Cox regression analysis identified the final candidates for the prognostic model. We then computed a risk score for every patient by linearly combining the expression values of these specific KcrlncRNAs weighted by their regression coefficients, each multiplied by its respective coefficient from the multivariate model:

Riskscore=∑i=1n(Coefficienti×Expressioni)[1]

where Coefficient_i signifies the regression coefficient of the i-th lncRNA, and Expression_i represents its normalized expression level. Drawing the line at the median risk score, participants were divided into high-risk and low-risk categories. The prognostic value of this signature was confirmed across both the training and validation datasets through Kaplan-Meier survival plots and time-dependent receiver operating characteristic (ROC) evaluations to gauge its forecasting capabilities.

Assessment of independent prognostic value

To determine whether our identified KcrlncRNA signature could independently predict patient outcomes, separate from standard clinical factors such as age, sex, tumor grade, and tumor-node-metastasis (TNM) stage, we applied Cox regression modeling with univariate and multivariate approaches. We set our threshold for statistical significance at P<0.05 identifying which variables were truly independent predictors of patient prognosis.

Construction of a predictive nomogram

We developed an extensive nomogram merging the KcrlncRNA risk assessment with critical clinical determinants affecting patient results, using the “rms” and “regplot” R packages as foundational tools. The innovative visual aid was crafted to offer medical professionals a user-friendly approach to predict survival rates over the next one, three, and five years for those afflicted with GC. To rigorously evaluate our nomogram, we employed calibration curves to assess how well our predictions aligned with actual survival data. Moreover, we computed the concordance statistics to assess the precision of our predictive tool for patient results quantitatively.

Functional enrichment analysis

In order to shed light on the prognostic signature’s rich biological processes and molecular pathways, we initiated a differential gene expression analysis. This was designed to pinpoint genes that were dramatically altered versus the low-risk cohort—a cutoff of a two-fold change in absolute log₂ values and a statistically significant P value below 0.05 was applied. The “clusterProfiler” package then enabled us to annotate Gene Ontology (GO) terms and perform Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses on the genes that showed significant differences. Taking it a step further, we applied Gene Set Variation Analysis (GSVA) to gauge the levels and enrichment of particular biological pathways among the different risk groups. This analysis provided us with a deeper understanding of the fundamental functional changes occurring within these groups.

Tumor microenvironment (TME) profiling

We thoroughly examined the TME environment by leveraging a suite of sophisticated bioinformatics tools. To assess the cellular composition, we used CIBERSORT, allowing measurement of 22 unique immune cell types present in tumor tissues among our samples. Simultaneously, we applied the ESTIMATE algorithm to generate scores for stromal and immune components, providing a clearer picture of sample purity and the extent of non-cancerous cell infiltration. Additionally, we employed the tumor immune dysfunction and exclusion (TIDE) framework to predict how GC patients might respond to immune checkpoint blockade therapy, noting that lower TIDE scores are generally associated with better immunotherapy outcomes.

Genomic mutation and drug sensitivity analysis

We delved deep into the somatic mutation profiles of GC patients, leveraging the “maftools” R package to meticulously analyze and visualize these genomic datasets. This approach enabled us to characterize the genomic features of cancer while calculating the tumor mutation burden (TMB) for each patient. We also examined how the KcrlncRNA risk score might correlate with TMB levels. To shed light on potential treatment options, we utilized the “oncoPredict” package to estimate half-maximal drug inhibitory concentration (IC₅₀) values for a broad spectrum of chemotherapy drugs. Finally, we performed Wilcoxon signed-rank tests to compare these differential IC₅₀ values across risk stratification cohorts, helping us identify which drugs might work differently for each patient subgroup.

Cell lines and transfection

We employed diverse human GC cell lines in our investigation—namely AGS, BGC-823, MGC-803, and SGC-7901—along with viable gastric epithelial cells from the GES-1 line, all obtained from trusted biological collections like the Cell Bank of the Chinese Academy of Sciences. Cells were grown in RPMI-1640 or DMEM enriched with 10% fetal bovine serum (FBS) and 1% antibiotics. The incubation conditions were strictly maintained at 37 °C with 5% CO₂ under humidified conditions. To modify gene expression, we designed small interfering RNAs (siRNAs) targeting the lncRNA UBOX5-AS1, along with corresponding control siRNAs. The UBOX5-AS1 sequence we targeted was TGGCCATTTGGTTCATCTACTTC. Transfection assays were executed employing Lipofectamine 3000 (Invitrogen) in strict adherence to product guidelines.

RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR)

Using TRIzol reagent, we isolated total RNA from the GC cell culture with a high-quality RNA isolation kit. RNA concentration and purity were assessed via a NanoDrop 2000c (Thermo Scientific). Subsequently, reverse transcription was performed to synthesize cDNA using the PrimeScript RT Reagent Kit (TaKaRa), followed by qRT-PCR utilizing TaKaRa’s SYBR Green Master Mix on a real-time detection system. The levels of the specified lncRNAs were adjusted against the housekeeping gene GAPDH and quantified through the 2^−ΔΔCT analysis technique. To guarantee the specificity and efficacy of the primers, we meticulously designed and validated them for the lncRNAs. The forward primer for UBOX5-AS1 is 5'-GAAAACAGGCCAGGGTTAG-3', and its reverse primer is 5'-GGACTCGGGAGGGATGAAG-3'. For GAPDH, the forward primer is 5'-TTAAAAGCAGCCCTGGTGAC-3', and the reverse primer is 5'-CTCTGCTCCTCCTGTTCGAC-3'.

Cell proliferation and migration assays

In order to investigate cell proliferation in gastric cancer, we introduced alterations in the lncRNA expression of the cells and placed them in 96-well plates. Next, we took cell viability readings across a variety of time points, including 0, 24, 48, 72, and 96 hours, with either the Cell Counting Kit-8 (CCK-8) kit or colony assays to get the results. To investigate cell migration, the wound healing assay was employed. Initially, cells were plated into 6-well dishes to establish a dense monolayer. Using a sterile pipette tip, we made a clean scratch across the cell layer. We photographed the cells at both 0 and 48 hours after the scratch to measure the distance they had migrated. Moreover, we cultured cells in the upper chambers of Transwell inserts without Matrigel coating, with serum-free medium in the lower chambers and complete medium containing FBS in the upper chambers. After 24 or 48 hours, we removed the non-migrated cells, fixed with 4% paraformaldehyde for 15 minutes at room temperature, and subsequently stained with 0.1% crystal violet solution for 20 minutes and counted the cells that had traversed the membrane using a microscope.

