Unveiling Lymph Node Metastasis (LNM) in pleural-attached lung mucinous adenocarcinoma: A Predictive Model Using Pleural Contact Parameters

Unveiling Lymph Node Metastasis (LNM) in pleural-attached lung mucinous adenocarcinoma: A Predictive Model Using Pleural Contact Parameters | 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 Unveiling Lymph Node Metastasis (LNM) in pleural-attached lung mucinous adenocarcinoma: A Predictive Model Using Pleural Contact Parameters Junjie Zhang, Ligang Hao, Zhenbin Qiu, Fengxiao Gao, Wenzhao Zhong, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6676702/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 Objective: This study seeks to evaluate the prognostic significance of pleural-attached parameters and computed tomography (CT) imaging characteristics in predicting lymph node metastasis (LNM) in cases of pleural-attached invasive mucinous adenocarcinoma (IMA). Methodology: A retrospective analysis was conducted on a cohort of 276 IMA patients from three public tertiary hospitals in China, covering the period from January 2017 to December 2022. A comparative analysis was performed on pleural-attached parameters and CT imaging characteristics between different patient groups. Based on variables that showed statistical significance, nine machine learning models were developed, and the model with the highest area under the curve (AUC) value was identified as the optimal model. Patients were randomly allocated into training and testing groups in a 7:3 ratio. The 5-fold cross-validation technique was employed to evaluate the receiver operating characteristic (ROC) curve AUC value of the most effective machine learning model. Results: Both univariate and multivariate logistic regression analyses identified pleural contact length, pleural contact surface area (CSA), the skirt-like sign, cavity sign, angiogram sign, and lymph node enlargement as significant predictive factors for lymph node metastasis (LNM) in pleural-attached invasive mucinous adenocarcinoma (IMA). The logistic regression model demonstrated superior performance, achieving ROC-AUC values of 0.856 and 0.803 in the training and test groups, respectively. Further analysis of the model in patients without lymph node enlargement indicated that it maintained superior performance, with AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort, and accuracy rates of 0.826 and 0.838, respectively. Conclusion: The logistic regression model, which incorporates pleural contact parameters and imaging characteristics, demonstrated substantial diagnostic value for assessing LNM in pleural-attached IMA, particularly in patients without lymph node enlargement. It exhibited excellent diagnostic efficacy and provides a non-invasive evaluation method for clinical practice. Lymph node metastasis (LNM) lung invasive mucinous adenocarcinoma (IMA) pleural contact computed tomography (CT) machine learning model Figures Figure 1 Figure 2 Figure 3 Introduction In the 2011 classification by the International Association for the Study of Lung Cancer (IASLC), the American Thoracic Society (ATS), and the European Respiratory Society (ERS), as well as the 2015 World Health Organization (WHO) publication, primary pulmonary invasive mucinous adenocarcinoma (IMA) is identified as a variant subtype of lung adenocarcinoma, comprising 2–5% of adenocarcinoma cases[ 1 , 2 ]. The clinical, imaging, pathological, and genetic characteristics of IMA are distinct from those of non-mucinous adenocarcinoma. According to the literature, only less than 10% of IMA patients exhibit lymph node metastasis (LNM), while more than 90% patients are classified within the nodal stage N0 group[ 3 – 5 ]. Nevertheless, the presence of LNM significantly impacts the prognosis of IMA patients. Therefore, understanding and predicting LNM in IMA is crucial for improving patient outcomes. Research suggests that IMA frequently occurs in a subpleural location, which increases its potential to disseminate through the pleural lymphatic network[ 6 , 7 ]. Tumors located beneath the pleura, due to their direct contact with this membrane, may elevate the risk of metastasis via pleural lymphatics, consequently leading to an increased incidence of LNM. The lymphatic drainage pathways of the pleura are pivotal in the propagation of lung cancer. Research indicates that the lymphatic drainage of lung cancer extends beyond the routes adjacent to the bronchi to include those within the pleura[ 8 ]. This drainage pathway is frequently segmental, especially when the tumor is near the pleura. Such segmental lymphatic drainage may increase the likelihood of tumor cell metastasis via the lymphatic system, thereby affecting patient prognosis[ 9 ]. IMA, characterized by its mucin secretion, may promote the dissemination of tumor cells along the pleural surface, thus elevating the risk of LNM. The relationship between pleural involvement and LNM has been thoroughly investigated in numerous studies. A systematic review and meta-analysis have identified several prognostic factors for IMA, notably pleural metastasis and lymph node involvement, which are significant risk factors for poor prognosis[ 10 , 11 ]. Furthermore, it was recommended that the presence of LNM should lead to the upstaging of small tumors (≤ 3 cm), potentially impacting surgical planning. Accurate preoperative assessment of LNM is essential for patients with clinically early-stage subpleural lung cancers. In this context, visceral pleural invasion (VPI) has been identified using immunocytochemistry on paraffin-embedded sections rather than through intraoperative frozen section analysis. We hypothesize that pleural-contacted IMA may similarly serve as a predictor of LNM. Currently, no effective method exists to predict LNM in patients with subpleural IMA. Therefore, this study aims to investigate the correlation between pleural-contacted parameters and lymph node metastasis in cases of pleural-contacted IMA. Materials and methods This study was conducted in accordance with the Helsinki Declaration and received approval from the Ethics Committee of our hospital (Ethics Committee of our hospitals, reference number: 2023【124】, dated 2023.11.15). For this retrospective study, informed consent was not required. Patient selection This study complies with the STROBE guidelines. We performed a retrospective review and analysis of all patients diagnosed with peripheral IMA at three public tertiary hospitals in our country: The Fourth Hospital of Hebei Medical University, XingTai People's Hospital, and Guangdong Province People's Hospital, spanning the period from January 2017 to December 2022. The inclusion criteria for this study were as follows: (1) surgical pathology-confirmed IMA, encompassing both pure invasive mucinous adenocarcinoma and mixed mucinous and non-mucinous adenocarcinoma, with systematic lymph node dissection conducted; (2) availability of computed tomography (CT) images in thin sections of 1.25 mm or less, demonstrating lung cancer as a single lesion; and (3) presence of solid nodules directly adjacent to the costal, mediastinal, or diaphragmatic pleura. The exclusion criteria included: (1) receipt of antitumor therapy prior to CT examination and pathological diagnosis; (2) presence of other types of cancer or incomplete clinical and imaging data, and (3) the availability of CT images obtained more than two weeks before the pathological diagnosis. This retrospective analysis received approval from our hospital's ethical review board, which also granted a waiver for the requirement of informed consent. Chest CT Imaging and feature extraction The CT scans were re-evaluated to confirm the adjacency of the IMA to the pleura, utilizing lung window settings on Picture Archiving and Communication System (PACS) workstations. The examinations encompassed the entire thorax with patients positioned supinely and at full inspiration. All images were reconstructed in transverse, sagittal, and coronal planes, with section thicknesses ranging from 1 to 1.25 mm, and lung window settings set at 1500 Hounsfield Units (HU) width and -600 HU level. A radiologist with extensive experience in chest CT imaging (Z.JJ., 14 years) independently assessed all nodules, without access to clinical data or pathology findings, to evaluate specific CT characteristics. These characteristics included: (1) size, contact length, depth, and contact surface area. Tumor size was defined as the longest diameter measured on axial, sagittal, or coronal lung window images. The tumor-pleura contact length was determined as the maximum contact length measured on axial, sagittal, or coronal mediastinum window images. The tumor-pleura distance was defined as the shortest distance measured on axial, sagittal, or coronal mediastinum window images. For nodules with multiple pleural connections, the largest contact surface area (CSA) was recorded. Given the inherent convexity at the junction of lung nodules and the pleura, and the lack of a universally accepted method for CSA measurement, the method developed by Qi et al was utilized to evaluate the interface between lung nodules and the pleura[12]. The CT indicators assessed comprised pleural thickening, the skirt-like sign, and the jellyfish sign. The skirt-like sign was identified by the thickening of the peripheral pleura on both sides of the contact surface for nodules adherent to the pleura. The jellyfish sign was characterized in sections adjacent to the area of maximum contact surface, where multiple linear septations between the nodule and the costal pleura resembled jellyfish tentacles. Intra- and Interreader Agreement on Nodule Features A radiologist (Z.JJ.), who was blinded to the initial classifications, conducted an analysis of the CT features of each pleural-attached nodule during two separate sessions spaced three months apart. Subsequently, following a training session led by the first radiologist (Z.JJ.), a second radiologist (G.FX.), possessing 31 years of experience in chest CT imaging, independently evaluated the CT scans of the nodules. This evaluation was performed without access to any clinicopathologic information to assess the CT characteristics. Any disagreements between the two radiologists were collectively