Neoadjuvant therapy (NAT) is recommended for locally advanced gastric cancer (LAGC), but some patients respond poorly. We aimed to construct a multimodal model integrating CT images, transcriptomic sequencing, and clinicopathological data to assess prognosis in LAGC patients receiving NAT. This multicenter study included 505 LAGC patients who underwent NAT. Radiomic features were extracted from preoperative CT images of 505 patients. RNA-seq was performed on 277 post-NAT specimens, with additional data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases (n = 804). Patients were divided into training (168 cases), internal validation (72 cases), and external validation cohorts. Machine learning algorithms identified key radiomic, molecular, and clinical features associated with NAT response, which were then integrated into a multimodal model to predict overall survival (OS) and disease-free survival (DFS). Six radiomic and three molecular features significantly associated with NAT response were selected. Radiomic risk (hazard ratio [HR]: 4.0, P < 0.001) and molecular risk (HR: 7.1, P < 0.001) were independent prognostic factors. By integrating radiomic risk, molecular risk, and clinical characteristics, a multimodal model (MuMo) was constructed.The C-index results (OS, C-index = 0.855; DFS, C-index = 0.786) demonstrated that MuMo outperformed the single-modality models and ypTNM staging.Mechanistic analysis suggested that the efficacy of neoadjuvant therapy was significantly enriched in immune-inflammatory pathways. MuMo can effectively predict postoperative survival risk in LAGC patients receiving NAT, serving as a powerful tool for optimizing prognostic assessment.

Background:Immune checkpoint inhibitors (ICIs) revolutionised cancer treatment but may alter infection risk. The effect of ICIs on infection susceptibility remains unclear. Methods:We searched PubMed, the Cochrane Library, and EMBASE from database inception to October 10, 2025 for randomised controlled trials (RCTs). The primary endpoint was the risk ratio (RR) of all-grade infections; the secondary endpoint was the RR of severe (Grade 3-5) infections. Stratified analyses by ICI agent, regimen, treatment timing, infection site, and cancer type were performed separately for treatment-emergent adverse events (TEAEs) and treatment-related adverse events (TRAEs). Publication bias was assessed using funnel plots, Egger's test, and trim-and-fill; heterogeneity was evaluated by I 2 statistic. This study is registered with PROSPERO, number CRD420251231682. Findings:A total of 141 RCTs (87,484 participants) were included. In TEAE-based analyses, ICIs were associated with increased risks of any-grade infections [RR = 1.25; 95% confidence interval (CI): 1.12-1.39; I 2 = 71.1%] and severe infections (RR = 1.30; 95% CI: 1.17-1.44), driven by symmetric comparisons (ICI vs. Placebo and ICI plus chemotherapy vs. chemotherapy). For TRAEs, no significant increase was seen for any-grade infections (RR = 1.01; 95% CI: 0.84-1.22; I 2 = 67.3%) or severe infections (RR = 0.96; 95% CI: 0.81-1.14). However, symmetric comparisons again showed increased TRAE risks for ICI vs. placebo and ICI plus chemotherapy vs. chemotherapy. Publication bias was suggested for any-grade infectious TRAEs (Egger's P = 0.015); after trim-and-fill correction, the pooled RR was strengthened (RR = 1.24). Interpretation:The relationship between ICI therapy and infection risk is nuanced and context-dependent. In symmetric comparisons, ICIs are associated with modestly increased TEAE- and TRAE-based infection risks. This risk should be weighed against the expected survival benefit, underscoring the need for individualised assessment. Future prospective studies with time-to-event analyses and standardised infection surveillance are needed to clarify the temporal relationship between ICI exposure and infection risk. Funding:There was no funding source for this study.

Gastric cancer (GC) derives limited benefit from immune checkpoint blockade (ICB), in part due to stromal-mediated immune suppression. To investigate the underlying mechanisms, we performed single-cell RNA sequencing on 14 GC specimens from patients treated with neoadjuvant ICB. We identified a distinct subset of TRPA1+ cancer-associated fibroblasts (CAFs) maintained by tumor-derived midkine (MDK) through LEF1-dependent transcription under ICB pressure. These TRPA1+ CAFs drive immunosuppression by secreting WNT5A, which polarizes SPP1+ macrophages and subsequently promotes the accumulation of CD25 + FOXP3+ regulatory T cells (Tregs) in the tumor microenvironment. Disrupting this stromal-myeloid-lymphoid axis via WNT5A inhibition or SPP1+ macrophage depletion reduced Treg infiltration and enhanced the antitumor efficacy of anti-PD-L1 therapy in vivo. Our study unveils an MDK-TRPA1 + CAF-mediated mechanism of ICB resistance in GC and highlights the WNT5A-SPP1+ macrophage pathway as a promising target to overcome immunotherapy resistance.

