Machine Learning-Based Prediction of Lymphovascular Invasion in Superficial Esophageal Carcinoma: Model Development and Risk Factor Analysis

Machine Learning-Based Prediction of Lymphovascular Invasion in Superficial Esophageal Carcinoma: Model Development and Risk Factor Analysis | 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 Machine Learning-Based Prediction of Lymphovascular Invasion in Superficial Esophageal Carcinoma: Model Development and Risk Factor Analysis Zichen Luo, Xinrui Chen, Yutong Cui, Ji Zuo, Shiqi Liang, Guangbing Hu, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6591573/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 Lymphovascular invasion (LVI) represents a critical prognostic determinant in superficial esophageal carcinoma (SEC), significantly influencing therapeutic decision-making and clinical outcomes. Despite its clinical importance, reliable predictive tools for early LVI detection remain unavailable. The current study was designed to develop and validate a machine learning-based predictive model for accurate LVI risk stratification in SEC patients. Methods Predictive factor selection was conducted using least absolute shrinkage and selection operator (LASSO) regression followed by multivariable logistic regression analysis. Multiple machine learning algorithms were systematically evaluated, with model performance quantified through receiver operating characteristic (ROC) curve analysis. Model interpretability was enhanced through implementation of Shapley Additive Explanations (SHAP) methodology. Results Eight independent predictors of LVI were identified: neutrophil-to-lymphocyte ratio (NLR), esophageal wall thickness on computed tomography (CT), endoscopic ultrasound or magnifying endoscopy (EOM) findings, tumor diameter, multiple lesions, circumferential involvement proportion (CIP), consumption of pickled food and preoperative biopsy results. The logistic regression model demonstrated superior predictive performance, with area under the curve (AUC) values of 0.871 (training cohort), 0.852 (validation cohort), and 0.902 (test cohort). Conclusion The developed SHAP-interpretable logistic regression model provides an effective tool for early LVI detection in SEC, enabling personalized risk assessment and optimized clinical management strategies. This approach may significantly improve treatment decision-making for SEC patients. superficial esophageal carcinoma lymphovascular invasion LASSO regression logistic regression machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 INTRODUCTION Esophageal cancer is a common malignancy of the digestive tract, with a troubling increase in both incidence and mortality rates observed globally in recent years. This neoplasm has emerged as a significant public health challenge, particularly in certain geographic regions, notably China, where it represents a substantial threat to public health and life expectancy. Epidemiological data reveal a distinct distribution of disease burden, with high prevalence rates concentrated in specific areas, underscoring the need for focused research efforts and the development of targeted intervention strategies in these high-risk regions [ 1 ] .Esophageal cancer is histologically classified into two principal subtypes: squamous cell carcinoma and adenocarcinoma. In the Chinese population, squamous cell carcinoma is the predominant subtype, demonstrating distinct epidemiological patterns and unique clinical and biological behave ors [ 2 ] . Esophageal cancer is often diagnosed at advanced local or metastatic stages due to th e absence of distinct clinical symptoms in its early phases. Consequently, the majority of patients present with disease at a stage where treatment outcomes are suboptimal, leading to a generally unfavorable prognosis. Superficial esophageal carcinoma (SEC) is defined as an early-stage esophageal cancer in which the neoplasm remains restricted to the mucosal or submucosal layers of the esophagus, without invasion of the muscularis propria [ 3 ] . The rapid advancements in endoscopic diagnostic and therapeutic techniques have led to a significant increase in the detection rate of SEC, facilitating earlier intervention. However, it is crucial to recognize that even at this early stage, esophageal cancer can metastasize to distant sites through Lymphovascular invasion (LVI), a key pathological process. As a primary pathway for tumor dissemination, LVI substantially enhances the risk of both lymphatic and hematogenous metastasis, which is strongly associated with the worsening of patient prognosis [ 4 – 6 ] . Consequently, the establishment of a robust risk prediction system for LVI in patients with SEC is pivotal for refining treatment strategies and enhancing clinical prognosis. At present, the investigation into the pathological mechanisms of LVI in SEC is still in the exploratory phase, with a significant gap in the development of comprehensive risk prediction models. While several prognostic models for esophageal cancer have been proposed, the role of LVI as a central variable has not been extensively examined. Clinical evidence indicates a complex interplay between LVI and various biological characteristics, such as tumor volume, differentiation, and invasion depth; however, the underlying mechanisms and their quantitative relationships remain to be elucidated [ 7 – 9 ] . A critical challenge in current research is the integration of clinical phenotypes, imaging characteristics, and molecular pathological biomarkers to establish a comprehensive, multidimensional, and quantifiable risk assessment system. This study intends to conduct a comprehensive analysis of multidimensional clinical data, including imaging and pathological features, to construct an evidence-based predictive model for LVI risk. Through the application of multivariate analysis to identify critical risk factors, a predictive algorithm will be developed to offer quantitative decision support for clinicians. The goal is to optimize individualized treatment strategies, ultimately improving treatment outcomes and quality of life. MATERIALS AND METHODS Materials A retrospective cohort study was performed on 592 patients diagnosed with SEC at the Department of Gastroenterology and Thoracic Surgery, Affiliated Hospital of North Sichuan Medical College, between January 2017 and December 2024. Among these, 286 patients underwent endoscopic treatment, while 306 patients received surgical intervention. The inclusion criteria were: (1) confirmed diagnosis of superficial esophageal squamous cell carcinoma based on postoperative biopsy; (2) absence of regional lymph node involvement or distant metastasis on preoperative evaluation; and (3) no prior history of other malignancies. Exclusion criteria encompassed: (1) Postoperative pathological diagnosis of non-esophageal squamous cell carcinoma cases; (2) patients who had received preoperative chemotherapy or radiotherapy; (3) lack of histopathological assessment for LVI; and (4) patients with missing data exceeding 15%. The patient selection flowchart is summarized in Fig. 1 . The verification of vascular invasion. The pathological diagnosis of LVI is primarily based on histological examination of tumor tissue obtained through surgical resection or biopsy. Microscopic evaluation focuses on identifying tumor cell penetration through vascular or lymphatic basement membranes with luminal infiltration. Immunohistochemical stains (CD31 for blood vessels and D2-40 for lymphatic channels) are routinely employed to enhance diagnostic accuracy by precisely delineating endothelial structures, thereby improving the reliability of LVI detection. Study variables This study included 34 variables potentially influencing esophageal lesions, with 31 being preoperatively assessable. These variables encompass three primary dimensions: clinical indicators, histopathological characteristics, and lifestyle-related factors. Clinical factors: age, sex, preoperative neutrophil-lymphocyte ratio (NLR), preoperative platelet-lymphocyte ratio (PLR), preoperative platelet-neutrophil ratio (PNR), body mass index (BMI), hypertension, cardiac disease, diabetes mellitus, chronic kidney disease, chronic obstructive pulmonary disease (COPD), family history, brownish discoloration, chronic esophagitis, endoscopic ultrasound or magnifying endoscopy (EOM), esophageal wall thickness, esophageal stricture, tumor enhancement, lesion location, multiple lesions, tumor diameter, circumferential involvement proportion(CIP), Paris classification histopathological characteristics: depth of invasion, preoperative pathological type, LVI, differentiation grade. Lifestyle factors: long-term smoking, long-term alcohol consumption, pickled food intake, high-temperature food intake, fried food intake, fruit and vegetable intake, and oral hygiene. Definitions Lesion location: According to the current standard [ 10 ] , the esophagus is divided into four segments based on the distance from the upper incisors: cervical segment (15–20 cm), upper thoracic segment (20–25 cm), middle thoracic segment (25–30 cm), and lower thoracic segment (30–40 cm). Depth of tissue invasion:According to the 2002 Paris Classification for Gastrointestinal Tumors [ 11 ] , the histological depth of esophageal or gastrointestinal tumor invasion is classified as: (1) Mucosal layer (M): M1 (epithelium-limited), M2 (lamina propria invasion), M3 (muscularis mucosae involvement); (2) Submucosal layer (SM): SM1 (upper third, ≤ 200µm), SM2 (middle third), SM3 (lower third). Lifestyle assessment: Lifestyle factors were evaluated using previously established scoring systems [ 12 , 13 ] , including dietary habits (e.g., fruit/vegetable, pickled food, and fried food intake), oral hygiene (e.g., toothbrushing frequency), and hot water consumption. These were categorized by frequency as occasional (≤ 3 times/week) or frequent (> 3 times/week). Hot water intake was assessed based on self-perceived temperature thresholds. Chronic smoking was typically defined as sustained tobacco use for > 6 months, while chronic alcohol consumption was generally considered present with sustained use > 3 months [ 14 , 15 ] . Endoscopic ultrasound or magnifying endoscopy with narrow-band imaging (EUS or ME-NBI, EOM): refers to the evaluation of esophageal lesion invasion depth using either of these endoscopic modalities. Contrast-enhanced CT findings: Esophageal strictures primarily manifest as proximal luminal dilation with fluid-air levels [ 16 ] . Esophageal wall thickness is classified into three grades: mild ( 7 mm). Following intravenous contrast administration, tumor regions demonstrate more pronounced enhancement compared to normal esophageal tissue due to their rich vascular supply [ 17 , 18 ] . Conventional endoscopic findings: Under narrow-band imaging (NBI), esophageal lesions exhibiting brownish discoloration may indicate inflammation, vascular abnormalities, or neoplastic changes [ 19 , 20 ] . Chronic esophagitis typically presents with mucosal erythema, edema, erosion, or ulceration, potentially accompanied by white plaques, exudates, or a granular appearance [ 21 ] . Statistical analysis of variables Continuous variables were summarized as median with interquartile range (IQR) and compared using nonparametric Mann-Whitney U tests. Categorical data were expressed as frequencies (percentages) and analyzed with Pearson χ² or Fisher exact tests, as appropriate. A two-tailed α level of 0.05 defined statistical significance. All statistical computations were performed using R version 4.2.3 (gtsummary v1.7.2) and Python (scikit-learn v1.1.3) with standard parameter configurations. Model development and validation The analytical framework employed a 70:30 random division of patient data into training (n = 414) and validation (n = 178) cohorts for predictive modeling. Multiple machine learning algorithms were systematically evaluated using Shapley Additive exPlanations (SHAP) analysis to determine variable importance and enhance model interpretability. The preprocessing phase incorporated k-nearest neighbors imputation (scikit-learn v1.1.3) and dichotomization of continuous variables (e.g., age, PLR, NLR) based on receiver operating characteristic (ROC)-derived thresholds (e.g., age ≥ 67 years), with subsequent statistical analyses performed in R (v4.2.3) using the gtsummary package (v1.7.2) for dataset stratification and baseline characterization. Variable selection was conducted using least absolute shrinkage and selection operator (LASSO) regression (glmnet package, version 4.1.8) followed by multivariate logistic regression analysis (R software, version 4.2.3), retaining statistically significant variables (p < 0.05). This dual approach effectively addressed overfitting through coefficient shrinkage while resolving multicollinearity [ 22 ] . Nine machine learning algorithms were implemented: eXtreme Gradient Boosting (XGBoost), logistic regression, Light Gradient Boosting Machine (LightGBM), random forest, Adaptive Boosting (AdaBoost), decision tree, Gradient Boosting Decision Tree (GBDT), Gaussian Naïve Bayes (GNB), and Complement Naïve Bayes (CNB). Model optimization employed 10-fold cross-validation with repetition, with comprehensive evaluation incorporating receiver operating characteristic (ROC) curve analysis, decision curve analysis, and calibration plots to determine the optimal predictive model. Model Validation and Interpretation: The selected prediction model was rigorously evaluated through 5-fold cross-validation, with model stability assessed via learning curve analysis (Python scikit-learn v1.1.3). We employed SHapley Additive exPlanations (SHAP; Python package v0.43.0) to quantitatively evaluate feature importance and provide interpretable visualization of the model's decision architecture. For clinical validation, two paradigmatic cases were systematically analyzed to demonstrate the model's operational performance in authentic clinical practice. The nomogram was employed to visualize the machine learning prediction model. RESULTS Partition of clinical data The analysis included a final cohort of 592 eligible cases, among which pathological evaluation identified LVI in 66 cases (11.1%). Participants were randomly assigned to either training (70%, n = 414) or testing (30%, n = 178) cohorts using a computerized allocation system. Optimal cutoff values for continuous variables were derived from receiver operating characteristic (ROC) curve analysis (e.g., age ≥ 67 years as the optimal threshold). Baseline demographic and clinical characteristics of both cohorts are presented in Table 1 . Intergroup comparisons demonstrated no statistically significant differences in any baseline parameters (all p > 0.05), confirming appropriate randomization and balanced cohort characteristics. Table 1 Patient demographics and clinical characteristics Variable classification term All (n = 592) Training Set (n = 414) Testing Set (n = 178) p Age ,n(%) <67 299(50.507) 214(51.691) 85(47.753) 0.380 ≥ 67 293(49.493) 200(48.309) 93(52.247) Sex ,n(%) Female 215(36.318) 153(36.957) 62(34.831) 0.622 Male 377(63.682) 261(63.043) 116(65.169) BMI ,n(%) <22.21 164(27.703) 116(28.019) 48(26.966) 0.793 ≥ 22.21 428(72.297) 298(71.981) 130(73.034) Hypertension ,n(%) No 457(77.196) 316(76.329) 141(79.213) 0.443 Yes 135(22.804) 98(23.671) 37(20.787) Diabetes mellitus ,n(%) No 526(88.851) 372(89.855) 154(86.517) 0.237 Yes 66(11.149) 42(10.145) 24(13.483) Cardiac disease ,n(%) No 557(94.088) 389(93.961) 168(94.382) 0.842 Yes 35(5.912) 25(6.039) 10(5.618) Chronic kidney disease ,n(%) No 580(97.973) 407(98.309) 173(97.191) 0.376 Yes 12(2.027) 7(1.691) 5(2.809) COPD ,n(%) No 569(96.115) 397(95.894) 172(96.629) 0.671 Yes 23(3.885) 17(4.106) 6(3.371) Smoking history ,n(%) No 360(60.811) 248(59.903) 112(62.921) 0.490 Yes 232(39.189) 166(40.097) 66(37.079) Alcohol consumption ,n(%) No 422(71.284) 289(69.807) 133(74.719) 0.226 Yes 170(28.716) 125(30.193) 45(25.281) Family history ,n(%) No 551(93.074) 386(93.237) 165(92.697) 0.812 Yes 41(6.926) 28(6.763) 13(7.303) PLR ,n(%) <138.211 398(67.230) 282(68.116) 116(65.169) 0.484 ≥ 138.211 194(32.770) 132(31.884) 62(34.831) PNR ,n(%) <47.137 215(36.318) 144(34.783) 71(39.888) 0.236 ≥ 47.137 377(63.682) 270(65.217) 107(60.112) NLR ,n(%) <2.014 311(52.534) 228(55.072) 83(46.629) 0.059 ≥ 2.014 281(47.466) 186(44.928) 95(53.371) EOM findings, ,n(%) SM1 177(29.899) 117(28.261) 60(33.708) Brownish discoloration,n(%) No 279(47.128) 189(45.652) 90(50.562) 0.272 Yes 313(52.872) 225(54.348) 88(49.438) Tumor diamete ,n(%) <2.43cm 308(52.027) 217(52.415) 91(51.124) 0.773 ≥ 2.43cm 284(47.973) 197(47.585) 87(48.876) Multiple lesions ,n(%) No 522(88.176) 365(88.164) 157(88.202) 0.990 Yes 70(11.824) 49(11.836) 21(11.798) CIP,n(%) 3/4 181(30.574) 127(30.676) 54(30.337) Chronic esophagitis ,n(%) No 567(95.777) 398(96.135) 169(94.944) 0.509 Yes 25(4.223) 16(3.865) 9(5.056) Tumor enhancement ,n(%) No 348(58.784) 251(60.628) 97(54.494) 0.164 Yes 244(41.216) 163(39.372) 81(45.506) Esophageal stricture ,n(%) No 370(62.500) 269(64.976) 101(56.742) 0.058 Yes 222(37.500) 145(35.024) 77(43.258) Thickness,n(%) 7mm 103(17.399) 75(18.116) 