CT-based delta-radiomics nomogram to assess tumor regression grade in locally advanced gastric cancer patients following neoadjuvant chemotherapy

CT-based delta-radiomics nomogram to assess tumor regression grade in locally advanced gastric cancer patients following neoadjuvant chemotherapy | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article CT-based delta-radiomics nomogram to assess tumor regression grade in locally advanced gastric cancer patients following neoadjuvant chemotherapy Shulan Chen, Zhe Xiao, Jiayi Jin, Hanzhe Wang, Shouliang Miao, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7072201/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 17 Sep, 2025 Read the published version in Abdominal Radiology → Version 1 posted 7 You are reading this latest preprint version Abstract Purpose To develop and validate a delta computed tomography radiomics (delCT-RS) based nomogram for accurate preoperative prediction of tumor regression grade (TRG) in locally advanced gastric cancer (LAGC) patients following neoadjuvant chemotherapy (NAC). Methods This retrospective study enrolled 147 LAGC patients. Two delineation strategies were compared: 1) contouring both the primary tumor and the largest lymph node (P + L) as regions of interest (ROIs), and 2) contouring only the primary tumor (P). Subsequently, radiomic features were extracted to construct corresponding radiomic models. This study compared the predictive accuracy of delCT-RS signatures to conventional single-phase radiomic signatures for TRG assessment. Then, delCT-RS signatures and clinical variables were combined into a nomogram. Finally, the prediction performance of nomogram was comprehensively evaluated. Results In assessing tumor response, delCT-RS outperformed single-phase radiomic signatures. Notably, delta computed tomography delCT-RS P + L demonstrated superior accuracy to delCT-RS P (delCT-RS P + L vs delCT-RS area under the curve (AUC): training cohort: 0.805 vs 0.727; validation cohort: 0.795 vs 0.655). The nomogram, combining delCT-RS P + L and clinical factors, achieved optimal performance among all models (training cohort AUC = 0.841; validation cohort AUC = 0.817). (p < 0.05) Conclusion In this study, we innovatively employed a method that simultaneously delineated the primary tumor and the largest lymph node. This model can accurately predict TRG, effectively identify LAGC patients who can benefit from NAC, and provide scientific support for individualized treatment. Locally advanced gastric cancer Neoadjuvant chemotherapy Delta computed tomography radiomics Tumor regression grade Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Gastric cancer (GC) ranks as the fifth most common malignant tumor and the fifth leading cause of cancer-related mortality globally [ 1 ]. Although surgical resection remains the primary treatment modality, nonspecific clinical manifestations frequently delay diagnosis until the locally advanced gastric cancer (LAGC) stage, which is associated with poor prognoses. The 5-year overall survival rate following radical gastrectomy persists at 30–40% [ 2 ]. In recent years, neoadjuvant chemotherapy (NAC) has been recommended by the National Comprehensive Cancer Network (NCCN) as the preferred preoperative treatment for LAGC patients undergoing curative resection, based on its demonstrated clinical benefits: tumor volume reduction, clinical stage downstaging, improved R0 resection rates, elimination of potential micrometastases, and reduced risks of postoperative recurrence and metastasis [ 3 – 5 ]. However, due to tumor heterogeneity, not all LAGC patients derive therapeutic benefit from standardized NAC regimens, with some experiencing treatment-related toxicity accumulation and potential disease progression during therapy [ 6 , 7 ]. Consequently, the development of reliable early prediction methodologies is crucial for achieving precision medicine in LAGC management. Tumor regression grade (TRG) serves as a pivotal biomarker for assessing therapeutic response and prognostic outcomes in oncology. Its principal value lies in quantifying tumor regression through histopathological analysis of surgical specimens [ 8 ]. However, TRG demonstrates notable clinical limitations: its evaluation relies exclusively on postoperative tissue samples, precluding dynamic monitoring of individual responses during early treatment phases, thereby constraining its temporal relevance in guiding personalized therapeutic strategies. Consequently, the development of non-invasive predictive tools capable of early-phase treatment assessment holds significant clinical implications for optimizing NAC regimens and related therapeutic approaches. Currently, the Response Evaluation Criteria in Solid Tumors (RECIST) and World Health Organization (WHO) criteria are widely employed for evaluating NAC responses, primarily based on bidimensional measurements of target lesion dimensions. Although these criteria demonstrate high operational feasibility in clinical practice, their principal limitation resides in focusing solely on volumetric tumor changes while disregarding tumor heterogeneity [ 9 ]. Emerging evidence indicates that tumor heterogeneity not only constitutes a critical biological characteristic but may also directly influence therapeutic responsiveness and survival outcomes [ 10 ]. Radiomics, an emerging interdisciplinary domain in precision medicine, has attracted considerable attention in recent years by extracting quantitative features from medical images through high-throughput and transforming image data into high-dimensional features suitable for mining to reveal the micro-biological characteristics of tumors [ 11 ]. Clinical investigations in malignancies including colorectal cancer [ 12 , 13 ], breast cancer [ 14 ], and hepatocellular carcinoma [ 15 ] have demonstrated the marked superiority of radiomics in tumor phenotyping, validating its potential for oncological evaluation and clinical decision-making. Among them, delta radiomics demonstrates more dynamic evolution information compared to conventional single-phase (pre- or post-treatment) radiomics. In this study, we propose the following scientific hypothesis:delta computed tomography radiomics (delCT-RS) can dynamically quantify temporal changes in lesion biology during therapeutic interventions or natural disease progression. These radiomic alterations may manifest earlier than detectable changes in conventional imaging modalities or clinical symptom onset, thereby providing a basis for early assessment of tumor regression following NAC. In current research, the analysis of imaging features based on primary tumor lesions has become a crucial approach for evaluating pathological response after NAC in LAGC [ 16 , 17 ]. And some studies have utilized approaches involving either combined delineation of primary tumors and lymph nodes or exclusive delineation of primary tumors to predict post-treatment lymph node metastasis in cancer patients [ 18 , 19 ]. This study innovatively employed a method combining slice-by-slice delineation of primary tumors with maximum cross-sectional delineation of the largest lymph node as regions of interest (ROIs), thereby establishing a corresponding radiomics model to predict tumor regression in LAGC patients following NAC treatment. In recent years, nomograms integrating radiomics with clinical risk factors have been been widely used to evaluate treatment responses and predict tumor prognosis across diverse clinical scenarios [ 20 , 21 ]. Previous studies have demonstrated that such multimodal integration strategies can significantly enhance the precision of risk stratification, thereby providing quantitative foundations for personalized therapeutic decision-making [ 22 ]. However, there is still a lack of an efficient and simple preoperative scoring tool in current clinical practice to accurately assess the efficacy of NAC and effectively predict the potential benefits of patients. The primary objective of this study is to develop and validate a delCT-RS-based nomogram for achieving personalized quantitative assessment of tumor regression in LAGC patients, enabling stratification into high-response and low-response subgroups, thereby providing clinicians with decision basis for dynamic adjustment of treatment plans. Methods Patients This retrospective study has been formally approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University, with a waiver granted for the informed consent procedure. The entire research process has strictly adhered to the ethical guidelines for medical research established by the Declaration of Helsinki and its subsequent amendments. This study included 193 patients with LAGC (cT2-4NxM0) who underwent NAC followed by radical gastrectomy at our institution between January 2018 and October 2024. Detailed information on the enrollment procedure is displayed in Supplementary Appendix 1 and Fig. 1 . Ultimately, 147 eligible patients were divided into a training cohort (TC) (n = 104) and a validation cohort (VC) (n = 43). And Fig. 2 shows the workflow of this study. Neoadjuvant chemotherapy regimens and clinicopathological data collection According to gastric cancer treatment guidelines, all patients received at least two cycles of standardized NAC regimen prior to radical gastrectomy (Supplementary Appendix 2). This retrospective study analyzed the patients’ clinical and pathological characteristics, including gender, age, body mass index (BMI), Borrmann classification, tumor differentiation grade, tumor location, serum levels of carcinoembryonic antigen (CEA), carbohydrate antigen 199 (CA199), and alpha-fetoprotein (AFP). Additionally, based on the American Joint Committee on Cancer (AJCC) TNM Classification (8th edition), the clinical T (cT) and clinical N (cN) stages were recorded from the medical records. Tumor regression grade After neoadjuvant chemotherapy, all patients underwent gastrectomy. And two senior pathologists conducted a standardized evaluation of the surgical pathology specimens independently while blinded to clinical or imaging data. The tumor pathological response was evaluated by TRG after NAC. TRG scores were evaluated using Ryan criteria [ 23 ] (Supplementary Appendix 3). The study cohort was categorized into two groups based on TRG: TRG 0–1: good response (GR) to NAC ; TRG 2–3: poor response (PR) to NAC (Fig. 3 ). Radiomics feature extraction and reproducibility assessment Specific parameters of CT scanning are shown in Supplementary Appendix 4 and Table S1 . Radiologist 1 manually contoured the ROI on tumor slices using 3D slicer software Version5.8.0 ( http://www.slicer.org ), slice-by-slice along the tumor margin, excluding the first and last layers to avoid partial volume effects. In addition, for the largest metastatic lymph node, precise contour mapping was performed simultaneously on the anatomical section with the largest transverse diameter. For comparison, another delineation method was used in this study, where only the primary tumor (delineated layer by layer, with the first and last layers excluded) was delineated as the region of interest. Radiomic feature extraction was conducted via the open-source Pyradiomics package Version3.1.0 ( https://pyradiomics.readthedocs.io/en/3.1.0/ ), with a total of 851 radiomic features extracted from each ROI (Supplementary Appendix 5). All features complied with the standards of the Image Biomarker Standardization Initiative (IBSI). To ensure feature reproducibility, after one month, 20 cases were randomly selected for repeat segmentation by Radiologists 1 and 2, who were blinded to clinical and pathological information. Inter- and intra-observer consistency was assessed using the intraclass correlation coefficient (ICC), with features showing ICC > 0.80 considered robust and included in subsequent analyses. The change in radiomic features (delCT-RS) was calculated as the difference between post-neoadjuvant chemotherapy computed tomography radiomics (postCT-RS) and pre-neoadjuvant chemotherapy computed tomography radiomics (preCT-RS): delCT-RS = postCT-RS - preCT-RS Radiomics feature selection and radiomics models building In this study, radiomic models were developed and validated using Minimum Redundancy - Maximum Relevance (mRMR) and Least Absolute Shrinkage and Selection Operator (LASSO) feature selection methods. Initially, mRMR identified 20 radiomic features highly correlated with target variables and with low redundancy. Then, LASSO regression with 10-fold cross-validation determined the regularization parameter λ to optimize feature selection, implemented via the R package glmnet. Ultimately, robust radiomic features strongly associated with treatment response were selected to build the prediction model. (Figure S1 and Table S2) Clinical model and nomogram building First, univariate and multivariate logistic regression analyses were performed to evaluate all clinical and pathological factors in the training cohort, calculating the odds ratios (OR) and 95% confidence interval (CI) for each factor. Independent clinical predictors were selected based on statistical significance, and a clinical prediction model was constructed accordingly. Subsequently radiomic signature was combined with the above independent clinical predictors and then a comprehensive nomogram was constructed. The entire modeling process adhered to standardized clinical prediction model development procedures, guaranteeing the model's scientific rigor and interpretability. Performance evaluation The predictive performance of each model was first evaluated using Receiver operating characteristic (ROC) curves. Area under the curve (AUC) was calculated and compared across different cohorts using the DeLong test. In addition, key metrics including accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were recorded for each model. To further assess clinical applicability, calibration curves were plotted to evaluate the consistency between radiomic labels and actual observations, and the Hosmer-Lemeshow test was used to assess model fit. Additionally, decision curve analysis (DCA) was applied to evaluate the clinical net benefit of the radiomic model in identifying patients likely to benefit from NAC, using data from both the training and validation cohorts. Statistical analysis SPSS statistical software version 27.0 (IBM) and R software version 4.4.2 ( http://www.r-project.org ) were used for analysis (Supplementary Appendix 6). A two-sided p < 0.05 was considered statistically