RATIONALE AND OBJECTIVES:To evaluate whether baseline T2-mapping MRI is associated with pathologic response to neoadjuvant therapy (nT), lymphovascular invasion (LVI) and perineural invasion (PNI) in locally advanced esophageal squamous cell carcinoma (ESCC). MATERIALS AND METHODS:This single-center retrospective study enrolled 212 ESCC patients with pre-nT 3 T MRI including T2 mapping at a university-affiliated hospital. Patients were stratified by pathologic tumor regression grade (TRG) into pathologic complete response (pCR, TRG 0) vs non-pCR (npCR, TRG ≥ 1), good response (GR, TRG ≤ 1) vs poor response (PR, TRG ≥ 2), plus LVI/PNI status. Univariate/multivariable logistic regression identified predictors, with performance assessed via ROC analysis. RESULTS:Patients were grouped as pCR (n = 56), GR (n = 81), LVI (n = 67), PNI (n = 32). T2 values were markedly higher in pCR (98.0 ± 12.2 ms vs 82.6 ± 11.5 ms, P < .001; AUC = 0.822) and GR (94.8 ± 12.6 ms vs 81.7 ± 11.7 ms, P < .001; AUC = 0.734) vs npCR and PR. Conversely, T2 values were lower in LVI (79.0 ± 12.8 ms vs 90.5 ± 12.3 ms, P < .001; AUC = 0.840) and PNI (73.6 ± 9.7 ms vs 89.1 ± 12.8 ms, P < .001; AUC = 0.844). T2 value was independently associated with all endpoints in multivariate analysis. CONCLUSION:Pre-nT T2 values were associated with pathologic response, LVI, and PNI in locally advanced ESCC, which may inform personalized treatment planning in future studies.

Background Toward identifying new strategies to target gastric cancer stem-like cells (CSCs), we evaluated the function of the tumour suppressor CDK5 regulatory subunit-associated protein 3 (CDK5RAP3) in gastric CSC maintenance. Methods We examined the expression of CDK5RAP3 and CD44 in gastric cancer patients. The function and mechanisms of CDK5RAP3 were checked in human and mouse gastric cancer cell lines and in mouse xenograft. Results We show that CDK5RAP3 is weakly expressed in gastric CSCs and is negatively correlated with the gastric CSC marker CD44. CDK5RAP3 overexpression decreased expression of CSC markers, spheroid formation, invasion and migration, and reversed chemoresistance in gastric CSCs in vitro and vivo. CDK5RAP3 expression was found to be regulated by extracellular-related kinase (ERK) signalling. ERK inhibitors decreased spheroid formation, migration and invasion, and the expression of epithelial-to-mesenchymal transition (EMT)-related proteins in both GA cells and organoids derived from a genetically engineered mouse model of GA. Finally, CDK5RAP3 expression was associated with reduced lymph-node metastasis and better prognosis, even in the presence of high expression of the EMT transcription factor Snail, among patients with CD44-positive GA. Conclusions Our results demonstrate that CDK5RAP3 is suppressed by ERK signalling and negatively regulates the self-renewal and EMT of gastric CSCs.

Cancer-associated fibroblasts (CAFs) orchestrate immune-excluded tumor microenvironment (TME), but the CAF heterogeneity remains incompletely understood in gastric cancer (GC). In this study, we integrated multicohort single-cell RNA sequencing (scRNA-seq), spatial transcriptomics, and bulk transcriptomic data to construct a comprehensive atlas of the GC TME. Unsupervised clustering identified eight transcriptionally distinct CAF subpopulations, among which CTHRC1+ CAFs were selectively enriched in tumors and showed the strongest association with T cell exclusion. Pseudotemporal trajectory analysis, gene regulatory network inference, and cell-cell communication analysis revealed that basic helix-loop-helix family member e41 (BHLHE41) serves as a key transcription factor driving CTHRC1+ CAF differentiation, whereas spatial analyses demonstrated these fibroblasts contribute to fibrotic niches at the tumor-stroma interface through macrophage migration inhibitory factor (MIF)-mediated signaling. Finally, we developed and validated a CTHRC1+ cancer-associated fibroblast-related risk signature (CRS) that accurately predicts immunotherapy response across independent cohorts. These findings establish CTHRC1+ CAFs as a critical stromal determinant of immune exclusion in GC, suggesting that targeting the CTHRC1+ CAF-MIF axis or applying CRS-guided patient stratification may enhance immunotherapy efficacy.