28(15.730) LVI ,n(%) 无 526(88.851) 366(88.406) 160(89.888) 0.599 有 66(11.149) 48(11.594) 18(10.112) Paris classification ,n(%) II-a 153(25.845) 113(27.295) 40(22.472) 0.167 II-b 310(52.365) 212(51.208) 98(55.056) II-c 85(14.358) 54(13.043) 31(17.416) Others 44(7.432) 35(8.454) 9(5.056) Differentiation grade ,n(%) Well-differentiated 169(28.547) 129(31.159) 40(22.472) 0.098 Moderately-differentiated 397(67.061) 267(64.493) 130(73.034) Poorly-differentiated 26(4.392) 18(4.348) 8(4.494) Depth of invasion ,n(%) SM1 123(20.777) 77(18.599) 46(25.843) Treatment modality ,n(%) ESD 286(48.311) 202(48.792) 84(47.191) 0.721 Surgical operation 306(51.689) 212(51.208) 94(52.809) Lesion location,n(%) Cervical or Upper thoracic 92(15.541) 64(15.459) 28(15.730) 0.713 Middle thoracic 356(60.135) 253(61.111) 103(57.865) Lower thoracic 144(24.324) 97(23.430) 47(26.404) Preoperative pathology ,n(%) HGIN/LGIN 275(46.453) 191(46.135) 84(47.191) 0.813 ESC 317(53.547) 223(53.865) 94(52.809) High-temperature food intake ,n(%) <3 times per week 438(73.986) 308(74.396) 130(73.034) 0.729 ≥ 3 times per week 154(26.014) 106(25.604) 48(26.966) Fried food intake ,n(%) <3 times per week 519(87.669) 361(87.198) 158(88.764) 0.595 ≥ 3 times per week 73(12.331) 53(12.802) 20(11.236) Fruit/vegetable intake ,n(%) <3 times per week 57(9.628) 44(10.628) 13(7.303) 0.209 ≥ 3 times per week 535(90.372) 370(89.372) 165(92.697) Pickled food intake,n(%) <3 times per week 312(52.703) 214(51.691) 98(55.056) 0.452 ≥ 3 times per week 280(47.297) 200(48.309) 80(44.944) Oral hygiene ,n(%) 7 times per week 16(2.703) 12(2.899) 4(2.247) Identification of Risk Factors for LVI LASSO Regression Analysis Least absolute shrinkage and selection operator (LASSO) regression analysis was performed to identify robust predictors of LVI while addressing multicollinearity. As demonstrated in Fig. 1 , the coefficient shrinkage profile (Fig. 2 a) and 10-fold cross-validation (Fig. 2 b) identified an optimal penalty parameter (λ = 0.029) using the 1-standard error criterion. This yielded a parsimonious predictive model comprising nine clinically significant variables: PLR, NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative histopathological characteristics, and pickled food intake. Adjustment for confounding variables To control for potential confounding effects, multivariable logistic regression analysis was performed incorporating LASSO-selected variables. Among the nine candidate predictors initially identified, stepwise regression analysis confirmed eight statistically significant independent risk factors (Table 2 ): NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative histopathological characteristics, and pickled food intake. Table 2 Multivariable logistic regression analysis of LVI in superficial esophageal carcinoma Predictor Estimate SE Z p Odds Ratio Lower Upper PLR 0.593 0.341 1.74 0.082 1.809 0.924 3.536 NLR 0.731 0.343 2.132 0.033 2.077 1.071 4.135 Thickness = 5-7mm 0.385 0.44 0.877 0.381 1.47 0.62 3.524 Thickness>7mm 1.82 0.422 4.312 0.0 6.17 2.749 14.52 EOM findings, = SM1 0.914 0.517 1.769 0.077 2.495 0.914 7.07 EOM findings,>SM1 1.516 0.449 3.379 0.001 4.554 1.951 11.5 Tumor diamete ≥ 2.43cm 0.723 0.344 2.099 0.036 2.06 1.059 4.111 Multiple lesions 1.687 0.413 4.081 0.0 5.401 2.401 12.237 CIP = 1/2–3/4 0.703 0.492 1.429 0.153 2.02 0.799 5.638 CIP>3/4 1.394 0.5 2.788 0.005 4.032 1.578 11.447 Preoperative pathology 0.894 0.387 2.309 0.021 2.446 1.171 5.409 Pickled food intake 1.117 0.351 3.18 0.001 3.054 1.564 6.243 Model Evaluation and Comparative Analysis Through 10-fold cross-validation with repeated training, we evaluated multiple machine learning algorithms for classification tasks (Figs. 3 a, 3 b). Logistic regression emerged as the optimal model, demonstrating robust performance with training and validation AUC values of 0.882 and 0.865, respectively. These results surpassed competing models: XGBoost (0.990/0.826), LightGBM (0.986/0.842), and GBDT (0.970/0.866). The model's strong generalizability was evidenced by minimal AUC discrepancy between training and validation sets (Δ = 0.017), substantially lower than XGBoost (Δ = 0.164), suggesting better resistance to overfitting. Additional validation through calibration analysis showed superior prediction accuracy (deviation = 0.066 vs GBDT's 0.078) (Fig. 3 c). Decision curve analysis confirmed clinical utility with higher net benefit (0.04 at threshold = 0.5) (Fig. 3 d), while precision-recall metrics showed consistent performance across datasets (training = 0.657, validation = 0.635) (Figs. 3 e, 3 f). These comprehensive evaluations establish logistic regression as the preferred choice, offering an optimal balance of predictive accuracy, generalizability, and clinical interpretability. Development and Validation of the Prediction Model The logistic regression model demonstrated significant advantages in classification tasks. Logistic regression analysis and 5-fold cross-validation were performed on the training set. Given that the validation set performance under the AUC metric did not exceed the test set or showed a discrepancy of less than 10%, the model was considered successfully fitted, confirming its applicability for classification modeling in this dataset. As evidenced by the ROC curves (Figs. 4 a, 4 b, 4 c), the AUC values for the training, validation, and test sets reached 0.871, 0.852, and 0.902 respectively, indicating that the model exhibits high accuracy in distinguishing positive and negative samples. Furthermore, the learning curve (Fig. 4 d) revealed that as the number of training samples increased, both training and validation accuracy rates gradually stabilized, demonstrating favorable learning capability and generalization performance. These findings suggest that the logistic regression model not only achieves superior classification accuracy but also maintains satisfactory stability and reliability in practical applications, establishing it as an effective classification methodology. SHAP-based Interpretation of Logistic Regression Model SHAP analysis was utilized to quantify risk factor contributions for LVI. The heatmap (Fig. 5 a) displays directional effects of eight predictors (red/blue indicating high/low risk), with mean absolute SHAP values (Fig. 5 b) revealing the following hierarchy: esophageal wall thickness (most significant) > tumor diameter > NLR > CIP > EOM findings > multiple lesions > preoperative pathology > pickled food intake. All factors positively correlated with LVI risk, particularly esophageal wall thickness (strongest association), tumor diameter, and NLR. This interpretable approach enhances preoperative risk assessment in SEC. SHAP-based Interpretation of Individual Case Predictions Figure 6 a demonstrates accurate negative prediction of LVI (probability = 0.05), with Shapley values revealing primary contributions from elevated NLR, increased tumor diameter, and limited circumferential involvement. Conversely, Fig. 7 b illustrates correct positive identification (probability = 0.63), predominantly influenced by: (i) frequent pickled food consumption, (ii) extensive circumferential involvement (> 75%), (iii) elevated NLR (> 2.014), (iv) tumor diameter exceeding 2.43 cm, and (v) EOM findings meeting SM2 invasion criteria. Visualization of the Machine Learning Prediction Model for LVI The study developed a predictive nomogram incorporating eight independent risk factors for LVI: NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative pathological characteristics, and pickled food intake. Clinicians can apply this tool by assigning points for each parameter according to the nomogram scale and summing them to calculate the total predicted probability of LVI in SEC patients. DISCUSSION Superficial esophageal carcinoma, representing an early pathological stage of esophageal cancer, displays unique biological characteristics and clinical presentations when compared to advanced disease [ 23 ] . The evaluation of LVI during this stage holds significant clinical relevance, as it serves as a key parameter for assessing disease progression and predicting the risk of lymph node metastasis. Furthermore, it plays a crucial role in determining tailored treatment approaches. Precise determination of LVI status is of considerable clinical importance for enhancing patient outcomes and increasing survival rates. The Japanese "Esophageal Cancer Practice Guidelines" [ 24 ] specifically advocate for additional surgical procedures in patients with histologically confirmed LVI following endoscopic submucosal dissection (ESD). This often entails esophagectomy, potentially accompanied by extensive lymph node dissection, to reduce the likelihood of postoperative recurrence and distant metastasis. In cases where surgery is contraindicated, radiotherapy and chemotherapy emerge as feasible alternative adjuvant treatment options. Among 66 LVI-confirmed patients, 28 were diagnosed post-ESD (19 underwent surgery, 6 received chemoradiotherapy due to surgical contraindications, 3 declined treatment for personal reasons). Precise preoperative LVI risk stratification could optimize surgical candidate selection and potentially avoid unnecessary ESD. Current LVI prediction faces accuracy limitations, underscoring the urgent need for robust predictive models to enhance assessment reliability and guide evidence-based treatment decisions. Through LASSO and multivariable logistic regression analyses of 31 preoperative clinical variables, we identified eight significant predictors for LVI in SEC: NLR, EOM features, esophageal wall thickness, tumor size, multifocal lesions, CIP, pickled food intake, and preoperative pathology. Importantly, we provide the first evidence that elevated NLR independently predicts LVI in SEC (P = 0.024, OR = 2.197, 95%CI:1.121–4.434). Although the NLR-LVI mechanism requires further investigation, current evidence suggests NLR reflects the inflammatory/anti-tumor immune balance [ 25 ] , where neutrophils promote tumor progression via cytokine release while lymphopenia facilitates immune evasion, collectively creating a pro-LVI microenvironment [ 26 – 28 ] . A significant positive correlation exists between LVI incidence and tumor invasion depth in esophageal cancer. As SEC progresses from mucosal (T1a) to submucosal (T1b) invasion, particularly reaching deeper layers (SM2+), LVI frequency increases markedly [ 29 ] . A 516-patient cohort study reported LVI rates of 2.5% for T1a versus 15.7% for T1b lesions [ 29 ] , consistent with our findings: among 66 LVI-positive cases, 15 (5.32%) were T1a and 51 (16.78%) T1b. Stratified analysis revealed significantly higher LVI positivity in SM2 (10.52%) versus SM1 (6.25%) invasion (P < 0.05), indicating progressive risk escalation with depth. Endoscopic ultrasound/magnifying endoscopy with narrow-band imaging (EUS/ME-NBI) confirmed substantially elevated LVI risk in submucosal involvement, particularly SM2 (P 7 mm was significantly associated with LVI risk in SEC (P < 0.001, OR = 6.280, 95%CI:2.767–14.978). These findings corroborate previous reports [ 30 – 32 ] confirming wall thickness as an independent LVI predictor (OR = 16.32, 95%CI:5.89–45.17). Pathologically, increased thickness indicates deeper tumor invasion, promoting basement membrane penetration and submucosal vascular/lymphatic invasion. Tumor diameter > 2.43 cm independently predicted LVI (P = 0.049, OR = 1.992, 95%CI:1.013–4.015), in line with the findings reported by Lin et al. [ 33 ] , while multifocal lesions showed significantly higher LVI incidence (P < 0.001, OR = 5.345, 95%CI:2.371–12.165), suggesting more aggressive tumor biology that warrants further multicenter investigation. The circumferential involvement proportion represents the transverse infiltration range along the esophageal wall. While direct evidence remains limited, existing data suggest extensive CIP may elevate LVI risk through multiple pathways. Clinicopathological studies confirm a significant LVI-tumor extent association [ 34 ] . Notably, as esophageal submucosal lymphatic networks primarily follow longitudinal orientation, widely circumferential tumors show greater propensity for horizontal dissemination and interaction with dense vascular/lymphatic structures, facilitating vascular infiltration [ 35 , 36 ] . These observations position circumferential extent as a potentially valuable morphological marker for LVI risk assessment in SEC (P = 0.008, OR = 3.831, 95%CI:1.490–10.920), though its predictive utility requires further validation incorporating invasion depth and molecular subtypes. This investigation systematically incorporated lifestyle factors into the predictive model for LVI in SEC. The analysis demonstrated that long-term consumption of pickled foods significantly increases LVI risk (P = 0.002, odds ratio [OR] = 2.940, 95% confidence interval [CI]:1.502–6.006), independent of its established association with ESC development [ 37 , 38 ] . The pathophysiological mechanisms involve: (1) chronic inflammation induced by high salt and nitrosamine content, leading to nuclear factor kappa B (NF-κB) pathway activation that promotes tumor proliferation and vascular endothelial growth factor (VEGF)-mediated angiogenesis [ 39 – 43 ] ; (2) nitrite metabolite-induced DNA methylation changes resulting in p16 tumor suppressor gene silencing and histone modifications that enhance oncogenic activity [ 44 , 45 ] ; and (3) matrix metalloproteinase (MMP) secretion by activated cancer-associated fibroblasts (CAFs) that facilitates basement membrane degradation [ 37 , 46 , 47 ] . Our findings propose a comprehensive prevention framework: (1) Individual-level dietary modification: Restricting pickled food consumption to <3 times per week with daily antioxidant supplementation (500 mg vitamin C + 400 IU vitamin E) based on established chemoprevention evidence [ 48 , 49 ] ; (2) Enhanced surveillance: Biennial endoscopic screening for populations with prolonged pickled food exposure; (3) Culturally adapted public health campaigns: Implement community education in high-risk regions on pickled food carcinogenicity (notably nitrite-to-nitrosamine conversion), validated through risk perception surveys, with culturally tailored delivery [ 50 ] . Patients with preoperative pathological confirmation of ESC showed a significantly higher risk of LVI than those with high-grade (HGIN) or low-grade (LGIN) intraepithelial neoplasia (P = 0.029; odds ratio [OR] = 2.344; 95% confidence interval [CI]:1.114–5.219). These findings underscore the critical need for enhanced LVI evaluation when preoperative pathology confirms ESC diagnosis [ 51 ] . Among eight evaluated machine learning algorithms, logistic regression demonstrated optimal performance for predicting LVI. Using 5-fold cross-validation, the model achieved area under the curve (AUC) values of 0.871 (training set) and 0.902 (validation set), with consistent performance across multiple analyses. Comprehensive evaluation included receiver operating characteristic (ROC) curve analysis, decision curve analysis (DCA), calibration curves, learning curves and precision-recall (PR) curves.SHAP analysis was implemented to enhance model interpretability by quantifying feature contributions. This investigation has several limitations that should be acknowledged. First, as a single-center retrospective study with a limited sample size, the findings require validation through multicenter prospective studies with larger cohorts. Second, the assessment of lifestyle factors relied on subjective measures, highlighting the need for more objective quantitative methods to establish standardized evaluation protocols. Third, the current randomization approach may introduce selection bias, suggesting that stratified sampling or other rigorous methods could enhance study quality. Finally, while ROC curve-derived thresholds were employed, these may not fully capture clinically relevant cutoffs and could oversimplify data complexity, indicating the need to explore alternative analytical approaches. CONCLUSIONS A novel machine learning-based predictive model was developed, with the logistic regression algorithm demonstrating optimal performance upon comparative validation. By incorporating SHAP interpretability analysis, this model enables personalized risk assessment for LVI in SEC, addressing a critical limitation of existing predictive approaches. The model provides objective decision-support for clinicians to accurately identify high-risk patients, refine treatment strategies, and reduce unnecessary medical resource utilization from excessive interventions. Abbreviations lymphovascular invasion (LVI) superficial esophageal carcinoma (SEC) Shapley Additive Explanations (SHAP) body mass index (BMI) chronic obstructive pulmonary disease (COPD) neutrophil-to-lymphocyte ratio (NLR) preoperative platelet-lymphocyte ratio (PLR) preoperative platelet-neutrophil ratio (PNR) endoscopic or magnification (EOM) circumferential involvement proportion(CIP) receiver operating characteristic (ROC) decision curve analysis (DCA) precision-recall (PR) high-grade intraepithelial neoplasia(HGIN) low-grade ntraepithelial neoplasia(LGIN) endoscopic submucosal dissection (ESD) eXtreme Gradient Boosting (XGBoost) Light Gradient Boosting Machine (LightGBM) Adaptive Boosting (AdaBoost) Gradient Boosting Decision Tree (GBDT) Gaussian Naïve Bayes (GNB) Complement Naïve Bayes (CNB) Declarations Ethical approval and consent to participate The present study was performed at The Affiliated Hospital of North Sichuan Medical College under the Declaration of Helsinki.This retrospective analysis was approved by the The Affiliated Hospital of North Sichuan Medical College's Ethics Committee, which waived the requirement for informed consent due to the study's retrospective design.