significant. Notably, In the univariate logistic regression analysis, a lenient screening criterion of p < 0.20 was used to avoid missing clinically relevant factors. Results Clinicopathological characteristics The baseline clinicopathological characteristics of all 147 LAGC patients are summarized in Table 1. As shown in Table 1, the GR rates were comparable between TC and VC, with proportions of 36.5% and 34.6% respectively. In both cohorts, no statistically significant differences exist between the GR and PR groups in terms of demographic parameters (sex, age, and BMI) or tumor-related features (cN_stage, Borrmann type, differentiation, and tumor location). Similarly, no significant differences are observed in tumor markers (pre-neoadjucant chemotherapy (pre-NAC) CEA, pre-NAC CA199, pre-NAC AFP, post-neoadjuvant chemotherapy (post-NAC) CA199, and post-NAC AFP). Notably, cT_stage demonstrated significant intergroup differences in both TC and VC ( p < 0.05), whereas post-NAC CEA showed significant differences only in VC ( p < 0.05). Table 1 Clinicopathological characteristics of LAGC patients in the training and validation cohorts Variables Training cohort (n = 104) p Validation cohort (n = 43) p GR (n = 38) PR (n = 66) GR (n = 15) PR (n = 28) Sex (%) 0.374 1.000 Female 27 (71.1%) 52 (78.8%) 12 (80.0%) 22 (78.6%) Male 11 (28.9%) 14 (21.2%) 3 (20.0%) 6 (21.4%) Age (year, %) 0.103 0.964 ≤ 65 15 (39.5%) 37 (56.1%) 6 (40.0%) 11 (39.3%) > 65 23 (60.5%) 29 (43.9%) 9 (60.0%) 17 (60.7%) BMI (kg/m 2 , %) 0.429 1.000 < 24 23 (60.5%) 45 (68.2%) 11 (73.3%) 20 (71.4%) ≥ 24 15 (39.5%) 21 (31.8%) 4 (26.7%) 8 (28.6%) cT_stage (%) < 0.001*** 0.033* T2་T3 32 (84.2%) 32 (48.5%) 12 (80.0%) 13 (46.4%) T4 6 (15.8%) 34 (51.5%) 3 (20.0%) 15 (53.6%) cN_stage (%) 0.468 0.349 N0 0 (0.0%) 3 (4.5%) 1(6.7%) 0 (0.0%) N+ 38 (100.0%) 63 (95.5%) 14 (93.3%) 28 (100.0%) Borrmann (%) 0.559 0.317 Ⅰ 0 (0.0%) 3 (4.5%) 0 (0.0%) 3 (10.7%) Ⅱ 10 (26.3%) 14 (21.2%) 7 (46.7%) 7 (25.0%) Ⅲ 19 (50.0%) 43 (65.2%) 6 (40.0%) 14 (50.0%) Ⅳ 4 (10.5%) 6 (9.1%) 1 (6.7%) 4 (14.3%) Unknown 5 (13.2%) 0 (0.0%) 1(6.7%) 0 (0.0%) Differentiation (%) 0.131 0.282 Well 1(2.6%) 3 (4.5%) 0 (0.0%) 1 (3.6%) Moderately 15 (39.5%) 14 (21.2%) 5 (33.%) 4 (14.3%) Poorly 12 (31.6%) 47 (71.2%) 7 (46.7%) 23 (82.1%) Unknown 10 (26.3%) 2 (3.0%) 3 (20.0%) 0 (0.0%) Location (%) 0.833 0.920 Upper 1/3 5 (13.2%) 13 (19.7%) 3 (20.0%) 4 (14.3%) Middle1/3 14 (36.8%) 23 (34.8%) 4 (26.7%) 10 (35.7%) Lower 1/3 16 (42.1%) 24 (36.4%) 6 (40.0%) 10 (35.7%) Whole stomach 3 (7.9%) 6 (9.1%) 2 (13.3%) 4 (14.3%) Pre-NAC CEA (ng/mL, %) 0.797 0.454 ≤ 5 24 (63.2%) 40 (60.6%) 13 (86.7%) 20 (71.4%) > 5 14 (36.8%) 26 (39.4%) 2 (13.3%) 8 (28.6%) Pre-NAC CA199 (U/mL, %) 0.911 1.000 ≤ 37 29 (76.3%) 52 (77.3%) 13 (86.7%) 23 (82.1%) > 37 9 (23.7%) 15 (22.7%) 2 (13.3%) 5 (17.9%) Pre-NAC AFP (ng/mL, %) 0.281 0.080 ≤ 8.78 31 (81.6%) 60 (90.9%) 11 (73.3%) 27 (96.4%) > 8.78 7 (18.4%) 6 (9.1%) 4 (26.7%) 1 (3.6%) Post-NAC CEA (ng/mL, %) 0.892 0.001** ≤ 5 26 (68.4%) 46(69.7%) 14 (93.3%) 12 (42.9%) > 5 12 (31.6%) 20 (30.3%) 1 (6.7%) 16 (57.1%) Post-NAC CA199 (U/mL, %) 0.150 0.092 ≤ 37 35 (92.1%) 54 (81.8%) 15 (100.0%) 21 (75.0%) > 37 3 (7.9%) 12 (18.2%) 0 (0.0%) 7 (25.0%) Post-NAC AFP (ng/mL, %) 0.945 1.000 ≤ 8.78 33 (86.8%) 57 (86.4%) 14 (93.3%) 26 (92.9%) > 8.78 5 (13.2%) 9 (13.6%) 1 (6.7%) 2 (7.1%) Note: Chi-squared or Fisher's exact tests, were used to compare the differences in categorical variables, whereas student t or Mann-Whitney U test was used to compare the differences in continuous variables, as appropriate. * p < 0.05; ** p < 0.01; *** p < 0.001. LAGC, locally advanced gastric cancer; GR, good response; PR,poor response; BMI, body mass index; cT_stage, clinical T stage; cN_stage, clinical N stage; CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; AFP, alpha-fetoprotein; NAC, neoadjuvant chemotherapy Correlation between the TRG and clinicopathological characteristics In the univariate logistic regression analysis, this study adopted p < 0.20 as the variable selection criterion to minimize the risk of omitting potentially clinically relevant factors [24]. The ultimately selected variables identified as associated with TRG included age、cT_stage、pre-NAC AFP and post-NAC CA199, all of which were subsequently incorporated into the multivariate logistic regression analysis. Through multivariate logistic regression analysis, cT_stage was ultimately identified as a statistically significant independent predictive factor ( p < 0.05). Building upon this key predictor, a corresponding clinical prediction model was established, with detailed information presented in Table 2. Radiomics feature selection and comparison of radiomics models and clinical model As shown in Fig. 4, AUC analysis of the radiomics models revealed that the delCT-RS established using both delineation approaches demonstrated superior predictive performance in evaluating tumor regression following NAC treatment compared to single time-point models based solely on pre-therapy or post-therapy radiomics signatures: delCT-RS P + L (AUC, 0.805; 95% CI 0.721–0.888)、preCT-RS P + L (AUC, 0.778; 95% CI 0.687–0.868)、 postCT-RS P + L (AUC, 0.706; 95% CI 0.606–0.805); delCT-RS P (AUC, 0.727; 95% CI 0.627–0.828)、preCT-RS P (AUC, 0.714; 95% CI 0.610–0.817)、postCT-RS P (AUC, 0.631; 95% CI 0.525–0.737). Further validation in the validation cohort confirmed these results: delCT-RS P + L (AUC, 0.795; 95% CI 0. 658–0. 932)、preCT-RS P + L (AUC,0.581; 95% CI 0.400–0.762)、postCT-RS P + L (AUC, 0.721;95% CI 0.547–0.895); delCT-RS P (AUC, 0.655; 95% CI 0. 490–0. 820)、 preCT-RS P (AUC,0.638; 95% CI 0.473–0.804)、postCT-RS P (AUC, 0.624; 95% CI 0.452–0.795). (p < 0.05) In addition, a comparative analysis was conducted between the conventional approach involving sole delineation of primary tumors and our innovative method combining primary tumor delineation with maximum lymph node delineation. It is evident that the radiomics models established by including the largest lymph node as the region of interest demonstrated superior predictive performance. Notably, delCT-RS P + L not only outperformed the other six radiomics models but also demonstrated more robust performance compared to the clinical model (TC: AUC, 0.697; 95% CI 0.598–0.763. VC: AUC, 0.668; 95% CI 0.527–0.809.), demonstrating comprehensive superiority across all evaluation metrics. In all cohorts listed in Table 3, delCT-RS outperformed the other models in terms of accuracy, sensitivity, specificity, PPV, and NPV. (DeLong test, all p < 0.05) Table 3 The performance of models in predicting the response of LAGC after NAC Models Accuracy Sensitivity Specificity PPV NPV (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) PreCT-RS P + L Training 0.692 0.895 0.576 0.548 0.905 (0.598–0.773) (0.759–0.958) (0.456–0.688) (0.425–0.666) (0.779–0.962) Validation 0.558 0.600 0.536 0.409 0.714 (0.411–0.696) (0.357–0.802) (0.358–0.705) (0.233–0.613) (0.500-0.862) PreCT-RS P Training 0.750 0.421 0.939 0.800 0.738 (0.659–0.823) (0.279–0.578) (0.854–0.976) (0.584–0.919) (0.635–0.820) Validation 0.628 0.267 0.821 0.444 0.676 (0.479–0.756) (0.109–0.520) (0.644–0.921) (0.189–0.733) (0.508–0.809) PostCT-RS P + L Training 0.663 0.816 0.576 0.525 0.844 (0.568–0.747) (0.666–0.908) (0.456–0.688) (0.400-0.647) (0.712–0.923) Validation 0.558 0.667 0.500 0.417 0.737 (0.411–0.696) (0.417–0.848) (0.326–0.674) (0.245–0.612) (0.512–0.882) PostCT-RS P Training 0.606 0.789 0.500 0.476 0.805 (0.510–0.694) (0.637–0.889) (0.383–0.617) (0.358–0.597) (0.660–0.898) Validation 0.581 0.733 0.500 0.440 0.778 (0.433–0.716) (0.480–0.891) (0.326–0.674) (0.267–0.629) (0.548-0,910) DelCT-RS P + L Training 0.740 0.921 0.636 0.593 0.933 (0.649–0.815) (0.792–0.973) (0.516–0.742) (0.466–0.709) (0.821–0.977) Validation 0.674 0.933 0.536 0.519 0.938 (0.525–0.795) (0.702–0.988) (0.358–0.705) (0.340–0.693) (0.717–0.989) DelCT-RS P Training 0.673 0.763 0.621 0.537 0.820 (0.578–0.756) (0.608–0.870) (0.501–0.729) (0.406–0.663) (0.692–0.902) Validation 0.581 0.733 0.500 0.440 0.778 (0.433–0.716) (0.480–0.891) (0.326–0.674) (0.267–0.629) (0.548–0.910) Clinical Training 0.635 0.842 0.515 0.500 0.850 (0.539–0.721) (0.696–0.926) (0.397–0.632) (0.381–0.619) (0.709–0.929) Validation 0.628 0.800 0.536 0.480 0.833 (0.479–0.756) (0.548–0.930) (0.358–0.705) (0.300-0.665) (0.608–0.942) RS-CN Training 0.827 0.842 0.818 0.727 0.900 (0.743–0.888) (0.696–0.926) (0.709–0.893) (0.582–0.837) (0.799–0.953) Validation 0.744 0.800 0.714 0.600 0.870 (0.598–0.851) (0.548–0.930) (0.529–0.847) (0.387–0.781) (0.679–0.955) Note: LAGC, locally advanced gastric cancer; NAC, neoadjuvant chemotherapy; CI, confidence interval; P + L, primary tumor and the largest lymph node; P, primary; preCT-RS, pre-neoadjuvant chemothrapy computed tomography radiomics signatures; postCT-RS, post-neoadjuvant chemothrapy computed tomography radiomics signatures; delCT-RS, delta computed tomography radiomics signatures; RS-CN, radiomic signatures-clinical-nomogram Nomogram model performance In subsequent research, we integrated delCT-RS P + L with the clinical model to form a combined model (Figure S2). As indicated in Fig. 4, Fig. 5A-B, and Table 3, this integrated model demonstrated stable superiority in both cohorts. It achieved an AUC of 0.841 (95% CI: 0.763–0.919) in TC and an AUC of 0.817 (95% CI: 0.691–0.942) in VC, significantly outperforming each single-source model (TC: accuracy: 0.827, sensitivity: 0.842, specificity: 0.818, PPV: 0.727, NPV: 0.900; VC: accuracy: 0.744, sensitivity: 0.800, and specificity: 0.714, PPV: 0.600; NPV: 0.870). Further analysis revealed that the radiomic score exhibited significant differences between the GR and PR groups in all cohorts (all p < 0.05) (Figure S3). Moreover, the predictive scores from both the clinical and radiomics models demonstrated a consistent trend, with higher score gradients positively correlating with improved NAC treatment response. Additionally, according to the TRG criteria, the radiomic model's predictions of LAGC patients' responses to NAC treatment showed a high degree of consistency with the tumor pathology assessment results in both the training and validation sets. Furthermore, the calibration of the integrated model across different cohorts was evaluated. The calibration curves indicated a good fit between predicted and observed probabilities (Hosmer-Lemeshow test p > 0.05), suggesting reliable model calibration (Fig. 5C). DCA further validated that the combined model provided substantial net clinical benefit across a wide threshold probability range (0.2–0.7) for predicting NAC treatment response in all cohorts (Fig. 5D). Discussion NAC has demonstrated significant interindividual variability in clinical efficacy as a conventional treatment for LAGC patients. Currently, TRG is universally recognized as the gold standard for evaluating neoadjuvant therapeutic outcomes, and its prognostic significance in LAGC patients has been well validated [ 8 ]. As a prognostic predictor, TRG serves as a valuable reference for assessing potential therapeutic effects of NAC and facilitating individualized prognosis evaluation in LAGC patients [ 25 , 26 ]. However, current TRG assessment systems exhibit notable limitations: they rely on postoperative pathological specimen analysis, which prevents real-time acquisition of dynamic monitoring data during treatment. This temporal lag in evaluation substantially restricts their applicability in timely therapeutic regimen adjustments. Consequently, developing novel assessment systems capable of noninvasive TRG prediction at early treatment stages represents an urgent clinical need for enhancing precision therapy with NAC and optimizing comprehensive management strategies. This study incorporated relevant clinicopathological indicators into the analysis. In the training cohort, cT_stage demonstrated significant correlation between the GR and PR groups ( p = 0.001), with its stage progression exhibiting a positive correlation trend with TRG. This observation suggests that the degree of local tumor infiltration may influence NAC efficacy, a finding consistent with previous research conclusions [ 27 ]. Consequently, cT_stage was integrated into our clinical model. Although existing literature reports associations between Lauren classification, tumor differentiation grade, cN_stage and pathological response [ 28 , 29 ], these conclusions remain controversial, potentially due to insufficient support from large-scale samples or multicenter studies. This study systematically integrates cutting-edge advancements and innovative applications of radiomics technology. Unlike conventional imaging evaluations that focus on macroscopic morphological features, radiomics provides novel insights into therapeutic response assessment by extracting deep-texture information from medical images, garnering increasing attention. Furthermore, by utilizing postoperative pathological findings as reference standards, radiomics establishes a critical bridge between imaging and histopathology, enabling precise identification of potential beneficiaries of NAC. Published evidence indicates that pre-NAC CT-based radiomic models effectively predict NAC efficacy in LAGC patients [ 30 , 31 ]. For instance, Song et al [ 17 ] extracted radiomic features from portal venous-phase CT images of 279 GC patients prior to NAC and constructed an radiomic model. This model demonstrated exceptional predictive performance, achieving an AUC of 0.790 in the training set, with corresponding internal and external validation cohort AUCs of 0.784 and 0.803, respectively. Recent advances in delta-radiomics—which quantifies tumor heterogeneity changes by analyzing texture features from pre- and post-treatment images—have shown superior accuracy over single-timepoint radiomics in predicting neoadjuvant therapeutic responses. In predicting major pathological response (MPR) following neoadjuvant chemoimmunotherapy for non-small cell lung cancer (NSCLC) [ 32 ], delta-radiomics models achieved AUCs of 0.768, 0.732, 0.833, and 0.716 in the training, test, and two external validation cohorts, significantly outperforming pretreatment radiomics models (AUCs: 0.644, 0.616, 0.475, and 0.608). Similarly, in preoperative assessment of high-grade osteosarcoma NAC response [ 33 ], delta-radiomics signatures exhibited higher AUCs than single-phase CT-based radiomics in both training and validation sets. However, limited research has been conducted in the field of delta radiomics to evaluate the tumor regression following