AIMS:To accurately evaluate the response to chemotherapy in primary gastric diffuse large B-cell lymphoma (DLBCL), which is essential for improving patient prognosis. Consequently, our objective was to develop and validate a point-based scoring system (PSS), utilising clinicopathological and quantitative computed tomography (CT) features, to assess the efficacy of chemotherapy in primary gastric DLBCL patients. MATERIALS AND METHODS:A total of 161 gastric DLBCL patients from two hospitals were retrospectively analysed. The quantitative CT features and clinicopathological of their tumours were systematically gathered. Subsequently, sequential univariate and multivariate analyses identified predictive factors associated with chemotherapy response. A PSS was developed using multivariate logistic regression analysis. The performance of the model was assessed using area under the curve (AUC), accuracy, sensitivity, specificity, recall, and F1 score. RESULTS:Post-chemotherapy lactate dehydrogenase, β2-microglobulin levels, lesion length, and the short-axis diameter of lymph nodes, along with staging, were identified as predictive factors in the scoring system for assessing chemotherapy efficacy in gastric DLBCL. Patients with a score ≥68 are considered to have metabolic progression of the disease; those with scores ranging from 4.0 to 68.0 are considered to show no metabolic response; scores from 0 to 4.0 indicate a partial metabolic response; and scores ≤0 signify a complete metabolic response. Utilising the scoring system, an AUC of 0.975 was recorded, accompanied by a 95% CI ranging from 0.940 to 1.000. CONCLUSION:This study presents a potentially practical and reliable PSS for assessing chemotherapy response in patients with gastric DLBCL.

BackgroundThe clinical efficacy of neoadjuvant immunotherapy combined with chemotherapy (NICT) in patients with locally advanced gastric cancer (LAGC) exhibits considerable variability. In this study, we developed and validated a radiomics-based scoring system to inform personalized neoadjuvant therapy strategies.MethodsA retrospective analysis was conducted on a multicenter cohort comprising 633 LAGC patients who underwent neoadjuvant therapy between 2019 and 2023. The primary endpoint was major pathological response (MPR). Radiomic features were extracted and selected from contrast-enhanced CT images obtained before neoadjuvant therapy to construct a radiomics scoring system, which stratified patients into high-, intermediate-, and low-score groups. The therapeutic outcomes across different groups were compared to discern patients who are likely to benefit from NICT as opposed to those who would derive more benefit from neoadjuvant chemotherapy alone (NACT). Additionally, analysis of tumor microenvironment immune infiltration and functional enrichment were performed to explore the potential biological mechanisms underlying the radiomic signatures.ResultsIn the intermediate-score group, the MPR rates were significantly higher among patients receiving NICT compared to those receiving NACT (training set: 49.2% vs. 26.0%, p = 0.005; validation set: 51.6% vs. 19.2%, p = 0.002). In contrast, no significant differences were observed in the high- or low-score groups. The analysis of immune infiltration indicated that the intermediate-score group exhibited significantly higher baseline levels of mast cells and dendritic cells compared to the combined high- and low-score groups (p < 0.05). Post-treatment evaluation following NICT revealed a significant increase in CD8+ T-cell infiltration relative to baseline levels within the intermediate-score group (p = 0.033). The gene KRT5 was identified as differentially expressed and correlated with dendritic cell abundance.ConclusionThe radiomics scoring system exhibits potential for stratifying patients who may be appropriate candidates for NICT or for whom the de-escalation of immunotherapy might be considered in future prospective studies, thereby contributing to the optimization of neoadjuvant treatment strategies.