(ethical batch number: 2024ER143-1) Consent for publication Not applicable. Availability of data and materials Data is provided within the manuscript or supplementary information files. Competing Interests All authors declare that they have no competing interests. Funding This research was funded by the Nanchong Science and Technology Program ((grant number: 23JCYJPT0060) and the Key Project of Research and Development Program of Affiliated Hospital of North Sichuan Medical College ((grant number: 2023ZD009). Authors’ Contributions Xianfei Wang, Zichen Luo designed the study; Xianfei Wang Zichen Luo analyzed the data; Xianfei Wang, Zichen Luo drafted the article; Xianfei Wang, Zichen Luo critically revised the article; Xianfei Wang finally approved the article; Zichen Luo, Xinrui Chen, Yutong Cui, Ji Zuo, Shiqi Liang, Guangbing Hu, Chengyu Zhang, Haorui Li , Xuemei Hou provided the study materials or patients. All authors read and approved the final manuscript. Acknowledgments Not applicable. References HOU H, MENG Z, ZHAO X, et al. Survival of Esophageal Cancer in China: A Pooled Analysis on Hospital-Based Studies From 2000 to 2018[J/OL]. Front Oncol. 2019;9:548. 10.3389/fonc.2019.00548 . SHAH M A, KENNEDY E B, CATENACCI D V, et al. Treatment of Locally Advanced Esophageal Carcinoma: ASCO Guideline[J/OL]. J Clin Oncol. 2020;38(23):2677–94. 10.1200/JCO.20.00866 . HEALTH COMMISSION OF THE PRC N, NATIONAL HEALTH COMMISSION OF THE PEOPLE’S REPUBLIC, OF CHINA. National guidelines for diagnosis and treatment of esophageal carcinoma 2022 in China (English version)[J/OL]. Chin J Cancer Res. 2022;34(3):309–34. 10.21147/j.issn.1000-9604.2022.04.01 . OYAMA T, INOUE H, ARIMA M, et al. Prediction of the invasion depth of superficial squamous cell carcinoma based on microvessel morphology: magnifying endoscopic classification of the Japan Esophageal Society[J/OL]. Esophagus. 2017;14(2):105–12. 10.1007/s10388-016-0527-7 . WANG A, TAN Y, ZHANG Y, et al. The prognostic role of angiolymphatic invasion in N0 esophageal carcinoma: a meta-analysis and systematic review[J/OL]. J Thorac Disease. 2019;11(8):3276–83. 10.21037/jtd.2019.08.50 . WANG A, TAN Y, WANG S, et al. The prognostic value of separate lymphatic invasion and vascular invasion in oesophageal squamous cell carcinoma: a meta-analysis and systematic review[J/OL]. BMC Cancer. 2022;22(1):1329. 10.1186/s12885-022-10441-6 . LI P, LING Y H, ZHU C M et al. Vascular invasion as an independent predictor of poor prognosis in nonmetastatic gastric cancer after curative resection[J]. TAO Y, CHEN S, YU J, et al. Risk factors of lymph node metastasis or lymphovascular invasion for superficial esophageal squamous cell carcinoma: A practical and effective predictive nomogram based on a cancer hospital data[J/OL]. Front Med. 2022;9:1038097. 10.3389/fmed.2022.1038097 . KOOK MC. Risk Factors for Lymph Node Metastasis in Undifferentiated-Type Gastric Carcinoma[J/OL]. Clin Endoscopy. 2019;52(1):15–20. 10.5946/ce.2018.193 . YEH R JAYAPRAKASAMVS, KU G Y, et al. Role of Imaging in Esophageal Cancer Management in 2020: Update for Radiologists[J/OL]. Am J Roentgenol. 2020;215(5):1072–84. 10.2214/AJR.20.22791 . PARTICIPANTS IN THE PARIS WORKSHOP. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon[J/OL]. Gastrointest Endosc. 2003;58(6):S3–43. 10.1016/S0016-5107(03)02159-X . ORLOFF NC, FLAMMER A, HARTNETT J, et al. Food cravings in pregnancy: Preliminary evidence for a role in excess gestational weight gain[J/OL]. Appetite. 2016;105:259–65. 10.1016/j.appet.2016.04.040 . KIKUCHI Y. Personality and Dietary Habits.[J/OL]. J Epidemiol. 2000;10(3):191–8. 10.2188/jea.10.191 . WEST R. Tobacco smoking: Health impact, prevalence, correlates and interventions[J/OL]. Psychol Health. 2017;32(8):1018–36. 10.1080/08870446.2017.1325890 . GARNETT C, OLDHAM M. Prevalence and characteristics of co-occurrence of smoking and increasing-and-higher-risk drinking: A population survey in England[J/OL]. Addict Behav. 2024;150:107928. 10.1016/j.addbeh.2023.107928 . BERKOVICH G Y, LEVINE M S, MILLER W T. CT Findings in Patients with Esophagitis[J/OL]. Am J Roentgenol. 2000;175(5):1431–4. 10.2214/ajr.175.5.1751431 . JANG K M, LEE K S, LEE SJ, et al. The Spectrum of Benign Esophageal Lesions: Imaging Findings[J/OL]. Korean J Radiol. 2002;3(3):199. 10.3348/kjr.2002.3.3.199 . HALLINAN J T P D, VENKATESH SK. Gastric carcinoma: imaging diagnosis, staging and assessment of treatment response[J/OL]. Cancer Imaging. 2013;13(2):212–27. 10.1102/1470-7330.2013.0023 . YANG Q, LIU Z. A narrative review: narrow-band imaging endoscopic classifications[J/OL]. Quant Imaging Med Surg. 2023;13(2):1138–63. 10.21037/qims-22-728 . CHIAM K H, SHIN S H, CHOI K C, et al. Current Status of Mucosal Imaging with Narrow-Band Imaging in the Esophagus[J/OL]. Gut Liver. 2021;15(4):492–9. 10.5009/gnl20031 . KURIBAYASHI S, HOSAKA H. Usefulness of Endoscopy for the Detection and Diagnosis of Primary Esophageal Motility Disorders and Diseases Relating to Abnormal Esophageal Motility[J/OL]. Diagnostics. 2023;13(4):695. 10.3390/diagnostics13040695 . ANDRIOPOULOS V. LASSO Regression with Multiple Imputations for the Selection of Key Variables Affecting the Fatty Acid Profile of Nannochloropsis oculata[J/OL]. Mar Drugs. 2023;21(9):483. 10.3390/md21090483 . DEMEESTER SR. Evaluation and Treatment of Superficial Esophageal Cancer[J/OL]. J Gastrointest Surg. 2010;14:S94–100. 10.1007/s11605-009-1025-1 . KITAGAWA Y, ISHIHARA R, ISHIKAWA H, et al. Esophageal cancer practice guidelines 2022 edited by the Japan esophageal society: part 1[J/OL]. Esophagus. 2023;20(3):343–72. 10.1007/s10388-023-00993-2 . ZHANG B, DU W. Significance of the neutrophil-to-lymphocyte ratio in young patients with oral squamous cell carcinoma[J/OL]. Cancer Manage Res. 2019;11:7597–603. 10.2147/CMAR.S211847 . BUONACERA A, STANCANELLI B, COLACI M, et al. Neutrophil to Lymphocyte Ratio: An Emerging Marker of the Relationships between the Immune System and Diseases[J/OL]. Int J Mol Sci. 2022;23(7):3636. 10.3390/ijms23073636 . HESHMAT-GHAHDARIJANI K, SARMADI V, HEIDARI A, et al. The neutrophil-to-lymphocyte ratio as a new prognostic factor in cancers: a narrative review[J/OL]. Front Oncol. 2023;13:1228076. 10.3389/fonc.2023.1228076 . MIN K W, KWON M J, KIM D H, et al. Persistent elevation of postoperative neutrophil-to-lymphocyte ratio: A better predictor of survival in gastric cancer than elevated preoperative neutrophil-to-lymphocyte ratio[J/OL]. Sci Rep. 2017;7(1):13967. 10.1038/s41598-017-13969-x . RUAN R, CHEN S, TAO Y, et al. A Nomogram for Predicting Lymphovascular Invasion in Superficial Esophageal Squamous Cell Carcinoma[J/OL]. Front Oncol. 2021;11:663802. 10.3389/fonc.2021.663802 . LI Y, YU M, WANG G, et al. Corrigendum: Contrast-Enhanced CT-Based Radiomics Analysis in Predicting Lymphovascular Invasion in Esophageal Squamous Cell Carcinoma[J/OL]. Front Oncol. 2021;11:712493. 10.3389/fonc.2021.712493 . LI Y, SU H. Can lymphovascular invasion be predicted by contrast-enhanced CT imaging features in patients with esophageal squamous cell carcinoma? A preliminary retrospective study[J/OL]. BMC Med Imaging. 2022;22(1):93. 10.1186/s12880-022-00804-7 . WANG Y, BAI G, HUANG W, et al. A radiomics nomogram based on contrast-enhanced CT for preoperative prediction of Lymphovascular invasion in esophageal squamous cell carcinoma[J/OL]. Front Oncol. 2023;13:1208756. 10.3389/fonc.2023.1208756 . MIN S K, LEE S K, WOO J, et al. Relation Between Tumor Size and Lymph Node Metastasis According to Subtypes of Breast Cancer[J/OL]. J Breast Cancer. 2021;24(1):75. 10.4048/jbc.2021.24.e4 . HUANG S, ZHU Y, CAI H, et al. Impact of lymphovascular invasion in oral squamous cell carcinoma: A meta-analysis[J/OL]. Oral Surgery, Oral Medicine. Oral Pathol Oral Radiol. 2021;131(3):319–e3281. 10.1016/j.oooo.2020.10.026 . WANG Y, ZHU L, XIA W, et al. Anatomy of lymphatic drainage of the esophagus and lymph node metastasis of thoracic esophageal cancer[J/OL]. Cancer Manage Res. 2018;10:6295–303. 10.2147/CMAR.S182436 . VOGT CD, PANOSKALTSIS-MORTARI A. Tissue engineering of the gastroesophageal junction[J/OL]. J Tissue Eng Regen Med. 2020;14(6):855–68. 10.1002/term.3045 . NILAND S, RISCANEVO A X, EBLE JA. Matrix Metalloproteinases Shape the Tumor Microenvironment in Cancer Progression[J/OL]. Int J Mol Sci. 2021;23(1):146. 10.3390/ijms23010146 . YE X Y, LAI Y T, SONG W P, et al. The research progress on the association between dietary habits and esophageal cancer: a narrative review[J/OL]. Annals Palliat Med. 2021;10(6):6948–56. 10.21037/apm-21-1467 . SALIHI A, AL–NAQSHABANDI M, KHUDHUR Z, et al. Gasotransmitters in the tumor microenvironment: Impacts on cancer chemotherapy (Review)[J/OL]. Mol Med Rep. 2022;26(1):233. 10.3892/mmr.2022.12749 . LI S, HOEFNAGEL S J M, KRISHNADATH KK. Molecular Biology and Clinical Management of Esophageal Adenocarcinoma[J/OL]. Cancers. 2023;15(22):5410. 10.3390/cancers15225410 . LIN S H, LI Y H, LEUNG K, et al. Salt Processed Food and Gastric Cancer in a Chinese Population[J/OL]. Asian Pac J Cancer Prev. 2014;15(13):5293–8. 10.7314/APJCP.2014.15.13.5293 . BOURAS E, TSILIDIS K K, TRIGGI M, et al. Diet and Risk of Gastric Cancer. Umbrella Review[J/OL] Nutrients. 2022;14(9):1764. 10.3390/nu14091764 . HSIEH H L, TSAI MM. Tumor progression-dependent angiogenesis in gastric cancer and its potential application[J/OL]. World J Gastrointest Oncol. 2019;11(9):686–704. 10.4251/wjgo.v11.i9.686 . MUTHUKUMARAN RB, BHATTACHARJEE P, BHOWMICK P, et al. Genetic and epigenetic instability induced by betel quid associated chemicals[J/OL]. Toxicol Rep. 2023;10:223–34. 10.1016/j.toxrep.2023.02.001 . DODD L E SENGUPTAS, CHEN I H, et al. Genes Involved in DNA Repair and Nitrosamine Metabolism and Those Located on Chromosome 14q32 Are Dysregulated in Nasopharyngeal Carcinoma[J/OL]. Cancer Epidemiol Biomarkers Prev. 2006;15(11):2216–25. 10.1158/1055-9965.EPI-06-0455 . HUA H, LI M, LUO T, et al. Matrix metalloproteinases in tumorigenesis: an evolving paradigm[J/OL]. Cell Mol Life Sci. 2011;68(23):3853–68. 10.1007/s00018-011-0763-x . SARKAR M, NGUYEN T. Cancer-associated fibroblasts: The chief architect in the tumor microenvironment[J/OL]. Front Cell Dev Biology. 2023;11:1089068. 10.3389/fcell.2023.1089068 . ERDÉLYI A, PÁLFI E. The Importance of Nutrition in Menopause and Perimenopause—A. Review[J/OL] Nutrients. 2023;16(1):27. 10.3390/nu16010027 . PANEL ON DIETARY ANTIOXIDANTS AND RELATED COMPOUNDS, SUBCOMMITTEE ON UPPER REFERENCE LEVELS OF NUTRIENTS, SUBCOMMITTEE ON INTERPRETATION AND USES OF DIETARY REFERENCE, INTAKES, Vitamin E, Selenium, Carotenoids [M/OL] et al. Washington, D.C.: National Academies, 2000[2025-05-01]. https://www.nap.edu/catalog/9810 . 10.17226/9810 NIU J, LI B, ZHANG Q, et al. Exploring the traditional Chinese diet and its association with health status—a systematic review[J/OL]. Nutr Rev. 2025;83(2):e237–56. 10.1093/nutrit/nuae013 . HATTA W, KOIKE T, UNO K, et al. Management of Superficial Esophageal Squamous Cell Carcinoma and Early Gastric Cancer following Non-Curative Endoscopic Resection[J/OL]. Cancers. 2022;14(15):3757. 10.3390/cancers14153757 . Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.docx 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-6591573","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":454983196,"identity":"156b8698-c050-4288-b35b-f54bfb326b5d","order_by":0,"name":"Zichen Luo","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Zichen","middleName":"","lastName":"Luo","suffix":""},{"id":454983197,"identity":"cc704d4f-a381-4dd9-915d-0af491d1f88c","order_by":1,"name":"Xinrui Chen","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Xinrui","middleName":"","lastName":"Chen","suffix":""},{"id":454983198,"identity":"4e2d44cf-7f36-40e7-af87-cd74a777fe99","order_by":2,"name":"Yutong Cui","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Yutong","middleName":"","lastName":"Cui","suffix":""},{"id":454983199,"identity":"e9ae1e9b-bc5d-4112-8799-d9fb370c2c33","order_by":3,"name":"Ji Zuo","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Ji","middleName":"","lastName":"Zuo","suffix":""},{"id":454983200,"identity":"eae317ce-fbdd-497a-87b4-eea4776bd766","order_by":4,"name":"Shiqi Liang","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Shiqi","middleName":"","lastName":"Liang","suffix":""},{"id":454983201,"identity":"4d0b0b5e-6a67-42d5-aabe-e72796bdeca8","order_by":5,"name":"Guangbing Hu","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Guangbing","middleName":"","lastName":"Hu","suffix":""},{"id":454983202,"identity":"6e2e1d74-09f9-4ac5-957f-00b95aa2b929","order_by":6,"name":"Chengyu Zhang","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Chengyu","middleName":"","lastName":"Zhang","suffix":""},{"id":454983203,"identity":"101b095b-3043-4e43-8d34-4c23fd5dc813","order_by":7,"name":"Haorui Li","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Haorui","middleName":"","lastName":"Li","suffix":""},{"id":454983204,"identity":"c4042550-fd6d-46b1-8f40-0a3ed80deb0d","order_by":8,"name":"Xuemei Hou","email":"","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":false,"prefix":"","firstName":"Xuemei","middleName":"","lastName":"Hou","suffix":""},{"id":454983205,"identity":"5527b50d-0eab-4f21-85f2-2f40b581f232","order_by":9,"name":"Xianfei Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2UlEQVRIie3RsQrCMBCA4chBukS6tlT0CYRAoNKlfRVFcMoDOAhGCnZR5xb6FkJxzJSp7hnt4iRodwdXnRo3wfz7x91xCNlsP5g7ul9k+1xBkqXSjPiC95pCKIcSNTUjVHJgfQEu9Tg13EyelV+ccDAhdatvKB6ORYfobQ4Lr60HLMr2x6hEcxbKDgKAQj/HeI7qcxUQJGdVF8EYhQHBsBaaX80IIYSx/haAao7NiOfhWZPXCvydYlFJDW5JNEj5WK7AddJG35bxsJN8jiSmr3kj3wqbzWb7i17woEQPE8quMwAAAABJRU5ErkJggg==","orcid":"","institution":"Affiliated Hospital of North Sichuan Medical College (University)","correspondingAuthor":true,"prefix":"","firstName":"Xianfei","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2025-05-05 06:08:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6591573/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6591573/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":82605911,"identity":"d43c5c20-0c9a-465c-b584-236160ea1798","added_by":"auto","created_at":"2025-05-13 10:04:10","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":64640,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of patients included in the analysis.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/19b0007dac0168dba89d3ae9.jpg"},{"id":82605912,"identity":"39a7c8a0-033d-4360-8beb-e2b8ec1aedf7","added_by":"auto","created_at":"2025-05-13 10:04:10","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":91001,"visible":true,"origin":"","legend":"\u003cp\u003eLASSO regression analysis was used to select characteristic factors. (a) Vertical line at optimal λ (0.029) from 10-fold CV identifies 9 key predictors. (b) LASSO coefficient paths for 31 variables. Dashed lines: λ at min MSE (0.048) and 1-SE rule (0.006)\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/ce0a82ab45bd3983ba6c867d.jpg"},{"id":82605915,"identity":"321b3307-b51a-4d5b-9eb0-5001e0dbe385","added_by":"auto","created_at":"2025-05-13 10:04:10","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":122906,"visible":true,"origin":"","legend":"\u003cp\u003eMultiple machine learning model comprehensive analysis. (a) Training ROC/AUC: Model discrimination for LVI(AUC→1 = better performance)(b) Validation ROC/AUC: Mean AUC from 10 stratified 7:3 samples evaluates generalizability(c) Calibration curve: Predicted vs actual probabilities (closer to diagonal = better; Brier score shown)(d) DCA: Net benefit between model (solid), treat-all (black dashed) and treat-none (red dashed) lines(e) Training PR: Precision-recall curve and AP (higher/covering more area = better ranking)(f) Validation PR: Generalizable ranking performance (top-right = better positive identification)\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/33a46a2dcd63d20b9cf88e83.jpg"},{"id":82605928,"identity":"db4ac34e-d58f-4445-b2e2-b1e24faf60b4","added_by":"auto","created_at":"2025-05-13 10:04:11","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":82724,"visible":true,"origin":"","legend":"\u003cp\u003eLogistic regression model training, validation, and testing. (a) Training set ROC curve with AUC(b) Validation set ROC/AUC from 30% random subset (5 replicates shown)(c) Test set ROC/AUC for 30% esophageal cancer cohort(d) Learning curves (training: red dashed; validation: blue dashed)\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/9968394e873e6a78caa3f883.jpg"},{"id":82605930,"identity":"a19e4043-b093-4b8e-9fd4-5cb39851ea2c","added_by":"auto","created_at":"2025-05-13 10:04:11","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":42660,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP values demonstrating individual feature importance and feature effect combinations based on the logistic regression model.