Establishment And Application Of Predictive Risk Model For Hepatocellular Carcinoma Based On The Combination Of Clinical Characteristics Of Hepatocytes And Liver Related Biomarkers

Background:Hepatocellular carcinoma(HCC)is the most common primary liver cancer,the sixth most common cancer in the world,and the second leading cause of cancer-related death.Early detection and early diagnosis can reduce the incidence of HCC.rate and case fatality rate.At present,there is no specific serological index or method to predict HCC,and early diagnosis of HCC remains a challenge,especially in China,where the morbidity and mortality of HCC are high.Objective:To find the influencing factors or indicators related to HCC,and to create a predictive model of hepatocellular carcinoma based on the combination of clinical characteristics and liver-related indicators,with a view to applying it to the clinic and benefiting patients.Methods:A retrospective study method was used to collect patients(n=2206)who visited our hospital from 2016 to 2020 as research objects according to the inclusion and exclusion criteria.The results of needle biopsy or surgical pathological diagnosis were used as the grouping criteria for the disease group and the benign control group in this study.The relevant information of the medical records was improved,and the patients were enrolled according to the time of admission.The patients from 2016 to 2019 were divided into a modeling group(n=1739),including 496 cases of HCC,1243 cases of benign liver diseases(liver hemangioma,liver cyst,liver abscess,liver hemangioma,liver cirrhosis,chronic hepatitis B,gallstones,cholecystitis,intrahepatic bile duct stones,etc).In 2020,the patients were divided into a validation group(n=467),including 156 cases of HCC and 311 cases of benign liver disease.(1)Descriptive statistical analysis was performed on the baseline data of the two groups,and whether there was a statistical difference between the two groups;(2)Correlation analysis was performed on the data of the two groups,and the correlation between the data was compared;(3)HCC was used as the The dependent variables were included in the logistic univariate and multivariate analysis to establish a HCC risk prediction model.(4)Based on the results of the HCC risk prediction model,the diagnostic performance and fit of the constructed model were evaluated by calibration,discrimination and Hosmer-Lemeshow(H-L)test.Result:1 The incidence of HCC in the modeling group and the validation group was basically the same(28.50%vs 33.41%,P>0.05).2 Logistic univariate regression analysis showed that the independent risk factors for HCC were:gender,age,alpha-fetoprotein(AFP),abnormal prothrombin(protein Induced by Vitamin K Absence or Antagonist-Ⅱ,PIVKA-Ⅱ),Gamma-glutamyltransferase isoenzyme(GGT),aspartate aminotransferase(AST),hepatitis B surface antigen(HBsAg).Among them,PIVKA-Ⅱ was most closely related to HCC,the partial regression coefficient was 1.879,and the corresponding OR value was 6.546;followed by AFP,HBsAg;AST,GGT again.PIVKA-Ⅱ,AFP,HBsAg,AST,and GGT were the five indicators most closely related to HCC.3 The risk prediction model constructed by Logistic multivariate regression analysis Logistic(P/1-P)=-7.115+1.879 X1+1.422 X2+1.537X3+1.115X4+1.133 X5+0.627 X6+0.051 X7-0.840 X8-1.464 X9-2.836 X 10,where PIVKA-Ⅱ(X1),AFP(X2),HbSAg(X3),GGT(X4),AST(X5),age(X6),gender(X7),TBA(X8),ALT(X9),TBIL(X10).4 Calibration degree of risk prediction model:After the HosmerLemeshow test,the χ2 value was 9.601,and P>0.05 had no statistical significance.It shows that the explanatory power of this model is no different from that of the saturated model,that is,the model fitting degree is better.5 Discrimination degree of risk prediction model:The area under the ROC curve(AUC)of the receiver operating curve characteristic of the modeling group was[0.960,95%CI:(0.950-0.971)],P<0.001 was statistically significant,when its model When the combined prediction Cut-off value was 0.8,the sensitivity of predictive diagnosis was 80.%and the specificity was 97.4%;the AUC of the validation group was[0.966,95%CI(0.945-0.986)],P<0.001 was statistically significant Significance,when the validation model jointly predicted a Cut-off value of 0.8,the sensitivity of the predictive diagnosis was 85.5%,and the specificity was 94.2%.6 The sensitivity and specificity of the model group were 76.6%and 97.5%respectively,and the accuracy of combined diagnosis was 93.5%;The diagnostic sensitivity and specificity of the validation group were 81.8%and 99.5%respectively,and the accuracy of combined diagnosis was 95.7%.Conclusion:1 Gender,age,AFP,PIVKA-II,TBIL,GGT,ALT,TBA,and HBsAg are independent risk factors for the diagnosis of HCC,while PIVKA-II,AFP,HBsAg,AST,and GGT are the most closely related to HCC patients5.an indicator.2 The HCC risk prediction model constructed based on the above independent risk factors in this study has good diagnostic performance and fit,and has strong diagnostic accuracy and clinical practicability.Further verification and calibration require long-term follow-up in the future.3 The model has certain guiding significance for predicting the early occurrence of HCC,and has strong clinical value and innovation.