Statistical analysis

R (version 4.2.1) and GraphPad Prism (version 9.0) served as platforms for statistical evaluation. Depending on normality, continuous data were compared using either Student’s t-test [reported as mean ± standard deviation (SD)] or the Wilcoxon rank-sum test [reported as median and interquartile range (IQR)]. Chi-squared tests were utilized to analyze categorical variables. Survival analysis involved Kaplan-Meier plotting and log-rank testing. Furthermore, univariate and multivariate Cox regression helped isolate independent survival predictors. Statistical significance was set at a threshold of P<0.05.

Results

Identification of crotonylation-associated lncRNAs

We recruited 407 GC cases from the TCGA-STAD database, all of whom possessed both RNA sequencing data and complete clinical records. This robust dataset served as the foundation for our investigative work. Previous research in the field had already pinpointed 18 genes linked to Kcr, including notable examples such as HDAC1, KAT8, and SIRT1. To unravel the potential relationship between crotonylation processes and lncRNAs, we dove deep into the relationship between crotonylation-associated genes and all documented lncRNAs in our GC patient set by running a Pearson correlation analysis to compare their expression profiles. By applying strict statistical thresholds (|R| >0.4 and P<0.001), we successfully identified 811 Kcr-associated lncRNAs that subsequently fueled the development of our prognostic modeling framework [Figure 1A and supplementary table 1 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2752-1.xlsx)].

Figure 1 Construction of the risk model based on crotonylation-related lncRNAs. (A) Sankey diagram of crotonylation genes and Kcr-related lncRNAs. (B) Forest plot of the 28 Kcr-lncRNAs identified through univariate Cox regression analysis. (C,D) LASSO regression analysis identifying 22 Kcr-lncRNAs. (E) Correlation heatmap displaying the expression correlations between the nine Kcr-lncRNAs included in the multivariate Cox regression model and crotonylation genes. *, P<0.05; **, P<0.01; ***, P<0.001. CI, confidence interval; HR, hazard ratio; Kcr, lysine crotonylation; LASSO, least absolute shrinkage and selection operator; lncRNA, long non-coding RNAs.

Construction of a prognostic KcrlncRNA signature in GC

To create a robust predictive model, we divided the 407 cases of GC patients into a training group (n=204) and a test group (n=203). Initially, in the training group, we performed univariate Cox regression analysis on the 811 detected KcrlncRNAs and identified 28 that substantially affected OS at the 0.05 level (Figure 1B). To refine our selection and avoid overfitting, we subjected these 28 lncRNAs to LASSO regression analysis (Figure 1C,1D), which further narrowed the list of potential lncRNAs. We then subjected the remaining 21 lncRNAs to multivariate Cox regression, resulting in a distinctive signature of nine KcrlncRNAs. Each of these nine lncRNAs—AC108693.2, STARD4-AS1, AL353804.2, UBOX5-AS1, DIP2A-IT1, AL359091.3, AC026412.3, CHROMR, and AP000873.4—had a unique coefficient and together formed our prognostic risk profile. The multivariate Cox model formula calculated individual patient risk score = AC108693.2 × −0.943975946656163 + STARD4-AS1 × 1.02431295617101 + AL353804.2 × −0.538249561448509 + UBOX5-AS1 × −1.44174891044707 + DIP2A-IT1 × 1.86486497971437 + AL359091.3 × −0.601688740427653 + AC026412.3 × −1.04668632163197 + CHROMR × 0.609534182332861 + AP000873.4× −1.07022605421171. Finally, we produced a correlation heatmap to illustrate the connections between 18 mitochondrial permeability transition (MPT)-related genes and 9 KcrlncRNAs (Figure 1E).

Survival analysis

Patients in the study were stratified into high- and low-risk groups based on the median value of the risk scores. When utilizing Kaplan-Meier analysis to evaluate survival results, it became evident that individuals falling into the high-risk category consistently demonstrated notably reduced overall survival times relative to their low-risk counterparts throughout the entire population studied (Figure 2A, P<0.001), training set (Figure 2B, P<0.001), and the testing set (Figure 2C, P=0.01). Mirroring these findings, progression-free survival (PFS) evaluations within the training set similarly indicated less favorable results for high-risk individuals (Figure 2D). Figure 2E-2G presents a comprehensive visualization of the prognostic model across different cohorts, illustrating the risk score distribution, survival outcomes, and predictive lncRNA expression profiles. As the risk scores increased, the distribution of patients shifted from the low-risk group (blue dots) to the high-risk group (red dots), accompanied by a markedly higher density of death events (red dots in the scatter plots) and shorter survival times, validating the model’s efficacy in identifying negative results. The heatmap analysis further delineated distinct expression profiles associated with risk stratification. Notably, STARD4-AS1, UBOK5-AS1, and DIP2A-IT1 were upregulated in the high-risk group, suggesting that their elevated expression may contribute to disease progression and poorer prognosis. Conversely, AC108693.2, AL353804.2, AL359091.3, AC026412.3, CHROMR, and AP000873.4 exhibited higher expression levels in the low-risk group, indicating potential protective roles in patient survival.

Figure 2 Survival analysis and validation. Kaplan-Meier curves for overall survival in the (A) overall set, (B) training set, and (C) test set. (D) Kaplan-Meier curves for progression-free survival in the overall set. Risk score plot, patient survival status, and expression heatmap of the nine lncRNAs in the (E) overall set, (F) training set, (G) test set. lncRNA, long non-coding RNAs.