reviewed and resolved through mutual consensus. In instances where consensus could not be achieved, a final decision was rendered by an additional radiologist (X.Q.), who has over 30 years of experience in chest CT imaging. Intraobserver agreement was assessed, and interobserver agreement was determined by comparing the characteristics of the subpleural nodules from the second review conducted by reader 1 (Z.JJ.) with those assessed by reader 2 (G.FX.). Statistical Analyses and Model Development Comparative and descriptive analyses were undertaken to examine the characteristics of patients with and without LNM. All statistical analyses were executed utilizing Python version 3.7. For continuous variables, the Student's t-test or the Wilcoxon rank-sum test was applied, whereas categorical variables were evaluated using the Pearson χ² test or Fisher's exact test. Independent CT features associated with LNM were identified through both univariable and multivariable logistic regression analyses. A p-value of less than 0.05 was deemed indicative of statistical significance. Furthermore, machine learning techniques were employed in this study to refine the construction of factors that exhibited statistical significance in the multivariate analysis. The methodology comprised several key steps: (1) Participants were randomly assigned to a training set and a test set in a 7:3 ratio. (2) Utilizing statistically significant factors identified through multivariate analysis, nine machine learning models were developed within the training set. These models included the Extreme Gradient Boosting (XGB) classifier, Light Gradient Boosting Machine (LGBM) classifier, Random Forest classifier, AdaBoost classifier, Gaussian Naive Bayes (GNB), Logistic Regression, Multilayer Perceptron (MLP) classifier, Polynomial Support Vector Machine (SVC), and k-Nearest Neighbor (KNeighbors) classifier. Optimal parameters for these models were retrospectively determined using 5-fold cross-validation. The performance of the nine machine learning models was assessed using the ROC curve. A 5-fold cross-validation was conducted to validate the most effective model. The primary evaluation metrics included the AUC, accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. Based on the ROC analysis, the most predictive machine learning classifier was selected. Results Patient characteristics In this study, a cohort of 276 patients diagnosed with IMA was recruited from three public tertiary hospitals within our country. Specifically, 188 patients were sourced from The Fourth Hospital of Hebei Medical University, 17 from XingTai People's Hospital, and 71 from Guangdong Province People's Hospital. The cohort comprised 27 patients with LNM and 249 without LNM, analyzed retrospectively. The clinical characteristics, including age, smoking habits, and sex, did not exhibit significant differences between the LNM and non-LNM groups. However, notable differences were observed between these groups concerning the presence of the Jellyfish sign (29.63% vs. 12.00%, P=0.019), the skirt-like sign (22.22% vs. 8.43%, P=0.035), spiculation (70.37% vs. 37.35%, P=0.002), cavity (22.22% vs. 54.62%, P=0.003), necrosis (22.22% vs. 54.62%, P=0.003), lymph node enlargement (2.41% vs. 22.22%, P<0.001), VPI (18.75% vs. 6.76%), and angiogram findings (40.74% vs. 22.22%, P=0.004). Additionally, the pleura contact length and CSA were significantly greater in patients with LNM compared to those without LNM (1.370 mm vs. 1.950 cm, P=0.036, and 0.6838 cm² vs. 1.665 cm², P<0.001, respectively).The detailed characteristics of patients in the training and testing cohorts are presented in Table 1. Feature selection and model construction The multivariate analysis identified significant differences between groups with and without LNM concerning CSA, contact length, lymph node enlargement, angiogram sign, cavities or vacuoles, and skirt sign, as detailed in Table 2. In the training dataset, nine machine learning prediction models were developed, including XGBoost (XGB), LightGBM (LGBM), Random Forest (RF), AdaBoost Classifier, Gaussian Naive Bayes (GNB), Logistic Regression (LR), Multi-Layer Perceptron (MLP), Decision Tree, and Gradient Boosting Decision Tree (GBDT), utilizing three selected radiomics features. The optimal parameters for these models were retrospectively determined using 5-fold cross-validation. Among these models, the Logistic Regression classifier demonstrated superior performance, achieving AUC values of 0.896 on the training dataset and 0.874 on the validation dataset, as detailed in Tables 3 and 4, and Figure 1. Furthermore, the AUC values for the Logistic Regression model in predicting MLP were 0.856 in the entire training cohort and 0.803 in the test cohort, with accuracies of 0.77 and 0.86, respectively, and the optimal cut-off value was 0.083. Additional information is presented in Figure 2. The AUC value of the logistic regression model was significantly higher than that for lymph node enlargement, which was 0.599. A subsequent analysis of the model in patients without lymph node enlargement demonstrated superior performance, with AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort. The accuracy rates were 0.826 and 0.838, respectively, Figure3. Discussion This study is the first to systematically evaluate the correlation between pleural-contacted parameters, imaging features, and LNM in pleural-contacted IMA. It identifies pleural contact length, CSA, the skirt-like sign, and the jellyfish sign as independent predictors of LNM. The prediction model developed from these six features offers a novel approach for preoperative non-invasive assessment of metastasis risk in patients with pleural-contacted IMA.. Unlike previous research that focused on qualitative assessments of pleural indentation or invasion[ 13 ], this study quantifies the predictive efficacy of pleural contact length and area, demonstrating their significant association with LNM. The findings suggest that pleural contact area may indicate the invasive biological behavior of tumors, with a larger contact area potentially reflecting a greater likelihood of tumor cells infiltrating the lymphatic network beneath the pleura, thus promoting lymphatic metastasis. Moreover, the incorporation of quantitative parameters can reduce subjective judgment errors and improve clinical reproducibility. Existing literature has demonstrated a correlation between VPI and LNM especially N2 lymph node in lung adenocarcinoma[ 14 – 16 ]. In the current study, it was found that 18.75% of patients with VPI exhibited LNM, a rate significantly higher than the 6.76% observed in patients without VPI, aligning with previous research findings in NSCLC[ 17 ]. The presence of VPI facilitates the migration of tumor cells to lymph nodes through lymphatic drainage, owing to the extensive lymphatic network of the visceral pleura that drains into the mediastinum[ 18 ]. Prior research has established a positive correlation between the length of pleural contact, CSA, and VPI in lung cancer[ 19 ]. As well as between VPI and LNM, our findings further substantiate the positive association between pleural contact length, CSA, and LNM in cases of IMA with pleural contact. And further large-scale studies are necessary to investigate these phenomenons in patients with IMA. The skirt sign may serve as an independent predictive factor linked to tumor stromal response and lymphatic invasion. Pathological investigations have demonstrated that IMA often exhibits significant connective tissue proliferation reactions. The skirt sign may correspond to active fibroblast proliferation and lymphatic vessel dilation at the tumor-stroma interface, potentially indicating tumor progression along the subpleural lymphatic vessels[ 12 , 17 ]. The presence of this phenomenon may act as an imaging marker indicative of the tumor microenvironment's pro-metastatic characteristics. Previous studies have identified the skirt sign in pleural attachment nodules as a valuable CT predictor of VPI in NSCLC with solid pleural attachments measuring 30 mm or smaller[ 12 ]. Our current research further corroborates that the skirt sign is an independent predictor of LNM in IMA. The presence of angiogram signs and cavities or vacuoles also demonstrates a significant association with LNM, although the underlying mechanisms require further investigation in future studies. Furthermore, the development of the Logistic Regression Classifier, which incorporates six distinct characteristics, represents a significant advancement in predictive modeling compared to the singular feature of lymph node enlargement. The prediction model achieved AUC values of 0.856 and 0.803 in the training and testing cohorts, respectively. Notably, in patients without lymph node enlargement, the model demonstrated superior performance, achieving AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort, with accuracy rates of 0.826 and 0.838, respectively. Through rigorous validation techniques, including ROC curve analysis, we have shown that our model can effectively stratify patients based on their risk of LNM, thereby informing treatment strategies. The implications of these findings are substantial, as they suggest that our predictive model could enhance clinical decision-making, leading to personalized surgical interventions and potentially improved prognostic outcomes for patients diagnosed with IMA. This study is subject to several limitations that must be acknowledged. Firstly, it was conducted as a retrospective analysis with a relatively small sample size. Despite utilizing data from three tertiary hospitals, the cohort included only 28 cases with positive lymph node metastasis, which limited the ability to divide the data into independent external validation groups. Secondly, the study's focus on patients with pleural contact and post-surgical pathological outcomes may introduce selection bias, necessitating cautious interpretation of the findings. Thirdly, the study primarily concentrated on nodule characteristics in the evaluation of CT images. Future research should consider the assessment of lymph nodes to provide a more comprehensive analysis. Furthermore, the models developed in this study underwent only