(a) Feature importance derived from SHAP algorithm.(b) Predictive correlation between features and LVI based on SHAP analysis.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/97d7cc5f8351bf6290edffef.jpg"},{"id":82607111,"identity":"8e2c702c-a9ed-488e-83dc-91df4b05d4bd","added_by":"auto","created_at":"2025-05-13 10:12:10","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":56483,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP feature contribution plots for predictions of LVI presence and absence.(a) SHAP plot for correct prediction of LVI presence.(b) SHAP plot for correct prediction of LVI absence.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/867c59796d6d3a02837daf3c.jpg"},{"id":82605931,"identity":"e6d5adf4-5f44-46d1-b727-17eff11ec476","added_by":"auto","created_at":"2025-05-13 10:04:11","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":39305,"visible":true,"origin":"","legend":"\u003cp\u003eNomogram of the machine learning prediction model for LVI in SEC.\u003c/p\u003e","description":"","filename":"7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/9d42eea34467721a00c38c20.jpg"},{"id":96803193,"identity":"7db20102-f823-498e-9d16-00c499d5db67","added_by":"auto","created_at":"2025-11-26 08:54:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1779454,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/31e9ab6d-d4ce-4935-a104-2cb09c6342c8.pdf"},{"id":82607110,"identity":"fc424c27-f31c-4403-8b2a-b372b3e237c7","added_by":"auto","created_at":"2025-05-13 10:12:10","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":27522,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-6591573/v1/60e9bc94b0f55c1c62c33efd.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning-Based Prediction of Lymphovascular Invasion in Superficial Esophageal Carcinoma: Model Development and Risk Factor Analysis","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eEsophageal cancer is a common malignancy of the digestive tract, with a troubling increase in both incidence and mortality rates observed globally in recent years. This neoplasm has emerged as a significant public health challenge, particularly in certain geographic regions, notably China, where it represents a substantial threat to public health and life expectancy. Epidemiological data reveal a distinct distribution of disease burden, with high prevalence rates concentrated in specific areas, underscoring the need for focused research efforts and the development of targeted intervention strategies in these high-risk regions \u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e.Esophageal cancer is histologically classified into two principal subtypes: squamous cell carcinoma and adenocarcinoma. In the Chinese population, squamous cell carcinoma is the predominant subtype, demonstrating distinct epidemiological patterns and unique clinical and biological behave ors \u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Esophageal cancer is often diagnosed at advanced local or metastatic stages due to th e absence of distinct clinical symptoms in its early phases. Consequently, the majority of patients present with disease at a stage where treatment outcomes are suboptimal, leading to a generally unfavorable prognosis.\u003c/p\u003e \u003cp\u003eSuperficial esophageal carcinoma (SEC) is defined as an early-stage esophageal cancer in which the neoplasm remains restricted to the mucosal or submucosal layers of the esophagus, without invasion of the muscularis propria \u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. The rapid advancements in endoscopic diagnostic and therapeutic techniques have led to a significant increase in the detection rate of SEC, facilitating earlier intervention. However, it is crucial to recognize that even at this early stage, esophageal cancer can metastasize to distant sites through Lymphovascular invasion (LVI), a key pathological process. As a primary pathway for tumor dissemination, LVI substantially enhances the risk of both lymphatic and hematogenous metastasis, which is strongly associated with the worsening of patient prognosis \u003csup\u003e[\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. Consequently, the establishment of a robust risk prediction system for LVI in patients with SEC is pivotal for refining treatment strategies and enhancing clinical prognosis.\u003c/p\u003e \u003cp\u003eAt present, the investigation into the pathological mechanisms of LVI in SEC is still in the exploratory phase, with a significant gap in the development of comprehensive risk prediction models. While several prognostic models for esophageal cancer have been proposed, the role of LVI as a central variable has not been extensively examined. Clinical evidence indicates a complex interplay between LVI and various biological characteristics, such as tumor volume, differentiation, and invasion depth; however, the underlying mechanisms and their quantitative relationships remain to be elucidated\u003csup\u003e[\u003cspan additionalcitationids=\"CR8\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e. A critical challenge in current research is the integration of clinical phenotypes, imaging characteristics, and molecular pathological biomarkers to establish a comprehensive, multidimensional, and quantifiable risk assessment system.\u003c/p\u003e \u003cp\u003eThis study intends to conduct a comprehensive analysis of multidimensional clinical data, including imaging and pathological features, to construct an evidence-based predictive model for LVI risk. Through the application of multivariate analysis to identify critical risk factors, a predictive algorithm will be developed to offer quantitative decision support for clinicians. The goal is to optimize individualized treatment strategies, ultimately improving treatment outcomes and quality of life.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eMaterials\u003c/h2\u003e \u003cp\u003eA retrospective cohort study was performed on 592 patients diagnosed with SEC at the Department of Gastroenterology and Thoracic Surgery, Affiliated Hospital of North Sichuan Medical College, between January 2017 and December 2024. Among these, 286 patients underwent endoscopic treatment, while 306 patients received surgical intervention. The inclusion criteria were: (1) confirmed diagnosis of superficial esophageal squamous cell carcinoma based on postoperative biopsy; (2) absence of regional lymph node involvement or distant metastasis on preoperative evaluation; and (3) no prior history of other malignancies. Exclusion criteria encompassed: (1) Postoperative pathological diagnosis of non-esophageal squamous cell carcinoma cases; (2) patients who had received preoperative chemotherapy or radiotherapy; (3) lack of histopathological assessment for LVI; and (4) patients with missing data exceeding 15%. The patient selection flowchart is summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe verification of vascular invasion.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe pathological diagnosis of LVI is primarily based on histological examination of tumor tissue obtained through surgical resection or biopsy. Microscopic evaluation focuses on identifying tumor cell penetration through vascular or lymphatic basement membranes with luminal infiltration. Immunohistochemical stains (CD31 for blood vessels and D2-40 for lymphatic channels) are routinely employed to enhance diagnostic accuracy by precisely delineating endothelial structures, thereby improving the reliability of LVI detection.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy variables\u003c/h3\u003e\n\u003cp\u003eThis study included 34 variables potentially influencing esophageal lesions, with 31 being preoperatively assessable. These variables encompass three primary dimensions: clinical indicators, histopathological characteristics, and lifestyle-related factors.\u003c/p\u003e \u003cp\u003eClinical factors: age, sex, preoperative neutrophil-lymphocyte ratio (NLR), preoperative platelet-lymphocyte ratio (PLR), preoperative platelet-neutrophil ratio (PNR), body mass index (BMI), hypertension, cardiac disease, diabetes mellitus, chronic kidney disease, chronic obstructive pulmonary disease (COPD), family history, brownish discoloration, chronic esophagitis, endoscopic ultrasound or magnifying endoscopy (EOM), esophageal wall thickness, esophageal stricture, tumor enhancement, lesion location, multiple lesions, tumor diameter, circumferential involvement proportion(CIP), Paris classification\u003c/p\u003e \u003cp\u003ehistopathological characteristics: depth of invasion, preoperative pathological type, LVI, differentiation grade.\u003c/p\u003e \u003cp\u003eLifestyle factors: long-term smoking, long-term alcohol consumption, pickled food intake, high-temperature food intake, fried food intake, fruit and vegetable intake, and oral hygiene.\u003c/p\u003e\n\u003ch3\u003eDefinitions\u003c/h3\u003e\n\u003cp\u003eLesion location: According to the current standard\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e, the esophagus is divided into four segments based on the distance from the upper incisors: cervical segment (15\u0026ndash;20 cm), upper thoracic segment (20\u0026ndash;25 cm), middle thoracic segment (25\u0026ndash;30 cm), and lower thoracic segment (30\u0026ndash;40 cm).\u003c/p\u003e \u003cp\u003eDepth of tissue invasion:According to the 2002 Paris Classification for Gastrointestinal Tumors\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e, the histological depth of esophageal or gastrointestinal tumor invasion is classified as: (1) Mucosal layer (M): M1 (epithelium-limited), M2 (lamina propria invasion), M3 (muscularis mucosae involvement); (2) Submucosal layer (SM): SM1 (upper third, \u0026le;\u0026thinsp;200\u0026micro;m), SM2 (middle third), SM3 (lower third).\u003c/p\u003e \u003cp\u003eLifestyle assessment: Lifestyle factors were evaluated using previously established scoring systems\u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e, including dietary habits (e.g., fruit/vegetable, pickled food, and fried food intake), oral hygiene (e.g., toothbrushing frequency), and hot water consumption. These were categorized by frequency as occasional (\u0026le;\u0026thinsp;3 times/week) or frequent (\u0026gt;\u0026thinsp;3 times/week). Hot water intake was assessed based on self-perceived temperature thresholds. Chronic smoking was typically defined as sustained tobacco use for \u0026gt;\u0026thinsp;6 months, while chronic alcohol consumption was generally considered present with sustained use\u0026thinsp;\u0026gt;\u0026thinsp;3 months\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eEndoscopic ultrasound or magnifying endoscopy with narrow-band imaging (EUS or ME-NBI, EOM): refers to the evaluation of esophageal lesion invasion depth using either of these endoscopic modalities.\u003c/p\u003e \u003cp\u003eContrast-enhanced CT findings: Esophageal strictures primarily manifest as proximal luminal dilation with fluid-air levels\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e. Esophageal wall thickness is classified into three grades: mild (\u0026lt;\u0026thinsp;5 mm), moderate (5\u0026ndash;7 mm), and severe (\u0026gt;\u0026thinsp;7 mm). Following intravenous contrast administration, tumor regions demonstrate more pronounced enhancement compared to normal esophageal tissue due to their rich vascular supply\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eConventional endoscopic findings: Under narrow-band imaging (NBI), esophageal lesions exhibiting brownish discoloration may indicate inflammation, vascular abnormalities, or neoplastic changes\u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e. Chronic esophagitis typically presents with mucosal erythema, edema, erosion, or ulceration, potentially accompanied by white plaques, exudates, or a granular appearance\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eStatistical analysis of variables\u003c/h3\u003e\n\u003cp\u003eContinuous variables were summarized as median with interquartile range (IQR) and compared using nonparametric Mann-Whitney U tests. Categorical data were expressed as frequencies (percentages) and analyzed with Pearson χ\u0026sup2; or Fisher exact tests, as appropriate. A two-tailed α level of 0.05 defined statistical significance. All statistical computations were performed using R version 4.2.3 (gtsummary v1.7.2) and Python (scikit-learn v1.1.3) with standard parameter configurations.\u003c/p\u003e\n\u003ch3\u003eModel development and validation\u003c/h3\u003e\n\u003cp\u003eThe analytical framework employed a 70:30 random division of patient data into training (n\u0026thinsp;=\u0026thinsp;414) and validation (n\u0026thinsp;=\u0026thinsp;178) cohorts for predictive modeling. Multiple machine learning algorithms were systematically evaluated using Shapley Additive exPlanations (SHAP) analysis to determine variable importance and enhance model interpretability. The preprocessing phase incorporated k-nearest neighbors imputation (scikit-learn v1.1.3) and dichotomization of continuous variables (e.g., age, PLR, NLR) based on receiver operating characteristic (ROC)-derived thresholds (e.g., age\u0026thinsp;\u0026ge;\u0026thinsp;67 years), with subsequent statistical analyses performed in R (v4.2.3) using the gtsummary package (v1.7.2) for dataset stratification and baseline characterization.\u003c/p\u003e \u003cp\u003eVariable selection was conducted using least absolute shrinkage and selection operator (LASSO) regression (glmnet package, version 4.1.8) followed by multivariate logistic regression analysis (R software, version 4.2.3), retaining statistically significant variables (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). This dual approach effectively addressed overfitting through coefficient shrinkage while resolving multicollinearity\u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e. Nine machine learning algorithms were implemented: eXtreme Gradient Boosting (XGBoost), logistic regression, Light Gradient Boosting Machine (LightGBM), random forest, Adaptive Boosting (AdaBoost), decision tree, Gradient Boosting Decision Tree (GBDT), Gaussian Na\u0026iuml;ve Bayes (GNB), and Complement Na\u0026iuml;ve Bayes (CNB). Model optimization employed 10-fold cross-validation with repetition, with comprehensive evaluation incorporating receiver operating characteristic (ROC) curve analysis, decision curve analysis, and calibration plots to determine the optimal predictive model.\u003c/p\u003e \u003cp\u003eModel Validation and Interpretation: The selected prediction model was rigorously evaluated through 5-fold cross-validation, with model stability assessed via learning curve analysis (Python scikit-learn v1.1.3). We employed SHapley Additive exPlanations (SHAP; Python package v0.43.0) to quantitatively evaluate feature importance and provide interpretable visualization of the model's decision architecture. For clinical validation, two paradigmatic cases were systematically analyzed to demonstrate the model's operational performance in authentic clinical practice. The nomogram was employed to visualize the machine learning prediction model.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003ePartition of clinical data\u003c/h2\u003e \u003cp\u003eThe analysis included a final cohort of 592 eligible cases, among which pathological evaluation identified LVI in 66 cases (11.1%). Participants were randomly assigned to either training (70%, n\u0026thinsp;=\u0026thinsp;414) or testing (30%, n\u0026thinsp;=\u0026thinsp;178) cohorts using a computerized allocation system. Optimal cutoff values for continuous variables were derived from receiver operating characteristic (ROC) curve analysis (e.g., age\u0026thinsp;\u0026ge;\u0026thinsp;67 years as the optimal threshold). Baseline demographic and clinical characteristics of both cohorts are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Intergroup comparisons demonstrated no statistically significant differences in any baseline parameters (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05), confirming appropriate randomization and balanced cohort characteristics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePatient demographics and clinical characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eclassification term\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAll (n\u0026thinsp;=\u0026thinsp;592)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTraining Set (n\u0026thinsp;=\u0026thinsp;414)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTesting Set (n\u0026thinsp;=\u0026thinsp;178)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e299(50.507)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e214(51.691)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e85(47.753)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.380\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e293(49.493)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e200(48.309)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e93(52.247)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e215(36.318)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e153(36.957)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e62(34.831)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.622\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e377(63.682)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e261(63.043)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e116(65.169)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;22.