Development Of A Nomogram Based On Clinicopathological And Biological Features To Predict Neck Lymph Node Metastasis In Hypopharyngeal Squamous Cell Carcinoma

BACKGROUND:Lymph node metastasis(LNM)in the neck is an important clinical feature of hypopharyngeal squamous cell carcinoma(HSCC)and an important factor influencing the prognosis and treatment decision selection of HSCC,but there are great differences in neck LNM among different patients.The outcomes of numerous studies have demonstrated that the probability of developing neck LNM in HSCC patients is associated with multiple risk factors,which include both clinical features such as tumor size and pathological and biological features such as lymph vascular invasion or not,and Stathminl(STMN1)expression level.Therefore,it is difficult to accurately predict neck LNM in HSCC patients based on a single factor,and the establishment of a nomogram that includes risk factors related to clinicopathological and biological characteristics will undoubtedly improve the accuracy of predicting neck LNM in HSCC patients,which can provide a basis for selecting the scope of radical surgical lymph node dissection or the irradiation area of lymph node drainage area for radical radiotherapy in HSCC.OBJECTIVE:To construct a nomogram containing clinicopathological and biological features based on the analysis of risk factors associated with HSCC neck LNM,and to validate its sensitivity and specificity for predicting HSCC neck LNM to provide a reference for making clinical treatment decisions.METHODS:This retrospective study screened 117 patients with HSCC who underwent hypopharyngeal carcinoma mass resection bilateral or ipsilateral cervical lymph node dissection(training cohort 64;validation cohort 53).Clinical data such as patients’ age,gender,and preoperative computed tomography(CT)reports were collected;immunohistochemical methods were applied to assess the pathological and biological characteristics of patients such as lymph vascular invasion,vascular invasion,and Ki67 expression,and STMN1 expression levels.In the training cohort,the risk factors for neck LNM screened by univariate analysis(P<0.2)were screened again in multivariate logistic regression analysis(P<0.2)to finally identify independent risk factors affecting HSCC neck LNM,and then the rms package in R software was used to construct a nomogram for HSCC neck LNM,while a CT diagnostic nomogram relying only on CT to diagnose LNM was built,and then the sensitivity and specificity as well as clinical utility of the new nomogram and the CT diagnostic nomogram in predicting neck LNM were compared while the new nomogram was internally validated.Finally,the new nomogram was externally validated in the validation cohort by plotting operating characteristic(ROC)curves and calibration curves to determine the discriminatory ability and accuracy,and by plotting decision curve analysis(DCA)curves to determine its clinical utility.RESULTS:In the training cohort,five independent risk factors associated with HSCC neck LNM were screened by univariate analysis and multi-factor logistic regression analyses,among which,clinical features included the maximum tumor diameter(cm)under CT and CT diagnostic LNM status,pathological and molecular biological features included the tumor cell differentiation degree,lymph vascular invasion or not and STMN1 expression level.Based on the above factors,we established a new nomogram that could predict LNM in the neck of HSCC,and also established a CT diagnostic nomogram.In the internal validation,the area under the ROC curve(AUC)was 0.916[95%confidence interval(CI),0.833,1.000],and the sensitivity and specificity of predicting LNM were 82%and 84%,respectively,and were greater than the sensitivity(61%)and specificity(78%)of the CT diagnostic prediction model;the calibration curve showed Brier score was 0.079;the DCA curves showed higher net benefit than the CT diagnostic nomogram when applying this new prediction model when the threshold probability of LNM transfer was greater than 10%.In external validation,the AUC was 0.928(95%CI,0.864,1.000),and the sensitivity and specificity of predicting LNM were 92%and 93%,respectively;the calibration curve showed Brier score was 0.072;the DCA curve showed that the new nomogram provided a good net benefit when the threshold probability of LNM in patients was greater than 20%.CONCLUSIONS:The neck LNM of HSCC is influenced by various factors such as clinicopathological features and biomolecules.The nomogram established on the basis of CT diagnosis,combined with pathological and molecular expression features such as the maximum tumor diameter(cm)under CT,degree of tumor cell differentiation,lymph vascular invasion or not and STMN1 expression level,can greatly improve the sensitivity,specificity and accuracy of predicting the neck LNM of HSCC and has good clinical utility.