preliminary evaluation and validation; their performance is anticipated to improve and undergo external validation in future research endeavors. Advancements in radiomics have enabled the transformation of imaging data into quantitative features, which may facilitate a more objective and quantitative characterization of tumors. As a result, it is expected that future research will investigate prognostic predictions and survival analyses for patients with LNM through the application of radiomics. In conclusion, this study introduces the development of an innovative model aimed at evaluating the correlation between pleural-contacted parameters, imaging characteristics, and LNM in instances of pleural-contacted IMA. The imaging predictor model exhibited significant diagnostic efficacy for LNM in pleural-contacted IMA, particularly in patients without LNM, indicating that CT semantic features can enhance preoperative clinical practice and possess considerable potential for clinical application. Abbreviations VPI: Visceral Pleural Invasion LNM: Lymph Node Metastasis NSCLC: non–small cell lung cancer CT: computed tomography ROC: receiver operating characteristic curve IASLC: International Association for the Study of Lung Cancer IMA: Invasive mucinous adenocarcinoma PIMA: pure Invasive mucinous adenocarcinoma TNM: Tumor, Node, Metastasis CSA: contact surface area GLCM: gray level co-occurrence matrix GLRLM: gray level run length matrix GLSZM: gray level size zone matrix NGTDM: neighboring gray tone difference matrix GLDM: gray level dependence matrix XGB: EXtreme Gradient Boosting Classifier LGBM: Light Gradient Boosting Machine Classifier, RF: RandomForest Classifier GNB: Gaussian naive bayes LR: Logistic Regression MLP: Multilayer Perceptron Classifier, GBDT: Gradient Boosting Decision Tree Declarations Ethics approval and consent to participate The Xing Tai People's Hospital ethical review board approved this retrospective analysis and waived informed consent requirements. Ethics Committee of Xing Tai People’s Hospital, reference number: 2023【124】, dated 2023.11.15. Consent for publication We confirm that the manuscript has been submitted solely to this journal and is not published, in press, or submitted elsewhere. Written informed consent was obtained from the individuals for the publication of any potentially identifiable images or data included in this article. Availability of data and materials The datasets generated and analyzed during the current study are not publicly available because the dataset will be further studied to publish other works but are available from the corresponding author on reasonable request. Competing interests The authors have no conflicts of interest to declare. Funding Key development plan of Xingtai (2023ZC049) Authors' contributions JJ Z, LG H and ZB Q performed the experiments and wrote the manuscript. FX G was responsible for the data collection and analysis, Q X and WZ Z were responsible for designing the experiments. All authors read and approved the final version of this submitted manuscript. Acknowledgments None References Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger KR, Yatabe Y, Beer DG, Powell CA, Riely GJ, Van Schil PE et al : International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma . Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer 2011, 6 (2):244-285. Travis WD, Brambilla E, Nicholson AG, Yatabe Y, Austin JHM, Beasley MB, Chirieac LR, Dacic S, Duhig E, Flieder DB et al : The 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification . Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer 2015, 10 (9):1243-1260. Moon SW, Choi SY, Moon MH: Effect of invasive mucinous adenocarcinoma on lung cancer-specific survival after surgical resection: a population-based study . Journal of thoracic disease 2018, 10 (6):3595-3608. Lee HY, Cha MJ, Lee KS, Lee HY, Kwon OJ, Choi JY, Kim HK, Choi YS, Kim J, Shim YM: Prognosis in Resected Invasive Mucinous Adenocarcinomas of the Lung: Related Factors and Comparison with Resected Nonmucinous Adenocarcinomas . Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer 2016, 11 (7):1064-1073. Kadota K, Yeh YC, D'Angelo SP, Moreira AL, Kuk D, Sima CS, Riely GJ, Arcila ME, Kris MG, Rusch VW et al : Associations between mutations and histologic patterns of mucin in lung adenocarcinoma: invasive mucinous pattern and extracellular mucin are associated with KRAS mutation . The American journal of surgical pathology 2014, 38 (8):1118-1127. Zhang J, Hao L, Xu Q, Gao F: Radiomics and Clinical Characters Based Gaussian Naive Bayes (GNB) Model for Preoperative Differentiation of Pulmonary Pure Invasive Mucinous Adenocarcinoma From Mixed Mucinous Adenocarcinoma . Technology in cancer research & treatment 2024, 23 :15330338241258415. Zhang J, Hao L, Li M, Xu Q, Shi G: CT Radiomics Combined With Clinicopathological Features to Predict Invasive Mucinous Adenocarcinoma in Patients With Lung Adenocarcinoma . Technology in cancer research & treatment 2023, 22 :15330338231174306. Fukumura D, Kloepper J, Amoozgar Z, Duda DG, Jain RK: Enhancing cancer immunotherapy using antiangiogenics: opportunities and challenges . Nature reviews Clinical oncology 2018, 15 (5):325-340. Berzenji L, Van Schil PE, Carp L: The eighth TNM classification for malignant pleural mesothelioma . Translational lung cancer research 2018, 7 (5):543-549. Shen F, Wu X, Geng J, Guo W, Duan J: Prognostic factors for resected invasive mucinous lung adenocarcinoma: a systematic review and meta-analysis . BMC cancer 2024, 24 (1):1317. Zhao T, Yi J, Luo D, Liu J, Fan X, Wu Q, Wang W: Prognostic factors for invasive mucinous adenocarcinoma of the lung: systematic review and meta-analysis . World journal of surgical oncology 2024, 22 (1):41. Sun Q, Li P, Zhang J, Yip R, Zhu Y, Yankelevitz DF, Henschke CI: CT Predictors of Visceral Pleural Invasion in Patients with Non-Small Cell Lung Cancers 30 mm or Smaller . Radiology 2024, 310 (1):e231611. Zhang T, Zhang JT, Li WF, Lin JT, Liu SY, Yan HH, Yang JJ, Yang XN, Wu YL, Nie Q et al : Visceral pleural invasion in T1 tumors ( ≤3 cm), particularly T1a, in the eighth tumor-node-metastasis classification system for non-small cell lung cancer: a population-based study . Journal of thoracic disease 2019, 11 (7):2754-2762. Liu J, Li J, Lin G, Long Z, Li Q, Liu B: Risk factors of lobar lymph node metastases in non-primary tumor-bearing lobes among the patients of non-small-cell lung cancer . PloS one 2020, 15 (9):e0239281. Zhang H, Lu C, Lu Y, Yu B, Lv F, Zhu Z: The predictive and prognostic values of factors associated with visceral pleural involvement in resected lung adenocarcinomas . OncoTargets and therapy 2016, 9 :2337-2348. Shimizu K, Yoshida J, Nagai K, Nishimura M, Ishii G, Morishita Y, Nishiwaki Y: Visceral pleural invasion is an invasive and aggressive indicator of n on-small cell lung cancer . J Thorac Cardiovasc Surg , 130 (1):160-165. Zhang C, Wang L, Cai X, Li M, Sun D, Wang P: Tumour-pleura relationship on CT is a risk factor for occult lymph node metastasis in peripheral clinical stage IA solid adenocarcinoma . European radiology 2023, 33 (5):3083-3091. Lee E, Biko DM, Sherk W, Masch WR, Ladino-Torres M, Agarwal PP: Understanding Lymphatic Anatomy and Abnormalities at Imaging . Radiographics : a review publication of the Radiological Society of North America, Inc 2022, 42 (2):487-505. Minamoto F, Araújo P, D'Ambrosio P, Dela Vega A, Lauricella L, Pêgo-Fernandes P, Terra R: The association of visceral pleural invasion with skip N2 metastasis on clinical stage IA NSCLC . Clinics (Sao Paulo, Brazil) 2024, 79 :100334. Tables Table 1 Clinical characteristics of the patients in training group and a test group. Characters Training cohort Testing cohort LN(-) LN(+) P LN(-) LN(+) P Gender Female 79 11 0.913 28 2 1.000 Male 92 11 50 3 Jellyfish No 150 15 0.023 69 4 0.483 Yes 21 7 9 1 Skirt No 157 16 0.015 71 5 1.000 Yes 14 6 7 0 Smoking No 121 16 1.000 57 4 1.000 Yes 50 6 21 1 Lobul No 24 1 0.319 9 0 1.000 Yes 147 21 69 5 Spicul No 110 6 0.002 46 2 0.646 Yes 61 16 32 3 Cavity No 75 16 0.020 38 5 0.056 Yes 96 6 40 0 Necrosis No 160 18 0.032 73 4 0.320 Yes 11 4 5 1 AS No 149 13 0.003 60 3 0.590 Yes 22 9 18 2 LNE No 168 16 <0.001 75 5 1.000 Yes 3 6 3 0 Age 62.00 60.00 0.104 63.000 65.000 0.328 D-max 2.250 3.150 0.001 2.450 3.000 0.320 CSA median 0.654 1.538 <0.001 0.756 1.849 0.011 Contact length 1.350 1.900 0.069 1.585 2.050 0.193 Table 2 Multivariate analysis to identify significant factors for LNM Predictors p OR Lower Upper Contact length 0.014 0.374 0.156 0.755 Skirt sign 0.092 4.382 0.74 24.89 Cavity 0.028 0.153 0.022 0.703 Angiogram sign 0.003 9.878 2.272 49.963 Lymph node enlargement 0.0 76.571 8.139 1138.941 CSA 0.013 3.072 1.389 8.052 Table 3. Performance metrics for nine models in the training dataset Model AUC(SD) Accuracy(SD) Sensitivity(SD) Specificity(SD) XGBoost 1.000(0.000) 1.000(0.000) 1.000(0.000) 1.000(0.000) logistic 0.896(0.003) 0.834(0.003) 0.875(0.063) 0.830(0.011) LightGBM 0.999(0.001) 0.981(0.019) 1.000(0.000) 0.978(0.022) RandomForest 1.000(0.000) 1.000(0.000) 1.000(0.000) 1.000(0.000) AdaBoost 1.000(0.000) 1.000(0.000) 1.000(0.000) 1.000(0.000) DecisionTree 1.000(0.000) 1.000(0.000) 1.000(0.000) 1.000(0.000) GBDT 1.000(0.000) 1.000(0.000) 1.000(0.000) 1.000(0.000) GNB 0.884(0.005) 0.831(0.019) 0.805(0.041) 0.835(0.026) MLP 0.340(0.002) 0.916(0.000) 0.133(0.000) 1.000(0.000) AUC: area under the curve; EXtreme Gradient Boosting (XGB) Classifier, Light Gradient Boosting Machine (LGBM) Classifier, RandomForest Classifier（RF）, AdaBoost Classifier, Gaussian naive bayes (GNB), Logistic Regression（LR）, Multilayer Perceptron (MLP) Classifier, Decision Tree, Gradient Boosting Decision Tree (GBDT). Table 4. Performance metrics for nine models in the validation dataset Model AUC(SD) Accuracy(SD) Sensitivity(SD) Specificity(SD) XGBoost 0.693(0.107) 0.846(0.000) 0.125(0.125) 0.929(0.014) logistic 0.874(0.030) 0.833(0.013) 0.667(0.167) 0.864(0.045) LightGBM 0.732(0.061) 0.808(0.038) 0.375(0.375) 0.912(0.055) RandomForest 0.833(0.001) 0.833(0.038) 0.292(0.042) 0.887(0.030) AdaBoost 0.840(0.025) 0.885(0.013) 0.350(0.150) 0.970(0.000) DecisionTree 0.555(0.097) 0.859(0.013) 0.167(0.167) 0.943(0.027) GBDT 0.807(0.004) 0.885(0.064) 0.000(0.000) 0.985(0.015) GNB 0.825(0.008) 0.744(0.026) 0.533(0.133) 0.778(0.045) MLP 0.261(0.002) 0.821(0.000) 0.000(0.000) 1.000(0.000) AUC: area under the curve; EXtreme Gradient Boosting (XGB) Classifier, Light Gradient Boosting Machine (LGBM) Classifier, RandomForest Classifier（RF）, AdaBoost Classifier, Gaussian naive bayes (GNB), Logistic Regression（LR）, Multilayer Perceptron (MLP) Classifier, Decision Tree, Gradient Boosting Decision Tree (GBDT). Additional Declarations No competing interests reported. 