Construction And Validation Of The Clinical Prediction Models For Deep Vein Thrombosis,Metastasis And Recurrence In Patients With Hepatocellular Carcinoma After Hepatectomy

Background: The incidence of deep vein thrombosis(DVT)is unclear in hepatocellular carcinoma(HCC)patients after laparoscopic hepatectomy,and no effective technique to assess DVT risk in these patients.Methods: 355 consecutive HCC patients,including the training set with 243 and the validation set with 112,who underwent laparoscopic hepatectomy were identified.A DVT risk algorithm was developed in the training set,and its predictive performance was evaluated in both the two sets.The DVT risk nomogram based on the model was created.Results:In the present study,the incidence rate of DVT was 18.6%,including 40(60.6%)unilateral distal DVT,24(36.4%)bilateral distal DVT,2(3.0%)unilateral proximal DVT and 0 bilateral proximal DVT.No pulmonary embolism(PE)occurred.The model consisted of age,sex,body mass index(BMI),comorbidities and operative position,which were independent risk factors of DVT in the training set.The model had a good predictive ability with area under the receiver operating characteristic curve(AUC)0.861 with 95%confidence intervals(CI)0.802-0.920,H-L test p value 0.626,the optimal cut off 0.280,sensitivity of 44.4% with 95% CI 30.0%-59.9%,specificity of 96.5% with 95% CI92.6%-98.4%,positive predictive value(PPV)of 74.1% with 95% CI 53.4%-88.1%,negative predictive value(NPV)of 88.4% with 95% CI 83.2%-92.2% and the accuracy of86.8% with 95% CI 82.5%-91.1% in the training set;and AUC of 0.818 with 95% CI0.716-0.919,H-L test p value 0.259,sensitivity of 38.1% with 95% CI 19.0%-61.3%,specificity of 98.9% with 95% CI 93.2%-99.9%,PPV of 88.9% with 95% CI 50.7%-99.4%,NPV of 87.4% with 95% CI 79.0%-92.8% and the accuracy of 87.5% with 95% CI81.4%-93.6% in the validation set.The DVT risk nomogram performed well in both the internal validation and the external validation,and had a good clinical application value.The DVT risk nomogram had a better predictive value of DVT than the Caprini score(p <0.001).Conclusion: The incidence of DVT was high and should not be neglected in HCC patients after laparoscopic hepatectomy.The DVT risk nomogram provides an efficacious method for DVT risk evaluation and selection of pharmacologic thromboprophylaxis.Background: At present,the high recurrence rate and the extrahepatic metastasis in hepatocellular carcinoma(HCC)patients after hepatectomy are important reasons for poor prognosis.However,there are few effective tools to predict the risk of recurrence and metastasis.Methods: Consecutive Barcelona clinic liver cancer(BCLC)0-A stage HCC patients who underwent R0 hepatectomy at the Liver Surgery Center of Tongji Hospital Affiliated to Tongji Medical College of Huazhong University of Science and Technology from January 1,2014 and December 31,2016 were identified,and they were divided into extrahepatic and non-extrahepatic metastasis groups based on whether the site of the first recurrence contained an extrahepatic organ.Meanwhile,the non-extrahepatic metastasis group was divided into early recurrence and non-early recurrence groups based on whether the time of first recurrence was longer than two years after surgery.Then,on the basis of preoperative clinical datas,the models for predicting the risk of early recurrence and extrahepatic metastasis were developed by the Logistic regression,and the risk nomograms were developed as well and their predictive performance was evaluated.Finally,the clinical decision curve was used to evaluate the clinical application value of the risk prediction model.Results:351 consecutive HCC patients,including extrahepatic metastasis group with 36(10.3%)and non-extrahepatic metastasis group with 315(89.7%)(early recurrence group with 120(38.1%)and non-early recurrence group with 195(61.9%)).By the Logistic regression,the independent risk factors for extrahepatic metastasis were identified,including alpha fetoprotein(AFP)grade(p = 0.002,odds ratio(OR)= 1.580,95% confidence intervals(CI): 1.182-2.144),tumor diameter(p = 0.039,OR = 1.103,95%CI: 1.002-1.209)and tumor necrosis(yes)(p = 0.034,OR = 2.379,95%CI: 1.042-5.222).And the independent risk factors for early recurrence were identified,including: habitual drinking(yes)(p = 0.019,OR = 1.994,95%CI: 1.123-3.561),viral hepatitis(yes)(p= 0.004,OR = 4.480,95%CI: 1.736-13.707),Ln AFP(p = 0.039,OR = 1.094,95%CI: 1.005-1.191),number of tumors(multiple)(p < 0.001,OR = 6.468,95%CI:2.702-16.543),tumor diameter(p = 0.001,OR = 1.170,95%CI: 1.082-1.271)and tumor capsule(yes)(p = 0.004,OR = 0.448,95%CI: 0.258-0.773).Subsequently,the risk prediction models and nomograms were constructed based on these independent risk factors.And the models had a good predictive ability: the area under the receiver operating characteristic curve(AUC)was 0.740 with 95%CI 0.663-0.816,the optimal cut off 0.095,and H-L test p value 0.653 for the extrahepatic metastasis risk model,and AUC was 0.754 with 95%CI 0.700-0.808,the optimal cut off 0.560,and Hosmer-Lemeshow(H-L)test p value 0.096 for the early recurrence risk model.Overall,the models had good predictive ability.Conclusion: The independent risk factors for extrahepatic metastasis in BCLC 0-A stage HCC patients after R0 hepatectomy included AFP grade,tumor diameter and tumor necrosis.And habitual drinking,viral hepatitis,Ln AFP,number of tumor,tumor diameter and tumor capsule were independent risk factors for early recurrence.The extrahepatic metastasis risk model and early recurrence risk model based on the independent risk factors showed good ability to identify high risk patients.With the help of the models,aggressive interventions for the high risk patients might further improve the prognosis.