Development and Validation of a General Clinical Model for Predicting Neoadjuvant Chemotherapy Efficacy in Locally Advanced Gastric Cancer: Evidence from Meta-Analysis and Real-World Study

Development and Validation of a General Clinical Model for Predicting Neoadjuvant Chemotherapy Efficacy in Locally Advanced Gastric Cancer: Evidence from Meta-Analysis and Real-World Study | 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 Development and Validation of a General Clinical Model for Predicting Neoadjuvant Chemotherapy Efficacy in Locally Advanced Gastric Cancer: Evidence from Meta-Analysis and Real-World Study Lu Wang, Xiaohu Sun, Siru Nie, Yingying Wang, Rui Guo, Shuwen Zheng, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6136117/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Neoadjuvant chemotherapy (NCT) is a cornerstone treatment for locally advanced gastric cancer (LAGC), yet patient responses vary significantly. This study aimed to develop and validate a general clinical model to predict NCT efficacy in LAGC patients. Methods A systematic review and meta-analysis were performed to identify independent clinical features associated with NCT efficacy. Using β coefficients, a risk score-based predictive model was constructed. Model performance was validated in 3 real-world cohorts using Area Under Curve (AUC) metrics. Prognostic utility was analyzed via Kaplan-Meier analysis. Additionally, an online NCT response prediction calculator was developed using R Shiny . Results A total of 4,014 patients from 25 high-quality cohort studies were included in the meta-analysis. Nine clinical features—CEA, tumor location, Lauren classification, histological grade, depth of invasion, lymph node metastasis, clinical stage, HER-2 status (IHC score), and Ki67—were incorporated into the final prediction model for NCT efficacy in LAGC. The present model demonstrated robust predictive performance, with AUCs of 0.760 (95% CI: 0.725–0.795), 0.786 (95% CI: 0.691–0.880), and 0.796 (95% CI: 0.718–0.875) across validation cohorts. NCT response was stratified into 4 levels based on risk scores, with increasing risk levels correlated with a progressive decline in treatment efficacy and poorer prognosis ( P < 0.001). The response rates in low-risk groups were 2.44- and 3.96-fold higher than those in high-risk and very high-risk groups, respectively. Conclusions This study establishes a robust and validated clinical model for predicting NCT efficacy and prognosis in LAGC patients. The accompanying online calculator provides a practical tool for personalized treatment planning. Future efforts will focus on expanding validation cohorts and refining the model to further optimize therapeutic decision-making for LAGC patients undergoing NCT. Trial registration: The protocol for the systematic review and meta-analysis was prospectively registered on PROSPERO (CRD42023483908) on March 12, 2023, prior to data collection. The validation cohorts (Cohorts 1–3) were derived from retrospective real-world data. As this study analyzed existing clinical records without prospective intervention, trial registration was not required for these cohorts. Stomach Neoplasms Neoadjuvant chemotherapy Response evaluation Meta-Analysis Clinical prediction model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Gastric cancer (GC) is an aggressive malignancy and ranks fourth among cancer-related deaths ( 1 , 2 ). Radical resection remains the optimal treatment for GC( 3 ), however, due to its insidious onset, many patients are diagnosed with an advanced stage, resulting in the occurrence of locally advanced gastric cancer (LAGC) and the loss of eligibility for initial radical surgery ( 4 , 5 ). For these patients, the combination of fluorouracil-based neoadjuvant chemotherapy (NCT) and surgery has been proven effective and recommended as a first-line treatment in current guidelines ( 6 , 7 ). Preoperative NCT for LAGC reduces tumor burden, increases R0 surgical resection rates and pathological complete response rates, and decreases recurrence and metastasis risks, ultimately improving both survival rate and quality of life ( 8 , 9 ). However, due to tumor heterogeneity, patients with similar clinical profiles may exhibit variable responses to NCT ( 10 ). Therefore, accurately assessing NCT efficacy and identifying patients most likely to benefit from NCT are critical for optimizing personalized treatment strategies. In recent years, the prediction of chemotherapy efficacy has gained increasing attention ( 11 – 13 ). A promising approach is developing predictive models based on general clinical features, including demographic data, biochemical markers, clinically relevant variables, and immunohistochemical indicators. These models exhibit several advantages, such as easy parameter accessibility, straightforward analysis, data stability, robust interpretability, and reliable results. Additionally, their independence from complex technical requirements and equipment constraints enhances the applicability in various healthcare settings, making these models cost-effective, simple to implement, and scalable( 14 ). Although prediction models based on general clinical characteristics have been widely applied across various fields ( 15 – 18 ), research focused on predicting NCT efficacy in LAGC patients remains limited. Among the few existing studies, Liu et al . developed a nomogram-based model incorporating tumor location, histological grade, clinical T stage, and carbohydrate antigen 724 to predict NCT treatment response in LAGC patients ( 19 ). However, existing models face several key limitations:( 1 ) Limited exploration of predictive factors—most models rely on a narrow set of clinical characteristics, failing to comprehensively capture the complexity of NCT response and limiting predictive accuracy;( 2 ) Small sample sizes and limited generalizability—many studies were based on single-institution cohort, which may not adequately represent the broader GC population;( 3 ) Variability in NCT regimens—differences in chemotherapy protocols across studies complicate the identification of universally applicable predictive markers, leading to inconsistent findings. Overall, the existing models for predicting NCT efficacy in LAGC patients lack clinical applicability, reliability, and stability, highlighting the need for a more comprehensive and validated predictive tool to guide clinical decision-making. The present study aims to bridge these gaps by conducting a systematic review and meta-analysis to identify key clinical features for predicting NCT efficacy in LAGC patients, followed by the development and validation of a general predictive model using 3 real-world cohorts. The research process is organized into 4 steps, as shown in Fig. 1 . Ultimately, this study seeks to provide valuable insights for optimizing NCT response prediction, refining treatment strategies, and advancing personalized therapeutic approaches for LAGC patients. Methods Population of the study Derivation cohort The derivation cohort was established through a comprehensive search of electronic databases (PubMed, Web of Science, Embase, Cochrane, CNKI, Wanfang, VIP) up to August 2023. A combined search strategy incorporating text terms and MeSH terms was implemented. To ensure scientific rigor, the study was preregistered in PROSPERO (CRD42023483908) prior to data collection and followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines( 20 ). The cohort included prospective and retrospective cohort studies without language restrictions. Search terms included "Gastric cancer", "Neoadjuvant Therapy", and "Risk Factor" with enrolled patients having clinical stages of cT1 ~ 4b, any N, and M0. All included studies reported risk ratios (RRs) or standardized mean differences (SMDs) with 95% confidence intervals (CIs)( 21 ), evaluated by the Newcastle-Ottawa Scale( 18 ). A flowchart illustrating the study selection process is provided in Fig. 2 . Additional details on inclusion criteria, search strategy, data extraction, and quality assessment are available in the Supplemental Information (Supplemental Materials and Methods and Table S1 ). Validation cohort Three independent validation cohorts were included in the study. Cohort 1 comprised LAGC patients from Liaoning Cancer Hospital (June 1, 2014, to July 31, 2021), Cohort 2 included patients from Liaoning Cancer Hospital (August 1, 2021, to January 31, 2022), and Cohort 3 included newly diagnosed LAGC patients from the First Hospital of China Medical University (August 21, 2023, to October 22, 2024). Patients were classified according to the AJCC 8th edition GC cTNM staging system( 22 ), with clinical stages cT1 ~ 4b, any N, and M0. Inclusion criteria required patients to be aged 18 to 75 years, have been histologically confirmed gastric adenocarcinoma via gastroscopy biopsy, and have received at least 2 cycles of NCT prior to surgery, followed by radical resection. Exclusion criteria included patients with concurrent primary malignancies or those diagnosed with other malignant tumors within the past 3 years. In cohort 1, overall survival (OS) was defined as the time from enrollment to death from any cause. Survival status was tracked through outpatient visits, telephone interviews, and electronic medical records. Follow-ups were conducted every 3 months during the first 2 years and every 6 months thereafter, with the final follow-up date on March 1, 2023. Patients lost to follow-up or alive at the end of the study period were censored in the analysis. The study was approved by the Medical Ethics Committees of Liaoning Cancer Hospital and the First Hospital of China Medical University.The requirement for informed consent was waived due to the retrospective nature of the analysis. Assessment of NCT treatment response Validation cohorts 1 and 2 used the Mandard tumor regression grading (TRG) system to assess pathological efficacy of NCT in postoperative specimens. The TRG grades were defined as: grade 1 (no residual cancer cells, total fibrosis), grade 2 (rare residual cancer cells, scattered through the fibrosis), grade 3 (more residual cancer cells, but outgrown by fibrosis), grade 4 (residual cancer cells outgrowing fibrosis), and grade 5 (absence of regressive changes)( 23 ). Grades 1–3 were considered responsive, while grades 4–5 were non-responsive. Validation cohort 3 employed the AJCC 8th edition staging and College of American Pathologists (CAP) TRG system( 24 ), which includes grade 0 (no viable cancer cells), grade 1 (single cells or rare small groups of cancer cells), grade 2 (more residual cancer with evident tumor regression), and grade 3 (extensive residual cancer with no evident tumor regression). AJCC/CAP grades 0–1 were classified as responsive, and grades 2–3 as non-responsive. Two senior pathologists evaluated the pathological responses independently under double-blind conditions, with a third pathologist mediating any disagreements. The Response Evaluation Criteria In Solid Tumors (RECIST v1.1) was used to assess radiological responses. Complete response (CR) is defined as disappearance of all target lesions, partial response (PR) as a ≥ 30% decrease in the sum of target lesions, progressive disease (PD) as a ≥ 20% increase, and stable disease (SD) when neither PR nor PD criteria are met( 25 , 26 ). Patients achieving CR or PR were classified as responsive, and those with SD or PD as non-responsive.. Imaging evaluations were performed by 2 senior radiologists independently, with a third radiologist mediating disagreements. The NCT response rate is defined as the proportion of patients who exhibited responsive outcomes to NCT among the total population. Specifically: Statistical analyses Systematic review and Meta-analysis We conducted a systematic review and meta-analysis by extracting RRs and 95% CIs for features associated with NCT efficacy from studies in the derivation cohort. To minimize the impact of varying efficacy evaluation standards across studies, we standardized the clinical features, without distinguishing between TRG and RECIST criteria. TRG was prioritized as the primary evaluation criterion when both standards were reported in a study. For studies using only TRG or RECIST as the sole efficacy evaluation criterion in at least 2 studies, subgroup analyses were performed based on the respective criterion. Statistical analyses were performed using Stata SE 15.1, and forest plots were generated. Heterogeneity was assessed using the Q test combined with the I² statistic. A fixed-effects model was applied if P ≥ 0.10 for the Q test and I² ≤ 50%, indicating sufficient homogeneity across studies. In cases of significant heterogeneity (defined as either a Q test P 50%), a random-effects model was employed( 27 ). Statistical significance was determined by a pooled RR with 95%CI not encompassing 1, or P ≤ 0.05, indicating meaningful differences between groups. Subgroup analyses were performed when features were categorized differently across studies (e.g., age cutoffs of 55 vs. 60 years), following the predefined methods based on TRG and RECIST efficacy criteria. Model development We developed a prediction model for NCT efficacy following the methodology outlined by Chen D et al .