Independent prognostic value

The prognostic independence of the nine-KcrlncRNA signature was assessed via univariate and multivariate Cox regression, incorporating clinical parameters like age, sex, tumor stage, and grade. According to the univariate Cox regression analysis, multiple clinicopathological characteristics showed statistically significant associations with patient prognosis. In detail, the hazard ratios (HR) derived from the univariate evaluation were 1.024 for age [95% confidence interval (CI): 1.008–1.041, P=0.004], 1.591 for stage (95% CI: 1.295–1.956, P<0.001), and 1.055 for the risk score (95% CI: 1.030–1.080, P<0.001), which are presented in Figure 3A. Following this, a multivariate approach demonstrated that age (HR =1.035, 95% CI: 1.017–1.052, P<0.001), clinical stage (HR =1.696, 95% CI: 1.362–2.113, P<0.001), and the risk model (HR =1.052, 95% CI: 1.027–1.078, P<0.001) all maintained their roles as independent prognostic factors (Figure 3B). These results reinforce the clinical relevance of our risk assessment model, demonstrating its ability to predict patient outcomes independently of conventional parameters. The established model demonstrates significant clinical utility for identifying high-risk patients who may benefit from intensive therapeutic interventions. As depicted in the ROC analysis (Figure 3C), the risk scoring system exhibited exceptional predictive accuracy with an area under the curve (AUC) of 0.720. This performance distinctly surpassed that of traditional clinical metrics, where tumor stage was the best predictor (AUC =0.613), followed by age (AUC =0.578), grade (AUC =0.556), and gender (AUC =0.514). Furthermore, time-dependent AUC evaluations yielded values of 0.720, 0.723, and 0.738 for 1-, 3-, and 5-year survival, respectively, confirming the model’s robust stability (Figure 3D). Finally, the concordance index (C-index) analysis (Figure 3E) corroborated the superiority of the risk model, which persistently outperformed standard clinical factors throughout the entire follow-up period. When dividing patients into groups based on the cancer’s stage—ranging from stage I to II, and stage III to IV—the risk assessment tool proved to be a powerful predictor of how long they’d live in gastric cancer populations. It notably marked the difference in the outlook for both the early and later stages (P<0.001, Figure 3F) and advanced stages (P<0.001, Figure 3G), solidifying its universality in risk stratification.

Figure 3 Validation of the independence of the constructed model. (A) Forest plot for univariate Cox regression analysis. (B) Forest plot for multivariate Cox regression analysis. (C) Chi-ROC curves for the risk score and other clinical risk factors. (D) Time-dependent ROC curves showing the predictive accuracy for 1-, 3-, and 5-year overall survival. (E) C-indexes for the risk score and other clinical risk factors. (F) Kaplan-Meier analysis of OS in patients with stages I–II disease. (G) Kaplan-Meier analysis of OS in patients with stages III–IV disease. AUC, area under the curve; C-index, concordance index; CI, confidence interval; HR, hazard ratio; OS, overall survival; ROC, receiver operating characteristic.

Principal component analysis (PCA) and establishment of the nomogram

PCA visualization revealed that while global gene expression and crotonylation-related sets resulted in intermixed risk profiles (Figure 4A-4C), our prognostic lncRNA signature successfully segregated patients into discrete clusters (Figure 4D). For clinical utility, a predictive nomogram was generated, combining the risk score with standard clinicopathological factors (Figure 4E). The summation of points in this system correlates with survival likelihood. The calibration curves (Figure 4F) confirmed that the nomogram’s predictions for 1-, 3-, and 5-year survival closely matched observed data, demonstrating strong predictive performance.

Figure 4 PCA and establishment of the nomogram. PCA plots (A) all genes, (B) Kcr-related genes, (C) all Kcr-related lncRNAs, (D) model-related lncRNAs. (E) Clinical nomogram. (F) Calibration curves for the nomogram. Kcr, lysine crotonylation; lncRNA, long non-coding RNAs; M, metastasis; N, node; OS, overall survival; PC, principal component; PCA, principal component analysis; T, tumor.

Functional enrichment analysis

Differentially expressed genes (DEGs) were screened using the ‘limma’ package based on the risk model and subsequently subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. The GO analysis (Figure 5A,5B) highlighted enrichment in several key areas: biological processes included extracellular structure/matrix organization and Wnt signaling; cellular components featured the sarcomere, sarcoplasmic reticulum, and contractile fibers; and molecular functions involved binding activities for heparin, sulfur compounds, and glycosaminoglycans. Concurrently, KEGG analysis (Figure 5C) identified relevant pathways such as the renin-angiotensin system, retinol metabolism, and vascular smooth muscle contraction. Gene set enrichment analysis (GSEA) results provided further insight (Figure 5D,5E): pathways like dilated cardiomyopathy and focal adhesion were upregulated in high-risk patients, while the low-risk subset was characterized by processes including DNA replication, protein export, and one-carbon metabolism.

Figure 5 Gene function enrichment analysis. (A) Bar graphs and (B) chord diagrams illustrating significant GO enrichment outcomes. (C) Bar plot displaying significant KEGG enrichment findings. GSVA showing enriched pathways in the (D) high-risk group and (E) low-risk group. BP, biological process; CC, cellular component; GO, Gene Ontology; GSVA, Gene Set Variation Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function.

Immune microenvironment analysis

The TME is a critical factor influencing both the development and therapeutic response of GC. To explore these dynamics across different risk categories, we applied several computational approaches. The ESTIMATE algorithm analysis revealed that low-risk patients exhibited substantially diminished stromal (P<0.001), immune (P<0.01), and overall ESTIMATE composite scores (P<0.001) when pitted against their high-risk peers (Figure 6A). Further visualization via boxplots (Figure 6B) revealed distinct patterns of immune infiltration. The high-risk cohort was characterized by significantly lower levels of memory B cells, regulatory T cells (Tregs), and M0 macrophage infiltration. Finally, employing the TIDE algorithm to assess immunotherapy response patterns, we found that high-risk patients had substantially higher TIDE scores (P<0.001), indicating a greater probability of immune evasion and potentially inferior clinical outcomes (Figure 6C).