Perioperative chemotherapy is an important treatment option for resectable locally advanced gastric cancer (LAGC). However, completing planned cycles remains challenging for patients. This study aimed to investigate the effect of treatment completion on the survival of patients with poor tumor regression following perioperative chemotherapy. This multicenter retrospective study reviewed the data of patients with LAGC who underwent perioperative chemotherapy and curative resection and had a tumor regression grade of 2 or 3 (i.e., at least 10

Loss of adipose tissue and systemic inflammation are recognized hallmarks of poor prognosis in advanced cancer. However, it is still uncertain whether changes in visceral (VAT) and subcutaneous adipose tissue (SAT) during neoadjuvant chemotherapy (NACT) (beyond static body composition) carry independent prognostic relevance, especially when interpreted alongside the evolving inflammatory response. Between February 2010 and January 2024, patients with LAGC receiving NACT followed by radical gastrectomy were screened across five tertiary centres (n = 587). Visceral and subcutaneous adipose index changes (ΔVAI, ΔSAI) were derived from L3-level computed tomography images, and inflammatory variation was assessed using Δneutrophil-to-lymphocyte ratio (ΔNLR). Transcriptomic profiling of 94 pretreatment tumours explored biological correlates. Visceral fat-preserved group conferred a marked survival advantage exclusively in patients with ΔNLR-high(3-year OS: 62.1

Aberrant alternative splicing (AS) in cancer generates oncogenic proteomic diversity that drives tumor progression. Given the suboptimal efficacy of immune checkpoint inhibitors (ICIs) in gastric cancer (GC), the therapeutic potential of modulating RNA splicing to augment immunotherapy remains unclear. Here, we demonstrate that the splicing factor SRSF10 is progressively upregulated during gastric tumorigenesis and exhibits elevated expression in ICIs-resistant GC. Utilizing multiple mouse models, we confirmed that SRSF10 ablation with a selective inhibitor 1C8 robustly inhibits GC growth and enhances CD8+ T-cell infiltration via CCL2-mediated reprogramming of tumor-associated macrophages (TAMs). Notably, SRSF10 blockade restricts pre-neoplastic metaplastic cells re-entry the cell cycle and the TAMs reprogramming. Mechanistically, cell-autonomous SRSF10 activates mTOR signaling primarily through inclusion of exon 2 in the BCAA transaminase 2 (BCAT2) mRNA. Pharmacological antagonism of SRSF10 potentiated the therapeutic effect of anti-PD-1 antibody in Tff1-CreERT2; Apcfl/fl; p53fl/fl orthotopic GC models. Collectively, our findings revealed that SRSF10 orchestrates mTOR-CCL2 signaling by alternative RNA splicing of BCAT2 to reprogram TAMs, proposing SRSF10 as a tempting therapeutic target for GC immunotherapy.

Prognostic models for resectable gastric cancer rarely integrate adipose tissue and immune-inflammatory parameters. This study evaluated the predictive value of preoperative relative fat mass (RFM) and the modified systemic inflammation score (mSIS), and explored their potential synergy with nanomedicine-based theranostics. We retrospectively analyzed 518 patients who underwent radical gastrectomy across three tertiary centers in China (overall enrollment period: 2010–2019). Preoperative CT-derived waist circumference at L3 was used to calculate RFM, with the optimal cutoff determined by ROC analysis using the Youden index. Univariate and multivariate Cox regressions identified independent prognostic factors and constructed the nutritional-inflammatory composite parameter RFMmSIS. Predictive performance was assessed using AUC, AIC, BIC, NRI, IDI, and DCA. Postoperative follow-up was conducted every 3 months during the first 2 years, every 6 months during years 3 through 5, and annually thereafter. Patients were randomly allocated to training (n = 362) and validation (n = 156) cohorts. RFM outperformed BMI for survival prediction (AUC: 0.584 vs. 0.503, P < 0.05) and was associated with worse 3-year overall survival (OS) (64.1

