| Background and purpose : Intrahepatic cholangiocarcinoma(ICC)is a malignant tumor second only to hepatocellular carcinoma(HCC)in primary liver cancer(PLC).In recent years,the ICC incidence has been rising in the world,and the mortality rate is high.Currently,surgical resection is the only potential treatment method,however ICC lacks effective tumor markers,in addition to this HCC and ICC with primary liver cancer,it is easy to confuse.The current identification of HCC and ICC is still depending on the pathological examination after hepatectomy,but the contradiction is that accurate preoperative diagnosis is important for clinical decision-making.In view of the fact ICC diagnostics is large difficult,MRI,CT,and ultrasound have certain limitations,Laboratory indicators have relatively objective,easy access.In addition,the nomogram also has a certain advantage in disease diagnosis by visualizing complex equations and using intuitive graphics to quantify the contribution of different influencing factors to the outcome.Therefore,the purpose of this study is to develop and verify the ICC and HCC differential diagnosis nomogram prediction model based on patients’ basic conditions and laboratory indicators,in order to find a practical and easy to popularize prediction model.Method: In this study,the data of patients diagnosed as ICC and HCC by pathology from January 2017 to October 2020 were collected.The first 75% of the data is defined as the training cohort and the last 25% of the data as the validation cohort.Analysis of its general information,blood routine,coagulation function and tumor marker,then use the LASSO regression to reduce the design,and finally use binary logistics regression to screen the indicator of modeling.Establish a nomogram model as a model 1,and establish two conventional tumor marker joint models,respectively model 2(AFP + CA19-9),Model 3(CA19-9 + CEA).Three models were compared with the ROC curve and the clinical decision curve analysis(DCA)to compare 3 models in ICC and HCC diagnostic performance,And then Use the calibration curve judgment model 1(nomogram)fitting.Finally,the data of the validation cohort is further verified the repeatability of the three model.Results:1.The total number of cases included in the study was 1068,After univariate analysis,LASSO regression and binary Logistics regression analysis,the nomogram was established by gender,NLR,AFP,CA15-3,CA19-9 and CEA as model 1.2.When the total points of the nomogram exceed 160 points,the probability of differential diagnosis of ICC is 50%.When the total points reache300 points,the probability of differential diagnosis of ICC is 99%.3.The calibration curves show that Model 1(nomogram)has good consistency among the modeling groups.The area AUC under C-index and ROC curve was used as the diagnostic efficiency index to evaluate the differential diagnosis of ICC and HCC,C-index and AUC of model 1(nomogram)are 0.899 and 0.900,respectively,which are superior to model2(AFP+CA19-9)and model 3(CA19-9+CEA),Clinical decision curve analysis(DCA)showed that Model 1 had more net benefits than other models.4.Model 1 was also superior to other models by using the training cohort’s own data and the validation cohort’s data for verification.And The calibration curves in the validation cohort also indicated that Model 1 had good consistency.Conclusion:1.Model 1(including gender,NLR,AFP,CA15-3,CA19-9,CEA)combines laboratory indicators and gender,quantifies the contribution of each index to differential diagnosis,each index gradient has its corresponding points,and the total points can quantify the probability of differential diagnosis of ICC.The result is more intuitive,which is more helpful to the clinical differential diagnosis of ICC and HCC.2.The model 1(nomogram)established in this study has good consistency in the training cohort and verification cohort,and in the differential diagnosis of ICC and HCC,the diagnostic efficiency is better than the conventional model 2(AFP+CA19-9)and model 3(CA19-9+CEA). |
PDF Full Text Request