Analysis Of Risk Factors For Lymph Node Metastasis In Patients With Superficial Esophageal Carcinoma And Early Gastric Cancer And Establishment Of Prediction Models

Part I Analysis of Risk Factors for Lymph Node Metastasis in Patients with Superficial Esophageal Carcinoma and Establishment of Prediction ModelBackgroundEsophageal carcinoma is a common malignancy originating from esophageal epithelium,including squamous carcinoma and adenocarcinoma.The morbidity and mortality of esophageal carcinoma were the eighth and fifth respectively in malignant neoplasms worldwide.Esophageal cancer occurs frequently in China,with squamous carcinoma more than 90%.In 2015,there were estimated 470,000 new cases and 370,000 deaths of esophageal cancer in China,which were the highest in the world,the third and fourth in Chinese malignant tumors respectively.The definition of early esophageal cancer(EEC)was updated with the development of medicine.In the past,superficial esophageal carcinoma(SEC)was referred to as EEC.At present,EEC,generally recognized in China,is defined as esophageal cancer,which invades mucosal layer and submucosal layer,without lymph node metastasis(LNM).However,SEC is defined as an esophageal carcinoma confined to the mucosal layer and submucosal layer,whether or not with LNM.In addition to surgical and lymph node dissection,endoscopic treatment,including multiband mucosectomy(MBM),endoscopic mucosal resection(EMR)and endoscopic submucosal dissection(ESD),was gradually accepted for SEC without LNM.Endoscopic treatment had so many advantages,such as less trauma,fewer complications,and less costs.In addition,it could preserve the integrity of the esophagus and have a small impact on patients' survival quality.The 5-year survival rate of patients with endoscopic treatment was not lower than those with surgical treatment.The endoscopic treatment is an option for SEC without LNM.When the SEC occurred with LNM,lesions surgical resection and regional lymph node cleaning normally should be performed in accordance with the therapy of advanced esophageal cancer,because endoscopic treatment couldn't completely remove the lesions on account of inability to accomplish lymph node dissection.Therefore,it is very important to determine LNM whether or not in SEC to select the appropriate therapy method.Nowadays,endoscopic resection is mainly used to complete lesion resection in SEC patients with low risk of LNM.Due to the limitations of preoperative endoscopic ultrasonography(EUS),computed tomography(CT)scan,and other relevant imaging technologies in the assessment of infiltration depth and LNM,the analysis of postoperative clinical pathological characteristics is still the most effective method to forecast LNM in SEC.Endoscopic treatment is very popular at present,and the follow-up treatment and evaluation are essential for the prognosis of patients.ObjectiveWe aimed to establish a prediction model for LNM in SEC by analyzing the related risk factors to provide a guide for the choice of treatment.Furthermore,we aimed to evaluate whether additional surgery or chemoradiotherapy is required after endoscopic treatment.MethodsWe retrospectively analyzed the postoperative pathological reports of patients with esophageal cancer undergone thoracic surgery in Shandong Provincial Hospital Affiliated to Shandong University from March 2010 to March 2017.A total of 128 cases of SEC were involved with pathological stages of Tis or T1.Tumors' infiltration depth was divided into 6 groups:m1,m2,m3,sml,sm2 and sm3 after accurate measurement.We combined age,sex,tumor size,lesion location,macroscopic type,differentiation type,infiltration depth,lymphovascular invasion,with the condition of lymph node metastasis to measure the related risk factors and to explore the correlation between risk factors and LNM.ResultsIn this study,128 patients with SEC were enrolled.Among them,19 cases had LNM,with a LNM rate of 14.84%.The LNM rates in SEC were 0%,0%,4.0%,1034%,26.09%and 40.91%,respectively,for the infiltration depths of ml,m2,m3,sm1,sm2 and sm3.The LNM rates were 0%,19.23%and 32.14%,respectively corresponding to high differentiation,medium differentiation and low differentiation.Besides,the LNM rate for SEC was 34.21%in positive lymphovascular invasion patients,but 6.67%in the negative.The LNM rates in tumor diameter less than and equal to 10mm,between 11 to 20mm,between 21 to 30mm,and above 30mm groups were 0%,13.94%,15.63%,and 47.06%,respectively.We found LNM was associated with depth of invasion,differentiation type,tumor size,lymphovascular invasion,and macroscopic type by univariate analysis.Multivariate logistic regression analysis showed that infiltration depth,differentiation type and lymphovascular invasion were independent risk factors for LNM,according to which,a prediction model for LNM of SEC was established,P = ex/1+ex,-8.948+1.018*differentiation type +0.656*infiltration depth + 1.472*lymphovascular invasion,to calculate LNM probability.The area under ROC curve was 0.853(95%CI:0.762-0.945).Conclusions1.The infiltration depth,differentiation type,lymphovascular invasion,macroscopic type and tumor size were significantly related to LNM in SEC;meanwhile the former three were independent risk factors.2.We established a prediction model of LMN in SEC,with P = ex/1+ex,x ?