neoadjuvant chemotherapy for gastric cancer. Our study specifically compared delCT-RS with preCT-RS and postCT-RS. The results revealed that delCT-RS demonstrated better evaluation efficacy in assessing tumor regression following NAC treatment, with its AUC values in both the training and validation cohorts exceeding those of the other two single time-point radiomics models. Notably, prior studies predominantly focused on delineating primary tumors in LAGC patients while neglecting lymph node analysis [ 16 ]. Innovatively, our study implemented slice-by-slice delineation of primary tumors alongside maximum cross-sectional delineation of the largest lymph node. Compared to the conventional approach centered solely on the primary tumor, this integrative mapping strategy has shown better predictive power. It has enhanced the accuracy of assessing tumor regression in LAGC patients following NAC. This approach comprehensively captures tumor infiltration depth, spatial extent, and morphological evolution within gastric walls and adjacent tissues, while lymph node delineation facilitates tracking of dimensional, densitometric, and morphological alterations. Effective chemotherapy typically reduces primary tumor volume with sharper boundaries and lower density, accompanied by concurrent decreases in lymph node size and density. Through comparative delineation, the dynamic changes of tumor and lymph nodes can be accurately quantified, so as to comprehensively and accurately evaluate the regression of tumor and examine the efficacy of chemotherapy on tumor. Subsequently, this study developed a combined model integrating delCT-RS P + L with clinical indicators (cT_stage), which demonstrated superior predictive performance among all evaluated models. In TC, the model achieved an AUC of 0.841 (95% CI: 0.763–0.919), and in VC, the model achieved an AUC of 0.817 (95% CI: 0.691–0.942). The calibration curve and DCA further demonstrated that the combined model provided substantial net clinical benefit for predicting neoadjuvant chemotherapy response over a wide threshold probability range (0.2–0.7), increasing the clinical utility of radiomics in advising patients on whether they should receive NAC treatment. This study has several limitations: 1) As a retrospective single-center investigation, it may be susceptible to confounding factors, thereby limiting the generalizability of the findings. 2) The ROI delineation process inherently carries subjective interpretation, which may introduce measurement bias. 3) Heterogeneity in NAC regimens could introduce outcome bias. Conclusions In summary, this study demonstrates that delta radiomics—a methodology quantifying tumor heterogeneity changes through pre- and post-treatment image analysis—provides more reliable predictive capability for tumor regression assessment compared to conventional single-timepoint radiomics. Furthermore, we innovatively developed and validated a nomogram model integrating delCT-RS with clinical predictors, utilizing a novel methodology that combines simultaneous delineation of primary tumors and the largest lymph nodes. This tool effectively stratifies patients into high-response and low-response subgroups, thereby offering clinicians actionable evidence for optimizing therapeutic strategies during treatment adaptation. Abbreviations DelCT-RS Delta computed tomography radiomics TRG Tumor regression grade LAGC Locally advanced gastric cancer NAC Neoadjuvant chemotherapy P+L Primary tumor and the largest lymph node P Primary tumor ROIs Regions of interest AUC Area under the curve GC Gastric cancer NCCN National Comprehensive Cancer Network RECIST Response Evaluation Criteria in Solid Tumors WHO World Health Organization TC Training cohort VC Validation cohort BMI Body mass index CEA Carcinoembryonic antigen CA199 Carbohydrate antigen 199 AFP Alpha-fetoprotein AJCC American Joint Committee on Cancer cT Clinical T cN Clinical N GR Good response PR Poor response IBSI Image Biomarker Standardization Initiative ICC Intraclass correlation coefficient PostCT-RS Post-neoadjuvant chemotherapy computed tomography radiomics PreCT-RS Pre-neoadjuvant chemotherapy computed tomography radiomics mRMR Minimum Redundancy - Maximum Relevance LASSO Least Absolute Shrinkage and Selection OperatorOR odds ratios CI Confidence interval ROC Receiver operating characteristic PPV Positive predictive value NPV Negative predictive value DCA Decision curve analysis Pre-NAC Pre-neoadjuvant chemotherapy Post-NAC Post-neoadjuvant chemotherapy RS-CN Radiomic signatures-clinical-nomogram MPR Major pathological response NACLC Non-small cell lung cancer Declarations Author Contribution Shulan. and Zhe. wrote the main manuscript text , Jiayi. Han. Shouliang. prepared tables, Kun. Yimin. and Linfeng. prepared figures. All authors reviewed the manuscript. References Bray F, Laversanne M, Sung H, et al (2024) Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA A Cancer J Clinicians 74:229–263. https://doi.org/10.3322/caac.21834 Machlowska J, Baj J, Sitarz M, et al (2020) Gastric cancer: Epidemiology, risk factors, classification, genomic characteristics and treatment strategies. IJMS 21:4012. https://doi.org/10.3390/ijms21114012 Al-Batran S-E, Homann N, Pauligk C, et al (2019) Perioperative chemotherapy with fluorouracil plus leucovorin, oxaliplatin, and docetaxel versus fluorouracil or capecitabine plus cisplatin and epirubicin for locally advanced, resectable gastric or gastro-oesophageal junction adenocarcinoma (FLOT4): A randomised, phase 2/3 trial. The Lancet 393:1948–1957. https://doi.org/10.1016/S0140-6736(18)32557-1 Al-Batran S-E, Hofheinz RD, Pauligk C, et al (2016) Histopathological regression after neoadjuvant docetaxel, oxaliplatin, fluorouracil, and leucovorin versus epirubicin, cisplatin, and fluorouracil or capecitabine in patients with resectable gastric or gastro-oesophageal junction adenocarcinoma (FLOT4-AIO): Results from the phase 2 part of a multicentre, open-label, randomised phase 2/3 trial. The Lancet Oncology 17:1697–1708. https://doi.org/10.1016/S1470-2045(16)30531-9 Wang X-Z, Zeng Z-Y, Ye X, et al (2020) Interpretation of the development of neoadjuvant therapy for gastric cancer based on the vicissitudes of the NCCN guidelines. WJGO 12:37–53. https://doi.org/10.4251/wjgo.v12.i1.37 Li Z, Gao X, Peng X, et al (2020) Multi-omics characterization of molecular features of gastric cancer correlated with response to neoadjuvant chemotherapy. Sci Adv 6:eaay4211. https://doi.org/10.1126/sciadv.aay4211 Zhang X, Liang H, Li Z, et al (2025) Perioperative or postoperative adjuvant oxaliplatin with S-1 versus adjuvant oxaliplatin with capecitabine in patients with locally advanced gastric or gastro-oesophageal junction adenocarcinoma undergoing D2 gastrectomy (RESOLVE): Final report of a randomised, open-label, phase 3 trial. The Lancet Oncology 26:312–319. https://doi.org/10.1016/S1470-2045(24)00676-4 Tong Y, Zhu Y, Zhao Y, et al (2021) Evaluation and comparison of predictive value of tumor regression grades according to mandard and becker in locally advanced gastric adenocarcinoma. Cancer Res Treat 53:112–122. https://doi.org/10.4143/crt.2020.516 Eisenhauer EA, Therasse P, Bogaerts J, et al (2009) New response evaluation criteria in solid tumours: Revised RECIST guideline (version 1.1). European Journal of Cancer 45:228–247. https://doi.org/10.1016/j.ejca.2008.10.026 Sicklick JK, Kato S, Okamura R, et al (2019) Molecular profiling of cancer patients enables personalized combination therapy: The I-PREDICT study. Nat Med 25:744–750. https://doi.org/10.1038/s41591-019-0407-5 Mu W, Schabath MB, Gillies RJ (2022) Images are data: Challenges and opportunities in the clinical translation of radiomics. Cancer Research 82:2066–2068. https://doi.org/10.1158/0008-5472.CAN-22-1183 Shin J, Seo N, Baek S-E, et al (2022) MRI radiomics model predicts pathologic complete response of rectal cancer following chemoradiotherapy. Radiology 303:351–358. https://doi.org/10.1148/radiol.211986 Abbaspour E, Karimzadhagh S, Monsef A, et al (2024) Application of radiomics for preoperative prediction of lymph node metastasis in colorectal cancer: A systematic review and meta-analysis. International Journal of Surgery 110:3795–3813. https://doi.org/10.1097/JS9.0000000000001239 Qi Y-J, Su G-H, You C, et al (2024) Radiomics in breast cancer: Current advances and future directions. Cell Reports Medicine 5:101719. https://doi.org/10.1016/j.xcrm.2024.101719 Feng Z, Li H, Liu Q, et al (2023) CT radiomics to predict macrotrabecular-massive subtype and immune status in hepatocellular carcinoma. Radiology 307:e221291. https://doi.org/10.1148/radiol.221291 Hu C, Chen W, Li F, et al (2023) Deep learning radio-clinical signature for predicting neoadjuvant chemotherapy response and prognosis from pretreatment CT images of locally advanced gastric cancer patients. International Journal of Surgery Publish Ahead of Print: https://doi.org/10.1097/JS9.0000000000000432 Song R, Cui Y, Ren J, et al (2022) CT-based radiomics analysis in the prediction of response to neoadjuvant chemotherapy in locally advanced gastric cancer: A dual-center study. Radiotherapy and Oncology 171:155–163. https://doi.org/10.1016/j.radonc.2022.04.023 Yu Y, Tan Y, Xie C, et al (2020) Development and validation of a preoperative magnetic resonance imaging radiomics–based signature to predict axillary lymph node metastasis and disease-free survival in patients with early-stage breast cancer. JAMA Netw Open 3:e2028086. https://doi.org/10.1001/jamanetworkopen.2020.28086 Jia H, Jiang X, Zhang K, et al (2022) A nomogram of combining IVIM‐DWI and MRI radiomics from the primary lesion of rectal adenocarcinoma to assess nonenlarged lymph node metastasis preoperatively. Magnetic Resonance Imaging 56:658–667. https://doi.org/10.1002/jmri.28068 Zheng Y, Xu W, Hao D, et al (2021) A CT-based radiomics nomogram for differentiation of lympho-associated benign and malignant lesions of the parotid gland. Eur Radiol 31:2886–2895. https://doi.org/10.1007/s00330-020-07421-4 Xu B, Zheng H-L, Chen C, et al (2024) Development and validation of a preoperative radiomics-based nomogram to identify patients who can benefit from splenic hilar lymphadenectomy: A pooled analysis of three prospective trials. International Journal of Surgery 110:4053–4061. https://doi.org/10.1097/JS9.0000000000001337 Lin P, Xie W, Li Y, et al (2024) Intratumoral and peritumoral radiomics of MRIs predicts pathologic complete response to neoadjuvant chemoimmunotherapy in patients with head and neck squamous cell carcinoma. J Immunother Cancer 12:e009616. https://doi.org/10.1136/jitc-2024-009616 Ryan R, Gibbons D, Hyland JMP, et al (2005) Pathological response following long‐course neoadjuvant chemoradiotherapy for locally advanced rectal cancer. Histopathology 47:141–146. https://doi.org/10.1111/j.1365-2559.2005.02176.x Lin H, Hua J, Wang Y, et al (2025) Prognostic and predictive values of a multimodal nomogram incorporating tumor and peritumor morphology with immune status in resectable lung adenocarcinoma. J Immunother Cancer 13:e010723. https://doi.org/10.1136/jitc-2024-010723 Wang Y, Xu H, Hu C, et al (2022) Prognostic value and clinicopathological correlation of the tumor regression grade in neoadjuvant chemotherapy for gastric adenocarcinoma: A retrospective cohort study. J Gastrointest Oncol 13:1046–1057. https://doi.org/10.21037/jgo-22-537 Chen Y, Xu W, Li Y-L, et al (2022) CT-based radiomics showing generalization to predict tumor regression grade for advanced gastric cancer treated with neoadjuvant chemotherapy. Front Oncol 12:758863. https://doi.org/10.3389/fonc.2022.758863 Cui Y, Zhang J, Li Z, et al (2022) A CT-based deep learning radiomics nomogram for predicting the response to neoadjuvant chemotherapy in patients with locally advanced gastric cancer: A multicenter cohort study. eClinicalMedicine 46:101348. https://doi.org/10.1016/j.eclinm.2022.101348 Liu C, Li L, Chen X, et al (2024) Intratumoral and peritumoral radiomics predict pathological response after neoadjuvant chemotherapy against advanced gastric cancer. Insights Imaging 15:23. https://doi.org/10.1186/s13244-023-01584-6 Chen Y-H, Xiao J, Chen X-J, et al (2020) Nomogram for predicting pathological complete response to neoadjuvant chemotherapy in patients with advanced gastric cancer. WJG 26:2427–2439. https://doi.org/10.3748/wjg.v26.i19.2427 Sun K-Y, Hu H-T, Chen S-L, et al (2020) CT-based radiomics scores predict response to neoadjuvant chemotherapy and survival in patients with gastric cancer. BMC Cancer 20:468. https://doi.org/10.1186/s12885-020-06970-7 Wang W, Peng Y, Feng X, et al (2021) Development and validation of a computed tomography–based radiomics signature to predict response to neoadjuvant chemotherapy for locally advanced gastric cancer. JAMA Netw Open 4:e2121143. https://doi.org/10.1001/jamanetworkopen.2021.21143 Han X, Wang M, Zheng Y, et al (2023) Delta-radiomics features for predicting the major pathological response to neoadjuvant chemoimmunotherapy in non-small cell lung cancer. Eur Radiol 34:2716–2726. https://doi.org/10.1007/s00330-023-10241-x Lin P, Yang P-F, Chen S, et al (2020) A delta-radiomics model for preoperative evaluation of neoadjuvant chemotherapy response in high-grade osteosarcoma. Cancer Imaging 20:7. https://doi.org/10.1186/s40644-019-0283-8 Table 2 Table 2 is available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformationfor.docx Table2Univariableandmultivariablelogsticana.docx Cite Share Download PDF Status: Published Journal Publication published 17 Sep, 2025 Read the published version in Abdominal Radiology → Version 1 posted Editorial decision: Revision requested 30 Jul, 2025 Reviews received at journal 29 Jul, 2025 Reviewers agreed at journal 20 Jul, 2025 Reviewers invited by journal 09 Jul, 2025 Editor assigned by journal 08 Jul, 2025 Submission checks completed at journal 08 Jul, 2025 First submitted to journal 08 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7072201","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":483489353,"identity":"df3f03d7-f6b6-4b3a-9665-cc16a832ed95","order_by":0,"name":"Shulan Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shulan","middleName":"","lastName":"Chen","suffix":""},{"id":483489354,"identity":"614131b4-d1b0-402e-8abf-f5b39dcbf5a6","order_by":1,"name":"Zhe Xiao","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhe","middleName":"","lastName":"Xiao","suffix":""},{"id":483489355,"identity":"5fbf544a-ce48-47ac-ba7e-c1de0802cfc9","order_by":2,"name":"Jiayi