( 28 ). First, features for the model were selected from the systematic reviews and meta-analyses described above. Combined RRs and their corresponding 95%CIs were extracted for each feature. The β coefficient for each feature was calculated using the formula: β = ln(RR). The β values were then multiplied by 10 and rounded to one decimal place. Values between 0 and 0.2 were recorded as 0, those between 0.3 and 0.7 as 0.5, and values between 0.8 and 0.9 as 1.0, yielding scores for each feature. Features were then classified based on meta-analysis results, the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology, and the Japanese Gastric Cancer Association (JGCA) guidelines for GC( 29 ). Each category was assigned a corresponding score, and the overall score was calculated by summing all feature scores within the prediction model. Model validation To validate the predictive performance of the model, we assessed its generalizability using 3 real-world cohorts. Each patient's features were scored according to the model, and a total score was calculated for each individual. Sensitivity, specificity, and area under the curve (AUC) were computed across various cutoff values. To compare the differences in AUC between the TRG and RECIST subgroups, we performed DeLong's test( 30 ). This analysis was implemented using the pROC package in R 4.2.0, with a two-sided P < 0.05 considered statistically significant( 31 ). The optimal cutoff value was determined by the maximum Youden index( 32 )for classifying patients into low-to-moderate and high-risk groups( 33 ). Receiver operating characteristic curves(ROC) for each group were generated to further stratify these groups into 4 risk levels—relatively low, moderate, high, and very high risk. The linear relationship between NCT treatment response rates and risk stratification was quantitatively evaluated by Pearson correlation coefficients, with values ranging from − 1 (perfect negative correlation) to 1 (perfect positive correlation), with 0 indicating absence of linear association. Kaplan-Meier survival curves were constructed for each risk level to examine the relationship between risk stratification and survival outcomes. Analyses were conducted using SPSS 26.0 and R 4.2.0. Model Deployment in a Web-Based Interface To provide a more convenient tool for evaluating NCT efficacy in clinical settings, we utilized the Shiny framework in R to transform the data into a user-friendly web interface for both clinicians and patients. First, we defined a score mapping table which could associate clinical indicator options with their corresponding scores. Next, we employed the `fluidPage` function to create a concise and clear user interface. In this interface, the title panel effectively displays the function of the calculator, and a selection input box in the sidebar facilitates users to select various clinical indicator options. Additionally, the sidebar includes 2 action buttons: "Calculate Risk Level" and "One-Click Clear Option", which simplify operations for doctors and allow for easy re-selection. Subsequently, we defined the `reactiveValues` object `user_scores` to store the scores selected by the user. As users select different options, the calculator listens for events via `observeEvent`, ensuring that the value of `user_scores` is updated promptly and accurately. Finally, we employed the `calculate_score` function to derive the corresponding score from the score mapping table based on the options selected by the user. This allowed us to calculate the total score and determine the patient's risk level accordingly. Results The baseline characteristics of the cohorts The derivation cohort The derivation cohort was established from 2 prospective cohort studies and 23 retrospective cohort studies, with a flowchart illustrating the study selection process presented in Fig. 2 .The baseline characteristics of the cohort included in the meta-analysis were detailed in Table S2 . According to the Newcastle-Ottawa scale, all 25 studies were classified as high-quality, each achieving scores exceeding 8 points (Table S3 ). Among them, 2,190 patients were evaluated as having an effective response to chemotherapy, resulting in an response rate of approximately 54.6%. A total of 62 clinical features related to the NCT efficacy in patients with LAGC were identified from the original literature. Among these, 20 features were reported in more than 2 studies, including sex, age, smoking history, carcinoembryonic antigen (CEA), carbohydrate antigen 19 − 9 (CA19-9), tumor location, Borrmann classification, Lauren classification, histological grade, depth of invasion(cT), lymph node metastasis(cN), clinical stage, NCT regimen, NCT cycles, human epidermal growth factor receptor 2(HER-2) status (IHC score), epidermal growth factor receptor (EGFR), Ki67, tumor protein 53(P53), topoisomerase II (TOPOII), and neural invasion. Notably, with the exception of the NCT regimen and cycles, information on the other features were collected prior to the initiation of the first NCT treatment. Detailed informations regarding the included features were presented in Tables S4 and S5. A systematic review and meta-analysis concerning the NCT efficacy in patients with LAGC was conducted based on these 20 features. The validation cohort Validation cohort 1 comprised 753 patients, cohort 2 consisted of 100 patients, and cohort 3 included 127 patients. Patient selection is summarized in Figure S1 . In validation cohort 1, 27.0% were female, with a mean age of 59.05 ± 8.89 years. Of these patients, 99.6% received fluorouracil-based NCT regimens. Among them, 63.9% received the SOX regimen, 8.8% received XELOX, and 27.3% received other fluorouracil-based regimens such as FOLFOX. Notably, 9.0%, 27.4%, and 45.3% underwent more than 2 cycles of NCT. According to TRG classification, 496 patients (65.9%) were responders, while 461 patients (61.2%) met the RECIST criteria for response. In validation cohort 2, 32.0% were female, with a mean age of 64 ± 8.49 years. All of the patients received fluorouracil-based NCT regimens. Of these, 60% received the SOX regimen, 11% received XELOX, and 29% received other fluorouracil-based regimens such as FOLFOX. Additionally, 46% underwent more than 2 cycles of NCT. Based on TRG classification, 65 patients (65%) were responders, while 66 patients (66%) were classified as responders according to RECIST. In validation cohort 3, 27.6% were female, with a mean age of 61.83 ± 8.7 years. All of the patients received fluorouracil-based NCT regimens. Among these patients, 25.2% received the SOX regimen, 53.5% received XELOX, and 21.3% received other fluorouracil-based regimens such as FOLFOX. Additionally, 37.8% underwent more than 2 cycles of NCT. Based on TRG classification, 33 patients (26.0%) were identified as responders, while 66 patients (45.7%) were classified as responders according to RECIST. The baseline characteristics of patients in validation cohorts 1, 2, and 3 are presented in Table S6 -1. Among these cohorts, there were no significant differences in sex, age, Borrmann classification, NCT regimen, or NCT cycle between the responsive and the non-responsive group ( P > 0.05). A comparison of the baseline characteristics across the 3 validation cohorts is illustrated in Table S6 -2, showing no significant differences observed in sex, Borrmann classification, histological grade, or HER-2 status among 3 cohorts ( P > 0.05). Development of the predicting model for NCT efficacy in LAGC using Meta-Analysis Identification of clinical features associated with the NCT efficacy We first stratified the 20 clinical features according to their status in various studies. The features requiring stratification included age (≤ 55 years old, > 55 years old, ≤ 60 years old, and > 60 years old), tumor location (upper, middle, lower, and total stomach)( 34 ), Lauren classification (intestinal type, diffuse type, and mixed type), histological grade (well, moderately, poorly differentiated and undifferentiated), depth of invasion (T2, T2-T3, T3, and T4), lymph node involvement (N0, N1, N2, and N3), clinical stage (stage II, stage I + II, and stage III), NCT regimen (SOX, XELOX, and FOLFOX), and NCT cycle (2 cycles, 3 cycles, and 4 cycles). First, we conducted a pooled analysis of these clinical features. Subsequently, we performed subgroup analyses on the features assessed by TRG or RECIST criteria in at least 2 studies. In the TRG subgroup, we analyzed features including sex, tumor location, Borrmann classification, histological grade, depth of invasion, and lymph node metastasis. In the RECIST subgroup, the analyzed features included sex, tumor location, Borrmann classification, and histological grade.The features across different strata and subgroups are presented in Table S7 . The results indicated that 9 out of the 20 features were associated with the NCT efficacy in LAGC. These features, along with their respective combined RRs, were as follows: CEA ≤ 5 µg/L (RR = 1.10), upper GC (RR = 1.77), lower GC (RR = 0.87), intestinal-type GC (RR = 1.28), diffuse-type GC (RR = 1.16), well-differentiated GC (RR = 1.70), moderately-differentiated GC (RR = 1.41), T2 GC (RR = 1.90), T3 GC (RR = 1.52), T4 GC (RR = 0.78), N0 lymph node status (RR = 1.75), N1 lymph node status (RR = 1.35), N3 lymph node status (RR = 0.62), clinical stage I + II (RR = 1.42), HER-2 (3+) (IHC score) (RR = 4.16), and Ki67 ≥ 10% (RR = 2.29) (Fig. 3 ). The meta-analysis results of these features across different strata and subgroups are presented in Supplemental Materials and Methods. Development of the predicting model To further develop the prediction model for NCT efficacy based on clinical features, we converted the RR values into β coefficients and assigned the risk scores for each clinical feature. Detailed information, including the number of studies, sample size, pooled RRs, 95%CIs, β coefficients, and risk scores, was provided in Table S8 . The final model for predicting NCT efficacy in LAGC patients, based on general clinical characteristics, is summarized in Table 1.The 9 parameters and their corresponding risk scores are as follows: CEA : ≤5 µg/L (1 point), > 5 µg/L (0 points) Tumor location : Upper (5.5 points), Lower (− 1.5 points), Other regions (0 points) Lauren classification : Intestinal type (2.5 points), Diffuse type (1.5 points), Mixed type (0 points) Histological grade : Well differentiated (5.5 points), Moderately differentiated (3.5 points), Poorly differentiated (0 points) Depth of invasion : T2 (6.5 points), T3 (4 points), T4 (− 2.5 points) Lymph node metastasis : N0 (5.5 points), N1 (3 points), N2 (0 points), N3 (− 5 points) Clinical stage : Stages I/II (3.5 points), Stage III (0 points) HER-2 status : HER-2 3+ (IHC score) (14.5 points), HER-2 1 + ~ 2+ (IHC score) (0 points) Ki67 :10% (0 points), ≥ 10% (− 2.5 points) This prediction model was designed for LAGC patients aged 18 to 75 years and applicable to both white and Asian populations. Validation of the predicting model for NCT efficacy in LAGC using real-world data To further evaluate the performance of the predictive model for NCT efficacy, the model was validated in cohorts 1, 2, and 3. The ROC curves for each cohort were illustrated in Fig. 4 . In validation cohort 1, the AUC was 0.760 (95%CI: 0.725–0.795) for the TRG subgroup, and 0.646 (95%CI: 0.606–0.686) for the RECIST subgroup. In validation cohort 2, the AUC was 0.786 (95%CI: 0.691–0.880) for the TRG subgroup, and 0.639 (95%CI: 0.523–0.755) for the RECIST subgroup. In validation cohort 3, the AUC was 0.796 (95%CI: 0.718–0.875) for the TRG subgroup, and 0.659 (95%CI: 0.564–0.755) for the RECIST subgroup. To further evaluate the performance of the predictive model for NCT efficacy, the model was validated in cohorts 1, 2, and 3. The ROC curves for each cohort were illustrated in Fig. 4 . In validation cohort 1, the AUC was 0.760 (95%CI: 0.725–0.795) for the TRG subgroup, and 0.646 (95%CI: 0.606–0.686) for the RECIST subgroup (DeLong's test: D = -4.172, P = 3.195e-05). In validation cohort 2, the AUC was 0.786 (95%CI: 0.691–0.880) for the TRG subgroup, and 0.639 (95%CI: 0.523–0.755) for the RECIST subgroup (DeLong's test: D = -2.1696, P = 0.031). In validation cohort 3, the AUC was 0.796 (95%CI: 0.718–0.875) for the TRG subgroup, and 0.659 (95%CI: 0.564–0.755) for the RECIST subgroup (DeLong's test: D = -1.9266, P = 0.056). These results indicated that the TRG subgroup demonstrated statistically greater accuracy than the RECIST subgroup in predicting NCT treatment response (significant in cohorts 1 and 2, with borderline significance in cohort 3). Consequently, we calculated the maximum Youden index for the TRG subgroup, identifying 7.75 as the optimal cutoff value, which yielded a sensitivity of 0.704 and a specificity of 0.739. Sensitivity and specificity for various cutoff risk scores are presented in Table S9 . Evaluation of NCT efficacy and prognosis using risk scores To further evaluate the relationship between patients' risk scores and NCT efficacy, we used the optimal cutoff value (7.75) to classify patients in the validation cohort. Patients with a total score exceeding 7.75 were categorized into the low-to-moderate risk group, while those with a score below 7.75 were placed in the high-risk group. ROC curves were plotted for each group, and the maximum Youden index was used to identify additional optimal cutoff points. This analysis allowed us to further stratify patients into 4 risk levels: very high risk (< -1.5), high risk (-1.5 to 7.5), intermediate risk (8 to 15), and low risk (≥ 15). In the 3 independent validation cohorts, the distribution of patients across these risk groups was as follows: low-risk group (n = 198, 18, 6), moderate-risk group (n = 218, 26, 25), high-risk group (n = 185, 38, 29), and very high-risk group (n = 152, 18, 67). Across all 3 validation cohorts, the NCT efficacy demonstrated a gradual decrease with ascending risk stratification, with statistically significant differences among 4 risk groups ( P < 0.05) (Fig. 5 A-C). Pearson correlation analysis confirmed a significant negative correlation between elevated risk levels and NCT response rate, with correlation coefficients of R1 = -0.98, R2 = -0.86, and R3 = -0.93 for validation cohorts 1–3, respectively (Fig. 5 D-F). In the validation cohorts, the NCT response rates for patients in the low-risk groups were 2.44-, 1.31-, and 1.51-fold higher compared with those in the high-risk groups,and were 3.96-, 1.67-, and 1.94-fold higher compared with those in the very high-risk groups, respectively(Fig. 5 G). Furthermore, Kaplan-Meier survival curves were used to explore the relationship between risk levels and survival outcomes in patients from validation cohort 1. The Kaplan-Meier survival analysis demonstrated that higher risk levels were significantly associated with shorter overall survival ( P < 0.05) (Fig. 5 H). Network deployment of the predictive model Furthermore, we developed a risk calculator for the NCT efficacy model. By inputting 9 general clinical characteristics of the patient, this calculator can provide risk levels about the NCT efficacy. Additionally, we have deployed the calculator online to offer a personalized evaluation tool to support clinical treatment decisions. (The URL is: https://sunxiaohu.shinyapps.io/tigersun/ ). Discussion NCT significantly improves the surgical success rate and survival of LAGC patients ( 35 ). Accurate prediction of NCT efficacy is essential for optimizing personalized treatment strategies( 36 ). The present study integrated systematic review, meta-analysis, and real-world cohort data to develop and validate a prediction model for NCT efficacy in LAGC, enabling precise risk stratification for NCT response and prognosis. To enhance its practicality, we deployed an intuitive web-based interface that could provide real-time risk assessments for both clinicians and patients. This research enhances the accuracy of NCT efficacy prediction, optimizes the allocation of medical resources, and promotes the development of personalized precision medicine. Furthermore, the model offers a foundation for deeper insights into GC biological characteristics and its response to chemotherapy. In recent years, predicting responses to preoperative chemotherapy has attracted more attention, with several studies developing NCT efficacy prediction models for LAGC patients based on clinical characteristics( 37 , 38 ). However, these models have some limitations, including restricted population representativeness, incomplete clinical characteristics, small sample sizes that may lead to overfitting, and short follow-up periods that hinder long-term prognosis analysis( 39 ). To address these limitations, we conducted a systematic review and meta-analysis of 25 high-quality cohort studies involving 4,014 LAGC patients, from which independent clinical features of NCT efficacy were identified and quantified. These features were then used to develop a 9-feature prediction model for LAGC patients. Furthermore, our