Development and validation of nomograms to intraoperatively predict metastatic patterns in regional lymph nodes in patients diagnosed with esophageal cancer

Abstract

Background An accurate intraoperative prediction of lymph node metastatic risk can help surgeons in choosing precise surgical procedures. We aimed to develop and validate nomograms to intraoperatively predict patterns of regional lymph node (LN) metastasis in patients with esophageal cancer. Methods The prediction model was developed in a training cohort consisting of 487 patients diagnosed with esophageal cancer who underwent esophagectomy with complete LN dissection from January 2016 to December 2016. Univariate and multivariable logistic regression were used to identify independent risk factors that were incorporated into a prediction model and used to construct a nomogram. Contrast-enhanced computed tomography reported LN status and was an important comparative factor of clinical usefulness in a validation cohort. Nomogram performance was assessed in terms of calibration, discrimination, and clinical usefulness. An independent validation cohort comprised 206 consecutive patients from January 2017 to December 2017. Results Univariate analysis and multivariable logistic regression revealed three independent predictors of metastatic regional LNs, three independent predictors of continuous regional LNs, and two independent predictors of skipping regional LNs. Independent predictors were used to build three individualized prediction nomograms. The models showed good calibration and discrimination, with area under the curve (AUC) values of 0.737, 0.738, and 0.707. Application of the nomogram in the validation cohort yielded good calibration and discrimination, with AUC values of 0.728, 0.668, and 0.657. Decision curve analysis demonstrated that the three nomograms were clinically useful in the validation cohort. Conclusion This study presents three nomograms that incorporate clinicopathologic factors, which can be used to facilitate the intraoperative prediction of metastatic regional LN patterns in patients with esophageal cancer.

A nomogram model for predicting prognosis of obstructive colorectal cancer

“…8 Department of General Surgery, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200011, China. 9 Department of General Surgery, The First Affiliated Hospital of Hebei Medical University, No. 89 Donggang Road, Shijiazhuang 050000, Hebei, China.…”

“…These models are based on multivariate analysis and integrate the results of logistic or Cox regression to a great extent to predict the probability of a certain clinical event in patients along with intuitive graphical presentations. An increasing amount of literature has reported the advantages of these models in predicting tumor recurrence and metastasis, death, and other prognostic outcomes [8,9]. In comparison to conventional evaluation methods, the nomogram model can produce more accurate and intuitive predictions [10].…”

Radiomics Analysis of Lymph Nodes with Esophageal Squamous Cell Carcinoma Based on Deep Learning

“…The CT parameters are described in Section 1 of Appendix A1. As the location of the LNs was variable and complex, the maximum cross-sectional area of suspicious mediastinal LNs of one or several slices was selected for each patient according to previous reports [ 7 ]. The LNs as the region of interest (ROI) in each 2D image were automatically segmented according to the labels.…”

“…Therefore, whether LNM is judged or not is still primarily based on postoperative pathological examination results [ 1 ]. This has a certain influence on the choice of precise surgical procedures and the clinical stage of chemoradiotherapy [ 7 ].…”

“…is has a certain influence on the choice of precise surgical procedures and the clinical stage of chemoradiotherapy [7].…”