Figure 6 Analysis of immunological relevance. (A) Violin plots depicting the distributions of stromal, immune, and ESTIMATE scores. (B) Box plot characterizes the variation in immune cell infiltration between high- and low-risk gastric cancer cohorts; blue denotes low-risk samples, and red denotes high-risk samples. (C) TIDE scores for patients in the high-risk versus low-risk groups. *, P<0.05; **, P<0.01; ***, P<0.001. TIDE, tumor immune dysfunction and exclusion; TME, tumor microenvironment.

TMB analysis

The maftools algorithm’s genomic analysis brought to light some intriguing insights, pinpointing TTN, TP53, and MUC16 as the genes most frequently mutated across both study groups. Moreover, while the high-risk group demonstrated a mutation rate of 87.63% (Figure 7A), the low-risk cohort showcased a notably higher mutation frequency, reaching an impressive 92.08% (Figure 7B). Within the low-risk group, TTN and TP53 mutations were especially prevalent. These findings were mirrored by the TMB comparisons, which demonstrated elevated TMB levels in low-risk versus high-risk patients (P=4.4×10⁻⁵, Figure 7C). An interesting twist came when we looked at overall survival times: patients with a higher TMB tended to live longer than those with a lower TMB (P=0.009, Figure 7D). In terms of prognosis, the “high-TMB/low-risk” cohort exhibited superior outcomes, contrasting with the “low-TMB/high-risk” cohort’s poorest results (P<0.001, Figure 7E).

Figure 7 Analysis of TMB and somatic mutation frequency in gastric cancer. Waterfall plots displaying the top 15 mutated genes in the (A) high-risk group (n=194) and (B) low-risk group (n=202). (C) Comparative analysis of TMB levels between groups. (D) Kaplan-Meier analysis of OS stratified by high versus low TMB. (E) Kaplan-Meier analysis of OS combined with TMB status and risk score. H, high; L, low; OS, overall survival; TMB, tumor mutation burden.

Drug sensitivity analysis

Drug sensitivity analysis aims to identify medications that may be effective against tumors in gastric cancer patients, utilizing the “oncoPredict” software package to accomplish this task. This approach identified nine potential candidates: afatinib, dasatinib, dihydrorotenone, gallibiscoquinazole, gefitinib, ibrutinib, lapatinib, osimertinib, and sinularin. The data revealed that afatinib demonstrated markedly lower IC₅₀ values in the low-risk cohort versus the high-risk group (Figure 8A). In contrast, dasatinib deviated from this trend, displaying considerably lower IC₅₀ values among critically ill individuals (Figure 8B). Furthermore, other compounds—namely dihydrorotenone, gallibiscoquinazole, gefitinib, ibrutinib, lapatinib, osimertinib, and sinulari—followed the same pattern as afatinib, with significantly lower IC₅₀ values in the low-risk cohort (Figure 8C-8I). Conclusively, the data indicate that while this group of medications may be more effective for GC patients with lower risk profiles, dasatinib could be particularly beneficial for those with higher risk scores.

Figure 8 Drug sensitivity analysis in high- and low-risk groups. Comparison of estimated drug sensitivities (IC₅₀ values) to (A) afatinib, (B) dasatinib, (C) dihydrorotenone, (D) gallibiscoquinazole, (E) gefitinib, (F) ibrutinib, (G) lapatinib, (H) osimertinib, and (I) sinularin. IC₅₀, half-maximal drug inhibitory concentration.

UBOX5-AS1 promotes the progression of GC in vitro

We selected UBOX5-AS1 for further experimental validation. To delve into UBOX5-AS1’s impact on the aggressive behavior of GC, we initially looked at its expression in the standard gastric cell line (GES-1) and various GC lines, including AGS, BGC-823, MGC-803, and HGC27. Our findings showed that BGC-823 cells had the highest levels of UBOX5-AS1 expression (Figure 9A). Based on this, we successfully created cell lines with stably reduced UBOX5-AS1 expression through small interfering RNA (siRNA)-mediated transfection. qRT-PCR analysis revealed a notably diminished UBOX5-AS1 level in the knockdown cohort relative to the unmodified control (Figure 9B). As shown in Figure 9C, knocking down UBOX5-AS1 exerted a major impact on slowing down cell growth. Wound-healing assays further demonstrated that the wound closure rate was markedly impaired in the knockdown group after 48 hours (Figure 9D), and the cells’ ability to form colonies was also significantly reduced (Figure 9E,9F). In line with this, we conducted Transwell experiments to gauge how UBOX5-AS1 influences cell movement. The data suggested that silencing UBOX5-AS1 potently curbs the movement of GC cells (Figure 9G,9H). Taken together, these insights strongly suggest that UBOX5-AS1 is a key oncogenic factor propelling GC advancement in a laboratory setting.

Figure 9 UBOX5-AS1 promotes the progression in GC cell lines. (A) The UBOX5-AS1 levels in normal and GC cell lines. (B) The UBOX5-AS1 levels in BGC-823 following UBOX5-AS1 knockdown. (C) CCK-8 assay was used to detect BGC-823 cell growth. (D) Wound healing assay was used to detect BGC-823 cell migration at 48 h (magnification ×10). (E,F) Cell colony formation assay was used to detect BGC-823 cell growth and the number of colonies was counted (magnification unspecified). (G,H) The Transwell invasion assay detected the migration of the BGC-823 cells and the number of migrated cells was counted (magnification ×10). For both the cell colony formation and Transwell invasion assays, the cells were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. **, P<0.01; ***, P<0.001. CCK-8, Cell Counting Kit-8; GC, gastric cancer; OD, optical density; si-UBOX5-AS1, small interfering RNA targeting UBOX5-AS; siNC, small interfering RNA targeting negative control.

Discussion

Globally, gastric cancer presents a major health burden with high morbidity and mortality rates. Its insidious nature frequently results in late-stage diagnoses and a scarcity of effective treatments (2), necessitating the discovery of novel biomarkers for early detection and individualized therapies. LncRNAs are recognized for their key regulatory roles in physiological processes, including carcinogenesis and tumor advancement, positioning them as strong biomarker candidates (21,22). Furthermore, Kcr, a newly identified post-translational modification, has been implicated in gene regulation, cellular function, and cancer (23,24). This study aims to integrate these two fields by developing and validating a KcrlncRNA-based prognostic signature in GC patients, with the goal of clarifying its clinical relevance and biological underpinnings.