meta-analysis has been registered with PROSPERO, ensuring the transparency, reliability, and reproducibility of the findings. Among the 9 features in the prediction model, tumor location in the gastric antrum, invasion beyond the serosa layer, more than 7 lymph node metastases, and high tumor cell activity (Ki67 ≥ 10%) were associated with an ineffective NCT response. In contrast, NCT effective-related features included normal CEA levels, upper stomach tumors, intestinal-type, well differentiation, limited invasion depth to the submucosal layer, absence of lymph node metastasis, clinical stages I/II, and high HER-2 expression. Αmong these features, HER-2 expression is the most significant factor in the prediction model, serving as a key predictor of NCT efficacy. Similarly, a retrospective study showed that high HER-2 expression could independently predict drug sensitivity in LAGC patients undergoing NCT( 40 ). This finding is also consistent with previous breast cancer researches, in which high HER-2 expression was associated with higher sensitivity to fluorouracil-based chemotherapy (e.g., FEC, CAF, CMF)( 41 , 42 ) and better prognosis( 43 ). Although HER-2 is a conventional target for therapies like trastuzumab, its role in predicting chemotherapy efficacy and outcomes is still unclear. This uncertainty may due to the involvement of the HER-2 signaling pathway in regulating the cell cycle, promoting proliferation, and inhibiting apoptosis( 44 ). After activation, HER-2 could trigger downstream pathways, including PI3K-AKT and RAS-MAPK signaling pathway, which are critical for tumor cell sensitivity to chemotherapy( 45 ). Future researches should focus on clarifying how elevated HER-2 expression enhances chemotherapy efficacy, paving the way for more precise NCT strategies in LAGC. In 3 real-world validation cohorts, the models achieved AUC values of 0.760, 0.786, and 0.796, outperforming the existing NCT prediction model for LAGC based on general clinical features (AUC 0.65)( 15 ). Notably, models using TRG as the evaluation criteria outperformed those based on RECIST. In the TRG subgroup, with 7.75 as the optimal cutoff value, the sensitivity and specificity for predicting NCT efficacy in LAGC patients were 70.4% and 73.9%, respectively. Several factors may explain why TRG performs better than RECIST: First, the TRG criteria provides a more comprehensive evaluation by assessing tumor tissue shrinkage alongside microstructural changes, such as tumor cell density and fibrosis, whereas RECIST focuses solely on tumor size without accounting for cytological changes. Second, NCT may induce histological changes in tumors during early treatment, even without significant size changes. The RECIST criteria, however, focus more on size changes, failing to capture those early biological responses. Finally, RECIST relies on tumor size measurements, which are prone to errors due to inconsistent techniques and operator variability. In contrast, TRG is less affected by such errors, providing higher accuracy and reliability in efficacy assessment. Notably, in validation cohorts, the NCT response rates in the high-risk group were significantly lower than those in the low-risk group. This trend was further exacerbated in the very high-risk group, with a significant decrease in drug response rates compared to the low-risk group. These findings demonstrate that NCT response rates decrease progressively as the risk stratification increases, particularly in the very high-risk group with the highest risk of resistance. Moreover, patient outcomes deteriorated progressively as the individuals risk increasing from low to high, highlighting the ability of the risk scoring system to effectively stratify efficacy and prognosis. This approach not only helps identify high-risk patients but also supports the rational allocation of resources, clinical decision-making, and patient management. Additionally, we developed an online predictive calculator for NCT response risk stratification in LAGC patients, providing clinicians a practical tool to predict NCT responses, assess risk levels, and facilitate personalized, stratified care. Compared to previous NCT efficacy models for LAGC patients based on general clinical feature, this study offers several advantages. The present study integrated extensive clinical data from multiple studies, using systematic reviews and meta-analysis to develop a more reliable and generalizable prediction model. The clinical features in this model were easy to obtain, cost-effective, clinically applicable, interpretable, stable, and easily updated. Unlike other existing models, which often rely on limited patient numbers and single NCT regimens, our model combined a broader range of clinical patients and multiple NCT regimens to better align with real-world clinical practice. Additionally, although many models only report calibration curves or C-indices, our approach quantified performance indicators, providing actionable insights for clinicians and researchers. Notably, our model included multiple validation cohorts, providing a more rigorous evaluation of its stability and reliability—an aspect that was often lacked in previous models. This study has several limitations. Firstly, although 66 features potentially related to NCT efficacy were identified across 25 derivation cohorts, some features—such as smoking history, lymphocyte counts, and tumor size—were reported in only one single study and could not be included in the meta-analysis. As more studies being published, future analyses will provide a stronger foundation for LAGC treatment decisions. Additionally, while HER-2 expression was found to be a key predictor of NCT response in LAGC patients, its impact on NCT efficacy in GC is not yet fully understood. Future research will help explore the correlation between HER-2 and NCT response, focusing on the signaling pathways and their interaction with chemotherapy agents to refine treatment strategies. In summary, this study presents a practical and user-friendly model for predicting NCT efficacy in LAGC patients, based on general clinical features. The model demonstrated robust predictive performance in terms of both efficacy and survival across 3 real-world populations. Looking ahead to the future, further refinement and expansion of validation cohorts will enhance the accuracy of the model, enable more personalized treatment strategies and ultimately improve survival outcomes of patients undergoing NCT for GC. Conclusions This study successfully developed and validated a general clinical prediction model for assessing NCT efficacy in LAGC patients. By integrating nine readily accessible clinical features—including tumor location, HER-2 status, and Ki67 expression—the model demonstrated robust predictive accuracy (AUC ranged from 0.760 to 0.796) across three independent real-world cohorts. Notably, the risk stratification system effectively categorized patients into four distinct risk groups, revealing a significant inverse correlation between elevated risk scores and NCT response rates (P < 0.001). Low-risk patients exhibited response rates 2.44- to 3.96-fold higher than high-risk counterparts, with survival outcomes further validating the model’s prognostic utility (P < 0.0001). The accompanying online calculator enhances clinical applicability, offering a practical tool for personalized treatment planning and resource allocation. While the current model addresses critical gaps in existing prediction tools, future studies should focus on expanding validation to diverse populations, integrating molecular biomarkers (e.g., genomic signatures), and refining the algorithm through prospective trials. These efforts will further optimize therapeutic decision-making, minimize unnecessary toxicity in non-responders, and ultimately improve survival outcomes for LAGC patients undergoing NCT. List of abbreviations NCT Neoadjuvant Chemotherapy LAGC Locally Advanced Gastric Cancer AUC Area Under the Curve TRG Tumor Regression Grade RECIST Response Evaluation Criteria in Solid Tumors HER2 Human Epidermal Growth Factor Receptor 2 CEA Carcinoembryonic Antigen CA19-9 Carbohydrate Antigen 19-9 AJCC American Joint Committee on Cance CAP College of American Pathologists OS Overall Survival RR Risk Ratio SMD Standardized Mean Difference CI Confidence Interval PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses NCCN National Comprehensive Cancer Network JGCA Japanese Gastric Cancer Association Declarations Conflict of interest The authors declare no potential conflicts of interest. Ethics approval and consent to participate This study was approved by the Ethics Committees of the First Hospital of China Medical University (Approval No. NF-SOP-07-1.2-01) and Liaoning Cancer Hospital (Approval No. 202208102). The requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized clinical data. Consent for publication Not applicable. Availability of data and materials The datasets generated and analyzed during the current study are not publicly available due to patient privacy restrictions but are available from the corresponding author (Y.H.G) on reasonable request. The meta-analysis data supporting the findings of this study are included in this published article and its supplementary information files. Competing interests The authors declare no potential conflicts of interest. Funding This work was supported by Noncommunicable Chronic Diseases-National Science and Technology Major Project (No. 2023ZD0501400) and the Education Department project of Liaoning Province (No. LJ212410159002). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Authors' contributions Conception and design: Y.H. Gong, L. Wang and X.H. Sun; Development of methodology: Y.Y. Wang, R. Guo and S.W. Zheng; Acquisition of data: Y.H. Gong, L. Wang, X.H. Sun and Y.K. Li; Analysis and interpretation of data: L. Wang, X.H. Sun, S.R. Nie; Writing, review, and/or revision of the manuscript: Y.H. Gong, L. Wang, X.H. Sun, J.J. Jing, Y.Y. Wang, R. Guo, S.W. Zheng; X.N. Qiu, T.T. Tao; Administrative, technical, or material support: Y.H. Gong; Study supervision: Y.H. Gong. Acknowledgements Not applicable. Authors' information Y.H.G. is a Professor in the Department of Tumor Etiology and Screening at the First Hospital of China Medical University, specializing in gastrointestinal oncology. L.W. and X.H.S. are research fellows focusing on cancer therapy. Other authors are affiliated with clinical and research departments in oncology. References Machlowska J, Baj J, Sitarz M, Maciejewski R, Sitarz R. Gastric Cancer: Epidemiology, Risk Factors, Classification, Genomic Characteristics and Treatment Strategies. International journal of molecular sciences. 2020;21(11). Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a cancer journal for clinicians. 2021;71(3):209-49. Elizabeth S, Magnus N, Heike G, Nicole CTvG, Florian L. Gastric cancer. The Lancet. 2020. Songun I, Putter H, Kranenbarg EM, Sasako M, van de Velde CJ. Surgical treatment of gastric cancer: 15-year follow-up results of the randomised nationwide Dutch D1D2 trial. The Lancet Oncology. 2010;11(5):439-49. Sasako M, Sano T, Yamamoto S, Kurokawa Y, Nashimoto A, Kurita A, et al. D2 lymphadenectomy alone or with para-aortic nodal dissection for gastric cancer. The New England journal of medicine. 2008;359(5):453-62. Ajani JA, D'Amico TA, Bentrem DJ, Chao J, Cooke D, Corvera C, et al. Gastric Cancer, Version 2.2022, NCCN Clinical Practice Guidelines in Oncology. Journal of the National Comprehensive Cancer Network : JNCCN. 2022;20(2):167-92. Wang FH, Zhang XT, Tang L, Wu Q, Cai MY, Li YF, et al. The Chinese Society of Clinical Oncology (CSCO): Clinical guidelines for the diagnosis and treatment of gastric cancer, 2023. Cancer communications (London, England). 2024;44(1):127-72. Joshi SS, Badgwell BD. Current treatment and recent progress in gastric cancer. CA: a cancer journal for clinicians. 2021;71(3):264-79. Mohamed Salah Emam Sayed H, Khaled Abdallah E, Ashraf Kamal A, Aly MA, Mohammed AH. Role of Neo-Adjuvant Chemotherapy in Decision Making for Surgical Management of Cancer Stomach. Qjm. 2024. Sato Y, Okamoto K, Kawaguchi T, Nakamura F, Miyamoto H, Takayama T. Treatment Response Predictors of Neoadjuvant Therapy for Locally Advanced Gastric Cancer: Current Status and Future Perspectives. Biomedicines. 2022;10(7). Imyanitov EN, Iyevleva AG. Molecular tests for prediction of tumor sensitivity to cytotoxic drugs. Cancer letters. 2022;526:41-52. Tang Z, Gu Y, Shi Z, Min L, Zhang Z, Zhou P, et al. Multiplex immune profiling reveals the role of serum immune proteomics in predicting response to preoperative chemotherapy of gastric cancer. Cell reports Medicine. 2023;4(2):100931. Dai M, Sayuri K, Keizaburo K, Ryuichi K, Yoshikazu I, Toyotsugu O, et al. P-125 Radiomic prediction model for pathological responses of neoadjuvant chemotherapy with S-1 plus oxaliplatin (G-SOX) in clinical stage III gastric cancer. Annals of Oncology. 2020. Laura B, Kym IES, Gary SC, Richard DR. Guide to presenting clinical prediction models for use in clinical settings. The BMJ. 2019. Hatta W, Tsuji Y, Yoshio T, Kakushima N, Hoteya S, Doyama H, et al. Prediction model of bleeding after endoscopic submucosal dissection for early gastric cancer: BEST-J score. Gut. 2021;70(3):476-84. William AJ, Christine L, Rebecca CL, Teresa M, Christopher M, Benjamin SW, et al. Abstract 14217: Clinical Predictive Models for Sudden Cardiac Arrest: A Systematic Review of the Literature. Circulation. 2017. Riley B, Gaurav G, Jason N, Benjamin K, Christine L, Jinny P, et al. THE GENERALIZABILITY OF CLINICAL PREDICTIVE MODELS FOR PRIMARY PREVENTION OF CARDIOVASCULAR DISEASE: RESULTS FROM INDEPENDENT EXTERNAL VALIDATIONS. Journal of the American College of Cardiology. 2020. Murtaza M, Richard WG, Vincent XL. Use of Sepsis Clinical Prediction Models to Improve Patient Care. JAMA Internal Medicine. 2023. Liu B, Xu YJ, Chu FR, Sun G, Zhao GD, Wang SZ. Development of a clinical nomogram for prediction of response to neoadjuvant chemotherapy in patients with advanced gastric cancer. World journal of gastrointestinal surgery. 2024;16(2):396-408. Page MJ, Moher D. Evaluations of the uptake and impact of the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) Statement and extensions: a scoping review. Systematic reviews. 2017;6(1):263. Esterman AJ. 1. Epidemiology, study design and data analysis (2nd edn). Mark Woodward, Chapman & Hall/CRC, Boca Raton, 2005. No. of pages: xxii + 849. Price: $79.95. ISBN: 1‐58488‐415‐0. Statistics in medicine. 2006;25:1439-40. AJCC Cancer Staging Manual. 8 ed: Springer Cham; 2016. XVII, 1032 p. Mandard AM, Dalibard F, Mandard JC, Marnay J, Henry-Amar M, Petiot JF, et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Clinicopathologic correlations. Cancer. 1994;73(11):2680-6. Dhanpat Jain WVC, Rondell P. Graham. Protocol for the Examination of Specimens from Patients with Well-Differentiated Neuroendocrine Tumors (Carcinoid Tumors) of the Stomach.2023. 2023. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European journal of cancer (Oxford, England : 1990). 