Jin","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiayi","middleName":"","lastName":"Jin","suffix":""},{"id":483489356,"identity":"5f29e02f-1d9b-4733-b096-0385b9d1a523","order_by":3,"name":"Hanzhe Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hanzhe","middleName":"","lastName":"Wang","suffix":""},{"id":483489357,"identity":"71fa7eba-05c8-4b0e-a170-7bebee257f41","order_by":4,"name":"Shouliang Miao","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shouliang","middleName":"","lastName":"Miao","suffix":""},{"id":483489358,"identity":"1c41ed41-5bd3-41e6-ab94-fdcaa416e1b7","order_by":5,"name":"Kun Tang","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kun","middleName":"","lastName":"Tang","suffix":""},{"id":483489359,"identity":"5242397d-bed0-4e88-9384-494ef7e1aee7","order_by":6,"name":"Yimin Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yimin","middleName":"","lastName":"Chen","suffix":""},{"id":483489360,"identity":"b2ce8cc8-ee15-41ab-b0e2-818838c779af","order_by":7,"name":"Linfeng Shao","email":"","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Linfeng","middleName":"","lastName":"Shao","suffix":""},{"id":483489361,"identity":"c35b9756-1591-4286-847a-d3dadae2344d","order_by":8,"name":"Xiangwu Zheng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAApklEQVRIiWNgGAWjYLCCBAMJOQZmkrQ8KLAwJk0L44MPFYkNRCuXn5FjJgF0WPr8dt6DHxhqbKIJ29BzBqwld8NhvmQJhmNpuQStY2bv3QbRwsxjIMHYcJiwFjZmXrCWdPlmHuMfRGnhgdqSwHCYx4w4WyR4zn+2AGox3ADUYpFAjF/kZ6Ql3vzxp05evv+M8Y0PNTaEtaCCBNKUj4JRMApGwSjABQDJvDYslbHDfAAAAABJRU5ErkJggg==","orcid":"","institution":"First Affiliated Hospital of Wenzhou Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xiangwu","middleName":"","lastName":"Zheng","suffix":""}],"badges":[],"createdAt":"2025-07-08 08:08:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7072201/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7072201/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00261-025-05171-9","type":"published","date":"2025-09-17T15:57:19+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":86662978,"identity":"79a84955-3d3c-4518-bef0-bcf0d4da2bcd","added_by":"auto","created_at":"2025-07-14 10:46:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2006810,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the inclusion and exclusion process. Note: LAGC, locally advanced gastric cancer; NAC, neoadjuvant chemotherapy; CT, computed tomography; GR, good response; PR, poor response\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/c6e3e663111a3ea66a878d50.png"},{"id":86664373,"identity":"18557442-4483-4d38-b61b-eabc37fd99a8","added_by":"auto","created_at":"2025-07-14 10:54:29","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":20140237,"visible":true,"origin":"","legend":"\u003cp\u003eThe overall flowchart of this study. Note: NAC, neoadjuvant chemotherapy; ROC, receiver operator characteristic; DCA, decision curve analysis\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/918da08fe435c927599da6c8.png"},{"id":86663001,"identity":"03b1bc5b-4d56-494c-af4d-250f527032b3","added_by":"auto","created_at":"2025-07-14 10:46:29","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":53566309,"visible":true,"origin":"","legend":"\u003cp\u003eRepresentative computed tomography images of four patients in GR and PR. Note: NAC, neoadjuvant chemotherapy; ROI, the regions of interest; TRG, tumor regression grade\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/19082acbc4d7e5bc01d4f65b.png"},{"id":86662983,"identity":"48b67a72-0f66-479d-b728-35c46f13355d","added_by":"auto","created_at":"2025-07-14 10:46:29","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":12014797,"visible":true,"origin":"","legend":"\u003cp\u003eThe performance of the five models. The receiver operating characteristic (ROC) curves of the preCT-RS P+L, preCT-RS P, postCT-RS P+L, postCT-RS P, delCT-RS P+L, delCT-RS P, clinical and RS-CN in the training cohort (A) and validation cohort (B). Note: P+L, primary tumor and the largest lymph node; P, primary; preCT-RS, pre-neoadjuvant chemothrapy computed tomography radiomics signatures; postCT-RS, post-neoadjuvant chemothrapy computed tomography radiomics signatures; delCT-RS, delta computed tomography radiomics signatures; RS-CN, radiomic signatures-clinical-nomogram\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/eab91c320794b51b17daef19.png"},{"id":86662991,"identity":"4e5eea44-8083-438d-92ba-56120145c3fd","added_by":"auto","created_at":"2025-07-14 10:46:29","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":5375324,"visible":true,"origin":"","legend":"\u003cp\u003ePredictive performance of RS-CN. Waterfall plots of RS-CN predicted probability in the training cohort (A) and validation cohort (B). The blue below the baseline indicates the correctly predicted PR, the orange above the baseline indicates the correctly predicted GR, and the cross section is wrongly predicted, with a favorable overall prediction. (C) Calibration curves for RS-CN in all cohorts. (D) Decision curve analysis of preCT-RS P +L, preCT-RS P, postCT-RS P+L, postCT-RS P, delCT-RS P+L, delCT-RS P, clinical, and RS-CN models. Note: RS-CN, radiomic signatures-clinical-nomogram; P+L, primary tumor and the largest lymph node; P, primary; preCT-RS, pre-neoadjuvant chemothrapy computed tomography radiomics signatures; postCT-RS, post-neoadjuvant chemothrapy computed tomography radiomics signatures; delCT-RS, delta computed tomography radiomics signatures; NAC, neoadjuvant chemotherapy; GR, good response; PR, poor response\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/098098f15bd66678d9a30d0a.png"},{"id":91889852,"identity":"6e7864b0-f4d8-4762-9bd2-601e7d75cc88","added_by":"auto","created_at":"2025-09-22 16:02:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":75200997,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/7e44c009-8534-493f-a04d-3c4128f952db.pdf"},{"id":86664374,"identity":"2abfd13c-c71a-490e-a8e4-bf7670d88f5f","added_by":"auto","created_at":"2025-07-14 10:54:29","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":9326333,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformationfor.docx","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/809860f3b20a859b82e512ad.docx"},{"id":86662981,"identity":"ae39f997-1542-4e56-9812-fdef2b225a45","added_by":"auto","created_at":"2025-07-14 10:46:28","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":20445,"visible":true,"origin":"","legend":"","description":"","filename":"Table2Univariableandmultivariablelogsticana.docx","url":"https://assets-eu.researchsquare.com/files/rs-7072201/v1/f5be441aec16b6a1d23e1429.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"CT-based delta-radiomics nomogram to assess tumor regression grade in locally advanced gastric cancer patients following neoadjuvant chemotherapy","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGastric cancer (GC) ranks as the fifth most common malignant tumor and the fifth leading cause of cancer-related mortality globally [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Although surgical resection remains the primary treatment modality, nonspecific clinical manifestations frequently delay diagnosis until the locally advanced gastric cancer (LAGC) stage, which is associated with poor prognoses. The 5-year overall survival rate following radical gastrectomy persists at 30–40% [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In recent years, neoadjuvant chemotherapy (NAC) has been recommended by the National Comprehensive Cancer Network (NCCN) as the preferred preoperative treatment for LAGC patients undergoing curative resection, based on its demonstrated clinical benefits: tumor volume reduction, clinical stage downstaging, improved R0 resection rates, elimination of potential micrometastases, and reduced risks of postoperative recurrence and metastasis [\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e–\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. However, due to tumor heterogeneity, not all LAGC patients derive therapeutic benefit from standardized NAC regimens, with some experiencing treatment-related toxicity accumulation and potential disease progression during therapy [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Consequently, the development of reliable early prediction methodologies is crucial for achieving precision medicine in LAGC management.\u003c/p\u003e\u003cp\u003eTumor regression grade (TRG) serves as a pivotal biomarker for assessing therapeutic response and prognostic outcomes in oncology. Its principal value lies in quantifying tumor regression through histopathological analysis of surgical specimens [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. However, TRG demonstrates notable clinical limitations: its evaluation relies exclusively on postoperative tissue samples, precluding dynamic monitoring of individual responses during early treatment phases, thereby constraining its temporal relevance in guiding personalized therapeutic strategies. Consequently, the development of non-invasive predictive tools capable of early-phase treatment assessment holds significant clinical implications for optimizing NAC regimens and related therapeutic approaches. Currently, the Response Evaluation Criteria in Solid Tumors (RECIST) and World Health Organization (WHO) criteria are widely employed for evaluating NAC responses, primarily based on bidimensional measurements of target lesion dimensions. Although these criteria demonstrate high operational feasibility in clinical practice, their principal limitation resides in focusing solely on volumetric tumor changes while disregarding tumor heterogeneity [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Emerging evidence indicates that tumor heterogeneity not only constitutes a critical biological characteristic but may also directly influence therapeutic responsiveness and survival outcomes [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eRadiomics, an emerging interdisciplinary domain in precision medicine, has attracted considerable attention in recent years by extracting quantitative features from medical images through high-throughput and transforming image data into high-dimensional features suitable for mining to reveal the micro-biological characteristics of tumors [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Clinical investigations in malignancies including colorectal cancer [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], breast cancer [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], and hepatocellular carcinoma [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] have demonstrated the marked superiority of radiomics in tumor phenotyping, validating its potential for oncological evaluation and clinical decision-making. Among them, delta radiomics demonstrates more dynamic evolution information compared to conventional single-phase (pre- or post-treatment) radiomics. In this study, we propose the following scientific hypothesis:delta computed tomography radiomics (delCT-RS) can dynamically quantify temporal changes in lesion biology during therapeutic interventions or natural disease progression. These radiomic alterations may manifest earlier than detectable changes in conventional imaging modalities or clinical symptom onset, thereby providing a basis for early assessment of tumor regression following NAC.\u003c/p\u003e\u003cp\u003eIn current research, the analysis of imaging features based on primary tumor lesions has become a crucial approach for evaluating pathological response after NAC in LAGC [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. And some studies have utilized approaches involving either combined delineation of primary tumors and lymph nodes or exclusive delineation of primary tumors to predict post-treatment lymph node metastasis in cancer patients [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. This study innovatively employed a method combining slice-by-slice delineation of primary tumors with maximum cross-sectional delineation of the largest lymph node as regions of interest (ROIs), thereby establishing a corresponding radiomics model to predict tumor regression in LAGC patients following NAC treatment.\u003c/p\u003e\u003cp\u003eIn recent years, nomograms integrating radiomics with clinical risk factors have been been widely used to evaluate treatment responses and predict tumor prognosis across diverse clinical scenarios [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Previous studies have demonstrated that such multimodal integration strategies can significantly enhance the precision of risk stratification, thereby providing quantitative foundations for personalized therapeutic decision-making [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. However, there is still a lack of an efficient and simple preoperative scoring tool in current clinical practice to accurately assess the efficacy of NAC and effectively predict the potential benefits of patients. The primary objective of this study is to develop and validate a delCT-RS-based nomogram for achieving personalized quantitative assessment of tumor regression in LAGC patients, enabling stratification into high-response and low-response subgroups, thereby providing clinicians with decision basis for dynamic adjustment of treatment plans.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003ePatients\u003c/b\u003e\u003c/p\u003e\u003cp\u003e This retrospective study has been formally approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University, with a waiver granted for the informed consent procedure. The entire research process has strictly adhered to the ethical guidelines for medical research established by the Declaration of Helsinki and its subsequent amendments.\u003c/p\u003e\u003cp\u003e This study included 193 patients with LAGC (cT2-4NxM0) who underwent NAC followed by radical gastrectomy at our institution between January 2018 and October 2024. Detailed information on the enrollment procedure is displayed in Supplementary Appendix 1 and Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Ultimately, 147 eligible patients were divided into a training cohort (TC) (n = 104) and a validation cohort (VC) (n = 43). And Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the workflow of this study.\u003c/p\u003e\u003cp\u003e\u003cb\u003eNeoadjuvant chemotherapy regimens and clinicopathological data collection\u003c/b\u003e\u003c/p\u003e\u003cp\u003e According to gastric cancer treatment guidelines, all patients received at least two cycles of standardized NAC regimen prior to radical gastrectomy (Supplementary Appendix 2).