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e164(27.703)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e116(28.019)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e48(26.966)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;22.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e428(72.297)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e298(71.981)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e130(73.034)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e457(77.196)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e316(76.329)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e141(79.213)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.443\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e135(22.804)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e98(23.671)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e37(20.787)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes mellitus ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e526(88.851)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e372(89.855)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e154(86.517)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.237\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e66(11.149)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e42(10.145)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e24(13.483)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCardiac disease ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e557(94.088)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e389(93.961)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e168(94.382)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.842\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e35(5.912)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e25(6.039)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e10(5.618)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChronic kidney disease ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e580(97.973)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e407(98.309)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e173(97.191)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.376\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12(2.027)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e7(1.691)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e5(2.809)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCOPD ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e569(96.115)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e397(95.894)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e172(96.629)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.671\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23(3.885)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e17(4.106)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e6(3.371)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoking history ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e360(60.811)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e248(59.903)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e112(62.921)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.490\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e232(39.189)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e166(40.097)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e66(37.079)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlcohol consumption ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e422(71.284)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e289(69.807)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e133(74.719)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.226\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e170(28.716)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e125(30.193)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e45(25.281)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFamily history ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e551(93.074)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e386(93.237)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e165(92.697)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e41(6.926)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e28(6.763)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13(7.303)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePLR ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;138.211\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e398(67.230)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e282(68.116)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e116(65.169)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.484\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;138.211\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e194(32.770)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e132(31.884)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e62(34.831)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePNR ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;47.137\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e215(36.318)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e144(34.783)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e71(39.888)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.236\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;47.137\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e377(63.682)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e270(65.217)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e107(60.112)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNLR ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;2.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e311(52.534)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e228(55.072)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e83(46.629)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.059\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;2.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e281(47.466)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e186(44.928)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e95(53.371)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEOM findings, ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e254(42.905)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e178(42.995)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e76(42.697)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.294\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e161(27.196)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e119(28.744)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e42(23.596)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e177(29.899)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e117(28.261)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e60(33.708)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBrownish discoloration,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e279(47.128)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e189(45.652)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e90(50.562)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.272\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e313(52.872)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e225(54.348)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e88(49.438)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor diamete ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;2.43cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e308(52.027)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e217(52.415)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e91(51.124)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.773\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;2.43cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e284(47.973)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e197(47.585)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e87(48.876)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMultiple lesions ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e522(88.176)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e365(88.164)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e157(88.202)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.990\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e70(11.824)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e49(11.836)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e21(11.798)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCIP,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;1/2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e183(30.912)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e131(31.643)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e52(29.213)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1/2\u0026ndash;3/4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e228(38.514)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e156(37.681)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e72(40.449)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;3/4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e181(30.574)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e127(30.676)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e54(30.337)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChronic esophagitis ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e567(95.777)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e398(96.135)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e169(94.944)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.509\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25(4.223)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e16(3.865)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e9(5.056)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor enhancement ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e348(58.784)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e251(60.628)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e97(54.494)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.164\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e244(41.216)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e163(39.372)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e81(45.506)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEsophageal stricture ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e370(62.500)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e269(64.976)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e101(56.742)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.058\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e222(37.500)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e145(35.024)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e77(43.258)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThickness,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;5mm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e310(52.365)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e212(51.208)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e98(55.056)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.655\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5-7mm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e179(30.236)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e127(30.676)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e52(29.213)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;7mm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e103(17.399)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e75(18.116)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e28(15.730)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLVI ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e无\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e526(88.851)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e366(88.406)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e160(89.888)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.599\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e有\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e66(11.149)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e48(11.594)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e18(10.112)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParis classification ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eII-a\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e153(25.845)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e113(27.295)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e40(22.472)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.167\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eII-b\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e310(52.365)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e212(51.208)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e98(55.056)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eII-c\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e85(14.358)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e54(13.043)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e31(17.416)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOthers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e44(7.432)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e35(8.454)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e9(5.056)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDifferentiation grade ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eWell-differentiated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e169(28.547)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e129(31.159)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e40(22.472)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.098\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModerately-differentiated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e397(67.061)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e267(64.493)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e130(73.034)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoorly-differentiated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e26(4.392)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e18(4.348)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8(4.494)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDepth of invasion ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e288(48.649)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e202(48.792)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e86(48.315)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.082\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e181(30.574)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e135(32.609)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e46(25.843)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e123(20.777)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e77(18.599)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e46(25.843)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTreatment modality ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eESD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e286(48.311)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e202(48.792)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e84(47.191)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.721\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSurgical operation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e306(51.689)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e212(51.208)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e94(52.809)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLesion location,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCervical or Upper thoracic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e92(15.541)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e64(15.459)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e28(15.730)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.713\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMiddle thoracic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e356(60.135)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e253(61.111)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e103(57.865)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLower thoracic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e144(24.324)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e97(23.430)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e47(26.404)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePreoperative pathology ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHGIN/LGIN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e275(46.453)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e191(46.135)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e84(47.191)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.813\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eESC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e317(53.547)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e223(53.865)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e94(52.809)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh-temperature food intake ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e438(73.986)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e308(74.396)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e130(73.034)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.729\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e154(26.014)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e106(25.604)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e48(26.966)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFried food intake ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e519(87.669)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e361(87.198)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e158(88.764)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.595\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e73(12.331)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e53(12.802)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e20(11.236)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFruit/vegetable intake ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e57(9.628)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e44(10.628)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13(7.303)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.209\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e535(90.372)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e370(89.372)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e165(92.697)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePickled food intake,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e312(52.703)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e214(51.691)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e98(55.056)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.452\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;3 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e280(47.297)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e200(48.309)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e80(44.944)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOral hygiene ,n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;4 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e52(8.784)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e37(8.937)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e15(8.427)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.882\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4\u0026ndash;7 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e524(88.514)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e365(88.164)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e159(89.326)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;7 times per week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e16(2.703)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e12(2.899)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4(2.247)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eIdentification of Risk Factors for LVI\u003c/h3\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eLASSO Regression Analysis\u003c/h2\u003e \u003cp\u003eLeast absolute shrinkage and selection operator (LASSO) regression analysis was performed to identify robust predictors of LVI while addressing multicollinearity. As demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the coefficient shrinkage profile (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea) and 10-fold cross-validation (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb) identified an optimal penalty parameter (λ\u0026thinsp;=\u0026thinsp;0.029) using the 1-standard error criterion. This yielded a parsimonious predictive model comprising nine clinically significant variables: PLR, NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative histopathological characteristics, and pickled food intake.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eAdjustment for confounding variables\u003c/h2\u003e \u003cp\u003eTo control for potential confounding effects, multivariable logistic regression analysis was performed incorporating LASSO-selected variables. Among the nine candidate predictors initially identified, stepwise regression analysis confirmed eight statistically significant independent risk factors (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e): NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative histopathological characteristics, and pickled food intake.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMultivariable logistic regression analysis of LVI in superficial esophageal carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePredictor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEstimate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eZ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eOdds Ratio\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eLower\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eUpper\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePLR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.593\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.341\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.082\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.809\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.924\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.536\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNLR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.731\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.343\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.132\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.033\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.077\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.071\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4.135\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThickness\u0026thinsp;=\u0026thinsp;5-7mm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.385\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.877\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.381\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.524\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThickness\u0026gt;7mm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.422\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.312\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e6.17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2.749\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e14.52\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEOM findings, = SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.914\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.517\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.077\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.495\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.914\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e7.07\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEOM findings,\u0026gt;SM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.516\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.449\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.379\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4.554\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.951\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e11.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor diamete\u0026thinsp;\u0026ge;\u0026thinsp;2.43cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.723\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.344\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.099\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.036\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.059\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4.111\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMultiple lesions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.687\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.413\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.081\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e5.401\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2.401\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e12.237\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCIP\u0026thinsp;=\u0026thinsp;1/2\u0026ndash;3/4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.703\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.492\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.429\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.153\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.799\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e5.638\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCIP\u0026gt;3/4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.394\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.788\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.005\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4.032\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.578\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e11.447\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePreoperative pathology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.894\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.387\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.309\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.446\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.171\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e5.409\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePickled food intake\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.117\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.351\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e3.054\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.564\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e6.243\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eModel Evaluation and Comparative Analysis\u003c/h2\u003e \u003cp\u003eThrough 10-fold cross-validation with repeated training, we evaluated multiple machine learning algorithms for classification tasks (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea,\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Logistic regression emerged as the optimal model, demonstrating robust performance with training and validation AUC values of 0.882 and 0.865, respectively. These results surpassed competing models: XGBoost (0.990/0.826), LightGBM (0.986/0.842), and GBDT (0.970/0.866). The model's strong generalizability was evidenced by minimal AUC discrepancy between training and validation sets (Δ\u0026thinsp;=\u0026thinsp;0.017), substantially lower than XGBoost (Δ\u0026thinsp;=\u0026thinsp;0.164), suggesting better resistance to overfitting. Additional validation through calibration analysis showed superior prediction accuracy (deviation\u0026thinsp;=\u0026thinsp;0.066 vs GBDT's 0.078) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). Decision curve analysis confirmed clinical utility with higher net benefit (0.04 at threshold\u0026thinsp;=\u0026thinsp;0.5) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed), while precision-recall metrics showed consistent performance across datasets (training\u0026thinsp;=\u0026thinsp;0.657, validation\u0026thinsp;=\u0026thinsp;0.635) (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). These comprehensive evaluations establish logistic regression as the preferred choice, offering an optimal balance of predictive accuracy, generalizability, and clinical interpretability.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment and Validation of the Prediction Model\u003c/h2\u003e \u003cp\u003eThe logistic regression model demonstrated significant advantages in classification tasks. Logistic regression analysis and 5-fold cross-validation were performed on the training set. Given that the validation set performance under the AUC metric did not exceed the test set or showed a discrepancy of less than 10%, the model was considered successfully fitted, confirming its applicability for classification modeling in this dataset. As evidenced by the ROC curves (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec), the AUC values for the training, validation, and test sets reached 0.871, 0.852, and 0.902 respectively, indicating that the model exhibits high accuracy in distinguishing positive and negative samples. Furthermore, the learning curve (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed) revealed that as the number of training samples increased, both training and validation accuracy rates gradually stabilized, demonstrating favorable learning capability and generalization performance. These findings suggest that the logistic regression model not only achieves superior classification accuracy but also maintains satisfactory stability and reliability in practical applications, establishing it as an effective classification methodology.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eSHAP-based Interpretation of Logistic Regression Model\u003c/h2\u003e \u003cp\u003eSHAP analysis was utilized to quantify risk factor contributions for LVI. The heatmap (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea) displays directional effects of eight predictors (red/blue indicating high/low risk), with mean absolute SHAP values (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb) revealing the following hierarchy: esophageal wall thickness (most significant)\u0026thinsp;\u0026gt;\u0026thinsp;tumor diameter\u0026thinsp;\u0026gt;\u0026thinsp;NLR\u0026thinsp;\u0026gt;\u0026thinsp;CIP\u0026thinsp;\u0026gt;\u0026thinsp;EOM findings\u0026thinsp;\u0026gt;\u0026thinsp;multiple lesions\u0026thinsp;\u0026gt;\u0026thinsp;preoperative pathology\u0026thinsp;\u0026gt;\u0026thinsp;pickled food intake. All factors positively correlated with LVI risk, particularly esophageal wall thickness (strongest association), tumor diameter, and NLR. This interpretable approach enhances preoperative risk assessment in SEC.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eSHAP-based Interpretation of Individual Case Predictions\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea demonstrates accurate negative prediction of LVI (probability\u0026thinsp;=\u0026thinsp;0.05), with Shapley values revealing primary contributions from elevated NLR, increased tumor diameter, and limited circumferential involvement. Conversely, Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eb illustrates correct positive identification (probability\u0026thinsp;=\u0026thinsp;0.63), predominantly influenced by: (i) frequent pickled food consumption, (ii) extensive circumferential involvement (\u0026gt;\u0026thinsp;75%), (iii) elevated NLR (\u0026gt;\u0026thinsp;2.014), (iv) tumor diameter exceeding 2.43 cm, and (v) EOM findings meeting SM2 invasion criteria.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eVisualization of the Machine Learning Prediction Model for LVI\u003c/h2\u003e \u003cp\u003eThe study developed a predictive nomogram incorporating eight independent risk factors for LVI: NLR, esophageal wall thickness, EOM findings, tumor diameter, multiple lesions, CIP, preoperative pathological characteristics, and pickled food intake. Clinicians can apply this tool by assigning points for each parameter according to the nomogram scale and summing them to calculate the total predicted probability of LVI in SEC patients.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eSuperficial esophageal carcinoma, representing an early pathological stage of esophageal cancer, displays unique biological characteristics and clinical presentations when compared to advanced disease\u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e. The evaluation of LVI during this stage holds significant clinical relevance, as it serves as a key parameter for assessing disease progression and predicting the risk of lymph node metastasis. Furthermore, it plays a crucial role in determining tailored treatment approaches. Precise determination of LVI status is of considerable clinical importance for enhancing patient outcomes and increasing survival rates. The Japanese \"Esophageal Cancer Practice Guidelines\"\u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e specifically advocate for additional surgical procedures in patients with histologically confirmed LVI following endoscopic submucosal dissection (ESD). This often entails esophagectomy, potentially accompanied by extensive lymph node dissection, to reduce the likelihood of postoperative recurrence and distant metastasis. In cases where surgery is contraindicated, radiotherapy and chemotherapy emerge as feasible alternative adjuvant treatment options.