2009;45(2):228-47. Litière S, Collette S, de Vries EG, Seymour L, Bogaerts J. RECIST - learning from the past to build the future. Nature reviews Clinical oncology. 2017;14(3):187-92. Deeks JJ AD. Cochrane Handbook for Systematic Reviews of Interventions. 2011. The Cochrane Collaboration. Chen D, Wang M, Shang X, Liu X, Liu X, Ge T, et al. Development and validation of an incidence risk prediction model for early foot ulcer in diabetes based on a high evidence systematic review and meta-analysis. Diabetes research and clinical practice. 2021;180:109040. Japanese classification of gastric carcinoma: 3rd English edition. Gastric cancer : official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2011;14(2):101-12. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837-45. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC bioinformatics. 2011;12:77. Cook NR. Use and misuse of the receiver operating characteristic curve in risk prediction. Circulation. 2007;115(7):928-35. Hassanzad M, Hajian-Tilaki K. Methods of determining optimal cut-point of diagnostic biomarkers with application of clinical data in ROC analysis: an update review. BMC medical research methodology. 2024;24(1):84. Japanese Gastric Cancer Association. Japanese classification of gastric cance. 2017. Das M. Neoadjuvant chemotherapy: survival benefit in gastric cancer. The Lancet Oncology. 2017;18(6):e307. David RB, James DB, Thomas PB, Daniel CS, Donna MG. Current and future cancer staging after neoadjuvant treatment for solid tumors. CA: a cancer journal for clinicians. 2020. Lin J, Yi-Hui T, Wen-Xing Z, Jacopo D, Amilcare P, Jian‐Wei X, et al. Body composition parameters predict pathological response and outcomes in locally advanced gastric cancer after neoadjuvant treatment: A multicenter, international study. Clinical Nutrition. 2021. Ziyu L, Shuangxi L, Xiangji Y, Lianhai Z, Fei S, Yongning J, et al. The clinical value and usage of inflammatory and nutritional markers in survival prediction for gastric cancer patients with neoadjuvant chemotherapy and D2 lymphadenectomy. Gastric cancer : official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2020. Orestis E, Michael S, Konstantina C, Thomas PAD, Matthias E, Georgia S. Developing clinical prediction models: a step-by-step guide. Bmj. 2024. Haibin C, Huai'e G, Xiyong B, Wei Z, Yuanyuan Z. Application of the predictors EGFR, HER2, Ki-67 and P53 on neoadjuvant chemotherapy in patients with advanced gastric cancer. China Medical Herald. 2014;11(2):35-8. Konecny G, Fritz M, Untch M, Lebeau A, Felber M, Lude S, et al. HER-2/neu overexpression and in vitro chemosensitivity to CMF and FEC in primary breast cancer. Breast cancer research and treatment. 2001;69(1):53-63. DiGiovanna MP, Stern DF, Edgerton S, Broadwater G, Dressler LG, Budman DR, et al. Influence of activation state of ErbB-2 (HER-2) on response to adjuvant cyclophosphamide, doxorubicin, and fluorouracil for stage II, node-positive breast cancer: study 8541 from the Cancer and Leukemia Group B. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008;26(14):2364-72. Zhang W, Tian H, Yang SH. The Efficacy of Neoadjuvant Chemotherapy for HER-2-Positive Locally Advanced Breast Cancer and Survival Analysis. Analytical cellular pathology (Amsterdam). 2017;2017:1350618. Zhou J, Zhang S, Luo M. LncRNA PCAT7 promotes the malignant progression of breast cancer by regulating ErbB/PI3K/Akt pathway. Future oncology (London, England). 2021;17(6):701-10. Vinod BS, Nair HH, Vijayakurup V, Shabna A, Shah S, Krishna A, et al. Resveratrol chemosensitizes HER-2-overexpressing breast cancer cells to docetaxel chemoresistance by inhibiting docetaxel-mediated activation of HER-2-Akt axis. Cell death discovery. 2015;1:15061. Additional Declarations No competing interests reported. Supplementary Files Supplementalinformation.docx SupplementalFigures.docx Supplementaltables1.xlsx Supplementaltables2.xlsx Supplementaltables3.xlsx Supplementaltables4.xlsx Supplementaltables5.xlsx Supplementaltables6.xlsx Supplementaltables7.xlsx Supplementaltables8.xlsx Cite Share Download PDF Status: Posted Version 1 posted 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-6136117","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":423347997,"identity":"258ddc33-5d2e-426f-aaf6-c6262bc64041","order_by":0,"name":"Lu Wang","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lu","middleName":"","lastName":"Wang","suffix":""},{"id":423347998,"identity":"ed03fb48-2d7c-49b0-b61c-6d0f255c84ba","order_by":1,"name":"Xiaohu Sun","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaohu","middleName":"","lastName":"Sun","suffix":""},{"id":423347999,"identity":"f8fca446-0a79-4eda-beb2-754b5ab5e803","order_by":2,"name":"Siru Nie","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Siru","middleName":"","lastName":"Nie","suffix":""},{"id":423348000,"identity":"3e47ef01-32da-44dd-b950-797fc4865f99","order_by":3,"name":"Yingying Wang","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yingying","middleName":"","lastName":"Wang","suffix":""},{"id":423348001,"identity":"675014b7-4acc-427c-9507-c047f174ac66","order_by":4,"name":"Rui Guo","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Rui","middleName":"","lastName":"Guo","suffix":""},{"id":423348002,"identity":"b79d4376-d35c-470d-a489-e327c2f54070","order_by":5,"name":"Shuwen Zheng","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shuwen","middleName":"","lastName":"Zheng","suffix":""},{"id":423348003,"identity":"58fcf499-8669-4cae-a5ef-db1ff8724ae2","order_by":6,"name":"Xunan Qiu","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xunan","middleName":"","lastName":"Qiu","suffix":""},{"id":423348004,"identity":"500ca23e-80a7-47e8-9f9c-19939108e35f","order_by":7,"name":"Tingting Tao","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Tingting","middleName":"","lastName":"Tao","suffix":""},{"id":423348005,"identity":"a2ac408b-046e-4c70-a6c0-82e06fe9cdc2","order_by":8,"name":"Jingjing Jing","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jingjing","middleName":"","lastName":"Jing","suffix":""},{"id":423348006,"identity":"ca275378-6eb7-4b5a-8b3b-53d7df0994a8","order_by":9,"name":"Yanke Li","email":"","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yanke","middleName":"","lastName":"Li","suffix":""},{"id":423348007,"identity":"edbe98de-d3e5-49a4-ab94-0bb04755c899","order_by":10,"name":"Yuehua Gong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwklEQVRIiWNgGAWjYDACCTBpA+HwkKAljXQth0nQYnC7/eLjgl/n5XVnJDA+eNvGIG9OUMudM8XGM/tuG267kcBsOLeNwXBnAyEtN3LSpHl7bieY3Uhgk+ZtY0gwOECclnMgLey/idSSfkya58cBsC3MRGmRvJHDbMzbkGy47czDZsk55yQMNxDSwncj/eFjnj928mbHkw9+eFNmI0/QFoUDPAYMjG0gJmMDAyya8AL5BvYHDAx/CCscBaNgFIyCEQwANW9Df9IipT8AAAAASUVORK5CYII=","orcid":"","institution":"the First Hospital of China Medical University","correspondingAuthor":true,"prefix":"","firstName":"Yuehua","middleName":"","lastName":"Gong","suffix":""}],"badges":[],"createdAt":"2025-03-01 18:23:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6136117/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6136117/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":78238494,"identity":"f7c9758a-d130-443c-b1dd-de761b0b0ef3","added_by":"auto","created_at":"2025-03-11 08:44:51","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1532905,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFlowchart of the predictive model development process\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e1. The conduction of systematic review and meta-analysis in derivation cohort.\u003c/p\u003e\n\u003cp\u003e2. Prediction model development for NCT efficacy in patients with LAGC.\u003c/p\u003e\n\u003cp\u003e3.NCT efficacy-related features collection and prediction model validation in 3 Real-World cohorts.\u003c/p\u003e\n\u003cp\u003e4.NCT efficacy and prognosis risk stratification, and deployment of the online NCT response prediction calculator.\u003c/p\u003e","description":"","filename":"ExtractPage11.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/deba497c40c268cc49842d2a.jpg"},{"id":78237223,"identity":"111be536-80b3-4223-8f7f-948bd7d7f01c","added_by":"auto","created_at":"2025-03-11 08:36:47","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":601594,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFlow diagram outlining the literature search and clinical features collection associated with NCT efficacy in patients with LAGC in systematic reviews and meta-analysis\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"ExtractPage12.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/33edc3ec1751fedcb58e10d2.jpg"},{"id":78236453,"identity":"98ea9a5f-b8c9-4f40-a0dd-a34fbf7e6bc8","added_by":"auto","created_at":"2025-03-11 08:28:47","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":624904,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePooled risk coefficients and heterogeneity of clinical features related to the NCT efficacy in LAGC patients based on systematic review and meta - analysis\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"ExtractPage13.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/e96266cfe2dfa766c7ecf9e8.jpg"},{"id":78237219,"identity":"6867a901-8b98-41a6-981d-b232bd11cc57","added_by":"auto","created_at":"2025-03-11 08:36:47","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":607525,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC curves of the NCT efficacy prediction model for LAGC patients in the validation cohorts\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA,C,E：The ROC curves for the TRG subgroup in validation cohort 1,2,3\u003c/p\u003e\n\u003cp\u003eB,D,F：The ROC curves for the RECIST subgroup in validation cohort 1,2,3\u003c/p\u003e","description":"","filename":"ExtractPage14.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/4e90d08e77b9559866e95354.jpg"},{"id":78238488,"identity":"16bc84be-ed23-42dc-ad14-04da241e7571","added_by":"auto","created_at":"2025-03-11 08:44:47","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1025885,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRisk stratification analysis of NCT response and survival outcomes in LAGC.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA,B,C：The patient distribution of NCT response in 4 risk groups of validation cohort 1,2,3. Treatment response was assessed using TRG criteria, demonstrating a progressive decline in NCT efficacy with ascending risk levels ( \u003cem\u003eP \u003c/em\u003e\u0026lt; 0.05).\u003c/p\u003e\n\u003cp\u003eD,E,F:Robust negative correlations between elevated risk levels and NCT response rates across all validation cohorts. Pearson correlation coefficients were R1 = -0.98 (Cohort 1), R2 = -0.86 (Cohort 2), and R3 = -0.93 (Cohort 3), with \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001 for all analyses.\u003c/p\u003e\n\u003cp\u003eG:Stratified analysis of NCT response rates across different risk groups in 3 validation cohorts revealed that, the low-risk groups achieved 4.96-fold, 2.67-fold, and 2.94-fold compared to patients in very high-risk groups, and 3.44-, 2.31-, and 2.51-fold higher compared with those in the high-risk groups..\u003c/p\u003e\n\u003cp\u003eH：Survival curves of patients with 4 risk levels in Validation Cohort 1 (P \u0026lt; 0.0001). The treatment response was judged using the TRG criteria. As the risk level increased, the prognosis of patients gradually decreased.\u003c/p\u003e","description":"","filename":"ExtractPage15.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/90942dadd319e00d4321d56b.jpg"},{"id":78240821,"identity":"07fb74ab-769f-4133-8ed3-30f5e4644c51","added_by":"auto","created_at":"2025-03-11 09:00:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5709827,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/4001809b-9fc7-4fc9-87fb-6d1d3930847b.pdf"},{"id":78237225,"identity":"0ad17402-cb91-49b5-b477-b0896822dd08","added_by":"auto","created_at":"2025-03-11 08:36:47","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":66914,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementalinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/92bb9bf03c6ff3d6bed6f146.docx"},{"id":78236462,"identity":"74258dbb-1703-4b61-9764-ef03b5f3be29","added_by":"auto","created_at":"2025-03-11 08:28:47","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2989209,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/23b2a035932453925c0a498f.docx"},{"id":78237222,"identity":"c7e8545a-9dff-4836-a918-6f124e60a2db","added_by":"auto","created_at":"2025-03-11 08:36:47","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":13742,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/cdc269afb9833cde29e37554.xlsx"},{"id":78239215,"identity":"43f25c5a-4b0b-4199-9f73-7ca5ba8568f9","added_by":"auto","created_at":"2025-03-11 08:52:47","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":17118,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/1ca2f8becbe52d89d3691238.xlsx"},{"id":78236457,"identity":"a712152f-a769-4977-9acd-4f5f5ceeeaaa","added_by":"auto","created_at":"2025-03-11 08:28:47","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":12052,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/485285386111eddbaa6363a1.xlsx"},{"id":78238498,"identity":"e66e6e93-a9e6-47ad-9aa7-e1db3b884bac","added_by":"auto","created_at":"2025-03-11 08:44:53","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":27584,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/85563a369c551a00598da686.xlsx"},{"id":78239214,"identity":"f5daa2fa-ddda-40b6-9383-293e26c57c49","added_by":"auto","created_at":"2025-03-11 08:52:47","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":11804,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/e63e3763d0b0a28d58b16689.xlsx"},{"id":78236451,"identity":"a362d3e1-f43e-4af7-8c52-45fd5f8cb7f8","added_by":"auto","created_at":"2025-03-11 08:28:47","extension":"xlsx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":18198,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables6.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/b7f8832a2f7ad24b89bfd9f5.xlsx"},{"id":78237233,"identity":"f1d35300-7272-49f5-8d66-25c4a898e7dd","added_by":"auto","created_at":"2025-03-11 08:36:47","extension":"xlsx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":15039,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables7.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/37809ec3f9588189cca13ea2.xlsx"},{"id":78238493,"identity":"e627c6af-751b-4127-828b-908b6c5a724a","added_by":"auto","created_at":"2025-03-11 08:44:48","extension":"xlsx","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":11651,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltables8.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6136117/v1/f1ed99227441a48db891814c.