-7.537+1.018*differentiation type +0.656*infiltration depth +1.472*lymphovascular invasion.The area under ROC curve was 0.853(95%CI:0.762-0.945).3.Prior treatment,the model could be applied to calculate the probability of LNM in SEC.The larger of P value,the higher of the risk of LNM,surgical operation was suggested.On the contrary,the smaller of P value,the lower of the risk of LNM,endoscopic treatment was suggested.4.The model could also be used after endoscopic treatment.A large P value meaned a high risk of LNM,which suggested an additional surgery or chemoradiatherapy.A follow-up observation was recommended when the P value was smaller suggesting a lower risk of LNM.Part ? Analysis of Risk Factors for Lymph Node Metastasis in Patients with Early Gastric Cancer and Establishment of Prediction ModelBackgroundGastric cancer is a malignant tumor originating from the gastric epithelium,ranking fifth in the incidence and third in mortality of malignant tumors in the world.China had a high incidence of gastric cancer,and the number of new cases and deaths ranked first in the world each year.Early gastric cancer(EGC)is defined as carcinoma confined to the mucosa or submucosa of the stomach,independent of lymph node metastasis(LNM).Apart from gastrectomy with lymphadenectomy,endoscopic treatment including of endoscopic mucosal resection(EMR)and endoscopic submucosal dissection(ESD)have gained increasing popularity and have been widely regarded as an alternate treatment for some EGC patients.Compared with surgery,endoscopic treatment could retain the normal structure of the stomach with a less trauma,and greatly improve patients' quality of life.Nevertheless,endoscopic treatment technologies couldn't accomplish regional lymph node dissection,which were suitable for patients with EGC without LNM.Therefore,judging the presence or absence of LNM in EGC was of great importance in the decision of final treatment.To date,both domestic and foreign guidelines for the diagnosis and treatment of gastric cancer have pointed out that endoscopic treatment was mainly applied to patients with a complete resection of the lesion and low risk of LNM.At present,preoperative endoscopic ultrasonography(EUS),CT and other related imaging techniques have their own limitations in the assessment of invasion depth and LNM.The analysis of postoperative clinicopathological features is still the most effective method to predict LNM in EGC.Nowadays,endoscopic treatment is very popular,and the follow-up treatment and evaluation are essential for the prognosis of patients.ObjectiveIn this study,we aimed to establish a prediction model of LNM in EGC by analyzing the related risk factors,to guide the choice of treatment and assess whether additional surgery or radiotherapy and chemotherapy is required after endoscopic treatment.MethodsThe postoperative pathological reports of patients with gastric cancer who underwent surgical treatment in Shandong Provincial Hospital Affiliated to Shandong University from March 2010 to March 2017 were reviewed.A total of 286 EGC cases with pathological stages Tis or T1 were involved.The depth of tumor invasion was again accurately measured.They were divided into 6 groups of m1,m2,m3,sm1,sm2,and sm3 according to the depth of invasion.We studied the related risk factors combined with age,sex,tumor size,lesion location,macroscopic type,differentiation type,depth of invasion,lymphovascular invasion,cleared lymph node metastasis and explored the correlation between risk factors and LNM.Results1.In this study,a total of 286 patients with EGC were included.Among them,54 cases had LNM,with LNM rate of 18.88%.2.The LNM rates of EGC were 0%,320%,430%,10.40%,31.80%,and 62.50%corresponding to infiltration depths of m1,m2,m3,sml,sm2,and sm3.The high differentiation,moderate differentiation,and low differentiation had a LNM rate of EGC at 3.30%,12.80%,and 26.50%,respectively.The LNM of EGC rates were 34.60%in a tumor diameter above 2cm,9.5%in tumor diameter less than or equal to 2cm,37%in positive lymphovascular invasion and 17%in negative lymphovascular invasion.3.Univariate analysis showed that tumor size,depth of invasion,differentiation type and lymphovascular invasion were significantly correlated to LNM.Multivariate logistic regression analysis showed that infiltration depth,differentiation type,and lymphovascular invasion were independent risk factors for LNM in EGC.4.We established a prediction model of LNM in EGC,P=ex/1+ex,x=-8.948+0.772*differentiation type+1.221*depth of invasion+1.447*lymphovascular invasion to calculate the probability of LNM.The ROC curve was used to validate the model prediction value.The area under the curve(AUC)was 0.878 and the asymptotic 95%confidence interval(95%CI)was 0.827-0.928.Conclusions1.LNM in EGC was significantly related to tumor size,depth of invasion,differentiation type,and lymvascular invasion,which were independent risk factors except tumor size.2.We established a prediction model for the prediction of LNM in EGC:P=ex/1+ex,x=8.948+0.772*differentiation type+1.221*depth of invasion +1.447*lymphovascular invasion.The area under the curve(AUC)was 0.878 and the asymptotic 95%confidence interval(95%CI)was 0.827-0.928.3.Preoperatively,this model was used to calculate the probability of LNM in EGC.The higher the P value,the higher the risk of LNM,and it was recommended that surgical treatment be performed.The smaller the P value,the lower the probability of lymph node metastasis.Then,endoscopic treatment was recommended.4.After endoscopic therapy,this model could also be used.Similarly,the higher the P value,the higher the risk of LNM,additional surgery or radiotherapy and chemotherapy should be considered.The smaller the P value,the lower the probability of lymph node metastasis,follow-up was recommended.