Prediction of microvascular invasion in hepatocellular carcinoma using a preoperative serum C-reactive protein-based nomogram

Abstract

Microvascular invasion (MVI) diagnosis relies on postoperative pathological examinations, underscoring the urgent need for a novel diagnostic method. C-Reactive Protein (CRP), has shown significant relevance to hepatocellular carcinoma (HCC) prognosis. This study aims to explore the relationship between preoperative serum CRP levels and microvascular invasion in hepatocellular carcinoma and develop a nomogram model for predicting MVI. Patients were categorized into MVI-positive and MVI-negative groups for analysis. Serum CRP levels were compared between the two groups. And then use LASSO regression to screen variables and build a nomogram. CRP levels showed significant differences between the MVI-positive and MVI-negative groups. Multivariable logistic regression analysis identified CRP (OR = 4.85, P < 0.001), lnAFP (OR = 3.11, P < 0.001), WBC count (OR = 2.73, P = 0.003), and tumor diameter (OR = 2.38, P = 0.01) as independent predictors of MVI. A nomogram based on these variables showed good predictive performance in both the training and validation cohorts with dual validation. The clinical prediction nomogram model, which includes serum CRP levels, WBC count, tumor diameter, and serum AFP levels, showed good performance in predicting MVI in both the training and validation cohorts.

Keywords: C-Reactive protein; Clinical prediction model; Hepatocellular carcinoma; Microvascular invasion.

Fig. 1

Flow diagram of the final diagnosis of patient enrollment.

Fig. 2

Diagnostic potential of CRP: (a) ROC curve of CRP in 3 cohorts. All: all patients set, Train: training set, Valid: validation set; (b) Histogram of difference analysis of CRP in 3 cohorts.

Fig. 3

Correlation diagnosis: (a) Heatmap of correlations between variables; Screening of variables based on Lasso regression: (b) The variation characteristics of the coefficient of variables; (c) The selection process of the optimum value of the parameter λ in the Lasso regression model by cross-validation method.

Fig. 4

Nomograph model constructed based on independent risk factors.

Fig. 5

Comprehensive evaluation of the model: (a,b) Discriminative ability of the model, ROC curve; (c,d) Evaluation of the fitting ability of the model, Calibration curves; (e,f) Evaluation of clinical decision-making ability of the model, DCA curve.