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-6676702","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":495236415,"identity":"52439efd-9037-4dfd-be1e-b507f299e764","order_by":0,"name":"Junjie Zhang","email":"","orcid":"","institution":"the Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Junjie","middleName":"","lastName":"Zhang","suffix":""},{"id":495236416,"identity":"dcfa3922-dbec-4484-8b47-3edc62a43653","order_by":1,"name":"Ligang Hao","email":"","orcid":"","institution":"Xing Tai People’s Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ligang","middleName":"","lastName":"Hao","suffix":""},{"id":495236417,"identity":"498367bd-4292-45da-bbbf-ef811c760e33","order_by":2,"name":"Zhenbin Qiu","email":"","orcid":"","institution":"Guangdong Lung Cancer Institute, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhenbin","middleName":"","lastName":"Qiu","suffix":""},{"id":495236418,"identity":"8dba9346-304a-4eb0-8fa7-37739ed13e08","order_by":3,"name":"Fengxiao Gao","email":"","orcid":"","institution":"Xing Tai People’s Hospital","correspondingAuthor":false,"prefix":"","firstName":"Fengxiao","middleName":"","lastName":"Gao","suffix":""},{"id":495236419,"identity":"8441e9f3-ff05-4c63-a240-4d1d2f70b3cc","order_by":4,"name":"Wenzhao Zhong","email":"","orcid":"","institution":"Guangdong Lung Cancer Institute, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wenzhao","middleName":"","lastName":"Zhong","suffix":""},{"id":495236420,"identity":"064918d3-9a2d-45df-91b9-eba28372b854","order_by":5,"name":"Qian Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6ElEQVRIiWNgGAWjYJCCA0CUwMAARB8MbOxI08I4oyAtmWiLwFqYeT4cYmwgpNbgRvLGAz933MkzOJ7+TNrG4AAzA/vhoxvwaZGckVZwsPfMs2KDM2/MpHMM7vAx8KSl3cCnhV8ix+AAb9vhxA03ctiAWp4xM0jwmOHVwgbUcvAvWAvQYRYGhxkbCGkB2XIYYkuCmTQDMVoke54VHJZtO1wseeaNsWWPQVoyGyG/GBxP3vzxbdvhPL7j6Q9v/PhjY8fPfvgYXi0gXWBS4QADiwTYdwSUI7TINzAwfyBC9SgYBaNgFIxAAADX1FZNvLBIJQAAAABJRU5ErkJggg==","orcid":"","institution":"the Fourth Hospital of Hebei Medical University","correspondingAuthor":true,"prefix":"","firstName":"Qian","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2025-05-16 03:23:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6676702/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6676702/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88492937,"identity":"30bf4fe8-5d31-4993-9953-d52fe57ec276","added_by":"auto","created_at":"2025-08-07 04:40:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":249527,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver operating characteristics (ROC) curves of the nine machine learning model in training set. Figure 1A and 1B were analysis of ROC in training set and validation set respectively.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6676702/v1/bf30b432696f5d40ef60ac8f.png"},{"id":88491770,"identity":"de4d02e4-d5ab-42b8-9181-5108a9e4851c","added_by":"auto","created_at":"2025-08-07 04:24:11","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":159271,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis of receiver operating characteristic (ROC) curves in the training cohorts (2A) and testing cohorts (2B).\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-6676702/v1/67f35c452cc644f718067d99.png"},{"id":88490805,"identity":"994a43e6-1c3f-4090-98b2-9fff33d25da3","added_by":"auto","created_at":"2025-08-07 04:16:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":156829,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis of receiver operating characteristic (ROC) curves in the training cohorts (2A) and testing cohorts (2B) in patients without LNE.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6676702/v1/da9f0b8aab669eea4a92b60f.png"},{"id":103213385,"identity":"7b5d24ad-b32b-45a2-b6c0-bbe9f0bb79fd","added_by":"auto","created_at":"2026-02-23 08:57:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2291573,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6676702/v1/247b3d1a-90e0-41c9-8d6e-503f344680b9.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Unveiling Lymph Node Metastasis (LNM) in pleural-attached lung mucinous adenocarcinoma: A Predictive Model Using Pleural Contact Parameters","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIn the 2011 classification by the International Association for the Study of Lung Cancer (IASLC), the American Thoracic Society (ATS), and the European Respiratory Society (ERS), as well as the 2015 World Health Organization (WHO) publication, primary pulmonary invasive mucinous adenocarcinoma (IMA) is identified as a variant subtype of lung adenocarcinoma, comprising 2\u0026ndash;5% of adenocarcinoma cases[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The clinical, imaging, pathological, and genetic characteristics of IMA are distinct from those of non-mucinous adenocarcinoma. According to the literature, only less than 10% of IMA patients exhibit lymph node metastasis (LNM), while more than 90% patients are classified within the nodal stage N0 group[\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Nevertheless, the presence of LNM significantly impacts the prognosis of IMA patients. Therefore, understanding and predicting LNM in IMA is crucial for improving patient outcomes.\u003c/p\u003e\u003cp\u003eResearch suggests that IMA frequently occurs in a subpleural location, which increases its potential to disseminate through the pleural lymphatic network[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Tumors located beneath the pleura, due to their direct contact with this membrane, may elevate the risk of metastasis via pleural lymphatics, consequently leading to an increased incidence of LNM. The lymphatic drainage pathways of the pleura are pivotal in the propagation of lung cancer. Research indicates that the lymphatic drainage of lung cancer extends beyond the routes adjacent to the bronchi to include those within the pleura[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. This drainage pathway is frequently segmental, especially when the tumor is near the pleura. Such segmental lymphatic drainage may increase the likelihood of tumor cell metastasis via the lymphatic system, thereby affecting patient prognosis[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. IMA, characterized by its mucin secretion, may promote the dissemination of tumor cells along the pleural surface, thus elevating the risk of LNM.\u003c/p\u003e\u003cp\u003eThe relationship between pleural involvement and LNM has been thoroughly investigated in numerous studies. A systematic review and meta-analysis have identified several prognostic factors for IMA, notably pleural metastasis and lymph node involvement, which are significant risk factors for poor prognosis[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Furthermore, it was recommended that the presence of LNM should lead to the upstaging of small tumors (\u0026le;\u0026thinsp;3 cm), potentially impacting surgical planning. Accurate preoperative assessment of LNM is essential for patients with clinically early-stage subpleural lung cancers. In this context, visceral pleural invasion (VPI) has been identified using immunocytochemistry on paraffin-embedded sections rather than through intraoperative frozen section analysis. We hypothesize that pleural-contacted IMA may similarly serve as a predictor of LNM. Currently, no effective method exists to predict LNM in patients with subpleural IMA. Therefore, this study aims to investigate the correlation between pleural-contacted parameters and lymph node metastasis in cases of pleural-contacted IMA.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003eThis study was conducted in accordance with the Helsinki Declaration and received approval from the Ethics Committee of our hospital (Ethics Committee of our hospitals, reference number: 2023【124】, dated 2023.11.15). For this retrospective study, informed consent was not required.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePatient\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study complies with the STROBE guidelines. We performed a retrospective review and analysis of all patients diagnosed with peripheral IMA at three public tertiary hospitals in our country: The Fourth Hospital of Hebei Medical University, XingTai People\u0026apos;s Hospital, and Guangdong Province People\u0026apos;s Hospital, spanning the period from January 2017 to December 2022. The inclusion criteria for this study were as follows: (1) surgical pathology-confirmed IMA, encompassing both pure invasive mucinous adenocarcinoma and mixed mucinous and non-mucinous adenocarcinoma, with systematic lymph node dissection conducted; (2) availability of computed tomography (CT) images in thin sections of 1.25 mm or less, demonstrating lung cancer as a single lesion; and (3) presence of solid nodules directly adjacent to the costal, mediastinal, or diaphragmatic pleura. The exclusion criteria included: (1) receipt of antitumor therapy prior to CT examination and pathological diagnosis; (2) presence of other types of cancer or incomplete clinical and imaging data, and (3) the availability of CT images obtained more than two weeks before the pathological diagnosis. This retrospective analysis received approval from our hospital\u0026apos;s ethical review board, which also granted a waiver for the requirement of informed consent.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eChest CT Imaging and feature extraction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe CT scans were re-evaluated to confirm the adjacency of the IMA to the pleura, utilizing lung window settings on Picture Archiving and Communication System (PACS) workstations. The examinations encompassed the entire thorax with patients positioned supinely and at full inspiration. All images were reconstructed in transverse, sagittal, and coronal planes, with section thicknesses ranging from 1 to 1.25 mm, and lung window settings set at 1500 Hounsfield Units (HU) width and -600 HU level.