In this comprehensive bioinformatics study, we first identified a set of 811 KcrlncRNAs in the TCGA-STAD cohort by establishing a strong co-expression relationship between crotonylation regulators and lncRNAs. This initial screening highlights the potential regulatory interplay between Kcr and lncRNAs in GC, a domain that is relatively underexplored (25). Using a rigorous analytical pipeline, we constructed a novel nine-KcrlncRNA prognostic signature. This signature, comprising AC108693.2, STARD4-AS1, AL353804.2, UBOX5-AS1, DIP2A-IT1, AL359091.3, AC026412.3, CHROMR, and AP000873.4, effectively stratified GC patients into high- and low-risk groups with significantly different overall survival outcomes across independent training and testing cohorts. This marker’s strong forecasting capability was further evidenced by high AUC values in time-dependent ROC curves (e.g., 0.72–0.77 for 1, 3, and 5-year OS), underscoring its potential clinical utility. The results align with earlier research demonstrating the power of lncRNA-based signatures in GC prognosis (26,27).

A detailed examination of the nine lncRNAs in our signature reveals diverse roles and varying degrees of prior investigation in cancer. STARD4-AS1 has been implicated in various cancers. For instance, it has been shown to promote OSCC cell line spread and invasive capacity by knocking down STARD4-AS1 (28); and STARD4-AS1 suppression diminished melanoma cell growth and compromised motility and invasive capacity (29); serum STARD4-AS1 is a new indicator for diagnosing GC and enhances its progression (30). These observations propose a possible oncogenic function for STARD4-AS1 in gastric cancer, likely through the control of cell proliferation and migration. DIP2A-IT1 has also gained attention in cancer research. Studies have shown that osteosarcoma tissue exhibits elevated DIP2A-IT1 levels compared to typical bone tissue (31). LncRNA CHROMR enhances cell growth, invasion, and plays a role in rituximab resistance in diffuse large B-cell lymphoma (32). For other lncRNAs in our signature including UBOX5-AS1, AC026412.3, AC108693.2, AL353804.2, AL359091.3, and AP000873.4, specific literature in gastric cancer or even broader cancer contexts is less abundant, highlighting the novelty and potential significance of our findings. Their inclusion in a robust prognostic signature based on crotonylation association strongly suggests novel, previously unrecognized roles in GC pathogenesis, potentially through mechanisms involving Kcr regulation. For our investigation, we began by putting forward the hypothesis that UBOX5-AS1 acts as a cancer-promoting agent in gastric cancer, with supporting evidence presented in Figure 2E,2G. Subsequently, our analysis revealed that UBOX5-AS1 expression levels were considerably higher across various gastric cancer cell lines, including BGC-823, when contrasted with the GES-1 cell line. Upon reducing UBOX5-AS1 expression in BGC-823 cells through knockdown techniques, we observed a significant decrease in these cancer cells’ replication and movement capabilities. These preliminary findings suggest that UBOX5-AS1 promotes gastric cancer progression, aligning with the predictions from our bioinformatics analysis. Future mechanistic studies are critically warranted to delineate the precise functions of each component lncRNA, particularly the less-studied ones, in GC progression and their connection to crotonylation. Several limitations should be noted. First, experimental validation was restricted to a single cell line; given the heterogeneity of GC, these findings require further verification in broader panels (e.g., AGS or HGC-27). Second, the absence of rescue experiments limits the direct confirmation of the mechanistic link. Future research will combine rescue experiments with multi-cell validation to refine the specific mechanism analysis of lncRNAs related to crotonylation metabolism in GC.

A cornerstone of our research involved establishing the standalone prognostic significance of the KcrlncRNA signature. Through both single-factor and multifactorial Cox regression models, the risk score functioned as a dependable indicator for predicting survival outcomes, maintaining its statistical relevance even when accounting for established clinical parameters like patient age, sex, tumor grading, and disease stage. This finding underscores how our signature offers extra prognostic insights beyond standard pathological assessments, which could prove invaluable in shaping clinical treatment strategies (33). To connect scholarly inquiry with practical application, we engineered a nomogram that combines the KcrlncRNA risk assessment with key clinical determinants—notably TNM staging. The calibration plots demonstrated remarkable agreement between predicted and observed survival results, while the substantial C-index of 0.783 highlighted the model’s robust discriminatory capabilities, indicating its potential to enhance tailored prognostic evaluations for gastric cancer patients (34). These integrated prediction tools have been gaining traction within oncology circles precisely because they effectively consolidate multiple prognostic variables into an accessible and clinically practical format (35,36).

To unravel the biological implications of our prognostic signature, functional enrichment analyses were performed. Intriguingly, when comparing the gene expression profiles of high-risk and low-risk groups, both GO and KEGG analyses pinpointed a notable increase in the areas of ECM reorganization and stromal activation. This encompasses the complex extracellular architecture, such as collagen-rich ECM, and the pivotal Wnt signaling cascade. Moreover, an unanticipated surge in muscle and vascular smooth muscle contraction-related terms suggests a prevalence of cancer-associated fibroblasts (CAFs) within the dataset. CAFs express α-smooth muscle actin and remodel the stromal architecture. Biffi and Tuveson [2021] emphasize that such myofibroblastic CAFs create a dense, fibrotic stroma that provides a physical barrier to the administration of drugs (37). The activation of the Wnt signaling pathway further elucidates the mechanism underlying this phenotype. According to Zhan et al. [2017], Wnt signaling plays a vital role in controlling ECM assembly and triggering epithelial-mesenchymal transition (EMT), which in turn facilitates tumor invasion and metastasis (38). Further analysis of GSEA results showed that complement and coagulation cascades and focal adhesion were enriched in the high-risk cohort. While complement activation is often associated with inflammation, Afshar-Kharghan [2017] highlighted that in cancer, complement components can summon myeloid-derived suppressor cells (MDSCs), assisting in immune evasion (39). Additionally, the enrichment of focal adhesion pathways aligns with findings by Eke and Cordes [2015], who reported that focal adhesion signaling is integral to radio resistance and ECM-mediated survival in aggressive tumors (40). The low-risk group, in contrast, exhibited significant activation of cell cycle, DNA replication, and folate-dependent one-carbon metabolism. This profile suggests a proliferative rather than invasive phenotype. Newman and Maddocks [2017] highlighted that one-carbon metabolism is the metabolic engine supporting the rapid nucleotide synthesis required for DNA replication in fast-growing tumors (41). Consistent with this, Hanahan [2022] in his updated Hallmarks of Cancer describes the deregulation of cellular energetics and sustained proliferative signaling as fundamental traits of rapidly dividing tumor cells, distinct from the invasion-metastasis cascade observed in the high-risk group (42). To sum up, the risk model successfully differentiates a stromal-rich environment, immunologically suppressed phenotype (high-risk) and a metabolically active, highly proliferative phenotype (low-risk), suggesting that the former may benefit from stromal-targeting therapies, while the latter targets anti-proliferative agents.