\u003c/p\u003e \u003cp\u003eAmong 66 LVI-confirmed patients, 28 were diagnosed post-ESD (19 underwent surgery, 6 received chemoradiotherapy due to surgical contraindications, 3 declined treatment for personal reasons). Precise preoperative LVI risk stratification could optimize surgical candidate selection and potentially avoid unnecessary ESD. Current LVI prediction faces accuracy limitations, underscoring the urgent need for robust predictive models to enhance assessment reliability and guide evidence-based treatment decisions.\u003c/p\u003e \u003cp\u003eThrough LASSO and multivariable logistic regression analyses of 31 preoperative clinical variables, we identified eight significant predictors for LVI in SEC: NLR, EOM features, esophageal wall thickness, tumor size, multifocal lesions, CIP, pickled food intake, and preoperative pathology. Importantly, we provide the first evidence that elevated NLR independently predicts LVI in SEC (P\u0026thinsp;=\u0026thinsp;0.024, OR\u0026thinsp;=\u0026thinsp;2.197, 95%CI:1.121\u0026ndash;4.434). Although the NLR-LVI mechanism requires further investigation, current evidence suggests NLR reflects the inflammatory/anti-tumor immune balance \u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e, where neutrophils promote tumor progression via cytokine release while lymphopenia facilitates immune evasion, collectively creating a pro-LVI microenvironment\u003csup\u003e[\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eA significant positive correlation exists between LVI incidence and tumor invasion depth in esophageal cancer. As SEC progresses from mucosal (T1a) to submucosal (T1b) invasion, particularly reaching deeper layers (SM2+), LVI frequency increases markedly\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e. A 516-patient cohort study reported LVI rates of 2.5% for T1a versus 15.7% for T1b lesions\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e, consistent with our findings: among 66 LVI-positive cases, 15 (5.32%) were T1a and 51 (16.78%) T1b. Stratified analysis revealed significantly higher LVI positivity in SM2 (10.52%) versus SM1 (6.25%) invasion (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), indicating progressive risk escalation with depth. Endoscopic ultrasound/magnifying endoscopy with narrow-band imaging (EUS/ME-NBI) confirmed substantially elevated LVI risk in submucosal involvement, particularly SM2 (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001, OR\u0026thinsp;=\u0026thinsp;5.971, 95%CI:2.417\u0026ndash;16.446), establishing objective predictive criteria.\u003c/p\u003e \u003cp\u003eThis study demonstrated through contrast-enhanced chest CT that esophageal wall thickness\u0026thinsp;\u0026gt;\u0026thinsp;7 mm was significantly associated with LVI risk in SEC (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001, OR\u0026thinsp;=\u0026thinsp;6.280, 95%CI:2.767\u0026ndash;14.978). These findings corroborate previous reports\u003csup\u003e[\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e confirming wall thickness as an independent LVI predictor (OR\u0026thinsp;=\u0026thinsp;16.32, 95%CI:5.89\u0026ndash;45.17). Pathologically, increased thickness indicates deeper tumor invasion, promoting basement membrane penetration and submucosal vascular/lymphatic invasion. Tumor diameter\u0026thinsp;\u0026gt;\u0026thinsp;2.43 cm independently predicted LVI (P\u0026thinsp;=\u0026thinsp;0.049, OR\u0026thinsp;=\u0026thinsp;1.992, 95%CI:1.013\u0026ndash;4.015), in line with the findings reported by Lin et al. \u003csup\u003e[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/sup\u003e, while multifocal lesions showed significantly higher LVI incidence (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001, OR\u0026thinsp;=\u0026thinsp;5.345, 95%CI:2.371\u0026ndash;12.165), suggesting more aggressive tumor biology that warrants further multicenter investigation.\u003c/p\u003e \u003cp\u003eThe circumferential involvement proportion represents the transverse infiltration range along the esophageal wall. While direct evidence remains limited, existing data suggest extensive CIP may elevate LVI risk through multiple pathways. Clinicopathological studies confirm a significant LVI-tumor extent association\u003csup\u003e[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]\u003c/sup\u003e. Notably, as esophageal submucosal lymphatic networks primarily follow longitudinal orientation, widely circumferential tumors show greater propensity for horizontal dissemination and interaction with dense vascular/lymphatic structures, facilitating vascular infiltration\u003csup\u003e[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]\u003c/sup\u003e. These observations position circumferential extent as a potentially valuable morphological marker for LVI risk assessment in SEC (P\u0026thinsp;=\u0026thinsp;0.008, OR\u0026thinsp;=\u0026thinsp;3.831, 95%CI:1.490\u0026ndash;10.920), though its predictive utility requires further validation incorporating invasion depth and molecular subtypes.\u003c/p\u003e \u003cp\u003eThis investigation systematically incorporated lifestyle factors into the predictive model for LVI in SEC. The analysis demonstrated that long-term consumption of pickled foods significantly increases LVI risk (P\u0026thinsp;=\u0026thinsp;0.002, odds ratio [OR]\u0026thinsp;=\u0026thinsp;2.940, 95% confidence interval [CI]:1.502\u0026ndash;6.006), independent of its established association with ESC development\u003csup\u003e[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]\u003c/sup\u003e. The pathophysiological mechanisms involve: (1) chronic inflammation induced by high salt and nitrosamine content, leading to nuclear factor kappa B (NF-κB) pathway activation that promotes tumor proliferation and vascular endothelial growth factor (VEGF)-mediated angiogenesis\u003csup\u003e[\u003cspan additionalcitationids=\"CR40 CR41 CR42\" citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]\u003c/sup\u003e; (2) nitrite metabolite-induced DNA methylation changes resulting in p16 tumor suppressor gene silencing and histone modifications that enhance oncogenic activity\u003csup\u003e[\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]\u003c/sup\u003e; and (3) matrix metalloproteinase (MMP) secretion by activated cancer-associated fibroblasts (CAFs) that facilitates basement membrane degradation\u003csup\u003e[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]\u003c/sup\u003e. Our findings propose a comprehensive prevention framework: (1) Individual-level dietary modification: Restricting pickled food consumption to \u0026lt;3 times per week with daily antioxidant supplementation (500 mg vitamin C\u0026thinsp;+\u0026thinsp;400 IU vitamin E) based on established chemoprevention evidence\u003csup\u003e[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]\u003c/sup\u003e ; (2) Enhanced surveillance: Biennial endoscopic screening for populations with prolonged pickled food exposure; (3) Culturally adapted public health campaigns: Implement community education in high-risk regions on pickled food carcinogenicity (notably nitrite-to-nitrosamine conversion), validated through risk perception surveys, with culturally tailored delivery\u003csup\u003e[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003ePatients with preoperative pathological confirmation of ESC showed a significantly higher risk of LVI than those with high-grade (HGIN) or low-grade (LGIN) intraepithelial neoplasia (P\u0026thinsp;=\u0026thinsp;0.029; odds ratio [OR]\u0026thinsp;=\u0026thinsp;2.344; 95% confidence interval [CI]:1.114\u0026ndash;5.219). These findings underscore the critical need for enhanced LVI evaluation when preoperative pathology confirms ESC diagnosis\u003csup\u003e[\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eAmong eight evaluated machine learning algorithms, logistic regression demonstrated optimal performance for predicting LVI. Using 5-fold cross-validation, the model achieved area under the curve (AUC) values of 0.871 (training set) and 0.902 (validation set), with consistent performance across multiple analyses. Comprehensive evaluation included receiver operating characteristic (ROC) curve analysis, decision curve analysis (DCA), calibration curves, learning curves and precision-recall (PR) curves.SHAP analysis was implemented to enhance model interpretability by quantifying feature contributions.\u003c/p\u003e \u003cp\u003eThis investigation has several limitations that should be acknowledged. First, as a single-center retrospective study with a limited sample size, the findings require validation through multicenter prospective studies with larger cohorts. Second, the assessment of lifestyle factors relied on subjective measures, highlighting the need for more objective quantitative methods to establish standardized evaluation protocols. Third, the current randomization approach may introduce selection bias, suggesting that stratified sampling or other rigorous methods could enhance study quality. Finally, while ROC curve-derived thresholds were employed, these may not fully capture clinically relevant cutoffs and could oversimplify data complexity, indicating the need to explore alternative analytical approaches.\u003c/p\u003e"},{"header":"CONCLUSIONS","content":"\u003cp\u003eA novel machine learning-based predictive model was developed, with the logistic regression algorithm demonstrating optimal performance upon comparative validation. By incorporating SHAP interpretability analysis, this model enables personalized risk assessment for LVI in SEC, addressing a critical limitation of existing predictive approaches. The model provides objective decision-support for clinicians to accurately identify high-risk patients, refine treatment strategies, and reduce unnecessary medical resource utilization from excessive interventions.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003elymphovascular invasion (LVI)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003esuperficial esophageal carcinoma (SEC)\u003c/p\u003e\n\u003cp\u003eShapley Additive Explanations (SHAP)\u003c/p\u003e\n\u003cp\u003ebody mass index (BMI)\u003c/p\u003e\n\u003cp\u003echronic obstructive pulmonary disease (COPD)\u003c/p\u003e\n\u003cp\u003eneutrophil-to-lymphocyte ratio (NLR)\u003c/p\u003e\n\u003cp\u003epreoperative platelet-lymphocyte ratio (PLR)\u003c/p\u003e\n\u003cp\u003epreoperative platelet-neutrophil ratio (PNR)\u003c/p\u003e\n\u003cp\u003eendoscopic or magnification (EOM)\u003c/p\u003e\n\u003cp\u003ecircumferential involvement proportion(CIP)\u003c/p\u003e\n\u003cp\u003ereceiver operating characteristic (ROC)\u003c/p\u003e\n\u003cp\u003edecision curve analysis (DCA)\u003c/p\u003e\n\u003cp\u003eprecision-recall (PR)\u003c/p\u003e\n\u003cp\u003ehigh-grade intraepithelial neoplasia(HGIN)\u003c/p\u003e\n\u003cp\u003elow-grade ntraepithelial neoplasia(LGIN)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eendoscopic submucosal dissection (ESD)\u003c/p\u003e\n\u003cp\u003eeXtreme Gradient Boosting (XGBoost)\u003c/p\u003e\n\u003cp\u003eLight Gradient Boosting Machine (LightGBM)\u003c/p\u003e\n\u003cp\u003eAdaptive Boosting (AdaBoost)\u003c/p\u003e\n\u003cp\u003eGradient Boosting Decision Tree (GBDT)\u003c/p\u003e\n\u003cp\u003eGaussian Na\u0026iuml;ve Bayes (GNB)\u003c/p\u003e\n\u003cp\u003eComplement Na\u0026iuml;ve Bayes (CNB)\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe present study was performed at The Affiliated Hospital of North Sichuan Medical College under the Declaration of Helsinki.This retrospective analysis was approved by the The Affiliated Hospital of North Sichuan Medical College\u0026apos;s Ethics Committee, which waived the requirement for informed consent due to the study\u0026apos;s retrospective design.(ethical batch number:\u0026nbsp;2024ER143-1)\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\u003eData is provided within the manuscript or supplementary information files.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by the Nanchong Science and Technology Program ((grant number: 23JCYJPT0060) and the Key Project of Research and Development Program of Affiliated Hospital of North Sichuan Medical College ((grant number: 2023ZD009).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eXianfei Wang, Zichen Luo designed the study; Xianfei Wang Zichen Luo analyzed the data; Xianfei Wang, Zichen Luo drafted the article; Xianfei Wang, Zichen Luo critically revised the article; Xianfei Wang finally approved the article; Zichen Luo,\u0026nbsp;Xinrui Chen,\u0026nbsp;Yutong Cui,\u0026nbsp;Ji Zuo, Shiqi Liang,\u0026nbsp;Guangbing Hu, Chengyu Zhang, Haorui Li , Xuemei Hou provided the study materials or patients. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHOU H, MENG Z, ZHAO X, et al. Survival of Esophageal Cancer in China: A Pooled Analysis on Hospital-Based Studies From 2000 to 2018[J/OL]. Front Oncol. 2019;9:548. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2019.00548\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2019.00548\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSHAH M A, KENNEDY E B, CATENACCI D V, et al. Treatment of Locally Advanced Esophageal Carcinoma: ASCO Guideline[J/OL]. J Clin Oncol. 2020;38(23):2677\u0026ndash;94. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1200/JCO.20.00866\u003c/span\u003e\u003cspan address=\"10.1200/JCO.20.00866\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHEALTH COMMISSION OF THE PRC N, NATIONAL HEALTH COMMISSION OF THE PEOPLE\u0026rsquo;S REPUBLIC, OF CHINA. National guidelines for diagnosis and treatment of esophageal carcinoma 2022 in China (English version)[J/OL]. Chin J Cancer Res. 2022;34(3):309\u0026ndash;34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21147/j.issn.1000-9604.2022.04.01\u003c/span\u003e\u003cspan address=\"10.21147/j.issn.1000-9604.2022.04.01\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOYAMA T, INOUE H, ARIMA M, et al. Prediction of the invasion depth of superficial squamous cell carcinoma based on microvessel morphology: magnifying endoscopic classification of the Japan Esophageal Society[J/OL]. Esophagus. 2017;14(2):105\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10388-016-0527-7\u003c/span\u003e\u003cspan address=\"10.1007/s10388-016-0527-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWANG A, TAN Y, ZHANG Y, et al. The prognostic role of angiolymphatic invasion in N0 esophageal carcinoma: a meta-analysis and systematic review[J/OL]. J Thorac Disease. 2019;11(8):3276\u0026ndash;83. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/jtd.2019.08.50\u003c/span\u003e\u003cspan address=\"10.21037/jtd.2019.08.50\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWANG A, TAN Y, WANG S, et al. The prognostic value of separate lymphatic invasion and vascular invasion in oesophageal squamous cell carcinoma: a meta-analysis and systematic review[J/OL]. BMC Cancer. 2022;22(1):1329. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12885-022-10441-6\u003c/span\u003e\u003cspan address=\"10.1186/s12885-022-10441-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLI P, LING Y H, ZHU C M et al. Vascular invasion as an independent predictor of poor prognosis in nonmetastatic gastric cancer after curative resection[J].\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTAO Y, CHEN S, YU J, et al. Risk factors of lymph node metastasis or lymphovascular invasion for superficial esophageal squamous cell carcinoma: A practical and effective predictive nomogram based on a cancer hospital data[J/OL]. Front Med. 2022;9:1038097. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fmed.2022.1038097\u003c/span\u003e\u003cspan address=\"10.3389/fmed.2022.1038097\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKOOK MC. Risk Factors for Lymph Node Metastasis in Undifferentiated-Type Gastric Carcinoma[J/OL]. Clin Endoscopy. 2019;52(1):15\u0026ndash;20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5946/ce.2018.193\u003c/span\u003e\u003cspan address=\"10.5946/ce.2018.193\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYEH R JAYAPRAKASAMVS, KU G Y, et al. Role of Imaging in Esophageal Cancer Management in 2020: Update for Radiologists[J/OL]. Am J Roentgenol. 2020;215(5):1072\u0026ndash;84. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2214/AJR.20.22791\u003c/span\u003e\u003cspan address=\"10.2214/AJR.20.22791\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePARTICIPANTS IN THE PARIS WORKSHOP. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon[J/OL]. Gastrointest Endosc. 2003;58(6):S3\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0016-5107(03)02159-X\u003c/span\u003e\u003cspan address=\"10.1016/S0016-5107(03)02159-X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eORLOFF NC, FLAMMER A, HARTNETT J, et al. Food cravings in pregnancy: Preliminary evidence for a role in excess gestational weight gain[J/OL]. Appetite. 