\u003c/p\u003e\u003cp\u003eThis retrospective study analyzed the patients’ clinical and pathological characteristics, including gender, age, body mass index (BMI), Borrmann classification, tumor differentiation grade, tumor location, serum levels of carcinoembryonic antigen (CEA), carbohydrate antigen 199 (CA199), and alpha-fetoprotein (AFP). Additionally, based on the American Joint Committee on Cancer (AJCC) TNM Classification (8th edition), the clinical T (cT) and clinical N (cN) stages were recorded from the medical records.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTumor regression grade\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAfter neoadjuvant chemotherapy, all patients underwent gastrectomy. And two senior pathologists conducted a standardized evaluation of the surgical pathology specimens independently while blinded to clinical or imaging data. The tumor pathological response was evaluated by TRG after NAC. TRG scores were evaluated using Ryan criteria [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] (Supplementary Appendix 3). The study cohort was categorized into two groups based on TRG: TRG 0–1: good response (GR) to NAC ; TRG 2–3: poor response (PR) to NAC (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eRadiomics feature extraction and reproducibility assessment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eSpecific parameters of CT scanning are shown in Supplementary Appendix 4 and Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. Radiologist 1 manually contoured the ROI on tumor slices using 3D slicer software Version5.8.0 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.slicer.org\u003c/span\u003e\u003cspan address=\"http://www.slicer.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), slice-by-slice along the tumor margin, excluding the first and last layers to avoid partial volume effects. In addition, for the largest metastatic lymph node, precise contour mapping was performed simultaneously on the anatomical section with the largest transverse diameter. For comparison, another delineation method was used in this study, where only the primary tumor (delineated layer by layer, with the first and last layers excluded) was delineated as the region of interest. Radiomic feature extraction was conducted via the open-source Pyradiomics package Version3.1.0 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pyradiomics.readthedocs.io/en/3.1.0/\u003c/span\u003e\u003cspan address=\"https://pyradiomics.readthedocs.io/en/3.1.0/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), with a total of 851 radiomic features extracted from each ROI (Supplementary Appendix 5). All features complied with the standards of the Image Biomarker Standardization Initiative (IBSI). To ensure feature reproducibility, after one month, 20 cases were randomly selected for repeat segmentation by Radiologists 1 and 2, who were blinded to clinical and pathological information. Inter- and intra-observer consistency was assessed using the intraclass correlation coefficient (ICC), with features showing ICC \u0026gt; 0.80 considered robust and included in subsequent analyses.\u003c/p\u003e\u003cp\u003eThe change in radiomic features (delCT-RS) was calculated as the difference between post-neoadjuvant chemotherapy computed tomography radiomics (postCT-RS) and pre-neoadjuvant chemotherapy computed tomography radiomics (preCT-RS):\u003c/p\u003e\u003cp\u003edelCT-RS = postCT-RS - preCT-RS\u003c/p\u003e\u003cp\u003e\u003cb\u003eRadiomics feature selection and radiomics models building\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn this study, radiomic models were developed and validated using Minimum Redundancy - Maximum Relevance (mRMR) and Least Absolute Shrinkage and Selection Operator (LASSO) feature selection methods. Initially, mRMR identified 20 radiomic features highly correlated with target variables and with low redundancy. Then, LASSO regression with 10-fold cross-validation determined the regularization parameter λ to optimize feature selection, implemented via the R package glmnet. Ultimately, robust radiomic features strongly associated with treatment response were selected to build the prediction model. (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e and Table S2)\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinical model and nomogram building\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFirst, univariate and multivariate logistic regression analyses were performed to evaluate all clinical and pathological factors in the training cohort, calculating the odds ratios (OR) and 95% confidence interval (CI) for each factor. Independent clinical predictors were selected based on statistical significance, and a clinical prediction model was constructed accordingly. Subsequently radiomic signature was combined with the above independent clinical predictors and then a comprehensive nomogram was constructed. The entire modeling process adhered to standardized clinical prediction model development procedures, guaranteeing the model's scientific rigor and interpretability.\u003c/p\u003e\u003cp\u003e\u003cb\u003ePerformance evaluation\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe predictive performance of each model was first evaluated using Receiver operating characteristic (ROC) curves. Area under the curve (AUC) was calculated and compared across different cohorts using the DeLong test. In addition, key metrics including accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were recorded for each model. To further assess clinical applicability, calibration curves were plotted to evaluate the consistency between radiomic labels and actual observations, and the Hosmer-Lemeshow test was used to assess model fit. Additionally, decision curve analysis (DCA) was applied to evaluate the clinical net benefit of the radiomic model in identifying patients likely to benefit from NAC, using data from both the training and validation cohorts.\u003c/p\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eSPSS statistical software version 27.0 (IBM) and R software version 4.4.2 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.r-project.org\u003c/span\u003e\u003cspan address=\"http://www.r-project.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) were used for analysis (Supplementary Appendix 6). A two-sided \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05 was considered statistically significant. Notably, In the univariate logistic regression analysis, a lenient screening criterion of \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.20 was used to avoid missing clinically relevant factors.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eClinicopathological characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe baseline clinicopathological characteristics of all 147 LAGC patients are summarized in Table\u0026nbsp;1. As shown in Table\u0026nbsp;1, the GR rates were comparable between TC and VC, with proportions of 36.5% and 34.6% respectively. In both cohorts, no statistically significant differences exist between the GR and PR groups in terms of demographic parameters (sex, age, and BMI) or tumor-related features (cN_stage, Borrmann type, differentiation, and tumor location). Similarly, no significant differences are observed in tumor markers (pre-neoadjucant chemotherapy (pre-NAC) CEA, pre-NAC CA199, pre-NAC AFP, post-neoadjuvant chemotherapy (post-NAC) CA199, and post-NAC AFP). Notably, cT_stage demonstrated significant intergroup differences in both TC and VC (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05), whereas post-NAC CEA showed significant differences only in VC (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05).\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 1\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eClinicopathological characteristics of LAGC patients in the training and validation cohorts\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eTraining cohort (n = 104)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eValidation cohort (n = 43)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eGR (n = 38)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePR (n = 66)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eGR (n = 15)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePR (n = 28)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSex (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e27 (71.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e52 (78.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (80.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22 (78.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11 (28.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (21.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (20.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (21.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge (year, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.103\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.964\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (39.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e37 (56.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11 (39.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23 (60.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e29 (43.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9 (60.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e17 (60.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI (kg/m\u003csup\u003e2\u003c/sup\u003e, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.429\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt; 24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23 (60.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e45 (68.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11 (73.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20 (71.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≥ 24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (39.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e21 (31.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (26.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8 (28.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ecT_stage (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt; 0.001***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.033*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eT2་T3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e32 (84.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e32 (48.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (80.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13 (46.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eT4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (15.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e34 (51.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (20.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (53.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ecN_stage (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.468\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.349\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (4.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1(6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN+\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e38 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e63 (95.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (93.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e28 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBorrmann (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.559\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.317\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eⅠ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (4.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (10.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eⅡ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10 (26.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (21.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7 (46.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7 (25.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eⅢ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e19 (50.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e43 (65.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (50.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eⅣ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (10.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (9.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1 (6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5 (13.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1(6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDifferentiation (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.131\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.282\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWell\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1(2.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (4.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1 (3.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerately\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (39.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (21.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5 (33.%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePoorly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (31.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e47 (71.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7 (46.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23 (82.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10 (26.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2 (3.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (20.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLocation (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.833\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.920\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUpper 1/3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5 (13.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13 (19.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (20.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMiddle1/3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (36.