The examination of the TME sheds light on a contradictory immune environment within the high-risk cohort. Although the ESTIMATE algorithm suggests an increased level of stromal and immune purity, the detailed analysis of immune cell composition reveals a state of disarray. The reduced infiltration of adaptive immune effectors like memory B cells, combined with significantly elevated TIDE scores, suggests a state of immune exclusion or exhaustion despite the high stromal presence. This discordance between high immune scores and poor predicted immunotherapy response is a phenomenon known as the “stromal-excluded” phenotype. Turley et al. [2015] describe how dense stromal fibrosis can trap immune cells in the peritumoral region, preventing effective tumor killing, which aligns with our high stromal score findings (43). Vertically, the elevated TIDE scores in the high-risk group corroborate this mechanism of immune evasion. Jiang et al. [2018] developed TIDE specifically to identify such resistance mechanisms, noting that high TIDE scores correlate with poor checkpoint inhibitor efficacy due to T-cell dysfunction (44). Furthermore, the specific reduction in memory B cells in high-risk patients is critical; Helmink et al. [2020] demonstrated that B cells are vital for organizing tertiary lymphoid structures (TLS) that support effective anti-tumor immunity (45). The lack thereof likely contributes to the inferior clinical outcomes predicted in our high-risk cohort.

Genomic analysis unveiled a distinct patterns in the relationship between TMB and patient outcomes, alongside markedly different drug responses across risk categories. Although TTN, TP53, and MUC16 consistently stood out as the most frequently mutated genes in both groups, the low-risk cohort surprisingly exhibited a higher mutational burden, with significantly greater overall mutation frequencies and TMB scores. Furthermore, elevated TMB was associated with notably longer survival times, with ‘high-TMB + low-risk’ patients achieving the most favorable clinical outcomes. This finding aligns with a growing body of studies highlighting TMB as a valuable predictor of improved survival and immunotherapy responses, particularly in the context of immune checkpoint blockade. Back in 2017, Goodman and his team presented compelling evidence showing how extremely high TMB across solid tumors predicts better outcomes for patients (46). Likewise, Scobie et al. (47) established TMB as an independent predictor of response to immune checkpoint inhibitors. In our view, the low-risk group’s TMB advantage likely generates a surge of neoantigens, providing the immune system with more targets to recognize—a clear example of reversing tumor immunogenicity, despite earlier reports suggesting reduced immune activity in these patients. Meanwhile, the ‘low-TMB + high-risk’ patients’ poor prognosis highlights how genomically stable, immune-evasive tumors can evade detection with lethal consequences. Additionally, it is important not to overlook the well-known TTN (48) and TP53 (49) mutations, which were prevalent in both groups and have long been recognized as key drivers of cancer progression.

Drug sensitivity analysis further elucidated therapeutic distinctions. Most tested drugs, including epidermal growth factor receptor (EGFR)/human epidermal growth factor receptor 2 (HER2) inhibitors (afatinib, gefitinib, lapatinib, osimertinib) and other agents, showed lower IC₅₀ in the low-risk group. This suggests that the proliferative, high-TMB low-risk tumors may be more responsive to targeted therapies against growth factor pathways or other essential cellular processes. This is consistent with Shitara et al. [2020] who highlighted the role of HER2 inhibition in gastric cancer, indicating the clinical relevance of these targeted therapies (50). The “oncoPredict” software package used for this analysis is a validated tool for such predictions (51). Conversely, dasatinib, a multi-targeted tyrosine kinase inhibitor (TKI), uniquely exhibited greater efficacy (lower IC₅₀) in the high-risk group. Dasatinib inhibits SRC family kinases, which are often implicated in invasion, metastasis, and resistance mechanisms in stromal-rich tumors. Zhang et al. [2012] extensively discussed the role of SRC kinases in driving tumor progression and drug resistance (52). Therefore, dasatinib could represent a specific therapeutic avenue for the high-risk cohort, potentially overcoming the immune exclusion and invasive phenotype characterized by their high stromal presence. This mirrors the success seen with other TKIs in cancer (53).

Although our study yielded encouraging results, it is not without its limitations. As a retrospective bioinformatics investigation relying mainly on publicly accessible TCGA information, external validation across diverse and independent patient cohorts is essential to confirm the broad applicability of our findings. Our reliance solely on RNA sequencing data limits a comprehensive understanding of the complexity of gastric cancer; integrating this with other omics data, including genomic, proteomic, or metabolomic information, could provide a more complete picture of the multifaceted nature of GC (54). Most importantly, while our work identified lncRNAs associated with crotonylation and conducted initial tests on UBOX5-AS1, we have only begun to explore how these lncRNAs actually interact with crotonylation-associated genes and their specific functions in the progression of gastric cancer. These areas require thorough experimental investigation. Finally, while our computational predictions regarding drug sensitivity offer a promising lead, they require robust empirical support through laboratory and animal studies, such as testing in cell cultures and using patient-derived xenograft models (55).