Objective: To compare three-year outcomes of camrelizumab plus apatinib combined with chemotherapy (CA-chemo), apatinib combined with chemotherapy (A-chemo), and chemotherapy alone (chemo-alone) as perioperative treatment for locally advanced gastric cancer (LAGC). Methods: A total of 152 patients with LAGC (cT2–4, N+, M0) enrolled between July 2017 and March 2022 from 10 centers were included from two prospective phase II trials. Patients received CA-chemo (n=51), A-chemo (n=48), or chemo-alone (n=53). Endpoints included overall survival (OS), progression-free survival (PFS), recurrence patterns, and safety. Results: In the modified intention-to-treat (mITT) population, the major pathologic response (MPR) rates were 33.3%, 25.0%, and 17.0% in the CA-chemo, A-chemo, and chemo-alone groups, respectively. The 3-year PFS rate was significantly higher with CA-chemo than with A-chemo (64.2% vs 41.7%, P=0.009), with a trend toward improved OS (62.8% vs 43.8%, P=0.052). No statistically significant differences were observed between A-chemo and chemo-alone in 3-year PFS (41.7% vs 48.3%, P=0.288) or OS (43.8% vs 61.2%, P=0.065). Similar trends were observed in the per-protocol (PP) population. Among patients who underwent gastrectomy, the recurrence rate was significantly lower in the CA-chemo group than in the A-chemo group (32.7% vs 57.9%, P=0.033). Recurrence rates were comparable between A-chemo and chemo-alone (49.0%; P>0.05). Across the three groups, adverse events were predominantly grade 1–2, with a low incidence of grade 3–4 events. Conclusion: ​In perioperative treatment of locally advanced gastric cancer, adding camrelizumab to apatinib plus chemotherapy is necessary to prolong PFS and reduce postoperative recurrence.

The causal relationship between proton pump inhibitor (PPI) use and bone health outcomes remains uncertain. This study employs a Mendelian randomization (MR) approach to investigate the potential causal association between PPI use and the risk of osteoporosis and fractures. We selected 4 representative PPIs, including omeprazole, esomeprazole, lansoprazole, and rabeprazole, for our study. Bone health outcomes were evaluated through femoral neck bone mineral density (BMD), lumbar spine BMD, and the prevalence of osteoporosis and fracture across various anatomical sites, including the upper arm and shoulder, wrist and hand, lumbar spine and pelvis, femur, and lower leg and ankle. To evaluate PPI exposure and bone health outcomes, we utilized summary statistics derived from genome-wide association studies conducted in European ancestry populations. Primary causal estimates were derived using the inverse-variance weighting (IVW) approach, supplemented by MR-Egger, weighted median, and Mendelian Randomization Pleiotropy Residual Sum and Outlier methods. To strengthen the robustness of our findings, we conducted sensitivity analyses encompassing assessments of heterogeneity, horizontal pleiotropy, and leave-one-out analyses. Lansoprazole demonstrated a significant positive causal effect on femoral neck BMD (IVW: β = 0.137, 95% confidence interval: 0.063-0.210, P = 2.73E-04), whereas no statistically significant effects were observed for lansoprazole on lumbar spine BMD, osteoporosis, or fracture risk. Esomeprazole showed a marginal causal association with an elevated risk of femur fracture (IVW: odds ratio = 1.049, 95% confidence interval: 1.004-1.096, P = .031); however, this association lost statistical significance following Bonferroni correction. Its effects on BMD, osteoporosis, and fractures at other anatomical sites remained nonsignificant. No causal associations with BMD, osteoporosis, or fracture risk were identified for either omeprazole or rabeprazole. Sensitivity analyses further reinforced the robustness and reliability of these findings. This MR analysis found no compelling evidence to support a causal association between PPI use and the risk of osteoporosis or fractures.

Overcoming resistance to immune checkpoint blockade (ICB) therapy in gastric cancer (GC) remains a major clinical challenge. Here, we apply multi-omics profiling, including single-cell RNA sequencing and spatial transcriptomics, to GC tissues from patients receiving neoadjuvant ICB therapy to identify drivers of resistance. We identify tumor-intrinsic Yes-associated protein 1 (YAP1) as a key regulator of immunosuppressive cellular communities that contribute to ICB non-responsiveness. To mitigate the off-target toxicity of verteporfin, a YAP1 inhibitor, we develop macrophage-membrane-camouflaged hollow mesoporous silica nanoparticles (M@O-VNPs) co-loaded with verteporfin and oxaliplatin. This nanoplatform selectively inhibits YAP1, suppresses the CXCL5-CXCR2 axis, and reduces the activity of SPP1+ macrophages. By inducing immunogenic cell death, M@O-VNPs remodel the tumor microenvironment and enhance ICB efficacy while minimizing systemic toxicity. The therapeutic potential of this strategy is supported by synergistic antitumor effects of M@O-VNPs combined with anti-PD-1 therapy in genetically engineered and syngeneic GC models.