2016;105:259\u0026ndash;65. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.appet.2016.04.040\u003c/span\u003e\u003cspan address=\"10.1016/j.appet.2016.04.040\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKIKUCHI Y. Personality and Dietary Habits.[J/OL]. J Epidemiol. 2000;10(3):191\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2188/jea.10.191\u003c/span\u003e\u003cspan address=\"10.2188/jea.10.191\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWEST R. Tobacco smoking: Health impact, prevalence, correlates and interventions[J/OL]. Psychol Health. 2017;32(8):1018\u0026ndash;36. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/08870446.2017.1325890\u003c/span\u003e\u003cspan address=\"10.1080/08870446.2017.1325890\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGARNETT C, OLDHAM M. Prevalence and characteristics of co-occurrence of smoking and increasing-and-higher-risk drinking: A population survey in England[J/OL]. Addict Behav. 2024;150:107928. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.addbeh.2023.107928\u003c/span\u003e\u003cspan address=\"10.1016/j.addbeh.2023.107928\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBERKOVICH G Y, LEVINE M S, MILLER W T. CT Findings in Patients with Esophagitis[J/OL]. Am J Roentgenol. 2000;175(5):1431\u0026ndash;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2214/ajr.175.5.1751431\u003c/span\u003e\u003cspan address=\"10.2214/ajr.175.5.1751431\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJANG K M, LEE K S, LEE SJ, et al. The Spectrum of Benign Esophageal Lesions: Imaging Findings[J/OL]. Korean J Radiol. 2002;3(3):199. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3348/kjr.2002.3.3.199\u003c/span\u003e\u003cspan address=\"10.3348/kjr.2002.3.3.199\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHALLINAN J T P D, VENKATESH SK. Gastric carcinoma: imaging diagnosis, staging and assessment of treatment response[J/OL]. Cancer Imaging. 2013;13(2):212\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1102/1470-7330.2013.0023\u003c/span\u003e\u003cspan address=\"10.1102/1470-7330.2013.0023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYANG Q, LIU Z. A narrative review: narrow-band imaging endoscopic classifications[J/OL]. Quant Imaging Med Surg. 2023;13(2):1138\u0026ndash;63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/qims-22-728\u003c/span\u003e\u003cspan address=\"10.21037/qims-22-728\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCHIAM K H, SHIN S H, CHOI K C, et al. Current Status of Mucosal Imaging with Narrow-Band Imaging in the Esophagus[J/OL]. Gut Liver. 2021;15(4):492\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5009/gnl20031\u003c/span\u003e\u003cspan address=\"10.5009/gnl20031\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKURIBAYASHI S, HOSAKA H. Usefulness of Endoscopy for the Detection and Diagnosis of Primary Esophageal Motility Disorders and Diseases Relating to Abnormal Esophageal Motility[J/OL]. Diagnostics. 2023;13(4):695. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/diagnostics13040695\u003c/span\u003e\u003cspan address=\"10.3390/diagnostics13040695\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eANDRIOPOULOS V. LASSO Regression with Multiple Imputations for the Selection of Key Variables Affecting the Fatty Acid Profile of Nannochloropsis oculata[J/OL]. Mar Drugs. 2023;21(9):483. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/md21090483\u003c/span\u003e\u003cspan address=\"10.3390/md21090483\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDEMEESTER SR. Evaluation and Treatment of Superficial Esophageal Cancer[J/OL]. J Gastrointest Surg. 2010;14:S94\u0026ndash;100. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11605-009-1025-1\u003c/span\u003e\u003cspan address=\"10.1007/s11605-009-1025-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKITAGAWA Y, ISHIHARA R, ISHIKAWA H, et al. Esophageal cancer practice guidelines 2022 edited by the Japan esophageal society: part 1[J/OL]. Esophagus. 2023;20(3):343\u0026ndash;72. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10388-023-00993-2\u003c/span\u003e\u003cspan address=\"10.1007/s10388-023-00993-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZHANG B, DU W. Significance of the neutrophil-to-lymphocyte ratio in young patients with oral squamous cell carcinoma[J/OL]. Cancer Manage Res. 2019;11:7597\u0026ndash;603. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/CMAR.S211847\u003c/span\u003e\u003cspan address=\"10.2147/CMAR.S211847\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBUONACERA A, STANCANELLI B, COLACI M, et al. Neutrophil to Lymphocyte Ratio: An Emerging Marker of the Relationships between the Immune System and Diseases[J/OL]. Int J Mol Sci. 2022;23(7):3636. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms23073636\u003c/span\u003e\u003cspan address=\"10.3390/ijms23073636\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHESHMAT-GHAHDARIJANI K, SARMADI V, HEIDARI A, et al. The neutrophil-to-lymphocyte ratio as a new prognostic factor in cancers: a narrative review[J/OL]. Front Oncol. 2023;13:1228076. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2023.1228076\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2023.1228076\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMIN K W, KWON M J, KIM D H, et al. Persistent elevation of postoperative neutrophil-to-lymphocyte ratio: A better predictor of survival in gastric cancer than elevated preoperative neutrophil-to-lymphocyte ratio[J/OL]. Sci Rep. 2017;7(1):13967. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-017-13969-x\u003c/span\u003e\u003cspan address=\"10.1038/s41598-017-13969-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRUAN R, CHEN S, TAO Y, et al. A Nomogram for Predicting Lymphovascular Invasion in Superficial Esophageal Squamous Cell Carcinoma[J/OL]. Front Oncol. 2021;11:663802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2021.663802\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2021.663802\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLI Y, YU M, WANG G, et al. Corrigendum: Contrast-Enhanced CT-Based Radiomics Analysis in Predicting Lymphovascular Invasion in Esophageal Squamous Cell Carcinoma[J/OL]. Front Oncol. 2021;11:712493. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2021.712493\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2021.712493\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLI Y, SU H. Can lymphovascular invasion be predicted by contrast-enhanced CT imaging features in patients with esophageal squamous cell carcinoma? A preliminary retrospective study[J/OL]. BMC Med Imaging. 2022;22(1):93. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12880-022-00804-7\u003c/span\u003e\u003cspan address=\"10.1186/s12880-022-00804-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWANG Y, BAI G, HUANG W, et al. A radiomics nomogram based on contrast-enhanced CT for preoperative prediction of Lymphovascular invasion in esophageal squamous cell carcinoma[J/OL]. Front Oncol. 2023;13:1208756. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2023.1208756\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2023.1208756\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMIN S K, LEE S K, WOO J, et al. Relation Between Tumor Size and Lymph Node Metastasis According to Subtypes of Breast Cancer[J/OL]. J Breast Cancer. 2021;24(1):75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4048/jbc.2021.24.e4\u003c/span\u003e\u003cspan address=\"10.4048/jbc.2021.24.e4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHUANG S, ZHU Y, CAI H, et al. Impact of lymphovascular invasion in oral squamous cell carcinoma: A meta-analysis[J/OL]. Oral Surgery, Oral Medicine. Oral Pathol Oral Radiol. 2021;131(3):319\u0026ndash;e3281. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.oooo.2020.10.026\u003c/span\u003e\u003cspan address=\"10.1016/j.oooo.2020.10.026\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWANG Y, ZHU L, XIA W, et al. Anatomy of lymphatic drainage of the esophagus and lymph node metastasis of thoracic esophageal cancer[J/OL]. Cancer Manage Res. 2018;10:6295\u0026ndash;303. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/CMAR.S182436\u003c/span\u003e\u003cspan address=\"10.2147/CMAR.S182436\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVOGT CD, PANOSKALTSIS-MORTARI A. Tissue engineering of the gastroesophageal junction[J/OL]. J Tissue Eng Regen Med. 2020;14(6):855\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/term.3045\u003c/span\u003e\u003cspan address=\"10.1002/term.3045\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNILAND S, RISCANEVO A X, EBLE JA. Matrix Metalloproteinases Shape the Tumor Microenvironment in Cancer Progression[J/OL]. Int J Mol Sci. 2021;23(1):146. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms23010146\u003c/span\u003e\u003cspan address=\"10.3390/ijms23010146\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYE X Y, LAI Y T, SONG W P, et al. The research progress on the association between dietary habits and esophageal cancer: a narrative review[J/OL]. Annals Palliat Med. 2021;10(6):6948\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/apm-21-1467\u003c/span\u003e\u003cspan address=\"10.21037/apm-21-1467\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSALIHI A, AL\u0026ndash;NAQSHABANDI M, KHUDHUR Z, et al. Gasotransmitters in the tumor microenvironment: Impacts on cancer chemotherapy (Review)[J/OL]. Mol Med Rep. 2022;26(1):233. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3892/mmr.2022.12749\u003c/span\u003e\u003cspan address=\"10.3892/mmr.2022.12749\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLI S, HOEFNAGEL S J M, KRISHNADATH KK. Molecular Biology and Clinical Management of Esophageal Adenocarcinoma[J/OL]. Cancers. 2023;15(22):5410. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cancers15225410\u003c/span\u003e\u003cspan address=\"10.3390/cancers15225410\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLIN S H, LI Y H, LEUNG K, et al. Salt Processed Food and Gastric Cancer in a Chinese Population[J/OL]. Asian Pac J Cancer Prev. 2014;15(13):5293\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7314/APJCP.2014.15.13.5293\u003c/span\u003e\u003cspan address=\"10.7314/APJCP.2014.15.13.5293\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBOURAS E, TSILIDIS K K, TRIGGI M, et al. Diet and Risk of Gastric Cancer. Umbrella Review[J/OL] Nutrients. 2022;14(9):1764. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu14091764\u003c/span\u003e\u003cspan address=\"10.3390/nu14091764\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHSIEH H L, TSAI MM. Tumor progression-dependent angiogenesis in gastric cancer and its potential application[J/OL]. World J Gastrointest Oncol. 2019;11(9):686\u0026ndash;704. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4251/wjgo.v11.i9.686\u003c/span\u003e\u003cspan address=\"10.4251/wjgo.v11.i9.686\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMUTHUKUMARAN RB, BHATTACHARJEE P, BHOWMICK P, et al. Genetic and epigenetic instability induced by betel quid associated chemicals[J/OL]. Toxicol Rep. 2023;10:223\u0026ndash;34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.toxrep.2023.02.001\u003c/span\u003e\u003cspan address=\"10.1016/j.toxrep.2023.02.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDODD L E SENGUPTAS, CHEN I H, et al. Genes Involved in DNA Repair and Nitrosamine Metabolism and Those Located on Chromosome 14q32 Are Dysregulated in Nasopharyngeal Carcinoma[J/OL]. Cancer Epidemiol Biomarkers Prev. 2006;15(11):2216\u0026ndash;25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1158/1055-9965.EPI-06-0455\u003c/span\u003e\u003cspan address=\"10.1158/1055-9965.EPI-06-0455\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHUA H, LI M, LUO T, et al. Matrix metalloproteinases in tumorigenesis: an evolving paradigm[J/OL]. Cell Mol Life Sci. 2011;68(23):3853\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00018-011-0763-x\u003c/span\u003e\u003cspan address=\"10.1007/s00018-011-0763-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSARKAR M, NGUYEN T. Cancer-associated fibroblasts: The chief architect in the tumor microenvironment[J/OL]. Front Cell Dev Biology. 2023;11:1089068. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fcell.2023.1089068\u003c/span\u003e\u003cspan address=\"10.3389/fcell.2023.1089068\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eERD\u0026Eacute;LYI A, P\u0026Aacute;LFI E. The Importance of Nutrition in Menopause and Perimenopause\u0026mdash;A. Review[J/OL] Nutrients. 2023;16(1):27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu16010027\u003c/span\u003e\u003cspan address=\"10.3390/nu16010027\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePANEL ON DIETARY ANTIOXIDANTS AND RELATED COMPOUNDS, SUBCOMMITTEE ON UPPER REFERENCE LEVELS OF NUTRIENTS, SUBCOMMITTEE ON INTERPRETATION AND USES OF DIETARY REFERENCE, INTAKES, Vitamin E, Selenium, Carotenoids [M/OL] et al. Washington, D.C.: National Academies, 2000[2025-05-01]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nap.edu/catalog/9810\u003c/span\u003e\u003cspan address=\"https://www.nap.edu/catalog/9810\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.17226/9810\u003c/span\u003e\u003cspan address=\"10.17226/9810\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNIU J, LI B, ZHANG Q, et al. Exploring the traditional Chinese diet and its association with health status\u0026mdash;a systematic review[J/OL]. Nutr Rev. 2025;83(2):e237\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nutrit/nuae013\u003c/span\u003e\u003cspan address=\"10.1093/nutrit/nuae013\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHATTA W, KOIKE T, UNO K, et al. Management of Superficial Esophageal Squamous Cell Carcinoma and Early Gastric Cancer following Non-Curative Endoscopic Resection[J/OL]. Cancers. 2022;14(15):3757. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cancers14153757\u003c/span\u003e\u003cspan address=\"10.3390/cancers14153757\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\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":"superficial esophageal carcinoma, lymphovascular invasion, LASSO regression, logistic regression, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-6591573/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6591573/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eLymphovascular invasion (LVI) represents a critical prognostic determinant in superficial esophageal carcinoma (SEC), significantly influencing therapeutic decision-making and clinical outcomes. Despite its clinical importance, reliable predictive tools for early LVI detection remain unavailable. The current study was designed to develop and validate a machine learning-based predictive model for accurate LVI risk stratification in SEC patients.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003ePredictive factor selection was conducted using least absolute shrinkage and selection operator (LASSO) regression followed by multivariable logistic regression analysis. Multiple machine learning algorithms were systematically evaluated, with model performance quantified through receiver operating characteristic (ROC) curve analysis. Model interpretability was enhanced through implementation of Shapley Additive Explanations (SHAP) methodology.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eEight independent predictors of LVI were identified: neutrophil-to-lymphocyte ratio (NLR), esophageal wall thickness on computed tomography (CT), endoscopic ultrasound or magnifying endoscopy (EOM) findings, tumor diameter, multiple lesions, circumferential involvement proportion (CIP), consumption of pickled food and preoperative biopsy results. The logistic regression model demonstrated superior predictive performance, with area under the curve (AUC) values of 0.871 (training cohort), 0.852 (validation cohort), and 0.902 (test cohort).\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThe developed SHAP-interpretable logistic regression model provides an effective tool for early LVI detection in SEC, enabling personalized risk assessment and optimized clinical management strategies. This approach may significantly improve treatment decision-making for SEC patients.\u003c/p\u003e","manuscriptTitle":"Machine Learning-Based Prediction of Lymphovascular Invasion in Superficial Esophageal Carcinoma: Model Development and Risk Factor Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-13 10:04:05","doi":"10.21203/rs.3.rs-6591573/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":"185bde5c-1f64-4e5a-af58-5672e2d1dd3d","owner":[],"postedDate":"May 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-11-26T08:53:57+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-13 10:04:05","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6591573","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6591573","identity":"rs-6591573","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}