Conclusions

In a nutshell, our team has successfully developed a groundbreaking predictive framework utilizing crotonylation-associated lncRNAs and thoroughly examined the immune environment within gastric cancer. This approach proved its mettle in forecasting patient prognoses and shows great promise as an instrumental resource for refining immunotherapy strategies and personalized treatment protocols. These results highlight the critical importance of crotonylation-associated lncRNAs in cancer progression while offering fresh perspectives on the immunological features of the gastric cancer microenvironment. That said, additional experimental validation and deeper mechanistic investigations are necessary to confirm the practical applications of these lncRNAs and pave the way for their implementation in tailored cancer care.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the MDAR and TRIPOD reporting checklists. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2752/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2752/prf

Funding: The study was supported by the Zunyi Technology and Big Data Bureau, Moutai Institute Joint Science and Technology Research and Development Project (ZSKHHZ [2024] No. 372); Guizhou Provincial Science and Technology Department Basic Research Program (Natural Science Category) Youth Guidance Project (QKHJC-QN [2025]301); and Research Foundation for Scientiffc Scholars of Moutai Institute (mygccrc [2022] 006, mygccrc [2023] 046 and mygccrc [2024] 007).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2752/coif). The authors have no 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.

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. 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]
2. Smyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet 2020;396:635-48. [Crossref] [PubMed]
3. Duan Y, Xu Y, Dou Y, et al. Helicobacter pylori and gastric cancer: mechanisms and new perspectives. J Hematol Oncol 2025;18:10. [Crossref] [PubMed]
4. Guan WL, He Y, Xu RH. Gastric cancer treatment: recent progress and future perspectives. J Hematol Oncol 2023;16:57. [Crossref] [PubMed]
5. Lordick F, Carneiro F, Cascinu S, et al. Gastric cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2022;33:1005-20. [Crossref] [PubMed]
6. Janjigian YY, Shitara K, Moehler M, et al. First-line nivolumab plus chemotherapy versus chemotherapy alone for advanced gastric, gastro-oesophageal junction, and oesophageal adenocarcinoma (CheckMate 649): a randomised, open-label, phase 3 trial. Lancet 2021;398:27-40. [Crossref] [PubMed]
7. Shitara K, Van Cutsem E, Bang YJ, et al. Efficacy and Safety of Pembrolizumab or Pembrolizumab Plus Chemotherapy vs Chemotherapy Alone for Patients With First-line, Advanced Gastric Cancer: The KEYNOTE-062 Phase 3 Randomized Clinical Trial. JAMA Oncol 2020;6:1571-80. [Crossref] [PubMed]
8. Mattick JS, Amaral PP, Carninci P, et al. Long non-coding RNAs: definitions, functions, challenges and recommendations. Nat Rev Mol Cell Biol 2023;24:430-47. [Crossref] [PubMed]
9. Statello L, Guo CJ, Chen LL, et al. Gene regulation by long non-coding RNAs and its biological functions. Nat Rev Mol Cell Biol 2021;22:96-118. [Crossref] [PubMed]
10. Serviss JT, Johnsson P, Grandér D. An emerging role for long non-coding RNAs in cancer metastasis. Front Genet 2014;5:234. [Crossref] [PubMed]
11. Li C, Xing S, Zhang D, et al. Exosomal long non-coding RNAs in gastrointestinal cancer: chemoresistance mediators and therapeutic targets. J Transl Med 2025;23:889. [Crossref] [PubMed]
12. Dragomir MP, Kopetz S, Ajani JA, et al. Non-coding RNAs in GI cancers: from cancer hallmarks to clinical utility. Gut 2020;69:748-63. [Crossref] [PubMed]
13. Yang J, Liu F, Wang Y, et al. LncRNAs in tumor metabolic reprogramming and immune microenvironment remodeling. Cancer Lett 2022;543:215798. [Crossref] [PubMed]
14. Huang Y, Zhang J, Hou L, et al. LncRNA AK023391 promotes tumorigenesis and invasion of gastric cancer through activation of the PI3K/Akt signaling pathway. J Exp Clin Cancer Res 2017;36:194. [Crossref] [PubMed]
15. Xie JY, Ju J, Zhou P, et al. The mechanisms, regulations, and functions of histone lysine crotonylation. Cell Death Discov 2024;10:66. [Crossref] [PubMed]
16. Contreras-de la Rosa PA, Aragón-Rodríguez C, Ceja-López JA, et al. Lysine crotonylation: A challenging new player in the epigenetic regulation of plants. J Proteomics 2022;255:104488. [Crossref] [PubMed]
17. Fang Y, Xu X, Ding J, et al. Histone crotonylation promotes mesoendodermal commitment of human embryonic stem cells. Cell Stem Cell 2021;28:748-763.e7. [Crossref] [PubMed]
18. Ntorla A, Burgoyne JR. The Regulation and Function of Histone Crotonylation. Front Cell Dev Biol 2021;9:624914. [Crossref] [PubMed]
19. Wang X, Qu Y, Li Z, et al. Histone crotonylation in tumors Mol Clin Oncol 2025;22:39. (Review). [Crossref] [PubMed]
20. Yang B, Wen F, Cui Y. Integrative transcriptome analysis identifies a crotonylation gene signature for predicting prognosis and drug sensitivity in hepatocellular carcinoma. J Cell Mol Med 2024;28:e70083. [Crossref] [PubMed]
21. Su X, Zhang J, Yang W, et al. Identification of the Prognosis-Related lncRNAs and Genes in Gastric Cancer. Front Genet 2020;11:27. [Crossref] [PubMed]
22. Ghorbani A, Hosseinie F, Khorshid Sokhangouy S, et al. The prognostic, diagnostic, and therapeutic impact of Long noncoding RNAs in gastric cancer. Cancer Genet 2024;282-283:14-26. [Crossref] [PubMed]