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development and Validation of a General Clinical Model for Predicting Neoadjuvant Chemotherapy Efficacy in Locally Advanced Gastric Cancer: Evidence from Meta-Analysis and Real-World Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGastric cancer (GC) is an aggressive malignancy and ranks fourth among cancer-related deaths (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Radical resection remains the optimal treatment for GC(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e), however, due to its insidious onset, many patients are diagnosed with an advanced stage, resulting in the occurrence of locally advanced gastric cancer (LAGC) and the loss of eligibility for initial radical surgery (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). For these patients, the combination of fluorouracil-based neoadjuvant chemotherapy (NCT) and surgery has been proven effective and recommended as a first-line treatment in current guidelines (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Preoperative NCT for LAGC reduces tumor burden, increases R0 surgical resection rates and pathological complete response rates, and decreases recurrence and metastasis risks, ultimately improving both survival rate and quality of life (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). However, due to tumor heterogeneity, patients with similar clinical profiles may exhibit variable responses to NCT (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). Therefore, accurately assessing NCT efficacy and identifying patients most likely to benefit from NCT are critical for optimizing personalized treatment strategies.\u003c/p\u003e \u003cp\u003eIn recent years, the prediction of chemotherapy efficacy has gained increasing attention (\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). A promising approach is developing predictive models based on general clinical features, including demographic data, biochemical markers, clinically relevant variables, and immunohistochemical indicators. These models exhibit several advantages, such as easy parameter accessibility, straightforward analysis, data stability, robust interpretability, and reliable results. Additionally, their independence from complex technical requirements and equipment constraints enhances the applicability in various healthcare settings, making these models cost-effective, simple to implement, and scalable(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAlthough prediction models based on general clinical characteristics have been widely applied across various fields (\u003cspan additionalcitationids=\"CR16 CR17\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e), research focused on predicting NCT efficacy in LAGC patients remains limited. Among the few existing studies, \u003cem\u003eLiu et al\u003c/em\u003e. developed a nomogram-based model incorporating tumor location, histological grade, clinical T stage, and carbohydrate antigen 724 to predict NCT treatment response in LAGC patients (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). However, existing models face several key limitations:(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) Limited exploration of predictive factors\u0026mdash;most models rely on a narrow set of clinical characteristics, failing to comprehensively capture the complexity of NCT response and limiting predictive accuracy;(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) Small sample sizes and limited generalizability\u0026mdash;many studies were based on single-institution cohort, which may not adequately represent the broader GC population;(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) Variability in NCT regimens\u0026mdash;differences in chemotherapy protocols across studies complicate the identification of universally applicable predictive markers, leading to inconsistent findings. Overall, the existing models for predicting NCT efficacy in LAGC patients lack clinical applicability, reliability, and stability, highlighting the need for a more comprehensive and validated predictive tool to guide clinical decision-making.\u003c/p\u003e \u003cp\u003eThe present study aims to bridge these gaps by conducting a systematic review and meta-analysis to identify key clinical features for predicting NCT efficacy in LAGC patients, followed by the development and validation of a general predictive model using 3 real-world cohorts. The research process is organized into 4 steps, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Ultimately, this study seeks to provide valuable insights for optimizing NCT response prediction, refining treatment strategies, and advancing personalized therapeutic approaches for LAGC patients.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePopulation of the study\u003c/h2\u003e \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e \u003ch2\u003eDerivation cohort\u003c/h2\u003e \u003cp\u003eThe derivation cohort was established through a comprehensive search of electronic databases (PubMed, Web of Science, Embase, Cochrane, CNKI, Wanfang, VIP) up to August 2023. A combined search strategy incorporating text terms and MeSH terms was implemented. To ensure scientific rigor, the study was preregistered in PROSPERO (CRD42023483908) prior to data collection and followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines(\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). The cohort included prospective and retrospective cohort studies without language restrictions. Search terms included \"Gastric cancer\", \"Neoadjuvant Therapy\", and \"Risk Factor\" with enrolled patients having clinical stages of cT1\u0026thinsp;~\u0026thinsp;4b, any N, and M0. All included studies reported risk ratios (RRs) or standardized mean differences (SMDs) with 95% confidence intervals (CIs)(\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e), evaluated by the Newcastle-Ottawa Scale(\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). A flowchart illustrating the study selection process is provided in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Additional details on inclusion criteria, search strategy, data extraction, and quality assessment are available in the Supplemental Information (Supplemental Materials and Methods and Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eValidation cohort\u003c/h3\u003e\n\u003cp\u003eThree independent validation cohorts were included in the study. Cohort 1 comprised LAGC patients from Liaoning Cancer Hospital (June 1, 2014, to July 31, 2021), Cohort 2 included patients from Liaoning Cancer Hospital (August 1, 2021, to January 31, 2022), and Cohort 3 included newly diagnosed LAGC patients from the First Hospital of China Medical University (August 21, 2023, to October 22, 2024).\u003c/p\u003e \u003cp\u003ePatients were classified according to the AJCC 8th edition GC cTNM staging system(\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e), with clinical stages cT1\u0026thinsp;~\u0026thinsp;4b, any N, and M0. Inclusion criteria required patients to be aged 18 to 75 years, have been histologically confirmed gastric adenocarcinoma via gastroscopy biopsy, and have received at least 2 cycles of NCT prior to surgery, followed by radical resection. Exclusion criteria included patients with concurrent primary malignancies or those diagnosed with other malignant tumors within the past 3 years.\u003c/p\u003e \u003cp\u003eIn cohort 1, overall survival (OS) was defined as the time from enrollment to death from any cause. Survival status was tracked through outpatient visits, telephone interviews, and electronic medical records. Follow-ups were conducted every 3 months during the first 2 years and every 6 months thereafter, with the final follow-up date on March 1, 2023. Patients lost to follow-up or alive at the end of the study period were censored in the analysis. The study was approved by the Medical Ethics Committees of Liaoning Cancer Hospital and the First Hospital of China Medical University.The requirement for informed consent was waived due to the retrospective nature of the analysis.\u003c/p\u003e\n\u003ch3\u003eAssessment of NCT treatment response\u003c/h3\u003e\n\u003cp\u003eValidation cohorts 1 and 2 used the Mandard tumor regression grading (TRG) system to assess pathological efficacy of NCT in postoperative specimens. The TRG grades were defined as: grade 1 (no residual cancer cells, total fibrosis), grade 2 (rare residual cancer cells, scattered through the fibrosis), grade 3 (more residual cancer cells, but outgrown by fibrosis), grade 4 (residual cancer cells outgrowing fibrosis), and grade 5 (absence of regressive changes)(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Grades 1\u0026ndash;3 were considered responsive, while grades 4\u0026ndash;5 were non-responsive.\u003c/p\u003e \u003cp\u003eValidation cohort 3 employed the AJCC 8th edition staging and College of American Pathologists (CAP) TRG system(\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e), which includes grade 0 (no viable cancer cells), grade 1 (single cells or rare small groups of cancer cells), grade 2 (more residual cancer with evident tumor regression), and grade 3 (extensive residual cancer with no evident tumor regression). AJCC/CAP grades 0\u0026ndash;1 were classified as responsive, and grades 2\u0026ndash;3 as non-responsive. Two senior pathologists evaluated the pathological responses independently under double-blind conditions, with a third pathologist mediating any disagreements.\u003c/p\u003e \u003cp\u003eThe Response Evaluation Criteria In Solid Tumors (RECIST v1.1) was used to assess radiological responses. Complete response (CR) is defined as disappearance of all target lesions, partial response (PR) as a\u0026thinsp;\u0026ge;\u0026thinsp;30% decrease in the sum of target lesions, progressive disease (PD) as a\u0026thinsp;\u0026ge;\u0026thinsp;20% increase, and stable disease (SD) when neither PR nor PD criteria are met(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). Patients achieving CR or PR were classified as responsive, and those with SD or PD as non-responsive.. Imaging evaluations were performed by 2 senior radiologists independently, with a third radiologist mediating disagreements.\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003c/span\u003e \u003cspan class=\"InlineEquation\"\u003e \u003c/span\u003eThe NCT response rate is defined as the proportion of patients who exhibited responsive outcomes to NCT among the total population. Specifically:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" width=\"461\" height=\"58\"\u003e\u003c/p\u003e\n\u003ch3\u003eStatistical analyses\u003c/h3\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eSystematic review and Meta-analysis\u003c/h2\u003e \u003cp\u003eWe conducted a systematic review and meta-analysis by extracting RRs and 95% CIs for features associated with NCT efficacy from studies in the derivation cohort. To minimize the impact of varying efficacy evaluation standards across studies, we standardized the clinical features, without distinguishing between TRG and RECIST criteria. TRG was prioritized as the primary evaluation criterion when both standards were reported in a study. For studies using only TRG or RECIST as the sole efficacy evaluation criterion in at least 2 studies, subgroup analyses were performed based on the respective criterion.\u003c/p\u003e \u003cp\u003eStatistical analyses were performed using Stata SE 15.1, and forest plots were generated. Heterogeneity was assessed using the Q test combined with the \u003cem\u003eI\u0026sup2;\u003c/em\u003e statistic. A fixed-effects model was applied if \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026ge;\u0026thinsp;0.10 for the Q test and \u003cem\u003eI\u0026sup2;\u003c/em\u003e \u0026le; 50%, indicating sufficient homogeneity across studies. In cases of significant heterogeneity (defined as either a Q test \u003cem\u003eP\u0026thinsp;\u0026lt;\u003c/em\u003e\u0026thinsp;0.10 or \u003cem\u003eI\u0026sup2; \u0026gt;\u003c/em\u003e 50%), a random-effects model was employed(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eStatistical significance was determined by a pooled RR with 95%CI not encompassing 1, or \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026le;\u0026thinsp;0.05, indicating meaningful differences between groups. Subgroup analyses were performed when features were categorized differently across studies (e.g., age cutoffs of 55 vs. 60 years), following the predefined methods based on TRG and RECIST efficacy criteria.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eModel development\u003c/h3\u003e\n\u003cp\u003eWe developed a prediction model for NCT efficacy following the methodology outlined by \u003cem\u003eChen D et al\u003c/em\u003e.(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). First, features for the model were selected from the systematic reviews and meta-analyses described above. Combined RRs and their corresponding 95%CIs were extracted for each feature. The \u003cem\u003eβ\u003c/em\u003e coefficient for each feature was calculated using the formula: \u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;ln(RR). The \u003cem\u003eβ\u003c/em\u003e values were then multiplied by 10 and rounded to one decimal place. Values between 0 and 0.2 were recorded as 0, those between 0.3 and 0.7 as 0.5, and values between 0.8 and 0.9 as 1.0, yielding scores for each feature. Features were then classified based on meta-analysis results, the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology, and the Japanese Gastric Cancer Association (JGCA) guidelines for GC(\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). Each category was assigned a corresponding score, and the overall score was calculated by summing all feature scores within the prediction model.\u003c/p\u003e\n\u003ch3\u003eModel validation\u003c/h3\u003e\n\u003cp\u003eTo validate the predictive performance of the model, we assessed its generalizability using 3 real-world cohorts. Each patient's features were scored according to the model, and a total score was calculated for each individual. Sensitivity, specificity, and area under the curve (AUC) were computed across various cutoff values. To compare the differences in AUC between the TRG and RECIST subgroups, we performed DeLong's test(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). This analysis was implemented using the pROC package in R 4.2.0, with a two-sided \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 considered statistically significant(\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). The optimal cutoff value was determined by the maximum Youden index(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e)for classifying patients into low-to-moderate and high-risk groups(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). Receiver operating characteristic curves(ROC) for each group were generated to further stratify these groups into 4 risk levels\u0026mdash;relatively low, moderate, high, and very high risk.\u003c/p\u003e \u003cp\u003eThe linear relationship between NCT treatment response rates and risk stratification was quantitatively evaluated by Pearson correlation coefficients, with values ranging from \u0026minus;\u0026thinsp;1 (perfect negative correlation) to 1 (perfect positive correlation), with 0 indicating absence of linear association.\u003c/p\u003e \u003cp\u003eKaplan-Meier survival curves were constructed for each risk level to examine the relationship between risk stratification and survival outcomes. Analyses were conducted using SPSS 26.0 and R 4.2.0.