Prognostic Factors Analysis And Prognostic Model Construction Of Epithelioid Hemangioendothelioma

Part I Prognostic factors analysis of epithelioid hemangioendothelioma based on SEER databaseObjective:This study was reported one case of pulmonary epithelioid hemangioendothelioma,and then to the "epithelioid hemangioendothelioma(EHE)" diagnosed of inclusion criteria,using the "Surveillance,Epidemiology,and End Results" database to analysis the latest incidence,epidemiological characteristics and survival and prognosis risk factors of EHE.And also provide data support for the subsequent study.Methods:Reported a confirmed case of pulmonary epithelioid hemangioendothelioma admitted to the department of oncology,The First Affiliated Hospital of Dalian Medical University.We collected about 512 patients of clinical datas who diagnosed with EHE from the SEER database between 1986 and 2018.Sex,age at diagnosis,race,primary site,tumor grade and stage,treatment(surgery,chemotherapy and radiotherapy),survival time and vital status were used for survival and prognosis analyses.The prognostic indicators of different groups were analyzed by the log-rank test and COX regression models.Results:A total of 512 EHE patients were identified from the databases.The median overall survival(OS)was 100 months.The 1-,5-and 10-year OS rates were 76.5%,57.4% and 48.2%,respectively,while the cancer-specific survival(CSS)rates were77.3%,59.0% and 52.4%,respectively.There were no significant difference between the OS and CSS.Six factors with statistical difference(P<0.05)were took in the COX regression model for OS and CSS prognostic analysis,including age,stage,degree of tissue differentiation,primary site,surgery,chemotherapy,radiotherapy.The multivariate analyses of OS and CSS showed that sex,race and primary site were not independent prognostic factors.Patients with distant metastasis or poor grade had a poor prognosis in terms of OS and CSS respectively.Surgery had a better prognosis,but chemotherapy and radiotherapy did not help patients achieve a longer survival time.Conclusion1.The patients with epithelioid hemangioendothelioma had a long survival period and the longest OS was more than 29.6 years.The 1-,5-and 10-year OS rates were76.5%,57.4% and 48.2%,respectively.2.Age,distant metastases,poor tissue differentiation,chemotherapy and radiotherapy were independent factors of poor prognosis,while surgery was an independent factor of good prognosis.3.There was no standard treatment for pulmonary epithelioid hemangioendothelioma,and conventional chemotherapy and other modalities were not effective.This case report suggested that anti-angiogenic targeted agents combined with immunotherapy may be a potential treatment option with acceptable disease control and no significant toxicities,but large sample studies were needed to confirm the safety and clinical effectiveness of this treatment option.Part II Construction and validation of a clinical prognostic model for epithelioid hemangioendothelioma based on the SEER databaseObjective:A line graph model for predicting 1-,5-and 10-year OS of patients with epithelioid hemangioendothelioma was constructed from independent prognostic factors affecting epithelioid hemangioendothelioma.Method A total of 512 patients diagnosed with epithelioid hemangioendothelioma from1986-2018 were collected from the SEER database.Gender,age at diagnosis,year at diagnosis,race,primary tumor site,tumor histologic grade,tumor stage,treatment modality(surgery,chemotherapy,and radiation therapy),survival time or last follow-up time,and survival status were statistically analyzed.A univariate analysis of the relationship between patient survival outcome(death)and clinical variables was used,and variables that were statistically different(p < 0.05)were included in a multi-factor COX regression analysis,and this was used to construct a column line graph survival prediction model.Bootstrap sampling was used to internally validate the model,and patient-related data were sampled 500 times in duplicate to obtain calibration curves.The accuracy of the column line graph was measured by the C-index index,the diagnostic power of the data was described by the ROC curve,and the ROC curve was quantified using the AUC values,and the clinical utility of the model was assessed by the DCA curve.Result The column line graph prediction model included six independent clinical features,namely age,degree of tissue differentiation,stage,whether or not surgical treatment was performed,whether or not chemotherapy was performed,and whether or not radiotherapy was performed.Its prediction accuracy was validated with a C-index of0.752 and 0.753 in both the training and validation sets,respectively,both >0.7,suggesting moderate prediction accuracy.The AUC predicted 1-,5-year and 10-year survival probability values of 0.771,0.811 and 0.830,0.828 and 0.874,0.846 respectively,all of which were approaching or greater than 0.8,thus the model had a high discriminatory power.The calibration curves suggested that the predicted survival rate was in good agreement with the actual survival rate.The decision curves showed a fairly large net benefit interval in predicting the prognosis of patients at 1,5 and 10 years in both the training and validation sets,indicating that the model’s predicted values benefit clinical decision making over a wide range.Therefore,the model was useful for clinical decision making.Conclusion The line graph model was a good predictor of 1-,5-year and 10-year survival in patients with epithelioid hemangioendothelioma with good consistency and accuracy,and can provide clinicians with survival and prognostic assistance in the personalized treatment of patients with epithelioid hemangioendothelioma.