Although immune checkpoint inhibitors benefit a subset of patients with gastric cancer, most tumors exhibit primary resistance characterized by an immune-desert microenvironment. Through integrated multi-omics analyses, we define immune-inflamed and immune-desert phenotypes and identify RASAL2 as a pivotal epithelial-intrinsic regulator associated with the immune-desert state. Mechanistically, RASAL2 binds to and suppresses PPP1CA activity, thereby activating a p38 MAPK–SMAD2/3 signaling axis that promotes cancer stemness and chemoresistance. This pathway concurrently impairs CD8⁺ T-cell recruitment and reinforces immune evasion by repressing the IFN-γ–STAT1–CXCL9/10 chemokine axis and inducing PD-L1 expression through a SMAD2/3–BACH1 transcriptional circuit. Collectively, our findings uncover an epithelial-intrinsic link between cancer stemness and immune evasion and provide a therapeutic rationale for targeting immune-desert gastric cancer.

Background:Robotic total gastrectomy (RTG) has been proposed as a promising alternative to laparoscopic total gastrectomy (LTG) for advanced middle-upper gastric cancer (AMUGC) due to its potential to overcome anatomical challenges. However, the comparative evidence regarding both safety and long-term oncologic efficacy remains limited. Methods:This retrospective cohort study included 1099 patients with AMUGC who underwent RTG/LTG at eight high-volume centers in China between 2015 and 2019. Propensity score matching (PSM = 1:1) was used to balance clinicopathological characteristics between the two groups. The primary outcome was 3-year disease-free survival (DFS); secondary outcomes included 3-year overall survival (OS), 3-year cumulative incidence of recurrence (CIR), recurrence patterns, and operative outcomes. Results:In the PSM cohort, 229 patients were included in each group. RTG demonstrated lower overall postoperative complications (16.2% vs. 27.1%, P < 0.05), medical complications (6.6% vs. 17.0%, P < 0.05), and pneumonia rates (6.6% vs. 16.2%, P < 0.05). No differences were observed in 3-year DFS (73.7% vs. 68.1%, P = 0.230), 3-year OS (76.8% vs. 72.4%, P = 0.326), CIR (24.4% vs. 26.8%, P = 0.580), and recurrence patterns. In stage III, 3-year DFS was higher in the RTG group (67.4% vs. 54.5%, P = 0.047), but no significant differences were found in 3-year OS or CIR (all P > 0.05). Multivariate analysis confirmed that the surgical method was not significantly associated with 3-year DFS [hazard ratio (HR), 0.767; 95% confidence interval (CI), 0.542-1.086; P = 0.135] or 3-year OS (HR, 0.774; 95% CI, 0.534-1.122; P = 0.177). Conclusions:For AMUGC patients not receiving neoadjuvant chemotherapy, RTG demonstrated potential advantages in certain operative outcomes and non-inferior 3-year oncological outcomes compared with LTG. Prospective trials are needed to validate these findings.

BACKGROUND:Spasmolytic polypeptide-expressing metaplasia (SPEM) arises in the gastric corpus in response to oxyntic atrophy, but its cellular origin and role in gastric cancer remain unclear. OBJECTIVE:To define the cellular origin of SPEM in the gastric corpus and its relationship to gastric dysplasia and cancer progression. DESIGN:Tff2-CreERT2 knock-in mice were used for lineage tracing and genetic ablation to characterise Tff2+ corpus progenitor cells. Acute injury, chief cell ablation, H. pylori infection and KrasG12D activation models were applied. Single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics were performed on human gastric tissues to validate differentiation trajectories. RESULTS:Highly proliferative Tff2+ progenitors were localised to the corpus isthmus and generated multiple secretory lineages including chief cells, but lacked long-term self-renewal. Following acute injury or chief cell loss, Tff2+ progenitors rapidly expanded to form transient SPEM. Genetic ablation of Tff2+ progenitors abolished SPEM formation, whereas ablation of Lgr5-DTR- or Gif-rtTA-labelled chief cells enhanced SPEM derived from Tff2+ progenitors. on H. pylori infection or KrasG12D activation, Tff2+ progenitors progressed to SPEM and dysplasia. Kras activation in Tff2+ progenitors promoted direct progression to dysplasia through acquisition of stem cell-like properties. In contrast, Kras-mutant SPEM and chief cells failed to progress to dysplasia. Human scRNA-seq and spatial transcriptomics revealed distinct differentiation trajectories from isthmus proliferating cells to SPEM or gastric cancer. CONCLUSIONS:Tff2+ corpus progenitors represent a common cellular origin for SPEM and gastric dysplasia, challenge the conventional stepwise model of gastric carcinogenesis and indicate divergent differentiation programmes from Tff2+ progenitors.