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eModel Deployment in a Web-Based Interface\u003c/h2\u003e \u003cp\u003eTo provide a more convenient tool for evaluating NCT efficacy in clinical settings, we utilized the Shiny framework in R to transform the data into a user-friendly web interface for both clinicians and patients. First, we defined a score mapping table which could associate clinical indicator options with their corresponding scores. Next, we employed the `fluidPage` function to create a concise and clear user interface. In this interface, the title panel effectively displays the function of the calculator, and a selection input box in the sidebar facilitates users to select various clinical indicator options. Additionally, the sidebar includes 2 action buttons: \"Calculate Risk Level\" and \"One-Click Clear Option\", which simplify operations for doctors and allow for easy re-selection. Subsequently, we defined the `reactiveValues` object `user_scores` to store the scores selected by the user. As users select different options, the calculator listens for events via `observeEvent`, ensuring that the value of `user_scores` is updated promptly and accurately. Finally, we employed the `calculate_score` function to derive the corresponding score from the score mapping table based on the options selected by the user. This allowed us to calculate the total score and determine the patient's risk level accordingly.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eThe baseline characteristics of the cohorts\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003eThe derivation cohort\u003c/h2\u003e \u003cp\u003eThe derivation cohort was established from 2 prospective cohort studies and 23 retrospective cohort studies, with a flowchart illustrating the study selection process presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e.The baseline characteristics of the cohort included in the meta-analysis were detailed in Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e. According to the Newcastle-Ottawa scale, all 25 studies were classified as high-quality, each achieving scores exceeding 8 points (Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e). Among them, 2,190 patients were evaluated as having an effective response to chemotherapy, resulting in an response rate of approximately 54.6%.\u003c/p\u003e \u003cp\u003eA total of 62 clinical features related to the NCT efficacy in patients with LAGC were identified from the original literature. Among these, 20 features were reported in more than 2 studies, including sex, age, smoking history, carcinoembryonic antigen (CEA), carbohydrate antigen 19\u0026thinsp;\u0026minus;\u0026thinsp;9 (CA19-9), tumor location, Borrmann classification, Lauren classification, histological grade, depth of invasion(cT), lymph node metastasis(cN), clinical stage, NCT regimen, NCT cycles, human epidermal growth factor receptor 2(HER-2) status (IHC score), epidermal growth factor receptor (EGFR), Ki67, tumor protein 53(P53), topoisomerase II (TOPOII), and neural invasion. Notably, with the exception of the NCT regimen and cycles, information on the other features were collected prior to the initiation of the first NCT treatment. Detailed informations regarding the included features were presented in Tables S4 and S5. A systematic review and meta-analysis concerning the NCT efficacy in patients with LAGC was conducted based on these 20 features.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eThe validation cohort\u003c/h2\u003e \u003cp\u003eValidation cohort 1 comprised 753 patients, cohort 2 consisted of 100 patients, and cohort 3 included 127 patients. Patient selection is summarized in Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eIn validation cohort 1, 27.0% were female, with a mean age of 59.05\u0026thinsp;\u0026plusmn;\u0026thinsp;8.89 years. Of these patients, 99.6% received fluorouracil-based NCT regimens. Among them, 63.9% received the SOX regimen, 8.8% received XELOX, and 27.3% received other fluorouracil-based regimens such as FOLFOX. Notably, 9.0%, 27.4%, and 45.3% underwent more than 2 cycles of NCT. According to TRG classification, 496 patients (65.9%) were responders, while 461 patients (61.2%) met the RECIST criteria for response.\u003c/p\u003e \u003cp\u003eIn validation cohort 2, 32.0% were female, with a mean age of 64\u0026thinsp;\u0026plusmn;\u0026thinsp;8.49 years. All of the patients received fluorouracil-based NCT regimens. Of these, 60% received the SOX regimen, 11% received XELOX, and 29% received other fluorouracil-based regimens such as FOLFOX. Additionally, 46% underwent more than 2 cycles of NCT. Based on TRG classification, 65 patients (65%) were responders, while 66 patients (66%) were classified as responders according to RECIST.\u003c/p\u003e \u003cp\u003eIn validation cohort 3, 27.6% were female, with a mean age of 61.83\u0026thinsp;\u0026plusmn;\u0026thinsp;8.7 years. All of the patients received fluorouracil-based NCT regimens. Among these patients, 25.2% received the SOX regimen, 53.5% received XELOX, and 21.3% received other fluorouracil-based regimens such as FOLFOX. Additionally, 37.8% underwent more than 2 cycles of NCT. Based on TRG classification, 33 patients (26.0%) were identified as responders, while 66 patients (45.7%) were classified as responders according to RECIST.\u003c/p\u003e \u003cp\u003eThe baseline characteristics of patients in validation cohorts 1, 2, and 3 are presented in Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003e-1. Among these cohorts, there were no significant differences in sex, age, Borrmann classification, NCT regimen, or NCT cycle between the responsive and the non-responsive group (\u003cem\u003eP\u0026thinsp;\u0026gt;\u003c/em\u003e\u0026thinsp;0.05). A comparison of the baseline characteristics across the 3 validation cohorts is illustrated in Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003e-2, showing no significant differences observed in sex, Borrmann classification, histological grade, or HER-2 status among 3 cohorts (\u003cem\u003eP\u0026thinsp;\u0026gt;\u003c/em\u003e\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment of the predicting model for NCT efficacy in LAGC using Meta-Analysis\u003c/h2\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003eIdentification of clinical features associated with the NCT efficacy\u003c/h2\u003e \u003cp\u003eWe first stratified the 20 clinical features according to their status in various studies. The features requiring stratification included age (\u0026le;\u0026thinsp;55 years old, \u0026gt;\u0026thinsp;55 years old, \u0026le;\u0026thinsp;60 years old, and \u0026gt;\u0026thinsp;60 years old), tumor location (upper, middle, lower, and total stomach)(\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e), Lauren classification (intestinal type, diffuse type, and mixed type), histological grade (well, moderately, poorly differentiated and undifferentiated), depth of invasion (T2, T2-T3, T3, and T4), lymph node involvement (N0, N1, N2, and N3), clinical stage (stage II, stage I\u0026thinsp;+\u0026thinsp;II, and stage III), NCT regimen (SOX, XELOX, and FOLFOX), and NCT cycle (2 cycles, 3 cycles, and 4 cycles). First, we conducted a pooled analysis of these clinical features. Subsequently, we performed subgroup analyses on the features assessed by TRG or RECIST criteria in at least 2 studies. In the TRG subgroup, we analyzed features including sex, tumor location, Borrmann classification, histological grade, depth of invasion, and lymph node metastasis. In the RECIST subgroup, the analyzed features included sex, tumor location, Borrmann classification, and histological grade.The features across different strata and subgroups are presented in Table \u003cspan refid=\"MOESM7\" class=\"InternalRef\"\u003eS7\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eThe results indicated that 9 out of the 20 features were associated with the NCT efficacy in LAGC. These features, along with their respective combined RRs, were as follows: CEA\u0026thinsp;\u0026le;\u0026thinsp;5 \u0026micro;g/L (RR\u0026thinsp;=\u0026thinsp;1.10), upper GC (RR\u0026thinsp;=\u0026thinsp;1.77), lower GC (RR\u0026thinsp;=\u0026thinsp;0.87), intestinal-type GC (RR\u0026thinsp;=\u0026thinsp;1.28), diffuse-type GC (RR\u0026thinsp;=\u0026thinsp;1.16), well-differentiated GC (RR\u0026thinsp;=\u0026thinsp;1.70), moderately-differentiated GC (RR\u0026thinsp;=\u0026thinsp;1.41), T2 GC (RR\u0026thinsp;=\u0026thinsp;1.90), T3 GC (RR\u0026thinsp;=\u0026thinsp;1.52), T4 GC (RR\u0026thinsp;=\u0026thinsp;0.78), N0 lymph node status (RR\u0026thinsp;=\u0026thinsp;1.75), N1 lymph node status (RR\u0026thinsp;=\u0026thinsp;1.35), N3 lymph node status (RR\u0026thinsp;=\u0026thinsp;0.62), clinical stage I\u0026thinsp;+\u0026thinsp;II (RR\u0026thinsp;=\u0026thinsp;1.42), HER-2 (3+) (IHC score) (RR\u0026thinsp;=\u0026thinsp;4.16), and Ki67\u0026thinsp;\u0026ge;\u0026thinsp;10% (RR\u0026thinsp;=\u0026thinsp;2.29) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The meta-analysis results of these features across different strata and subgroups are presented in Supplemental Materials and Methods.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment of the predicting model\u003c/h2\u003e \u003cp\u003eTo further develop the prediction model for NCT efficacy based on clinical features, we converted the RR values into \u003cem\u003eβ\u003c/em\u003e coefficients and assigned the risk scores for each clinical feature. Detailed information, including the number of studies, sample size, pooled RRs, 95%CIs, \u003cem\u003eβ\u003c/em\u003e coefficients, and risk scores, was provided in Table \u003cspan refid=\"MOESM8\" class=\"InternalRef\"\u003eS8\u003c/span\u003e. The final model for predicting NCT efficacy in LAGC patients, based on general clinical characteristics, is summarized in Table\u0026nbsp;1.The 9 parameters and their corresponding risk scores are as follows:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eCEA\u003c/b\u003e: \u0026le;5 \u0026micro;g/L (1 point), \u0026gt;\u0026thinsp;5 \u0026micro;g/L (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eTumor location\u003c/b\u003e: Upper (5.5 points), Lower (\u0026minus;\u0026thinsp;1.5 points), Other regions (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eLauren classification\u003c/b\u003e: Intestinal type (2.5 points), Diffuse type (1.5 points), Mixed type (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eHistological grade\u003c/b\u003e: Well differentiated (5.5 points), Moderately differentiated (3.5 points), Poorly differentiated (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eDepth of invasion\u003c/b\u003e: T2 (6.5 points), T3 (4 points), T4 (\u0026minus;\u0026thinsp;2.5 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eLymph node metastasis\u003c/b\u003e: N0 (5.5 points), N1 (3 points), N2 (0 points), N3 (\u0026minus;\u0026thinsp;5 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eClinical stage\u003c/b\u003e: Stages I/II (3.5 points), Stage III (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eHER-2 status\u003c/b\u003e: HER-2 3+ (IHC score) (14.5 points), HER-2 1\u0026thinsp;+\u0026thinsp;~\u0026thinsp;2+ (IHC score) (0 points)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eKi67\u003c/b\u003e:10% (0 points), \u0026ge;\u0026thinsp;10% (\u0026minus;\u0026thinsp;2.5 points)\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThis prediction model was designed for LAGC patients aged 18 to 75 years and applicable to both white and Asian populations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eValidation of the predicting model for NCT efficacy in LAGC using real-world data\u003c/h2\u003e \u003cp\u003eTo further evaluate the performance of the predictive model for NCT efficacy, the model was validated in cohorts 1, 2, and 3. The ROC curves for each cohort were illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e. In validation cohort 1, the AUC was 0.760 (95%CI: 0.725\u0026ndash;0.795) for the TRG subgroup, and 0.646 (95%CI: 0.606\u0026ndash;0.686) for the RECIST subgroup. In validation cohort 2, the AUC was 0.786 (95%CI: 0.691\u0026ndash;0.880) for the TRG subgroup, and 0.639 (95%CI: 0.523\u0026ndash;0.755) for the RECIST subgroup. In validation cohort 3, the AUC was 0.796 (95%CI: 0.718\u0026ndash;0.875) for the TRG subgroup, and 0.659 (95%CI: 0.564\u0026ndash;0.755) for the RECIST subgroup.\u003c/p\u003e \u003cp\u003eTo further evaluate the performance of the predictive model for NCT efficacy, the model was validated in cohorts 1, 2, and 3. The ROC curves for each cohort were illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e. In validation cohort 1, the AUC was 0.760 (95%CI: 0.725\u0026ndash;0.795) for the TRG subgroup, and 0.646 (95%CI: 0.606\u0026ndash;0.686) for the RECIST subgroup (DeLong's test: D = -4.172, P\u0026thinsp;=\u0026thinsp;3.195e-05). In validation cohort 2, the AUC was 0.786 (95%CI: 0.691\u0026ndash;0.880) for the TRG subgroup, and 0.639 (95%CI: 0.523\u0026ndash;0.755) for the RECIST subgroup (DeLong's test: D = -2.1696, P\u0026thinsp;=\u0026thinsp;0.031). In validation cohort 3, the AUC was 0.796 (95%CI: 0.718\u0026ndash;0.875) for the TRG subgroup, and 0.659 (95%CI: 0.564\u0026ndash;0.755) for the RECIST subgroup (DeLong's test: D = -1.9266, P\u0026thinsp;=\u0026thinsp;0.056).\u003c/p\u003e \u003cp\u003eThese results indicated that the TRG subgroup demonstrated statistically greater accuracy than the RECIST subgroup in predicting NCT treatment response (significant in cohorts 1 and 2, with borderline significance in cohort 3). Consequently, we calculated the maximum Youden index for the TRG subgroup, identifying 7.75 as the optimal cutoff value, which yielded a sensitivity of 0.704 and a specificity of 0.739. Sensitivity and specificity for various cutoff risk scores are presented in Table \u003cspan refid=\"MOESM9\" class=\"InternalRef\"\u003eS9\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation of NCT efficacy and prognosis using risk scores\u003c/h2\u003e \u003cp\u003eTo further evaluate the relationship between patients' risk scores and NCT efficacy, we used the optimal cutoff value (7.75) to classify patients in the validation cohort. Patients with a total score exceeding 7.75 were categorized into the low-to-moderate risk group, while those with a score below 7.75 were placed in the high-risk group. ROC curves were plotted for each group, and the maximum Youden index was used to identify additional optimal cutoff points. This analysis allowed us to further stratify patients into 4 risk levels: very high risk (\u0026lt; -1.5), high risk (-1.5 to 7.5), intermediate risk (8 to 15), and low risk (\u0026ge;\u0026thinsp;15).