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA radiologist with extensive experience in chest CT imaging (Z.JJ., 14 years) independently assessed all nodules, without access to clinical data or pathology findings, to evaluate specific CT characteristics. These characteristics included: (1) size, contact length, depth, and contact surface area. Tumor size was defined as the longest diameter measured on axial, sagittal, or coronal lung window images. The tumor-pleura contact length was determined as the maximum contact length measured on axial, sagittal, or coronal mediastinum window images. The tumor-pleura distance was defined as the shortest distance measured on axial, sagittal, or coronal mediastinum window images. For nodules with multiple pleural connections, the largest contact surface area (CSA) was recorded. Given the inherent convexity at the junction of lung nodules and the pleura, and the lack of a universally accepted method for CSA measurement, the method developed by Qi et al was utilized to evaluate the interface between lung nodules and the pleura[12]. The CT indicators assessed comprised pleural thickening, the skirt-like sign, and the jellyfish sign. The skirt-like sign was identified by the thickening of the peripheral pleura on both sides of the contact surface for nodules adherent to the pleura. The jellyfish sign was characterized in sections adjacent to the area of maximum contact surface, where multiple linear septations between the nodule and the costal pleura resembled jellyfish tentacles.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIntra- and Interreader Agreement on Nodule Features\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA radiologist (Z.JJ.), who was blinded to the initial classifications, conducted an analysis of the CT features of each pleural-attached nodule during two separate sessions spaced three months apart. Subsequently, following a training session led by the first radiologist (Z.JJ.), a second radiologist (G.FX.), possessing 31 years of experience in chest CT imaging, independently evaluated the CT scans of the nodules. This evaluation was performed without access to any clinicopathologic information to assess the CT characteristics.\u003c/p\u003e\n\u003cp\u003eAny disagreements between the two radiologists were collectively reviewed and resolved through mutual consensus. In instances where consensus could not be achieved, a final decision was rendered by an additional radiologist (X.Q.), who has over 30 years of experience in chest CT imaging. Intraobserver agreement was assessed, and interobserver agreement was determined by comparing the characteristics of the subpleural nodules from the second review conducted by reader 1 (Z.JJ.) with those assessed by reader 2 (G.FX.).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analyses and Model Development\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComparative and descriptive analyses were undertaken to examine the characteristics of patients with and without LNM. All statistical analyses were executed utilizing Python version 3.7. For continuous variables, the Student\u0026apos;s t-test or the Wilcoxon rank-sum test was applied, whereas categorical variables were evaluated using the Pearson\u0026nbsp;\u0026chi;\u0026sup2;\u0026nbsp;test or Fisher\u0026apos;s exact test. Independent CT features associated with LNM were identified through both univariable and multivariable logistic regression analyses. A p-value of less than 0.05 was deemed indicative of statistical significance.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Furthermore, machine learning techniques were employed in this study to refine the construction of factors that exhibited statistical significance in the multivariate analysis. The methodology comprised several key steps: (1) Participants were randomly assigned to a training set and a test set in a 7:3 ratio. (2) Utilizing statistically significant factors identified through multivariate analysis, nine machine learning models were developed within the training set. These models included the Extreme Gradient Boosting (XGB) classifier, Light Gradient Boosting Machine (LGBM) classifier, Random Forest classifier, AdaBoost classifier, Gaussian Naive Bayes (GNB), Logistic Regression, Multilayer Perceptron (MLP) classifier, Polynomial Support Vector Machine (SVC), and k-Nearest Neighbor (KNeighbors) classifier. Optimal parameters for these models were retrospectively determined using 5-fold cross-validation. The performance of the nine machine learning models was assessed using the ROC curve. A 5-fold cross-validation was conducted to validate the most effective model. The primary evaluation metrics included the AUC, accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. Based on the ROC analysis, the most predictive machine learning classifier was selected.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003ePatient characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, a cohort of 276 patients diagnosed with IMA was recruited from three public tertiary hospitals within our country. Specifically, 188 patients were sourced from The Fourth Hospital of Hebei Medical University, 17 from XingTai People\u0026apos;s Hospital, and 71 from Guangdong Province People\u0026apos;s Hospital. The cohort comprised 27 patients with LNM and 249 without LNM, analyzed retrospectively. The clinical characteristics, including age, smoking habits, and sex, did not exhibit significant differences between the LNM and non-LNM groups. However, notable differences were observed between these groups concerning the presence of the Jellyfish sign (29.63% vs. 12.00%, P=0.019), the skirt-like sign (22.22% vs. 8.43%, P=0.035), spiculation (70.37% vs. 37.35%, P=0.002), cavity (22.22% vs. 54.62%, P=0.003), necrosis (22.22% vs. 54.62%, P=0.003), lymph node enlargement (2.41% vs. 22.22%, P\u0026lt;0.001), VPI (18.75% vs. 6.76%), and angiogram findings (40.74% vs. 22.22%, P=0.004). Additionally, the pleura contact length and CSA were significantly greater in patients with LNM compared to those without LNM (1.370 mm vs. 1.950 cm, P=0.036, and 0.6838 cm\u0026sup2; vs. 1.665 cm\u0026sup2;, P\u0026lt;0.001, respectively).The detailed characteristics of patients in the training and testing cohorts are presented in Table 1.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature selection and model construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe multivariate analysis identified significant differences between groups with and without LNM concerning CSA, contact length, lymph node enlargement, angiogram sign, cavities or vacuoles, and skirt sign, as detailed in Table 2. In the training dataset, nine machine learning prediction models were developed, including XGBoost (XGB), LightGBM (LGBM), Random Forest (RF), AdaBoost Classifier, Gaussian Naive Bayes (GNB), Logistic Regression (LR), Multi-Layer Perceptron (MLP), Decision Tree, and Gradient Boosting Decision Tree (GBDT), utilizing three selected radiomics features. The optimal parameters for these models were retrospectively determined using 5-fold cross-validation.\u003c/p\u003e\n\u003cp\u003eAmong these models, the Logistic Regression classifier demonstrated superior performance, achieving AUC values of 0.896 on the training dataset and 0.874 on the validation dataset, as detailed in Tables 3 and 4, and Figure 1. Furthermore, the AUC values for the Logistic Regression model in predicting MLP were 0.856 in the entire training cohort and 0.803 in the test cohort, with accuracies of 0.77 and 0.86, respectively, and the optimal cut-off value was 0.083. Additional information is presented in Figure 2. The AUC value of the logistic regression model was significantly higher than that for lymph node enlargement, which was 0.599. A subsequent analysis of the model in patients without lymph node enlargement demonstrated superior performance, with AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort. The accuracy rates were 0.826 and 0.838, respectively, Figure3.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study is the first to systematically evaluate the correlation between pleural-contacted parameters, imaging features, and LNM in pleural-contacted IMA. It identifies pleural contact length, CSA, the skirt-like sign, and the jellyfish sign as independent predictors of LNM. The prediction model developed from these six features offers a novel approach for preoperative non-invasive assessment of metastasis risk in patients with pleural-contacted IMA..\u003c/p\u003e\u003cp\u003eUnlike previous research that focused on qualitative assessments of pleural indentation or invasion[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], this study quantifies the predictive efficacy of pleural contact length and area, demonstrating their significant association with LNM. The findings suggest that pleural contact area may indicate the invasive biological behavior of tumors, with a larger contact area potentially reflecting a greater likelihood of tumor cells infiltrating the lymphatic network beneath the pleura, thus promoting lymphatic metastasis. Moreover, the incorporation of quantitative parameters can reduce subjective judgment errors and improve clinical reproducibility. Existing literature has demonstrated a correlation between VPI and LNM especially N2 lymph node in lung adenocarcinoma[\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. In the current study, it was found that 18.75% of patients with VPI exhibited LNM, a rate significantly higher than the 6.76% observed in patients without VPI, aligning with previous research findings in NSCLC[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. The presence of VPI facilitates the migration of tumor cells to lymph nodes through lymphatic drainage, owing to the extensive lymphatic network of the visceral pleura that drains into the mediastinum[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Prior research has established a positive correlation between the length of pleural contact, CSA, and VPI in lung cancer[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. As well as between VPI and LNM, our findings further substantiate the positive association between pleural contact length, CSA, and LNM in cases of IMA with pleural contact. And further large-scale studies are necessary to investigate these phenomenons in patients with IMA.