| Backgrounds and objectiveEsophageal cancer is a malignant tumor with the seventh morbidity and the sixth mortality in the world.Although it has improved in tumor staging,the mortality of Chinese patients is about half of the whole world,making it a devastating disease.The preferred treatment of esophageal cancer is surgery,however,most patients have missed the best treatment time because of the obscure symptoms in the early stage,then concurrent CRT and combination chemotherapy are considered the standard treatment.Unfortunately,patients with esophageal cancer show a different treatment response even their pathological type,degree of differentiation,clinical and pathological staging and treatment are same,due to the heterogeneity.Radiomics can quantify tumor heterogeneity accurately and quantitatively,and have important predictive value in tumor staging,treatment response and prognosis.The purpose of this study was to investigate the potential relationship between texture parameters combined with hematological parameters and radiotherapy sensitivity of ESCC based on CT texture analysis,and establish multivariate prediction models to guide individualized treatment of ESCC finally.MethodsRetrospectively collected ESCC patients treated at Shandong Cancer Hospital Affiliated to Shandong University between January 2014 and December 2017.All patients had histological or pathological diagnosis by esophagoscopy and were received radiotherapy first time with total dose of 60Gy/59.4Gy and repositioned at the dose of 40Gy/41.4Gy.The clinical data and hematological parameters were complete.Patients underwent surgery on chest before include those who had esophageal stent and didn’t finish radiotherapy continuously were excluded.Summarize the clinical data,hematological parameters and CT images of localization and repositioning,and divided patients into two groups according to the Response Evaluation Criteria in Solid Tumors(RECIST1.1)with the repositioning CT images,and we considered CR and PR as responders,SD and PD as non-responders.LIFEx were used to analysis the CT images.Univariate analyses were used to explore the relationship between each parameter and sensitivity,and Multivariate analysis was used to establish the model of prediction of sensitivity of ESCC treated with radiotherapy.Use receiver operating characteristic(ROC)curve and area under the curve(AUC)to estimate the ability of classification about the model.ResultsAmong the 92 patients enrolled in our study,59 were responders and 33 were nonresponders.Univariate analyses showed that age,clinical stage,treatment mode and hemoglobin had significant difference between two groups(P<0.05).We used ROC curve to estimate the ability of each parameters,AUC(sex)was 0.602 and 95% CI was 0.495-0.703;AUC(clinical stage)was 0.686 and the 95%CI was 0.581-0.779;AUC(treatment mode)was 0.603 and 95%CI was 0.496-0.704;AUC(hemoglobin)was 0.675 and 95%CI was 0.570-0.769,it showed that the ability of each separate variable was poor.Take the variables which were found P <0.1 by univariate analyses into binary logistic regression with the way of “enter”,namely sex,age,clinical stage,treatment mode,drinking,hemoglobin,albumin,skewness and HGZE.The result showed that clinical stage,hemoglobin and skewness were independent predictors for the radio-chemotherapy sensitivity and the ROC curve indicated that AUC of the model was 0.886 with 95% CI of 0.803-0.943,the sensitivity was 94.9% and the specificity was 66.7%.The model has ability in predicting the sensitivity to some extent and it’s obvious that it’s more meaningful than each separate variable.ConclusionsThe model combined CT texture features and hematological parameters has certain ability in predicting sensitivity,concentrates on the prediction of radiotherapy sensitivity of ESCC Patients and help formulating individualized treatment plan,and then is beneficial to the survival and prognosis. |
PDF Full Text Request