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23 (34.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (26.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10 (35.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLower 1/3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16 (42.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e24 (36.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10 (35.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWhole stomach\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (7.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (9.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2 (13.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePre-NAC CEA (ng/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.797\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.454\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e24 (63.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e40 (60.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13 (86.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20 (71.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (36.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26 (39.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2 (13.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8 (28.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePre-NAC CA199 (U/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.911\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e29 (76.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e52 (77.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13 (86.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23 (82.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9 (23.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (22.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2 (13.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5 (17.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePre-NAC AFP (ng/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.281\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.080\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 8.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e31 (81.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e60 (90.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11 (73.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e27 (96.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 8.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7 (18.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6 (9.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4 (26.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1 (3.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePost-NAC CEA (ng/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.892\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26 (68.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e46(69.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (93.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (42.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (31.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20 (30.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1 (6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16 (57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePost-NAC CA199 (U/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e35 (92.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e54 (81.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e21 (75.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3 (7.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12 (18.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7 (25.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePost-NAC AFP (ng/mL, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.945\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 8.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e33 (86.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e57 (86.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14 (93.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26 (92.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 8.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5 (13.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9 (13.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1 (6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2 (7.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\"\u003eNote: Chi-squared or Fisher's exact tests, were used to compare the differences in categorical variables, whereas student t or Mann-Whitney U test was used to compare the differences in continuous variables, as appropriate. * \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05; ** \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01; *** \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001. LAGC, locally advanced gastric cancer; GR, good response; PR,poor response; BMI, body mass index; cT_stage, clinical T stage; cN_stage, clinical N stage; CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; AFP, alpha-fetoprotein; NAC, neoadjuvant chemotherapy\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eCorrelation between the TRG and clinicopathological characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn the univariate logistic regression analysis, this study adopted \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.20 as the variable selection criterion to minimize the risk of omitting potentially clinically relevant factors [24]. The ultimately selected variables identified as associated with TRG included age、cT_stage、pre-NAC AFP and post-NAC CA199, all of which were subsequently incorporated into the multivariate logistic regression analysis. Through multivariate logistic regression analysis, cT_stage was ultimately identified as a statistically significant independent predictive factor (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). Building upon this key predictor, a corresponding clinical prediction model was established, with detailed information presented in Table 2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRadiomics feature selection and comparison of radiomics models and clinical model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Fig.\u0026nbsp;4, AUC analysis of the radiomics models revealed that the delCT-RS established using both delineation approaches demonstrated superior predictive performance in evaluating tumor regression following NAC treatment compared to single time-point models based solely on pre-therapy or post-therapy radiomics signatures: delCT-RS P + L (AUC, 0.805; 95% CI 0.721–0.888)、preCT-RS P + L (AUC, 0.778; 95% CI 0.687–0.868)、 postCT-RS P + L (AUC, 0.706; 95% CI 0.606–0.805); delCT-RS P (AUC, 0.727; 95% CI 0.627–0.828)、preCT-RS P (AUC, 0.714; 95% CI 0.610–0.817)、postCT-RS P (AUC, 0.631; 95% CI 0.525–0.737). Further validation in the validation cohort confirmed these results: delCT-RS P + L (AUC, 0.795; 95% CI 0. 658–0. 932)、preCT-RS P + L (AUC,0.581; 95% CI 0.400–0.762)、postCT-RS P + L (AUC, 0.721;95% CI 0.547–0.895); delCT-RS P (AUC, 0.655; 95% CI 0. 490–0. 820)、 preCT-RS P (AUC,0.638; 95% CI 0.473–0.804)、postCT-RS P (AUC, 0.624; 95% CI 0.452–0.795). (p \u0026lt; 0.05) In addition, a comparative analysis was conducted between the conventional approach involving sole delineation of primary tumors and our innovative method combining primary tumor delineation with maximum lymph node delineation. It is evident that the radiomics models established by including the largest lymph node as the region of interest demonstrated superior predictive performance.\u003c/p\u003e\n\u003cp\u003eNotably, delCT-RS P + L not only outperformed the other six radiomics models but also demonstrated more robust performance compared to the clinical model (TC: AUC, 0.697; 95% CI 0.598–0.763. VC: AUC, 0.668; 95% CI 0.527–0.809.), demonstrating comprehensive superiority across all evaluation metrics. In all cohorts listed in Table\u0026nbsp;3, delCT-RS outperformed the other models in terms of accuracy, sensitivity, specificity, PPV, and NPV. (DeLong test, all \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05)\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 3\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eThe performance of models in predicting the response of LAGC after NAC\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eModels\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePPV\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eNPV\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePreCT-RS P + L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.692\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.895\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.548\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.905\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.598–0.773)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.759–0.958)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.456–0.688)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.425–0.666)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.779–0.962)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.558\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.409\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.411–0.696)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.357–0.802)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.358–0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.233–0.613)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.500-0.862)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePreCT-RS P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.738\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.659–0.823)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.279–0.578)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.854–0.976)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.584–0.919)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.635–0.820)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.628\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.444\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.479–0.756)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.109–0.520)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.644–0.921)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.189–0.733)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.508–0.809)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePostCT-RS P + L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.663\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.816\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.525\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.568–0.747)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.666–0.908)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.456–0.688)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.400-0.647)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.712–0.923)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.558\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.417\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.411–0.696)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.417–0.848)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.326–0.674)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.245–0.612)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.512–0.882)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePostCT-RS P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.606\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.805\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.510–0.694)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.637–0.889)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.383–0.617)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.358–0.597)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.660–0.898)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.581\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.733\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.440\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.433–0.716)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.480–0.891)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.326–0.674)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.267–0.629)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.548-0,910)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDelCT-RS P + L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.740\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.921\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.636\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.593\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.933\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.649–0.815)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.792–0.973)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.516–0.742)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.466–0.709)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.821–0.977)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.674\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.933\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.519\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.938\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.525–0.795)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.702–0.988)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.358–0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.340–0.693)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.717–0.989)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDelCT-RS