\u003c/p\u003e\u003cp\u003eThe skirt sign may serve as an independent predictive factor linked to tumor stromal response and lymphatic invasion. Pathological investigations have demonstrated that IMA often exhibits significant connective tissue proliferation reactions. The skirt sign may correspond to active fibroblast proliferation and lymphatic vessel dilation at the tumor-stroma interface, potentially indicating tumor progression along the subpleural lymphatic vessels[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. The presence of this phenomenon may act as an imaging marker indicative of the tumor microenvironment's pro-metastatic characteristics. Previous studies have identified the skirt sign in pleural attachment nodules as a valuable CT predictor of VPI in NSCLC with solid pleural attachments measuring 30 mm or smaller[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Our current research further corroborates that the skirt sign is an independent predictor of LNM in IMA. The presence of angiogram signs and cavities or vacuoles also demonstrates a significant association with LNM, although the underlying mechanisms require further investigation in future studies.\u003c/p\u003e\u003cp\u003eFurthermore, the development of the Logistic Regression Classifier, which incorporates six distinct characteristics, represents a significant advancement in predictive modeling compared to the singular feature of lymph node enlargement. The prediction model achieved AUC values of 0.856 and 0.803 in the training and testing cohorts, respectively. Notably, in patients without lymph node enlargement, the model demonstrated superior performance, achieving AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort, with accuracy rates of 0.826 and 0.838, respectively. Through rigorous validation techniques, including ROC curve analysis, we have shown that our model can effectively stratify patients based on their risk of LNM, thereby informing treatment strategies. The implications of these findings are substantial, as they suggest that our predictive model could enhance clinical decision-making, leading to personalized surgical interventions and potentially improved prognostic outcomes for patients diagnosed with IMA.\u003c/p\u003e\u003cp\u003eThis study is subject to several limitations that must be acknowledged. Firstly, it was conducted as a retrospective analysis with a relatively small sample size. Despite utilizing data from three tertiary hospitals, the cohort included only 28 cases with positive lymph node metastasis, which limited the ability to divide the data into independent external validation groups. Secondly, the study's focus on patients with pleural contact and post-surgical pathological outcomes may introduce selection bias, necessitating cautious interpretation of the findings. Thirdly, the study primarily concentrated on nodule characteristics in the evaluation of CT images. Future research should consider the assessment of lymph nodes to provide a more comprehensive analysis. Furthermore, the models developed in this study underwent only preliminary evaluation and validation; their performance is anticipated to improve and undergo external validation in future research endeavors. Advancements in radiomics have enabled the transformation of imaging data into quantitative features, which may facilitate a more objective and quantitative characterization of tumors. As a result, it is expected that future research will investigate prognostic predictions and survival analyses for patients with LNM through the application of radiomics.\u003c/p\u003e\u003cp\u003eIn conclusion, this study introduces the development of an innovative model aimed at evaluating the correlation between pleural-contacted parameters, imaging characteristics, and LNM in instances of pleural-contacted IMA. The imaging predictor model exhibited significant diagnostic efficacy for LNM in pleural-contacted IMA, particularly in patients without LNM, indicating that CT semantic features can enhance preoperative clinical practice and possess considerable potential for clinical application.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eVPI: Visceral Pleural Invasion\u003c/p\u003e\n\u003cp\u003eLNM:\u0026nbsp;Lymph Node Metastasis\u003c/p\u003e\n\u003cp\u003eNSCLC: non\u0026ndash;small cell lung cancer\u003c/p\u003e\n\u003cp\u003eCT: computed tomography\u003c/p\u003e\n\u003cp\u003eROC: receiver operating characteristic curve\u003c/p\u003e\n\u003cp\u003eIASLC: International Association for the Study of Lung Cancer\u003c/p\u003e\n\u003cp\u003eIMA: Invasive mucinous adenocarcinoma\u003c/p\u003e\n\u003cp\u003ePIMA: pure Invasive mucinous adenocarcinoma\u003c/p\u003e\n\u003cp\u003eTNM: Tumor, Node, Metastasis\u003c/p\u003e\n\u003cp\u003eCSA: contact surface area\u003c/p\u003e\n\u003cp\u003eGLCM: gray level co-occurrence matrix \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGLRLM: gray level run length matrix\u003c/p\u003e\n\u003cp\u003eGLSZM: gray level size zone matrix\u003c/p\u003e\n\u003cp\u003eNGTDM: neighboring gray tone difference matrix\u003c/p\u003e\n\u003cp\u003eGLDM: gray level dependence matrix\u003c/p\u003e\n\u003cp\u003eXGB: EXtreme Gradient Boosting Classifier\u003c/p\u003e\n\u003cp\u003eLGBM: Light Gradient Boosting Machine Classifier,\u003c/p\u003e\n\u003cp\u003eRF: RandomForest Classifier\u003c/p\u003e\n\u003cp\u003eGNB: Gaussian naive bayes \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLR: Logistic Regression\u003c/p\u003e\n\u003cp\u003eMLP: Multilayer Perceptron Classifier,\u003c/p\u003e\n\u003cp\u003eGBDT: Gradient Boosting Decision Tree\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Xing Tai People\u0026apos;s Hospital ethical review board approved this retrospective analysis and waived informed consent requirements. Ethics Committee of Xing Tai People\u0026rsquo;s Hospital, reference number: 2023【124】, dated 2023.11.15. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;We confirm that the manuscript has been submitted solely to this journal and is not published, in press, or submitted elsewhere. Written informed consent was obtained from the individuals for the publication of any potentially identifiable images or data included in this article.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analyzed during the current study are not publicly available because the dataset will be further studied to publish other works but are available from the corresponding author on reasonable request.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no conflicts of interest to declare.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKey development plan of Xingtai (2023ZC049)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJJ Z, LG H and ZB Q performed the experiments and wrote the manuscript. FX G was responsible for the data collection and analysis, Q X and WZ Z were responsible for designing the experiments. All authors read and approved the final version of this submitted manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNone\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eTravis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger KR, Yatabe Y, Beer DG, Powell CA, Riely GJ, Van Schil PE\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInternational association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma\u003c/strong\u003e. \u003cem\u003eJournal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer \u003c/em\u003e2011, \u003cstrong\u003e6\u003c/strong\u003e(2):244-285.\u003c/li\u003e\n\u003cli\u003eTravis WD, Brambilla E, Nicholson AG, Yatabe Y, Austin JHM, Beasley MB, Chirieac LR, Dacic S, Duhig E, Flieder DB\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification\u003c/strong\u003e. \u003cem\u003eJournal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer \u003c/em\u003e2015, \u003cstrong\u003e10\u003c/strong\u003e(9):1243-1260.\u003c/li\u003e\n\u003cli\u003eMoon SW, Choi SY, Moon MH: \u003cstrong\u003eEffect of invasive mucinous adenocarcinoma on lung cancer-specific survival after surgical resection: a population-based study\u003c/strong\u003e. \u003cem\u003eJournal of thoracic disease \u003c/em\u003e2018, \u003cstrong\u003e10\u003c/strong\u003e(6):3595-3608.\u003c/li\u003e\n\u003cli\u003eLee HY, Cha MJ, Lee KS, Lee HY, Kwon OJ, Choi JY, Kim HK, Choi YS, Kim J, Shim YM: \u003cstrong\u003ePrognosis in Resected Invasive Mucinous Adenocarcinomas of the Lung: Related Factors and Comparison with Resected Nonmucinous Adenocarcinomas\u003c/strong\u003e. \u003cem\u003eJournal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer \u003c/em\u003e2016, \u003cstrong\u003e11\u003c/strong\u003e(7):1064-1073.\u003c/li\u003e\n\u003cli\u003eKadota K, Yeh YC, D\u0026apos;Angelo SP, Moreira AL, Kuk D, Sima CS, Riely GJ, Arcila ME, Kris MG, Rusch VW\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAssociations between mutations and histologic patterns of mucin in lung adenocarcinoma: invasive mucinous pattern and extracellular mucin are associated with KRAS mutation\u003c/strong\u003e. \u003cem\u003eThe American journal of surgical pathology \u003c/em\u003e2014, \u003cstrong\u003e38\u003c/strong\u003e(8):1118-1127.\u003c/li\u003e\n\u003cli\u003eZhang J, Hao L, Xu Q, Gao F: \u003cstrong\u003eRadiomics and Clinical Characters Based Gaussian Naive Bayes (GNB) Model for Preoperative Differentiation of Pulmonary Pure Invasive Mucinous Adenocarcinoma From Mixed Mucinous Adenocarcinoma\u003c/strong\u003e. \u003cem\u003eTechnology in cancer research \u0026amp; treatment \u003c/em\u003e2024, \u003cstrong\u003e23\u003c/strong\u003e:15330338241258415.\u003c/li\u003e\n\u003cli\u003eZhang J, Hao L, Li M, Xu Q, Shi G: \u003cstrong\u003eCT Radiomics Combined With Clinicopathological Features to Predict Invasive Mucinous Adenocarcinoma in Patients With Lung Adenocarcinoma\u003c/strong\u003e. \u003cem\u003eTechnology in cancer research \u0026amp; treatment \u003c/em\u003e2023, \u003cstrong\u003e22\u003c/strong\u003e:15330338231174306.