Construction and validation of a nomogram for hepatocellular carcinoma patients based on HCC-GRIm score

Abstract

Purpose: To construct a nomogram for hepatocellular carcinoma (HCC) patients base on HCC-GRIm score.

Methods: Clinical cases of HCC patients diagnosed at Hunan Integrated Traditional Chinese and Western Medicine Hospital were included, and these were randomly divided into the training cohort (n = 219) and the validation cohort (n = 94), and those patients were divided into low GRIm-Score group (scores 0, 1, and 2) and high GRIm-Score group (scores 3, 4, and 5). In the training cohort, independent risk factors were determined by Cox regression analysis, and a nomogram was constructed by independent risk factors. The efficiency and the clinical applicability of nomograms were evaluated using ROC curves, calibration plot, and the decision curve (DCA), and the patients were divided into high-risk, middle-risk, and low-risk groups according to total score of nomogram.

Results: Compared to low HCC-GRIm score group, high HCC-GRIm score group with BCLC stage is more advanced (P < 0.001), and fewer patients received TACE (P = 0.005) and surgical treatment (P = 0.001). There was higher rate of the presence of vascular invasion (P < 0.001) and distant metastasis (P < 0.001). Multivariate Cox regression analysis screened 4 independent risk factors to construct a nomogram of HCC patients, including HCC-GRIm score, BCLC stage, albumin-to-globulin ratio (AGR), and glutamyl trans-peptidase (GGT). The consistency index (C-index) of the nomogram of the training was 0.843 (0.832-0.854) and the validation was 0.870 (0.856-0.885). The time-dependent parameter showed the AUC values of the training cohort at 1, 3, and 5 years were 0.954 (95% CI 0.929-0.980), 0.952 (95% CI 0.919-0.985), and 925 (95% CI 0.871-0.979), while the AUC values of validation cohort at 1, 3, and 5 years were 0.974 (95% CI 0.950-0.998), 0.965 (95% CI 0.931-0.999), and 0.959 (95% CI 0.898-1.021). The calibration plot showed the nomogram fits well onto perfect curves, and the DCA curve showed the net benefit of the nomogram at a certain probability threshold is significantly higher than the net benefit of the BCLC stage at the same threshold probability. Finally, all patients were divided into high-risk, middle-risk, and low-risk groups based on the total score of nomogram, and it showed effectively to identify high-risk patients.

Conclusion: The nomogram constructed by the independent risk factors can predict the prognosis of HCC patients, providing an effective tool with clinical workers to evaluate the prognosis and survival time of HCC patients.

Keywords: HCC-GRIm score; Hepatocellular carcinoma; Nomogram; Survival prognosis.

Fig. 1

Kaplan–Meier curves for OS stratified by HCC-GRIm score, a training cohort; b validation cohort

Fig. 2

Prognostic nomogram of HCC Patient

Fig. 3

a ROC curve of training cohort of 1-, 3-, and 5-year; b ROC curve of validation cohort of 1-, 3-, and 5-year. An AUC value of 0.5 indicates that the nomogram has no predictive effect. An AUC value of 1 indicates that the nomogram can completely distinguish patients with different survival rates. The larger the value between 0.5 and 1, the stronger the discriminant ability of the column chart

Fig. 4

a–c Calibration curve of training cohort of 1-, 3-, 5-year; d–f calibration curve of validation cohort of 1-, 3-, 5-year. The gray line represents an ideal reference line, and the closer the blue solid line is to the gray line, the more accurate the survival rate predicted by the model

Fig. 5

a–c DCA curve of training cohort of 1-, 3-, 5-year; d–f DCA of validation cohort of 1-, 3-, 5-year

Fig. 6

a Risk stratification of the training cohort; b Risk stratification of the validation cohort