Clinical Hematological-based Nomograms For Prediction Of Lymph Node Metastasis And Prognosis In Resectable Esophagogastric Junction Adenocarcinoma Patients

BackgroundEsophagogastric junction adenocarcinoma(EJA)is a common malignant tumor,which occurs between the lower esophagus and gastric cardia.In recent decades,the incidence rate of EJA has increased in western countries and East Asia,especially in China and Japan.Compared with esophageal and gastric cancer,the 5-year survival rate of EJA is relatively lower,which poses a great threat to the survival of patients.Lymph node metastasis is an important indicator of treatment strategy and prognosis evaluation of EJA patients.At present,preoperative evaluation of lymph node metastasis mainly depends on imaging techniques such as computed tomography and ultrasound,but their diagnostic coincidence rates are low.At the same time,TNM staging system based on anatomy has been proven to be limited in predicting the prognosis of patients.In recent years,increasing studies have found that the clinical hematological indicators are related to the lymph node metastasis and the survival of cancer patients,which are expected to be used as markers to predict the status of lymph node and prognosis of cancer patients.ObjectivesThe purpose of this study was to explore the value of preoperative clinical laboratory blood indicators in predicting lymph node metastasis and evaluating the prognosis of EJA patients.MethodsWe collected the preoperative complete blood cell count,biochemical and coagulation function indicators,demographic sociological indexes,TNM stages and follow-up data of 781 EJA patients who received surgical treatment from Shantou University Medical College Cancer Hospital.The patients were divided into training(465 cases from 2003 to 2010)and validation cohorts(316 cases from 2011 to 2017).First,in the training cohort,using Lasso regression analysis,52 indexes including demographic sociological data and hematological parameters were used to screen out the significant indicators to construct the Hematological score nomogram(H-score nomogram)to predict lymph node metastasis.Furthermore,the significant variables were screened from the above 52 indicators by univariate and multivariate Cox regression analysis to build another nomogram(OS nomogram)to predict the overall survival(OS)of EJA patients.Then the Harrel consistence index(C-index)was used to evaluate the discriminant performance of the two nomograms.Moreover,the net reclassification improvement(NRI)and the integrated discrimination improvement(IDI)were applied to evaluate the degree of the improvement of OS nomogram in predicting OS of EJA patients,compared with TNM staging.At the same time,the calibration curves were created by Bootstrap self-sampling,marginal estimation and model average prediction probability.In addition,the OS nomogram scores of the patients were calculated for risk stratification,and Kaplan-Meier survival curves were used to analyse the risk classification of prognosis.Finally,the above results were verified in the validation cohort.ResultsIn the first section,we used Lasso regression to construct H-score for predicting lymph node metastasis,which contained seven variables: platelet count(PLT),neutrophil ratio,fibrinogen,alkaline phosphatase(ALP),glucose,C-reactive protein to albumin ratio and systemic immune inflammation score(SII).In the training and validation cohorts,the C-index of H-score were 0.684(95% CI: 0.629 ～ 0.738)and0.630(95% CI: 0.556 ～ 0.703),respectively.Moreover,the calibration curves showed that the accuracy of H-score nomogram in predicting the positive probability of lymph node metastasis was good.In the second section,the OS nomogram to predict the OS of EJA patients included the following 12 indicators: age,body mass index,PLT,aspartate aminotransferase-alanine transaminase ratio,ALP,albumin,uric acid,immunoglobulin A,immunoglobulin G,complement 3,complement factor B and SII.In the training cohort,the C-index of OS nomogram was 0.652(95% CI: 0.620 ～0.684),higher than that of TNM staging system(0.623,95% CI: 0.598 ～ 0.647;p =0.044),and the C-index of the combination of OS nomogram and TNM staging was0.709(95% CI: 0.680 ～ 0.737).Both NRI and IDI suggested that it could significantly improve the predictive ability of TNM stage(NRI = 0.300,IDI = 0.114,p < 0.001).In the validation cohort,the C-index of the combination of OS nomogram and TNM staging was 0.674,better than that of TNM staging system(0.616;p = 0.005).And it could also improve the predictive ability of TNM staging system(NRI = 0.265,IDI =0.054,p < 0.05).The calibration curves showed that the accuracy of OS nomogram in predicting 5-year OS matched the actual situation of patients.Then,we divided the patients into high-risk group and low-risk group according to the OS nomogram scores.The result showed that compared with the low-risk group,the 5-year OS of the high-risk group was worse(training cohort: 59.6% vs.26.2%;validation cohort:58.6% vs.38.6%;all p < 0.0001).ConclusionThe H-score nomogram and OS nomogram constructed based on the clinical hematological indicators are the potential lymph node status and survival prediction models of EJA,which may help to predict the probability of lymph node metastasis and the expected OS rate of EJA patients before operation.