BACKGROUND:In recent years, the use of laparoscopic techniques in total gastrectomy has gradually increased. Strong evidence is urgently needed to confirm the long-term oncological outcomes of laparoscopic-assisted total gastrectomy for resectable gastric cancer. METHODS:The training set contained 811 gastric cancer patients from Fujian Medical University Union Hospital, and 267 gastric cancer patients in Putian University Affiliated Hospital were used as an external validation set. A minimum P value approach was used to evaluate the optimal threshold of recurrence-free survival to divide the patients into early and late recurrence cohorts based on the length of the postrecurrence survival. RESULTS:The training set had a 5-year actual overall survival rate of 62.5% and a 5-year actual recurrence-free survival rate of 60.3%. Of the 811 patients, 324 (40.0%) experienced recurrence, with the highest percentage of distant metastases (n = 199; 61.4%). Univariate and multivariate Cox regression analyses showed that pT stage, pN stage, nerve/lymphatic invasion, preoperative carcinoembryonic antigen, preoperative immunoinflammatory index (systemic immune-inflammatory index), and postoperative adjuvant chemotherapy were independent prognostic factors associated with overall survival and recurrence-free survival (all P < .05). The optimal cutoff time for distinguishing between early recurrence (n = 113; 34.9%) and late recurrence (n = 211; 65.1%) was the 12th postoperative month (P = .011). Factors associated with early recurrence included pT stage, pN stage, preoperative cancer antigen 72-4, cancer antigen 199, cancer antigen 125, and systemic immune-inflammatory index, based on which we constructed an early recurrence prediction model. The predictive evaluation indices of this model were significantly better than those of the American Joint Committee on Cancer Eighth Edition pTNM staging system; these results were also found in the external validation cohort. CONCLUSION:In this retrospective analysis, we found that laparoscopic-assisted total gastrectomy in patients with resectable gastric cancer had good long-term oncological outcomes, with an optimal cutoff time of 12 months postoperatively for early recurrence and late recurrence after laparoscopic-assisted total gastrectomy. The early recurrence prediction model constructed for patients with laparoscopic-assisted total gastrectomy may inform clinicians to develop more precise treatment and follow-up strategies.

Objective:To achieve accurate prediction of early recurrence (ER) in locally advanced gastric cancer (LAGC) patients after neoadjuvant therapy (NAT) and surgery, we constructed a deep learning fusion model integrating preoperative computed tomography (CT) imaging and perioperative clinicopathological features. Methods:We retrospectively enrolled 611 LAGC patients who received NAT from four tertiary teaching hospitals, including a training cohort (TC), an internal validation cohort (IVC), and an external validation cohort (EVC). ER was defined as recurrence occurring within 2 years post-surgery. Based on preoperative 2.5D CT images, we constructed a deep learning signature (DLS) using a ResNet50 architecture. In parallel, a clinical signature (CLIS) was developed through logistic regression analyses. To further improve predictive performance, a deep learning fusion signature (DLFS) was constructed by integrating the DLS and CLIS. Model performance was evaluated by discrimination, calibration, and clinical utility. Kaplan-Meier analysis was used to evaluate prognostic differences across risk groups. Bulk and single-cell transcriptomic analyses explored biological features. Results:Compared with the DLS and CLIS, the DLFS demonstrated superior performance in predicting ER, with area under the curve (AUC) values of 0.884 in the TC, 0.828 in the IVC, and 0.748 in the EVC. Calibration curves exhibited good agreement, and decision curve analysis indicated a higher net benefit. Risk stratification based on DLFS showed worse overall survival (OS) in the high-risk group (3-year OS: TC, 35.36% vs. 77.99%; IVC, 22.90% vs. 73.88%; EVC, 36.01% vs. 63.35%; all P<0.001). Moreover, proliferation-related pathways were enriched in the high-DLFS subgroup, accompanied by an increased proportion of malignant epithelial cells in single-cell analysis. Conclusions:The DLFS model, integrating preoperative CT imaging and perioperative clinicopathological variables, effectively predicts ER and survival in LAGC patients after NAT. Thus, it can serve as a useful tool to optimize prognostic monitoring.