\u003c/p\u003e \u003cp\u003eIn the 3 independent validation cohorts, the distribution of patients across these risk groups was as follows: low-risk group (n\u0026thinsp;=\u0026thinsp;198, 18, 6), moderate-risk group (n\u0026thinsp;=\u0026thinsp;218, 26, 25), high-risk group (n\u0026thinsp;=\u0026thinsp;185, 38, 29), and very high-risk group (n\u0026thinsp;=\u0026thinsp;152, 18, 67). Across all 3 validation cohorts, the NCT efficacy demonstrated a gradual decrease with ascending risk stratification, with statistically significant differences among 4 risk groups (\u003cem\u003eP\u0026thinsp;\u0026lt;\u003c/em\u003e\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e5\u003c/span\u003eA-C). Pearson correlation analysis confirmed a significant negative correlation between elevated risk levels and NCT response rate, with correlation coefficients of R1 = -0.98, R2 = -0.86, and R3 = -0.93 for validation cohorts 1\u0026ndash;3, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e5\u003c/span\u003eD-F).\u003c/p\u003e \u003cp\u003eIn the validation cohorts, the NCT response rates for patients in the low-risk groups were 2.44-, 1.31-, and 1.51-fold higher compared with those in the high-risk groups,and were 3.96-, 1.67-, and 1.94-fold higher compared with those in the very high-risk groups, respectively(Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e5\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eFurthermore, Kaplan-Meier survival curves were used to explore the relationship between risk levels and survival outcomes in patients from validation cohort 1. The Kaplan-Meier survival analysis demonstrated that higher risk levels were significantly associated with shorter overall survival (\u003cem\u003eP\u0026thinsp;\u0026lt;\u003c/em\u003e\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e5\u003c/span\u003eH).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eNetwork deployment of the predictive model\u003c/h2\u003e \u003cp\u003eFurthermore, we developed a risk calculator for the NCT efficacy model. By inputting 9 general clinical characteristics of the patient, this calculator can provide risk levels about the NCT efficacy. Additionally, we have deployed the calculator online to offer a personalized evaluation tool to support clinical treatment decisions. (The URL is: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://sunxiaohu.shinyapps.io/tigersun/\u003c/span\u003e\u003cspan address=\"https://sunxiaohu.shinyapps.io/tigersun/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eNCT significantly improves the surgical success rate and survival of LAGC patients (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e). Accurate prediction of NCT efficacy is essential for optimizing personalized treatment strategies(\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). The present study integrated systematic review, meta-analysis, and real-world cohort data to develop and validate a prediction model for NCT efficacy in LAGC, enabling precise risk stratification for NCT response and prognosis. To enhance its practicality, we deployed an intuitive web-based interface that could provide real-time risk assessments for both clinicians and patients. This research enhances the accuracy of NCT efficacy prediction, optimizes the allocation of medical resources, and promotes the development of personalized precision medicine. Furthermore, the model offers a foundation for deeper insights into GC biological characteristics and its response to chemotherapy.\u003c/p\u003e \u003cp\u003eIn recent years, predicting responses to preoperative chemotherapy has attracted more attention, with several studies developing NCT efficacy prediction models for LAGC patients based on clinical characteristics(\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). However, these models have some limitations, including restricted population representativeness, incomplete clinical characteristics, small sample sizes that may lead to overfitting, and short follow-up periods that hinder long-term prognosis analysis(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). To address these limitations, we conducted a systematic review and meta-analysis of 25 high-quality cohort studies involving 4,014 LAGC patients, from which independent clinical features of NCT efficacy were identified and quantified. These features were then used to develop a 9-feature prediction model for LAGC patients. Furthermore, our meta-analysis has been registered with PROSPERO, ensuring the transparency, reliability, and reproducibility of the findings.\u003c/p\u003e \u003cp\u003eAmong the 9 features in the prediction model, tumor location in the gastric antrum, invasion beyond the serosa layer, more than 7 lymph node metastases, and high tumor cell activity (Ki67\u0026thinsp;\u0026ge;\u0026thinsp;10%) were associated with an ineffective NCT response. In contrast, NCT effective-related features included normal CEA levels, upper stomach tumors, intestinal-type, well differentiation, limited invasion depth to the submucosal layer, absence of lymph node metastasis, clinical stages I/II, and high HER-2 expression. Αmong these features, HER-2 expression is the most significant factor in the prediction model, serving as a key predictor of NCT efficacy. Similarly, a retrospective study showed that high HER-2 expression could independently predict drug sensitivity in LAGC patients undergoing NCT(\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e). This finding is also consistent with previous breast cancer researches, in which high HER-2 expression was associated with higher sensitivity to fluorouracil-based chemotherapy (e.g., FEC, CAF, CMF)(\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e) and better prognosis(\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). Although HER-2 is a conventional target for therapies like trastuzumab, its role in predicting chemotherapy efficacy and outcomes is still unclear. This uncertainty may due to the involvement of the HER-2 signaling pathway in regulating the cell cycle, promoting proliferation, and inhibiting apoptosis(\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). After activation, HER-2 could trigger downstream pathways, including PI3K-AKT and RAS-MAPK signaling pathway, which are critical for tumor cell sensitivity to chemotherapy(\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e). Future researches should focus on clarifying how elevated HER-2 expression enhances chemotherapy efficacy, paving the way for more precise NCT strategies in LAGC.\u003c/p\u003e \u003cp\u003eIn 3 real-world validation cohorts, the models achieved AUC values of 0.760, 0.786, and 0.796, outperforming the existing NCT prediction model for LAGC based on general clinical features (AUC 0.65)(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Notably, models using TRG as the evaluation criteria outperformed those based on RECIST. In the TRG subgroup, with 7.75 as the optimal cutoff value, the sensitivity and specificity for predicting NCT efficacy in LAGC patients were 70.4% and 73.9%, respectively. Several factors may explain why TRG performs better than RECIST: First, the TRG criteria provides a more comprehensive evaluation by assessing tumor tissue shrinkage alongside microstructural changes, such as tumor cell density and fibrosis, whereas RECIST focuses solely on tumor size without accounting for cytological changes. Second, NCT may induce histological changes in tumors during early treatment, even without significant size changes. The RECIST criteria, however, focus more on size changes, failing to capture those early biological responses. Finally, RECIST relies on tumor size measurements, which are prone to errors due to inconsistent techniques and operator variability. In contrast, TRG is less affected by such errors, providing higher accuracy and reliability in efficacy assessment.\u003c/p\u003e \u003cp\u003eNotably, in validation cohorts, the NCT response rates in the high-risk group were significantly lower than those in the low-risk group. This trend was further exacerbated in the very high-risk group, with a significant decrease in drug response rates compared to the low-risk group. These findings demonstrate that NCT response rates decrease progressively as the risk stratification increases, particularly in the very high-risk group with the highest risk of resistance. Moreover, patient outcomes deteriorated progressively as the individuals risk increasing from low to high, highlighting the ability of the risk scoring system to effectively stratify efficacy and prognosis. This approach not only helps identify high-risk patients but also supports the rational allocation of resources, clinical decision-making, and patient management. Additionally, we developed an online predictive calculator for NCT response risk stratification in LAGC patients, providing clinicians a practical tool to predict NCT responses, assess risk levels, and facilitate personalized, stratified care.\u003c/p\u003e \u003cp\u003eCompared to previous NCT efficacy models for LAGC patients based on general clinical feature, this study offers several advantages. The present study integrated extensive clinical data from multiple studies, using systematic reviews and meta-analysis to develop a more reliable and generalizable prediction model. The clinical features in this model were easy to obtain, cost-effective, clinically applicable, interpretable, stable, and easily updated. Unlike other existing models, which often rely on limited patient numbers and single NCT regimens, our model combined a broader range of clinical patients and multiple NCT regimens to better align with real-world clinical practice. Additionally, although many models only report calibration curves or C-indices, our approach quantified performance indicators, providing actionable insights for clinicians and researchers. Notably, our model included multiple validation cohorts, providing a more rigorous evaluation of its stability and reliability\u0026mdash;an aspect that was often lacked in previous models.\u003c/p\u003e \u003cp\u003eThis study has several limitations. Firstly, although 66 features potentially related to NCT efficacy were identified across 25 derivation cohorts, some features\u0026mdash;such as smoking history, lymphocyte counts, and tumor size\u0026mdash;were reported in only one single study and could not be included in the meta-analysis. As more studies being published, future analyses will provide a stronger foundation for LAGC treatment decisions. Additionally, while HER-2 expression was found to be a key predictor of NCT response in LAGC patients, its impact on NCT efficacy in GC is not yet fully understood. Future research will help explore the correlation between HER-2 and NCT response, focusing on the signaling pathways and their interaction with chemotherapy agents to refine treatment strategies.\u003c/p\u003e \u003cp\u003eIn summary, this study presents a practical and user-friendly model for predicting NCT efficacy in LAGC patients, based on general clinical features. The model demonstrated robust predictive performance in terms of both efficacy and survival across 3 real-world populations. Looking ahead to the future, further refinement and expansion of validation cohorts will enhance the accuracy of the model, enable more personalized treatment strategies and ultimately improve survival outcomes of patients undergoing NCT for GC.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study successfully developed and validated a general clinical prediction model for assessing NCT efficacy in LAGC patients. By integrating nine readily accessible clinical features\u0026mdash;including tumor location, HER-2 status, and Ki67 expression\u0026mdash;the model demonstrated robust predictive accuracy (AUC ranged from 0.760 to 0.796) across three independent real-world cohorts. Notably, the risk stratification system effectively categorized patients into four distinct risk groups, revealing a significant inverse correlation between elevated risk scores and NCT response rates (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Low-risk patients exhibited response rates 2.44- to 3.96-fold higher than high-risk counterparts, with survival outcomes further validating the model\u0026rsquo;s prognostic utility (P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001). The accompanying online calculator enhances clinical applicability, offering a practical tool for personalized treatment planning and resource allocation.\u003c/p\u003e \u003cp\u003eWhile the current model addresses critical gaps in existing prediction tools, future studies should focus on expanding validation to diverse populations, integrating molecular biomarkers (e.g., genomic signatures), and refining the algorithm through prospective trials. These efforts will further optimize therapeutic decision-making, minimize unnecessary toxicity in non-responders, and ultimately improve survival outcomes for LAGC patients undergoing NCT.\u003c/p\u003e"},{"header":"List of abbreviations","content":"\u003cp\u003eNCT\u0026nbsp; \u0026nbsp;\u0026nbsp;Neoadjuvant Chemotherapy\u003c/p\u003e\n\u003cp\u003eLAGC\u0026nbsp;Locally Advanced Gastric Cancer\u003c/p\u003e\n\u003cp\u003eAUC\u0026nbsp; \u0026nbsp;Area Under the Curve\u003c/p\u003e\n\u003cp\u003eTRG\u0026nbsp; \u0026nbsp;\u0026nbsp;Tumor Regression Grade\u003c/p\u003e\n\u003cp\u003eRECIST\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Response Evaluation Criteria in Solid Tumors\u003c/p\u003e\n\u003cp\u003eHER2\u0026nbsp;\u0026nbsp;Human Epidermal Growth Factor Receptor 2\u003c/p\u003e\n\u003cp\u003eCEA\u0026nbsp; \u0026nbsp;\u0026nbsp;Carcinoembryonic Antigen\u003c/p\u003e\n\u003cp\u003eCA19-9\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Carbohydrate Antigen 19-9\u003c/p\u003e\n\u003cp\u003eAJCC\u0026nbsp;\u0026nbsp;American Joint Committee on Cance\u003c/p\u003e\n\u003cp\u003eCAP\u0026nbsp; \u0026nbsp;\u0026nbsp;College of American Pathologists\u003c/p\u003e\n\u003cp\u003eOS \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Overall Survival\u003c/p\u003e\n\u003cp\u003eRR \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Risk Ratio\u003c/p\u003e\n\u003cp\u003eSMD\u0026nbsp; \u0026nbsp;Standardized Mean Difference\u003c/p\u003e\n\u003cp\u003eCI \u0026nbsp;\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Confidence Interval\u003c/p\u003e\n\u003cp\u003ePRISMA\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Preferred Reporting Items for Systematic Reviews and Meta-Analyses\u003c/p\u003e\n\u003cp\u003eNCCN\u0026nbsp;National Comprehensive Cancer Network\u003c/p\u003e\n\u003cp\u003eJGCA \u0026nbsp;Japanese Gastric Cancer Association\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eConflict of interest\u003c/h2\u003e \u003cp\u003eThe authors declare no potential conflicts of interest.\u003c/p\u003e \u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Ethics Committees of the First Hospital of China Medical University (Approval No. NF-SOP-07-1.2-01) and Liaoning Cancer Hospital (Approval No. 202208102). The requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized clinical data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analyzed during the current study are not publicly available due to patient privacy restrictions but are available from the corresponding author (Y.H.G) on reasonable request. The meta-analysis data supporting the findings of this study are included in this published article and its supplementary information files.