\u003c/li\u003e\n\u003cli\u003eFukumura D, Kloepper J, Amoozgar Z, Duda DG, Jain RK: \u003cstrong\u003eEnhancing cancer immunotherapy using antiangiogenics: opportunities and challenges\u003c/strong\u003e. \u003cem\u003eNature reviews Clinical oncology \u003c/em\u003e2018, \u003cstrong\u003e15\u003c/strong\u003e(5):325-340.\u003c/li\u003e\n\u003cli\u003eBerzenji L, Van Schil PE, Carp L: \u003cstrong\u003eThe eighth TNM classification for malignant pleural mesothelioma\u003c/strong\u003e. \u003cem\u003eTranslational lung cancer research \u003c/em\u003e2018, \u003cstrong\u003e7\u003c/strong\u003e(5):543-549.\u003c/li\u003e\n\u003cli\u003eShen F, Wu X, Geng J, Guo W, Duan J: \u003cstrong\u003ePrognostic factors for resected invasive mucinous lung adenocarcinoma: a systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eBMC cancer \u003c/em\u003e2024, \u003cstrong\u003e24\u003c/strong\u003e(1):1317.\u003c/li\u003e\n\u003cli\u003eZhao T, Yi J, Luo D, Liu J, Fan X, Wu Q, Wang W: \u003cstrong\u003ePrognostic factors for invasive mucinous adenocarcinoma of the lung: systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eWorld journal of surgical oncology \u003c/em\u003e2024, \u003cstrong\u003e22\u003c/strong\u003e(1):41.\u003c/li\u003e\n\u003cli\u003eSun Q, Li P, Zhang J, Yip R, Zhu Y, Yankelevitz DF, Henschke CI: \u003cstrong\u003eCT Predictors of Visceral Pleural Invasion in Patients with Non-Small Cell Lung Cancers 30 mm or Smaller\u003c/strong\u003e. \u003cem\u003eRadiology \u003c/em\u003e2024, \u003cstrong\u003e310\u003c/strong\u003e(1):e231611.\u003c/li\u003e\n\u003cli\u003eZhang T, Zhang JT, Li WF, Lin JT, Liu SY, Yan HH, Yang JJ, Yang XN, Wu YL, Nie Q\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eVisceral pleural invasion in T1 tumors (\u003c/strong\u003e\u003cstrong\u003e\u0026le;3 cm), particularly T1a, in the eighth tumor-node-metastasis classification system for non-small cell lung cancer: a population-based study\u003c/strong\u003e. \u003cem\u003eJournal of thoracic disease \u003c/em\u003e2019, \u003cstrong\u003e11\u003c/strong\u003e(7):2754-2762.\u003c/li\u003e\n\u003cli\u003eLiu J, Li J, Lin G, Long Z, Li Q, Liu B: \u003cstrong\u003eRisk factors of lobar lymph node metastases in non-primary tumor-bearing lobes among the patients of non-small-cell lung cancer\u003c/strong\u003e. \u003cem\u003ePloS one \u003c/em\u003e2020, \u003cstrong\u003e15\u003c/strong\u003e(9):e0239281.\u003c/li\u003e\n\u003cli\u003eZhang H, Lu C, Lu Y, Yu B, Lv F, Zhu Z: \u003cstrong\u003eThe predictive and prognostic values of factors associated with visceral pleural involvement in resected lung adenocarcinomas\u003c/strong\u003e. \u003cem\u003eOncoTargets and therapy \u003c/em\u003e2016, \u003cstrong\u003e9\u003c/strong\u003e:2337-2348.\u003c/li\u003e\n\u003cli\u003eShimizu K, Yoshida J, Nagai K, Nishimura M, Ishii G, Morishita Y, Nishiwaki Y: \u003cstrong\u003eVisceral pleural invasion is an invasive and aggressive indicator of n on-small cell lung cancer\u003c/strong\u003e. \u003cem\u003eJ Thorac Cardiovasc Surg\u003c/em\u003e, \u003cstrong\u003e130\u003c/strong\u003e(1):160-165.\u003c/li\u003e\n\u003cli\u003eZhang C, Wang L, Cai X, Li M, Sun D, Wang P: \u003cstrong\u003eTumour-pleura relationship on CT is a risk factor for occult lymph node metastasis in peripheral clinical stage IA solid adenocarcinoma\u003c/strong\u003e. \u003cem\u003eEuropean radiology \u003c/em\u003e2023, \u003cstrong\u003e33\u003c/strong\u003e(5):3083-3091.\u003c/li\u003e\n\u003cli\u003eLee E, Biko DM, Sherk W, Masch WR, Ladino-Torres M, Agarwal PP: \u003cstrong\u003eUnderstanding Lymphatic Anatomy and Abnormalities at Imaging\u003c/strong\u003e. \u003cem\u003eRadiographics : a review publication of the Radiological Society of North America, Inc \u003c/em\u003e2022, \u003cstrong\u003e42\u003c/strong\u003e(2):487-505.\u003c/li\u003e\n\u003cli\u003eMinamoto F, Ara\u0026uacute;jo P, D\u0026apos;Ambrosio P, Dela Vega A, Lauricella L, P\u0026ecirc;go-Fernandes P, Terra R: \u003cstrong\u003eThe association of visceral pleural invasion with skip N2 metastasis on clinical stage IA NSCLC\u003c/strong\u003e. \u003cem\u003eClinics (Sao Paulo, Brazil) \u003c/em\u003e2024, \u003cstrong\u003e79\u003c/strong\u003e:100334.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 Clinical characteristics of the patients in training group and a test group.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"548\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.8571%;\"\u003eCharacters\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 38.6861%;\" colspan=\"3\"\u003eTraining cohort\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 38.6861%;\" colspan=\"3\"\u003eTesting cohort\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.8571%;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003eLN(-)\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003eLN(+)\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003eP\u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003eLN(-)\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\u003cspan style='color: rgb(0, 0, 0); font-family: \"Times New Roman\"; font-size: medium; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: normal; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial; display: inline !important; float: none;'\u003eLN(+)\u003c/span\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003eP\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eGender Female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eJellyfish \u0026nbsp; \u0026nbsp; No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e69\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e4\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.483\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e9\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eSkirt \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e157\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e7\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eSmoking \u0026nbsp; \u0026nbsp; No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e121\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e57\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e4\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e21\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eLobul \u0026nbsp; \u0026nbsp; No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.319\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e9\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e69\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e5\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eSpicul \u0026nbsp; \u0026nbsp;No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e46\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.646\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e32\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eCavity \u0026nbsp; \u0026nbsp;No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e38\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e5\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.056\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e40\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eNecrosis \u0026nbsp;No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e73\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e4\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.320\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e5\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eAS \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; No \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e149\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e60\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.590\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e18\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eLNE \u0026nbsp; \u0026nbsp; No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e168\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e75\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e5\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eAge \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e62.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e60.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.104\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e63.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e65.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.328\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eD-max\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2.250\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e3.150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2.450\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e3.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.320\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eCSA median\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.654\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.538\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.756\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1.849\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.8571%;\"\u003e\n \u003cp\u003eContact length\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1.350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e1.900\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e0.069\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e1.585\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 13.4694%;\"\u003e\n \u003cp\u003e2.050\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 11.6327%;\"\u003e\n \u003cp\u003e0.193\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2 Multivariate analysis to identify significant factors for LNM\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"389\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003ePredictors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003eLower\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eUpper\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eContact length\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e0.156\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.755\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eSkirt sign\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e4.382\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e0.74\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e24.89\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eCavity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e0.028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e0.153\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 56px;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eAngiogram sign\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e9.878\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 56px;\"\u003e\n \u003cp\u003e2.272\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e49.963\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eLymph node enlargement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e76.