RESUMO
Objective:To establish a convenient preoperative nomogram prediction model for early diagnosis of hepatocellular carcinoma (HCC)with microvascular invasion (MVI), and to evaluate the model through internal and external validations for use informulating reasonable and individualized treatment strategies for patients with early-staged HCC.Methods:The clinical data of 294 patients who underwent hepatectomy at the General Hospital of Ningxia Medical University from January 2017 to December 2020 were retrospectively collected and analyzed. Based on the different admission times, they were divided into the training group ( n=231) and the validation group ( n=63). Based on the results from previous published literatures and our relevant clinical experience, risk factors including γ-glutamyltranspeptidase (GGT), platelet-lymphocyte ratio (PLR), fibrinogen albumin ratio (FAR), lymphocyte monocyte count ratio (LMR) and ALT-platelet ratio (APRI) were subjected to multi-factor logistic regression analysis to determine independent risk factors of HCC with MVI, and a nomogram prediction model was then constructed. The validation group was applied to the model for validation. Results:Of 294 patients who were enrolled in this study, there were 231 patients in the training cohort, with an average age of (55.1±10.9) years. In the training group, 95 patients were MVI positive and 136 patients were MVI negative. In the validation group, 38 patients were MVI positive and 25 patients were MVI negative. Logistic regression analysis showed that FAR>0.06, GGT>50 U/L, APRI>0.16, tumor diameter>5 cm, LMR>3.57 and PLR>98.75 were independent risk factors ( P<0.05), and a nomogram prediction model was established. The correction curve of the nomogram showed that the actual prediction result was close to the ideal result of the prediction model. The internal validated results showed the C-indexes to be between 0.71 and 0.90, and the prediction model had good discrimination. DCA curve was used to evaluate the clinical net benefit of the predictive model. When the net benefit rate was above zero, the threshold of the prediction model was 4%-77%, indicating that the prediction model had good clinical practicability. Conclusion:The established nomogram prediction model based on preoperative clinical indexes of GGT, APRI, LMR, PLR, FAR and diameter of tumor could be used to predict early diagnosis of HCC with MVI. The nomogram has good clinical application values.