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no potential conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by Noncommunicable Chronic Diseases-National Science and Technology Major Project (No. 2023ZD0501400) and the Education Department project of Liaoning Province (No. LJ212410159002). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConception and design:\u003c/strong\u003eY.H. Gong,\u0026nbsp;L. Wang and X.H. Sun; \u003cstrong\u003eDevelopment of methodology:\u0026nbsp;\u003c/strong\u003eY.Y. Wang, R. Guo and S.W. Zheng; \u003cstrong\u003eAcquisition of data:\u0026nbsp;\u003c/strong\u003eY.H. Gong, L. Wang, X.H. Sun and Y.K. Li;\u0026nbsp;\u003cstrong\u003eAnalysis and interpretation of data:\u0026nbsp;\u003c/strong\u003eL. Wang, X.H. Sun, S.R. Nie;\u0026nbsp;\u003cstrong\u003eWriting, review, and/or revision of the manuscript:\u0026nbsp;\u003c/strong\u003eY.H. Gong,\u0026nbsp;L. Wang, X.H. Sun, J.J. Jing, Y.Y. Wang, R. Guo, S.W. Zheng;\u0026nbsp;X.N. Qiu, T.T. Tao; \u003cstrong\u003eAdministrative, technical, or material support:\u0026nbsp;\u003c/strong\u003eY.H. Gong;\u003cstrong\u003e\u0026nbsp;Study supervision:\u0026nbsp;\u003c/strong\u003eY.H. Gong.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eY.H.G. is a Professor in the Department of Tumor Etiology and Screening at the First Hospital of China Medical University, specializing in gastrointestinal oncology. L.W. and X.H.S. are research fellows focusing on cancer therapy. Other authors are affiliated with clinical and research departments in oncology.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMachlowska J, Baj J, Sitarz M, Maciejewski R, Sitarz R. Gastric Cancer: Epidemiology, Risk Factors, Classification, Genomic Characteristics and Treatment Strategies. International journal of molecular sciences. 2020;21(11).\u003c/li\u003e\n\u003cli\u003eSung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a cancer journal for clinicians. 2021;71(3):209-49.\u003c/li\u003e\n\u003cli\u003eElizabeth S, Magnus N, Heike G, Nicole CTvG, Florian L. Gastric cancer. The Lancet. 2020.\u003c/li\u003e\n\u003cli\u003eSongun I, Putter H, Kranenbarg EM, Sasako M, van de Velde CJ. Surgical treatment of gastric cancer: 15-year follow-up results of the randomised nationwide Dutch D1D2 trial. The Lancet Oncology. 2010;11(5):439-49.\u003c/li\u003e\n\u003cli\u003eSasako M, Sano T, Yamamoto S, Kurokawa Y, Nashimoto A, Kurita A, et al. D2 lymphadenectomy alone or with para-aortic nodal dissection for gastric cancer. The New England journal of medicine. 2008;359(5):453-62.\u003c/li\u003e\n\u003cli\u003eAjani JA, D\u0026apos;Amico TA, Bentrem DJ, Chao J, Cooke D, Corvera C, et al. Gastric Cancer, Version 2.2022, NCCN Clinical Practice Guidelines in Oncology. Journal of the National Comprehensive Cancer Network : JNCCN. 2022;20(2):167-92.\u003c/li\u003e\n\u003cli\u003eWang FH, Zhang XT, Tang L, Wu Q, Cai MY, Li YF, et al. The Chinese Society of Clinical Oncology (CSCO): Clinical guidelines for the diagnosis and treatment of gastric cancer, 2023. Cancer communications (London, England). 2024;44(1):127-72.\u003c/li\u003e\n\u003cli\u003eJoshi SS, Badgwell BD. Current treatment and recent progress in gastric cancer. CA: a cancer journal for clinicians. 2021;71(3):264-79.\u003c/li\u003e\n\u003cli\u003eMohamed Salah Emam Sayed H, Khaled Abdallah E, Ashraf Kamal A, Aly MA, Mohammed AH. Role of Neo-Adjuvant Chemotherapy in Decision Making for Surgical Management of Cancer Stomach. Qjm. 2024.\u003c/li\u003e\n\u003cli\u003eSato Y, Okamoto K, Kawaguchi T, Nakamura F, Miyamoto H, Takayama T. Treatment Response Predictors of Neoadjuvant Therapy for Locally Advanced Gastric Cancer: Current Status and Future Perspectives. Biomedicines. 2022;10(7).\u003c/li\u003e\n\u003cli\u003eImyanitov EN, Iyevleva AG. Molecular tests for prediction of tumor sensitivity to cytotoxic drugs. Cancer letters. 2022;526:41-52.\u003c/li\u003e\n\u003cli\u003eTang Z, Gu Y, Shi Z, Min L, Zhang Z, Zhou P, et al. Multiplex immune profiling reveals the role of serum immune proteomics in predicting response to preoperative chemotherapy of gastric cancer. Cell reports Medicine. 2023;4(2):100931.\u003c/li\u003e\n\u003cli\u003eDai M, Sayuri K, Keizaburo K, Ryuichi K, Yoshikazu I, Toyotsugu O, et al. P-125 Radiomic prediction model for pathological responses of neoadjuvant chemotherapy with S-1 plus oxaliplatin (G-SOX) in clinical stage III gastric cancer. Annals of Oncology. 2020.\u003c/li\u003e\n\u003cli\u003eLaura B, Kym IES, Gary SC, Richard DR. Guide to presenting clinical prediction models for use in clinical settings. The BMJ. 2019.\u003c/li\u003e\n\u003cli\u003eHatta W, Tsuji Y, Yoshio T, Kakushima N, Hoteya S, Doyama H, et al. Prediction model of bleeding after endoscopic submucosal dissection for early gastric cancer: BEST-J score. Gut. 2021;70(3):476-84.\u003c/li\u003e\n\u003cli\u003eWilliam AJ, Christine L, Rebecca CL, Teresa M, Christopher M, Benjamin SW, et al. Abstract 14217: Clinical Predictive Models for Sudden Cardiac Arrest: A Systematic Review of the Literature. Circulation. 2017.\u003c/li\u003e\n\u003cli\u003eRiley B, Gaurav G, Jason N, Benjamin K, Christine L, Jinny P, et al. THE GENERALIZABILITY OF CLINICAL PREDICTIVE MODELS FOR PRIMARY PREVENTION OF CARDIOVASCULAR DISEASE: RESULTS FROM INDEPENDENT EXTERNAL VALIDATIONS. Journal of the American College of Cardiology. 2020.\u003c/li\u003e\n\u003cli\u003eMurtaza M, Richard WG, Vincent XL. Use of Sepsis Clinical Prediction Models to Improve Patient Care. JAMA Internal Medicine. 2023.\u003c/li\u003e\n\u003cli\u003eLiu B, Xu YJ, Chu FR, Sun G, Zhao GD, Wang SZ. Development of a clinical nomogram for prediction of response to neoadjuvant chemotherapy in patients with advanced gastric cancer. World journal of gastrointestinal surgery. 2024;16(2):396-408.\u003c/li\u003e\n\u003cli\u003ePage MJ, Moher D. Evaluations of the uptake and impact of the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) Statement and extensions: a scoping review. Systematic reviews. 2017;6(1):263.\u003c/li\u003e\n\u003cli\u003eEsterman AJ. 1. Epidemiology, study design and data analysis (2nd edn). Mark Woodward, Chapman \u0026amp; Hall/CRC, Boca Raton, 2005. No. of pages: xxii + 849. Price: $79.95. ISBN: 1‐58488‐415‐0. Statistics in medicine. 2006;25:1439-40.\u003c/li\u003e\n\u003cli\u003eAJCC Cancer Staging Manual. 8 ed: Springer Cham; 2016. XVII, 1032 p.\u003c/li\u003e\n\u003cli\u003eMandard AM, Dalibard F, Mandard JC, Marnay J, Henry-Amar M, Petiot JF, et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Clinicopathologic correlations. Cancer. 1994;73(11):2680-6.\u003c/li\u003e\n\u003cli\u003eDhanpat Jain WVC, Rondell P. Graham. Protocol for the Examination of Specimens from Patients with Well-Differentiated Neuroendocrine Tumors (Carcinoid Tumors) of the Stomach.2023. 2023.\u003c/li\u003e\n\u003cli\u003eEisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European journal of cancer (Oxford, England : 1990). 2009;45(2):228-47.\u003c/li\u003e\n\u003cli\u003eLiti\u0026egrave;re S, Collette S, de Vries EG, Seymour L, Bogaerts J. RECIST - learning from the past to build the future. Nature reviews Clinical oncology. 2017;14(3):187-92.\u003c/li\u003e\n\u003cli\u003eDeeks JJ AD. Cochrane Handbook for Systematic Reviews of Interventions. 2011. The Cochrane Collaboration.\u003c/li\u003e\n\u003cli\u003eChen D, Wang M, Shang X, Liu X, Liu X, Ge T, et al. Development and validation of an incidence risk prediction model for early foot ulcer in diabetes based on a high evidence systematic review and meta-analysis. Diabetes research and clinical practice. 2021;180:109040.\u003c/li\u003e\n\u003cli\u003eJapanese classification of gastric carcinoma: 3rd English edition. Gastric cancer : official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2011;14(2):101-12.\u003c/li\u003e\n\u003cli\u003eDeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837-45.\u003c/li\u003e\n\u003cli\u003eRobin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC bioinformatics. 2011;12:77.\u003c/li\u003e\n\u003cli\u003eCook NR. Use and misuse of the receiver operating characteristic curve in risk prediction. Circulation. 2007;115(7):928-35.\u003c/li\u003e\n\u003cli\u003eHassanzad M, Hajian-Tilaki K. Methods of determining optimal cut-point of diagnostic biomarkers with application of clinical data in ROC analysis: an update review. BMC medical research methodology. 2024;24(1):84.\u003c/li\u003e\n\u003cli\u003eJapanese Gastric Cancer Association. Japanese classification of gastric cance. 2017.\u003c/li\u003e\n\u003cli\u003eDas M. Neoadjuvant chemotherapy: survival benefit in gastric cancer. The Lancet Oncology. 2017;18(6):e307.\u003c/li\u003e\n\u003cli\u003eDavid RB, James DB, Thomas PB, Daniel CS, Donna MG. Current and future cancer staging after neoadjuvant treatment for solid tumors. CA: a cancer journal for clinicians. 2020.\u003c/li\u003e\n\u003cli\u003eLin J, Yi-Hui T, Wen-Xing Z, Jacopo D, Amilcare P, Jian‐Wei X, et al. Body composition parameters predict pathological response and outcomes in locally advanced gastric cancer after neoadjuvant treatment: A multicenter, international study. Clinical Nutrition. 2021.\u003c/li\u003e\n\u003cli\u003eZiyu L, Shuangxi L, Xiangji Y, Lianhai Z, Fei S, Yongning J, et al. The clinical value and usage of inflammatory and nutritional markers in survival prediction for gastric cancer patients with neoadjuvant chemotherapy and D2 lymphadenectomy. Gastric cancer : official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2020.\u003c/li\u003e\n\u003cli\u003eOrestis E, Michael S, Konstantina C, Thomas PAD, Matthias E, Georgia S. Developing clinical prediction models: a step-by-step guide. Bmj. 2024.\u003c/li\u003e\n\u003cli\u003eHaibin C, Huai\u0026apos;e G, Xiyong B, Wei Z, Yuanyuan Z. Application of the predictors EGFR, HER2, Ki-67 and P53 on neoadjuvant chemotherapy in patients with advanced gastric cancer. China Medical Herald. 2014;11(2):35-8.\u003c/li\u003e\n\u003cli\u003eKonecny G, Fritz M, Untch M, Lebeau A, Felber M, Lude S, et al. HER-2/neu overexpression and in vitro chemosensitivity to CMF and FEC in primary breast cancer. Breast cancer research and treatment. 2001;69(1):53-63.\u003c/li\u003e\n\u003cli\u003eDiGiovanna MP, Stern DF, Edgerton S, Broadwater G, Dressler LG, Budman DR, et al. Influence of activation state of ErbB-2 (HER-2) on response to adjuvant cyclophosphamide, doxorubicin, and fluorouracil for stage II, node-positive breast cancer: study 8541 from the Cancer and Leukemia Group B. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008;26(14):2364-72.\u003c/li\u003e\n\u003cli\u003eZhang W, Tian H, Yang SH. The Efficacy of Neoadjuvant Chemotherapy for HER-2-Positive Locally Advanced Breast Cancer and Survival Analysis. Analytical cellular pathology (Amsterdam). 2017;2017:1350618.\u003c/li\u003e\n\u003cli\u003eZhou J, Zhang S, Luo M. LncRNA PCAT7 promotes the malignant progression of breast cancer by regulating ErbB/PI3K/Akt pathway. Future oncology (London, England). 2021;17(6):701-10.\u003c/li\u003e\n\u003cli\u003eVinod BS, Nair HH, Vijayakurup V, Shabna A, Shah S, Krishna A, et al. Resveratrol chemosensitizes HER-2-overexpressing breast cancer cells to docetaxel chemoresistance by inhibiting docetaxel-mediated activation of HER-2-Akt axis. Cell death discovery. 2015;1:15061.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Stomach Neoplasms, Neoadjuvant chemotherapy, Response evaluation, Meta-Analysis, Clinical prediction model","lastPublishedDoi":"10.21203/rs.3.rs-6136117/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6136117/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNeoadjuvant chemotherapy (NCT) is a cornerstone treatment for locally advanced gastric cancer (LAGC), yet patient responses vary significantly. This study aimed to develop and validate a general clinical model to predict NCT efficacy in LAGC patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA systematic review and meta-analysis were performed to identify independent clinical features associated with NCT efficacy. Using \u003cem\u003eβ\u003c/em\u003e coefficients, a risk score-based predictive model was constructed. Model performance was validated in 3 real-world cohorts using Area Under Curve (AUC) metrics. Prognostic utility was analyzed via Kaplan-Meier analysis. Additionally, an online NCT response prediction calculator was developed using \u003cem\u003eR Shiny\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 4,014 patients from 25 high-quality cohort studies were included in the meta-analysis. Nine clinical features—CEA, tumor location, Lauren classification, histological grade, depth of invasion, lymph node metastasis, clinical stage, HER-2 status (IHC score), and Ki67—were incorporated into the final prediction model for NCT efficacy in LAGC. The present model demonstrated robust predictive performance, with AUCs of 0.760 (95% CI: 0.725–0.795), 0.786 (95% CI: 0.691–0.880), and 0.796 (95% CI: 0.718–0.875) across validation cohorts. NCT response was stratified into 4 levels based on risk scores, with increasing risk levels correlated with a progressive decline in treatment efficacy and poorer prognosis (\u003cem\u003eP \u0026lt;\u003c/em\u003e 0.001). The response rates in low-risk groups were 2.44- and 3.96-fold higher than those in high-risk and very high-risk groups, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study establishes a robust and validated clinical model for predicting NCT efficacy and prognosis in LAGC patients. The accompanying online calculator provides a practical tool for personalized treatment planning. Future efforts will focus on expanding validation cohorts and refining the model to further optimize therapeutic decision-making for LAGC patients undergoing NCT.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTrial registration:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe protocol for the systematic review and meta-analysis was prospectively registered on PROSPERO (CRD42023483908) on March 12, 2023, prior to data collection.\u003c/p\u003e\n\u003cp\u003eThe validation cohorts (Cohorts 1–3) were derived from retrospective real-world data. As this study analyzed existing clinical records without prospective intervention, trial registration was not required for these cohorts.\u003c/p\u003e","manuscriptTitle":"Development and Validation of a General Clinical Model for Predicting Neoadjuvant Chemotherapy Efficacy in Locally Advanced Gastric Cancer: Evidence from Meta-Analysis and Real-World Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-11 08:28:41","doi":"10.21203/rs.3.rs-6136117/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f9a4a7e4-1e27-4e59-8c89-3da3967f5e56","owner":[],"postedDate":"March 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-03-11T08:28:45+00:00","versionOfRecord":[],"versionCreatedAt":"2025-03-11 08:28:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6136117","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6136117","identity":"rs-6136117","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}