P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.673\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.537\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.820\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.578–0.756)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.608–0.870)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.501–0.729)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.406–0.663)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.692–0.902)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.581\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.733\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.440\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.433–0.716)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.480–0.891)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.326–0.674)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.267–0.629)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.548–0.910)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eClinical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.635\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.515\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.850\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.539–0.721)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.696–0.926)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.397–0.632)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.381–0.619)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.709–0.929)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.628\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.480\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.833\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.479–0.756)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.548–0.930)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.358–0.705)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.300-0.665)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.608–0.942)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRS-CN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.827\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.818\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.727\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.900\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.743–0.888)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.696–0.926)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.709–0.893)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.582–0.837)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.799–0.953)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.744\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.870\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.598–0.851)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.548–0.930)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.529–0.847)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.387–0.781)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e(0.679–0.955)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003eNote: LAGC, locally advanced gastric cancer; NAC, neoadjuvant chemotherapy; CI, confidence interval; P + L, primary tumor and the largest lymph node; P, primary; preCT-RS, pre-neoadjuvant chemothrapy computed tomography radiomics signatures; postCT-RS, post-neoadjuvant chemothrapy computed tomography radiomics signatures; delCT-RS, delta computed tomography radiomics signatures; RS-CN, radiomic signatures-clinical-nomogram\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eNomogram model performance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn subsequent research, we integrated delCT-RS P + L with the clinical model to form a combined model (Figure S2). As indicated in Fig.\u0026nbsp;4, Fig.\u0026nbsp;5A-B, and Table\u0026nbsp;3, this integrated model demonstrated stable superiority in both cohorts. It achieved an AUC of 0.841 (95% CI: 0.763–0.919) in TC and an AUC of 0.817 (95% CI: 0.691–0.942) in VC, significantly outperforming each single-source model (TC: accuracy: 0.827, sensitivity: 0.842, specificity: 0.818, PPV: 0.727, NPV: 0.900; VC: accuracy: 0.744, sensitivity: 0.800, and specificity: 0.714, PPV: 0.600; NPV: 0.870). Further analysis revealed that the radiomic score exhibited significant differences between the GR and PR groups in all cohorts (all \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05) (Figure S3). Moreover, the predictive scores from both the clinical and radiomics models demonstrated a consistent trend, with higher score gradients positively correlating with improved NAC treatment response. Additionally, according to the TRG criteria, the radiomic model's predictions of LAGC patients' responses to NAC treatment showed a high degree of consistency with the tumor pathology assessment results in both the training and validation sets. Furthermore, the calibration of the integrated model across different cohorts was evaluated. The calibration curves indicated a good fit between predicted and observed probabilities (Hosmer-Lemeshow test \u003cem\u003ep\u003c/em\u003e \u0026gt; 0.05), suggesting reliable model calibration (Fig.\u0026nbsp;5C). DCA further validated that the combined model provided substantial net clinical benefit across a wide threshold probability range (0.2–0.7) for predicting NAC treatment response in all cohorts (Fig.\u0026nbsp;5D).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eNAC has demonstrated significant interindividual variability in clinical efficacy as a conventional treatment for LAGC patients. Currently, TRG is universally recognized as the gold standard for evaluating neoadjuvant therapeutic outcomes, and its prognostic significance in LAGC patients has been well validated [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. As a prognostic predictor, TRG serves as a valuable reference for assessing potential therapeutic effects of NAC and facilitating individualized prognosis evaluation in LAGC patients [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. However, current TRG assessment systems exhibit notable limitations: they rely on postoperative pathological specimen analysis, which prevents real-time acquisition of dynamic monitoring data during treatment. This temporal lag in evaluation substantially restricts their applicability in timely therapeutic regimen adjustments. Consequently, developing novel assessment systems capable of noninvasive TRG prediction at early treatment stages represents an urgent clinical need for enhancing precision therapy with NAC and optimizing comprehensive management strategies.\u003c/p\u003e\u003cp\u003eThis study incorporated relevant clinicopathological indicators into the analysis. In the training cohort, cT_stage demonstrated significant correlation between the GR and PR groups (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001), with its stage progression exhibiting a positive correlation trend with TRG. This observation suggests that the degree of local tumor infiltration may influence NAC efficacy, a finding consistent with previous research conclusions [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Consequently, cT_stage was integrated into our clinical model. Although existing literature reports associations between Lauren classification, tumor differentiation grade, cN_stage and pathological response [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e], these conclusions remain controversial, potentially due to insufficient support from large-scale samples or multicenter studies.\u003c/p\u003e\u003cp\u003eThis study systematically integrates cutting-edge advancements and innovative applications of radiomics technology. Unlike conventional imaging evaluations that focus on macroscopic morphological features, radiomics provides novel insights into therapeutic response assessment by extracting deep-texture information from medical images, garnering increasing attention. Furthermore, by utilizing postoperative pathological findings as reference standards, radiomics establishes a critical bridge between imaging and histopathology, enabling precise identification of potential beneficiaries of NAC. Published evidence indicates that pre-NAC CT-based radiomic models effectively predict NAC efficacy in LAGC patients [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. For instance, Song et al [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] extracted radiomic features from portal venous-phase CT images of 279 GC patients prior to NAC and constructed an radiomic model. This model demonstrated exceptional predictive performance, achieving an AUC of 0.790 in the training set, with corresponding internal and external validation cohort AUCs of 0.784 and 0.803, respectively. Recent advances in delta-radiomics\u0026mdash;which quantifies tumor heterogeneity changes by analyzing texture features from pre- and post-treatment images\u0026mdash;have shown superior accuracy over single-timepoint radiomics in predicting neoadjuvant therapeutic responses. In predicting major pathological response (MPR) following neoadjuvant chemoimmunotherapy for non-small cell lung cancer (NSCLC) [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e], delta-radiomics models achieved AUCs of 0.768, 0.732, 0.833, and 0.716 in the training, test, and two external validation cohorts, significantly outperforming pretreatment radiomics models (AUCs: 0.644, 0.616, 0.475, and 0.608). Similarly, in preoperative assessment of high-grade osteosarcoma NAC response [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], delta-radiomics signatures exhibited higher AUCs than single-phase CT-based radiomics in both training and validation sets. However, limited research has been conducted in the field of delta radiomics to evaluate the tumor regression following neoadjuvant chemotherapy for gastric cancer. Our study specifically compared delCT-RS with preCT-RS and postCT-RS. The results revealed that delCT-RS demonstrated better evaluation efficacy in assessing tumor regression following NAC treatment, with its AUC values in both the training and validation cohorts exceeding those of the other two single time-point radiomics models.\u003c/p\u003e\u003cp\u003eNotably, prior studies predominantly focused on delineating primary tumors in LAGC patients while neglecting lymph node analysis [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Innovatively, our study implemented slice-by-slice delineation of primary tumors alongside maximum cross-sectional delineation of the largest lymph node. Compared to the conventional approach centered solely on the primary tumor, this integrative mapping strategy has shown better predictive power. It has enhanced the accuracy of assessing tumor regression in LAGC patients following NAC. This approach comprehensively captures tumor infiltration depth, spatial extent, and morphological evolution within gastric walls and adjacent tissues, while lymph node delineation facilitates tracking of dimensional, densitometric, and morphological alterations. Effective chemotherapy typically reduces primary tumor volume with sharper boundaries and lower density, accompanied by concurrent decreases in lymph node size and density. Through comparative delineation, the dynamic changes of tumor and lymph nodes can be accurately quantified, so as to comprehensively and accurately evaluate the regression of tumor and examine the efficacy of chemotherapy on tumor.\u003c/p\u003e\u003cp\u003eSubsequently, this study developed a combined model integrating delCT-RS P\u0026thinsp;+\u0026thinsp;L with clinical indicators (cT_stage), which demonstrated superior predictive performance among all evaluated models. In TC, the model achieved an AUC of 0.841 (95% CI: 0.763\u0026ndash;0.919), and in VC, the model achieved an AUC of 0.817 (95% CI: 0.691\u0026ndash;0.942). The calibration curve and DCA further demonstrated that the combined model provided substantial net clinical benefit for predicting neoadjuvant chemotherapy response over a wide threshold probability range (0.2\u0026ndash;0.7), increasing the clinical utility of radiomics in advising patients on whether they should receive NAC treatment.\u003c/p\u003e\u003cp\u003eThis study has several limitations: 1) As a retrospective single-center investigation, it may be susceptible to confounding factors, thereby limiting the generalizability of the findings. 2) The ROI delineation process inherently carries subjective interpretation, which may introduce measurement bias. 3) Heterogeneity in NAC regimens could introduce outcome bias.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn summary, this study demonstrates that delta radiomics\u0026mdash;a methodology quantifying tumor heterogeneity changes through pre- and post-treatment image analysis\u0026mdash;provides more reliable predictive capability for tumor regression assessment compared to conventional single-timepoint radiomics. Furthermore, we innovatively developed and validated a nomogram model integrating delCT-RS with clinical predictors, utilizing a novel methodology that combines simultaneous delineation of primary tumors and the largest lymph nodes. This tool effectively stratifies patients into high-response and low-response subgroups, thereby offering clinicians actionable evidence for optimizing therapeutic strategies during treatment adaptation.