23. Zhao H, Han Y, Zhou P, et al. Protein lysine crotonylation in cellular processions and disease associations. Genes Dis 2024;11:101060. [Crossref] [PubMed]
24. Westerveld M, Besermenji K, Aidukas D, et al. Cracking Lysine Crotonylation (Kcr): Enlightening a Promising Post-Translational Modification. Chembiochem 2025;26:e202400639. [Crossref] [PubMed]
25. Wan J, Liu H, Ming L. Lysine crotonylation is involved in hepatocellular carcinoma progression. Biomed Pharmacother 2019;111:976-82. [Crossref] [PubMed]
26. Miao Y, Sui J, Xu SY, et al. Comprehensive analysis of a novel four-lncRNA signature as a prognostic biomarker for human gastric cancer. Oncotarget 2017;8:75007-24. [Crossref] [PubMed]
27. Wang Z, Cao L, Zhou S, et al. Construction and Validation of a Novel Pyroptosis-Related Four-lncRNA Prognostic Signature Related to Gastric Cancer and Immune Infiltration. Front Immunol 2022;13:854785. [Crossref] [PubMed]
28. Kong L, Wang C, Lu X, et al. Novel cuproptosis-related lncRNAs risk model to predicting prognosis and guiding immunotherapy for OSCC patients. Discov Oncol 2025;16:723. [Crossref] [PubMed]
29. Li G, Wu T, Li H, et al. Construction of a tumor mutational burden-derived LncRNA prognostic computational framework associated with therapy sensitivity in skin cutaneous melanoma. J Transl Med 2024;22:966. [Crossref] [PubMed]
30. Chu X, Cao M, Qin X, et al. Serum STARD4-AS1 as a Novel Marker for Gastric Cancer Diagnosis and Promotes Gastric Cancer Progression. Clin Transl Gastroenterol 2025;16:e00915. [Crossref] [PubMed]
31. Rothzerg E, Xu J, Wood D. Identification of Differentially Expressed Intronic Transcripts in Osteosarcoma. Noncoding RNA 2022;8:73. [Crossref] [PubMed]
32. Liu C, Zhao X, Wang Z, et al. LncRNA CHROMR/miR-27b-3p/MET axis promotes the proliferation, invasion, and contributes to rituximab resistance in diffuse large B-cell lymphoma. J Biol Chem 2024;300:105762. [Crossref] [PubMed]
33. Nie C, Zhai J, Wang Q, et al. Comprehensive Analysis of an Individualized Immune-Related lncRNA Pair Signature in Gastric Cancer. Front Cell Dev Biol 2022;10:805623. [Crossref] [PubMed]
34. Zheng X, Wang X, Zheng L, et al. Construction and Analysis of the Tumor-Specific mRNA-miRNA-lncRNA Network in Gastric Cancer. Front Pharmacol 2020;11:1112. [Crossref] [PubMed]
35. Balachandran VP, Gonen M, Smith JJ, et al. Nomograms in oncology: more than meets the eye. Lancet Oncol 2015;16:e173-80. [Crossref] [PubMed]
36. Wang Q, Shen K, Fei B, et al. Development and validation of a nomogram to predict cancer-specific survival of elderly patients with unresected gastric cancer who received chemotherapy. Sci Rep 2024;14:9008. [Crossref] [PubMed]
37. Biffi G, Tuveson DA. Diversity and Biology of Cancer-Associated Fibroblasts. Physiol Rev 2021;101:147-76. [Crossref] [PubMed]
38. Zhan T, Rindtorff N, Boutros M. Wnt signaling in cancer. Oncogene 2017;36:1461-73. [Crossref] [PubMed]
39. Afshar-Kharghan V. The role of the complement system in cancer. J Clin Invest 2017;127:780-9. [Crossref] [PubMed]
40. Eke I, Cordes N. Focal adhesion signaling and therapy resistance in cancer. Semin Cancer Biol 2015;31:65-75. [Crossref] [PubMed]
41. Newman AC, Maddocks ODK. One-carbon metabolism in cancer. Br J Cancer 2017;116:1499-504. [Crossref] [PubMed]
42. Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov 2022;12:31-46. [Crossref] [PubMed]
43. Turley SJ, Cremasco V, Astarita JL. Immunological hallmarks of stromal cells in the tumour microenvironment. Nat Rev Immunol 2015;15:669-82. [Crossref] [PubMed]
44. Jiang P, Gu S, Pan D, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med 2018;24:1550-8. [Crossref] [PubMed]
45. Helmink BA, Reddy SM, Gao J, et al. B cells and tertiary lymphoid structures promote immunotherapy response. Nature 2020;577:549-55. [Crossref] [PubMed]
46. Goodman AM, Kato S, Bazhenova L, et al. Tumor Mutational Burden as an Independent Predictor of Response to Immunotherapy in Diverse Cancers. Mol Cancer Ther 2017;16:2598-608. [Crossref] [PubMed]
47. Scobie MR, Zhou KI, Ahmed S, et al. Utility of Tumor Mutational Burden as a Biomarker for Response to Immune Checkpoint Inhibition in the VA Population. JCO Precis Oncol 2023;7:e2300176. [Crossref] [PubMed]
48. Djulbegovic MB, Uversky VN, Karp CL, et al. Functional impact of titin (TTN) mutations in ocular surface squamous neoplasia. Int J Biol Macromol 2022;195:93-101. [Crossref] [PubMed]
49. Kennedy MC, Lowe SW. Mutant p53: it’s not all one and the same. Cell Death Differ 2022;29:983-7. [Crossref] [PubMed]
50. Shitara K, Bang YJ, Iwasa S, et al. Trastuzumab Deruxtecan in Previously Treated HER2-Positive Gastric Cancer. N Engl J Med 2020;382:2419-30. [Crossref] [PubMed]
51. Maeser D, Gruener RF, Huang RS. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform 2021;22:bbab260. [Crossref] [PubMed]
52. Zhang S, Yu D. Targeting Src family kinases in anti-cancer therapies: turning promise into triumph. Trends Pharmacol Sci 2012;33:122-8. [Crossref] [PubMed]
53. Druker BJ, Guilhot F, O’Brien SG, et al. Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med 2006;355:2408-17. [Crossref] [PubMed]
54. Gao YX, Guo XJ, Lin B, et al. Targeting LHPP in neoadjuvant chemotherapy resistance of gastric cancer: insights from single-cell and multi-omics data on tumor immune microenvironment and stemness characteristics. Cell Death Dis 2025;16:306. [Crossref] [PubMed]
55. Wu AR, Kawahara TL, Rapicavoli NA, et al. High throughput automated chromatin immunoprecipitation as a platform for drug screening and antibody validation. Lab Chip 2012;12:2190-8. [Crossref] [PubMed]