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e8.139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1138.941\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eCSA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e3.072\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e1.389\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e8.052\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u003c/strong\u003e Performance metrics for nine models in the training dataset\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"104%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003eAUC(SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eAccuracy(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003eSensitivity(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003eSpecificity(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003elogistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e0.896(0.003)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e0.834(0.003)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.875(0.063)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.830(0.011)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e0.999(0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e0.981(0.019)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.978(0.022)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eRandomForest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eDecisionTree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eGBDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eGNB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e0.884(0.005)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e0.831(0.019)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.805(0.041)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.835(0.026)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eMLP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 18px;\"\u003e\n \u003cp\u003e0.340(0.002)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e0.916(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.133(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAUC: area under the curve; EXtreme Gradient Boosting (XGB) Classifier, Light Gradient Boosting Machine (LGBM) Classifier, RandomForest Classifier（RF）, AdaBoost Classifier, Gaussian naive bayes (GNB), Logistic Regression（LR）, Multilayer Perceptron (MLP) Classifier, Decision Tree, Gradient Boosting Decision Tree (GBDT).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4.\u003c/strong\u003e Performance metrics for nine models in the validation dataset\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"97%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 17px;\"\u003e\n \u003cp\u003eAUC(SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003eAccuracy(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003eSensitivity(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003eSpecificity(SD)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.693(0.107)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.846(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.125(0.125)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.929(0.014)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003elogistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.874(0.030)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.833(0.013)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.667(0.167)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.864(0.045)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.732(0.061)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.808(0.038)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.375(0.375)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.912(0.055)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eRandomForest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.833(0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.833(0.038)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.292(0.042)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.887(0.030)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.840(0.025)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.885(0.013)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.350(0.150)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.970(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eDecisionTree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.555(0.097)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.859(0.013)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.167(0.167)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.943(0.027)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eGBDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.807(0.004)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.885(0.064)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.985(0.015)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003eGNB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 17px;\"\u003e\n \u003cp\u003e0.825(0.008)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.744(0.026)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.533(0.133)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.778(0.045)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 20px;\"\u003e\n \u003cp\u003eMLP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 17px;\"\u003e\n \u003cp\u003e0.261(0.002)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e0.821(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e0.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 21px;\"\u003e\n \u003cp\u003e1.000(0.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAUC: area under the curve; EXtreme Gradient Boosting (XGB) Classifier, Light Gradient Boosting Machine (LGBM) Classifier, RandomForest Classifier（RF）, AdaBoost Classifier, Gaussian naive bayes (GNB), Logistic Regression（LR）, Multilayer Perceptron (MLP) Classifier, Decision Tree, Gradient Boosting Decision Tree (GBDT).\u003c/p\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":"Lymph node metastasis (LNM), lung invasive mucinous adenocarcinoma (IMA), pleural contact, computed tomography (CT), machine learning model","lastPublishedDoi":"10.21203/rs.3.rs-6676702/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6676702/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eThis study seeks to evaluate the prognostic significance of pleural-attached parameters and computed tomography (CT) imaging characteristics in predicting lymph node metastasis (LNM) in cases of pleural-attached invasive mucinous adenocarcinoma (IMA).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethodology: \u003c/strong\u003eA retrospective analysis was conducted on a cohort of 276 IMA patients from three public tertiary hospitals in China, covering the period from January 2017 to December 2022. A comparative analysis was performed on pleural-attached parameters and CT imaging characteristics between different patient groups. Based on variables that showed statistical significance, nine machine learning models were developed, and the model with the highest area under the curve (AUC) value was identified as the optimal model. Patients were randomly allocated into training and testing groups in a 7:3 ratio. The 5-fold cross-validation technique was employed to evaluate the receiver operating characteristic (ROC) curve AUC value of the most effective machine learning model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eBoth univariate and multivariate logistic regression analyses identified pleural contact length, pleural contact surface area (CSA), the skirt-like sign, cavity sign, angiogram sign, and lymph node enlargement as significant predictive factors for lymph node metastasis (LNM) in pleural-attached invasive mucinous adenocarcinoma (IMA). The logistic regression model demonstrated superior performance, achieving ROC-AUC values of 0.856 and 0.803 in the training and test groups, respectively. Further analysis of the model in patients without lymph node enlargement indicated that it maintained superior performance, with AUC values of 0.880 in the entire training cohort and 0.883 in the test cohort, and accuracy rates of 0.826 and 0.838, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThe logistic regression model, which incorporates pleural contact parameters and imaging characteristics, demonstrated substantial diagnostic value for assessing LNM in pleural-attached IMA, particularly in patients without lymph node enlargement. It exhibited excellent diagnostic efficacy and provides a non-invasive evaluation method for clinical practice.\u003c/p\u003e","manuscriptTitle":"Unveiling Lymph Node Metastasis (LNM) in pleural-attached lung mucinous adenocarcinoma: A Predictive Model Using Pleural Contact Parameters","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-07 04:16:06","doi":"10.21203/rs.3.rs-6676702/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":"2b82e159-a853-4c14-9ed9-f5cef9446ac9","owner":[],"postedDate":"August 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-23T08:56:18+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-07 04:16:06","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6676702","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6676702","identity":"rs-6676702","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}