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eDelCT-RS Delta computed tomography radiomics\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTRG Tumor regression grade\u003c/p\u003e\n\u003cp\u003eLAGC Locally advanced gastric cancer\u003c/p\u003e\n\u003cp\u003eNAC Neoadjuvant chemotherapy\u003c/p\u003e\n\u003cp\u003eP+L Primary tumor and the largest lymph node\u003c/p\u003e\n\u003cp\u003eP Primary tumor\u003c/p\u003e\n\u003cp\u003eROIs Regions of interest\u003c/p\u003e\n\u003cp\u003eAUC Area under the curve\u003c/p\u003e\n\u003cp\u003eGC Gastric cancer\u003c/p\u003e\n\u003cp\u003eNCCN National Comprehensive Cancer Network\u003c/p\u003e\n\u003cp\u003eRECIST Response Evaluation Criteria in Solid Tumors\u003c/p\u003e\n\u003cp\u003eWHO World Health Organization\u003c/p\u003e\n\u003cp\u003eTC Training cohort\u003c/p\u003e\n\u003cp\u003eVC Validation cohort\u003c/p\u003e\n\u003cp\u003eBMI Body mass index\u003c/p\u003e\n\u003cp\u003eCEA Carcinoembryonic antigen\u003c/p\u003e\n\u003cp\u003eCA199 Carbohydrate antigen 199\u003c/p\u003e\n\u003cp\u003eAFP Alpha-fetoprotein\u003c/p\u003e\n\u003cp\u003eAJCC American Joint Committee on Cancer\u003c/p\u003e\n\u003cp\u003ecT Clinical T\u003c/p\u003e\n\u003cp\u003ecN Clinical N\u003c/p\u003e\n\u003cp\u003eGR Good response\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePR Poor response\u003c/p\u003e\n\u003cp\u003eIBSI Image Biomarker Standardization Initiative\u003c/p\u003e\n\u003cp\u003eICC Intraclass correlation coefficient\u003c/p\u003e\n\u003cp\u003ePostCT-RS Post-neoadjuvant chemotherapy computed tomography radiomics\u003c/p\u003e\n\u003cp\u003ePreCT-RS Pre-neoadjuvant chemotherapy computed tomography radiomics\u003c/p\u003e\n\u003cp\u003emRMR Minimum Redundancy - Maximum Relevance\u003c/p\u003e\n\u003cp\u003eLASSO Least Absolute Shrinkage and Selection OperatorOR odds ratios\u003c/p\u003e\n\u003cp\u003eCI Confidence interval\u003c/p\u003e\n\u003cp\u003eROC Receiver operating characteristic\u003c/p\u003e\n\u003cp\u003ePPV Positive predictive value\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNPV Negative predictive value\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDCA Decision curve analysis\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePre-NAC Pre-neoadjuvant chemotherapy\u003c/p\u003e\n\u003cp\u003ePost-NAC Post-neoadjuvant chemotherapy\u003c/p\u003e\n\u003cp\u003eRS-CN Radiomic signatures-clinical-nomogram\u003c/p\u003e\n\u003cp\u003eMPR Major pathological response\u003c/p\u003e\n\u003cp\u003eNACLC Non-small cell lung cancer\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eShulan. and Zhe. wrote the main manuscript text , Jiayi. Han. Shouliang. prepared tables, Kun. Yimin. and Linfeng. prepared figures. All authors reviewed the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Laversanne M, Sung H, et al (2024) Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA A Cancer J Clinicians 74:229\u0026ndash;263. https://doi.org/10.3322/caac.21834\u003c/li\u003e\n\u003cli\u003eMachlowska J, Baj J, Sitarz M, et al (2020) Gastric cancer: Epidemiology, risk factors, classification, genomic characteristics and treatment strategies. IJMS 21:4012. https://doi.org/10.3390/ijms21114012\u003c/li\u003e\n\u003cli\u003eAl-Batran S-E, Homann N, Pauligk C, et al (2019) Perioperative chemotherapy with fluorouracil plus leucovorin, oxaliplatin, and docetaxel versus fluorouracil or capecitabine plus cisplatin and epirubicin for locally advanced, resectable gastric or gastro-oesophageal junction adenocarcinoma (FLOT4): A randomised, phase 2/3 trial. The Lancet 393:1948\u0026ndash;1957. https://doi.org/10.1016/S0140-6736(18)32557-1\u003c/li\u003e\n\u003cli\u003eAl-Batran S-E, Hofheinz RD, Pauligk C, et al (2016) Histopathological regression after neoadjuvant docetaxel, oxaliplatin, fluorouracil, and leucovorin versus epirubicin, cisplatin, and fluorouracil or capecitabine in patients with resectable gastric or gastro-oesophageal junction adenocarcinoma (FLOT4-AIO): Results from the phase 2 part of a multicentre, open-label, randomised phase 2/3 trial. The Lancet Oncology 17:1697\u0026ndash;1708. https://doi.org/10.1016/S1470-2045(16)30531-9\u003c/li\u003e\n\u003cli\u003eWang X-Z, Zeng Z-Y, Ye X, et al (2020) Interpretation of the development of neoadjuvant therapy for gastric cancer based on the vicissitudes of the NCCN guidelines. WJGO 12:37\u0026ndash;53. https://doi.org/10.4251/wjgo.v12.i1.37\u003c/li\u003e\n\u003cli\u003eLi Z, Gao X, Peng X, et al (2020) Multi-omics characterization of molecular features of gastric cancer correlated with response to neoadjuvant chemotherapy. Sci Adv 6:eaay4211. https://doi.org/10.1126/sciadv.aay4211\u003c/li\u003e\n\u003cli\u003eZhang X, Liang H, Li Z, et al (2025) Perioperative or postoperative adjuvant oxaliplatin with S-1 versus adjuvant oxaliplatin with capecitabine in patients with locally advanced gastric or gastro-oesophageal junction adenocarcinoma undergoing D2 gastrectomy (RESOLVE): Final report of a randomised, open-label, phase 3 trial. The Lancet Oncology 26:312\u0026ndash;319. https://doi.org/10.1016/S1470-2045(24)00676-4\u003c/li\u003e\n\u003cli\u003eTong Y, Zhu Y, Zhao Y, et al (2021) Evaluation and comparison of predictive value of tumor regression grades according to mandard and becker in locally advanced gastric adenocarcinoma. Cancer Res Treat 53:112\u0026ndash;122. https://doi.org/10.4143/crt.2020.516\u003c/li\u003e\n\u003cli\u003eEisenhauer EA, Therasse P, Bogaerts J, et al (2009) New response evaluation criteria in solid tumours: Revised RECIST guideline (version 1.1). European Journal of Cancer 45:228\u0026ndash;247. https://doi.org/10.1016/j.ejca.2008.10.026\u003c/li\u003e\n\u003cli\u003eSicklick JK, Kato S, Okamura R, et al (2019) Molecular profiling of cancer patients enables personalized combination therapy: The I-PREDICT study. Nat Med 25:744\u0026ndash;750. https://doi.org/10.1038/s41591-019-0407-5\u003c/li\u003e\n\u003cli\u003eMu W, Schabath MB, Gillies RJ (2022) Images are data: Challenges and opportunities in the clinical translation of radiomics. Cancer Research 82:2066\u0026ndash;2068. https://doi.org/10.1158/0008-5472.CAN-22-1183\u003c/li\u003e\n\u003cli\u003eShin J, Seo N, Baek S-E, et al (2022) MRI radiomics model predicts pathologic complete response of rectal cancer following chemoradiotherapy. Radiology 303:351\u0026ndash;358. https://doi.org/10.1148/radiol.211986\u003c/li\u003e\n\u003cli\u003eAbbaspour E, Karimzadhagh S, Monsef A, et al (2024) Application of radiomics for preoperative prediction of lymph node metastasis in colorectal cancer: A systematic review and meta-analysis. International Journal of Surgery 110:3795\u0026ndash;3813. https://doi.org/10.1097/JS9.0000000000001239\u003c/li\u003e\n\u003cli\u003eQi Y-J, Su G-H, You C, et al (2024) Radiomics in breast cancer: Current advances and future directions. Cell Reports Medicine 5:101719. https://doi.org/10.1016/j.xcrm.2024.101719\u003c/li\u003e\n\u003cli\u003eFeng Z, Li H, Liu Q, et al (2023) CT radiomics to predict macrotrabecular-massive subtype and immune status in hepatocellular carcinoma. Radiology 307:e221291. https://doi.org/10.1148/radiol.221291\u003c/li\u003e\n\u003cli\u003eHu C, Chen W, Li F, et al (2023) Deep learning radio-clinical signature for predicting neoadjuvant chemotherapy response and prognosis from pretreatment CT images of locally advanced gastric cancer patients. International Journal of Surgery Publish Ahead of Print: https://doi.org/10.1097/JS9.0000000000000432\u003c/li\u003e\n\u003cli\u003eSong R, Cui Y, Ren J, et al (2022) CT-based radiomics analysis in the prediction of response to neoadjuvant chemotherapy in locally advanced gastric cancer: A dual-center study. Radiotherapy and Oncology 171:155\u0026ndash;163. https://doi.org/10.1016/j.radonc.2022.04.023\u003c/li\u003e\n\u003cli\u003eYu Y, Tan Y, Xie C, et al (2020) Development and validation of a preoperative magnetic resonance imaging radiomics\u0026ndash;based signature to predict axillary lymph node metastasis and disease-free survival in patients with early-stage breast cancer. JAMA Netw Open 3:e2028086. https://doi.org/10.1001/jamanetworkopen.2020.28086\u003c/li\u003e\n\u003cli\u003eJia H, Jiang X, Zhang K, et al (2022) A nomogram of combining IVIM‐DWI and MRI radiomics from the primary lesion of rectal adenocarcinoma to assess nonenlarged lymph node metastasis preoperatively. Magnetic Resonance Imaging 56:658\u0026ndash;667. https://doi.org/10.1002/jmri.28068\u003c/li\u003e\n\u003cli\u003eZheng Y, Xu W, Hao D, et al (2021) A CT-based radiomics nomogram for differentiation of lympho-associated benign and malignant lesions of the parotid gland. Eur Radiol 31:2886\u0026ndash;2895. https://doi.org/10.1007/s00330-020-07421-4\u003c/li\u003e\n\u003cli\u003eXu B, Zheng H-L, Chen C, et al (2024) Development and validation of a preoperative radiomics-based nomogram to identify patients who can benefit from splenic hilar lymphadenectomy: A pooled analysis of three prospective trials. International Journal of Surgery 110:4053\u0026ndash;4061. https://doi.org/10.1097/JS9.0000000000001337\u003c/li\u003e\n\u003cli\u003eLin P, Xie W, Li Y, et al (2024) Intratumoral and peritumoral radiomics of MRIs predicts pathologic complete response to neoadjuvant chemoimmunotherapy in patients with head and neck squamous cell carcinoma. J Immunother Cancer 12:e009616. https://doi.org/10.1136/jitc-2024-009616\u003c/li\u003e\n\u003cli\u003eRyan R, Gibbons D, Hyland JMP, et al (2005) Pathological response following long‐course neoadjuvant chemoradiotherapy for locally advanced rectal cancer. Histopathology 47:141\u0026ndash;146. https://doi.org/10.1111/j.1365-2559.2005.02176.x\u003c/li\u003e\n\u003cli\u003eLin H, Hua J, Wang Y, et al (2025) Prognostic and predictive values of a multimodal nomogram incorporating tumor and peritumor morphology with immune status in resectable lung adenocarcinoma. J Immunother Cancer 13:e010723. https://doi.org/10.1136/jitc-2024-010723\u003c/li\u003e\n\u003cli\u003eWang Y, Xu H, Hu C, et al (2022) Prognostic value and clinicopathological correlation of the tumor regression grade in neoadjuvant chemotherapy for gastric adenocarcinoma: A retrospective cohort study. J Gastrointest Oncol 13:1046\u0026ndash;1057. https://doi.org/10.21037/jgo-22-537\u003c/li\u003e\n\u003cli\u003eChen Y, Xu W, Li Y-L, et al (2022) CT-based radiomics showing generalization to predict tumor regression grade for advanced gastric cancer treated with neoadjuvant chemotherapy. Front Oncol 12:758863. https://doi.org/10.3389/fonc.2022.758863\u003c/li\u003e\n\u003cli\u003eCui Y, Zhang J, Li Z, et al (2022) A CT-based deep learning radiomics nomogram for predicting the response to neoadjuvant chemotherapy in patients with locally advanced gastric cancer: A multicenter cohort study. eClinicalMedicine 46:101348. https://doi.org/10.1016/j.eclinm.2022.101348\u003c/li\u003e\n\u003cli\u003eLiu C, Li L, Chen X, et al (2024) Intratumoral and peritumoral radiomics predict pathological response after neoadjuvant chemotherapy against advanced gastric cancer. Insights Imaging 15:23. https://doi.org/10.1186/s13244-023-01584-6\u003c/li\u003e\n\u003cli\u003eChen Y-H, Xiao J, Chen X-J, et al (2020) Nomogram for predicting pathological complete response to neoadjuvant chemotherapy in patients with advanced gastric cancer. WJG 26:2427\u0026ndash;2439. https://doi.org/10.3748/wjg.v26.i19.2427\u003c/li\u003e\n\u003cli\u003eSun K-Y, Hu H-T, Chen S-L, et al (2020) CT-based radiomics scores predict response to neoadjuvant chemotherapy and survival in patients with gastric cancer. BMC Cancer 20:468. https://doi.org/10.1186/s12885-020-06970-7\u003c/li\u003e\n\u003cli\u003eWang W, Peng Y, Feng X, et al (2021) Development and validation of a computed tomography\u0026ndash;based radiomics signature to predict response to neoadjuvant chemotherapy for locally advanced gastric cancer. JAMA Netw Open 4:e2121143. https://doi.org/10.1001/jamanetworkopen.2021.21143\u003c/li\u003e\n\u003cli\u003eHan X, Wang M, Zheng Y, et al (2023) Delta-radiomics features for predicting the major pathological response to neoadjuvant chemoimmunotherapy in non-small cell lung cancer. Eur Radiol 34:2716\u0026ndash;2726. https://doi.org/10.1007/s00330-023-10241-x\u003c/li\u003e\n\u003cli\u003eLin P, Yang P-F, Chen S, et al (2020) A delta-radiomics model for preoperative evaluation of neoadjuvant chemotherapy response in high-grade osteosarcoma. Cancer Imaging 20:7. https://doi.org/10.1186/s40644-019-0283-8\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table 2","content":"\u003cp\u003eTable 2 is available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Locally advanced gastric cancer, Neoadjuvant chemotherapy, Delta computed tomography radiomics, Tumor regression grade","lastPublishedDoi":"10.21203/rs.3.rs-7072201/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7072201/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003ePurpose\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo develop and validate a delta computed tomography radiomics (delCT-RS) based nomogram for accurate preoperative prediction of tumor regression grade (TRG) in locally advanced gastric cancer (LAGC) patients following neoadjuvant chemotherapy (NAC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective study enrolled 147 LAGC patients. Two delineation strategies were compared: 1) contouring both the primary tumor and the largest lymph node (P + L) as regions of interest (ROIs), and 2) contouring only the primary tumor (P). Subsequently, radiomic features were extracted to construct corresponding radiomic models. This study compared the predictive accuracy of delCT-RS signatures to conventional single-phase radiomic signatures for TRG assessment. Then, delCT-RS signatures and clinical variables were combined into a nomogram. Finally, the prediction performance of nomogram was comprehensively evaluated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn assessing tumor response, delCT-RS outperformed single-phase radiomic signatures. Notably, delta computed tomography delCT-RS P + L demonstrated superior accuracy to delCT-RS P (delCT-RS P + L vs delCT-RS area under the curve (AUC): training cohort: 0.805 vs 0.727; validation cohort: 0.795 vs 0.655). The nomogram, combining delCT-RS P + L and clinical factors, achieved optimal performance among all models (training cohort AUC = 0.841; validation cohort AUC = 0.817). (p \u0026lt; 0.05)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, we innovatively employed a method that simultaneously delineated the primary tumor and the largest lymph node. This model can accurately predict TRG, effectively identify LAGC patients who can benefit from NAC, and provide scientific support for individualized treatment.\u003c/p\u003e","manuscriptTitle":"CT-based delta-radiomics nomogram to assess tumor regression grade in locally advanced gastric cancer patients following neoadjuvant chemotherapy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-14 10:46:24","doi":"10.21203/rs.3.rs-7072201/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-30T15:14:55+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-29T15:38:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"73668533356748556759308058239093312064","date":"2025-07-20T13:51:35+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-09T11:58:53+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-08T09:02:16+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-08T09:00:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"Abdominal Radiology","date":"2025-07-08T07:52:18+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"3a1996a3-2c73-4262-8985-466277c9c64d","owner":[],"postedDate":"July 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-09-22T15:59:04+00:00","versionOfRecord":{"articleIdentity":"rs-7072201","link":"https://doi.org/10.1007/s00261-025-05171-9","journal":{"identity":"abdominal-radiology","isVorOnly":false,"title":"Abdominal Radiology"},"publishedOn":"2025-09-17 15:57:19","publishedOnDateReadable":"September 17th, 2025"},"versionCreatedAt":"2025-07-14 10:46:24","video":"","vorDoi":"10.1007/s00261-025-05171-9","vorDoiUrl":"https://doi.org/10.1007/s00261-025-05171-9","workflowStages":[]},"version":"v1","identity":"rs-7072201","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7072201","identity":"rs-7072201","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}