Development and validation of nomogram for predicting survival outcomes in patients with small hepatocellular carcinoma

• Main Authors: LI Chuanhong, HU Peng, OU Yanjiao, WANG Hong, ZHANG Leida
• Format: Article
• Language: zho
• Published: Editorial Office of Journal of Third Military Medical University 2020-10-01
• Series: Di-san junyi daxue xuebao
• Subjects: small hepatocellular carcinoma; prediction model; nomogram; overall survival; cancer-specific survival
• Online Access: http://aammt.tmmu.edu.cn/Upload/rhtml/202007005.htm

Objective To develop nomograms for predicting the overall survival (OS) and cancer-specific survival (CSS) of patients with small hepatocellular carcinoma (HCC). Methods We retrieved the data of patients with small HCC diagnosed between 2000 and 2014 from the Surveillance, Epidemiology, and End Results (SEER) database, and then randomized the eligible patients into training group (n=958) and validation group (n=958). The prognosis-related risk factors were selected using univariate Cox regression analysis, and the independent prognostic factors of OS and CSS were identified using a lasso Cox regression model. Akaike's information criterion (AIC) was used to evaluate the quality of fit of the models. The nomograms for predicting 1-, 3- and 5-year OS and CSS of the patients were constructed based on the identified prognostic factors, and their prediction ability was evaluated using the concordance index (C-index), receiver operating characteristic (ROC) curve and calibration curve in both the training and validation cohorts. Results In the overall cohort, the 1-, 3-, and 5-year probabilities of OS rates were 81.3%, 60.7% and 49.2%, and the 1-, 3- and 5-year CSS rates were 85.2%, 66.9% and 56.9%, respectively. Six independent prognostic factors related to OS were selected to develop the OS nomogram, and 6 independent prognostic factors and 1 CSS-related factor were selected to develop the CSS nomogram. In the training group, the C-index for OS nomogram was 0.727 (95%CI: 0.705-0.750) and the areas under the ROC curve (AUCs) for 1-, 3- and 5-year OS were 0.754, 0.783 and 0.795 respectively. For the CSS nomogram, the C-index was 0.753 (95%CI: 0.729-0.777) and the AUCs for 1-, 3- and 5-year CSS were 0.801, 0.806 and 0.818, respectively. In the validation group, the C-index for OS nomogram was 0.725 (95%CI: 0.702-0.749) with AUCs for 1-, 3- and 5- year OS of 0.752, 0.770 and 0.799 respectively; The C-index for CSS nomogram was 0.772 (95%CI: 0.748-0.795) with AUCs for 1-, 3- and 5-year CSS of 0.806, 0.815 and 0.837 respectively. The calibration curves for the probabilities of 1-, 3-, and 5-year OS and CSS showed a good agreement between the nomogram-predicted survival and the actual survival outcomes in both the training and validation cohorts. We calculated the risk score for every patient according to the nomograms and accurately stratified the patients into high- and low-risk groups based on the median risk score. Conclusion The nomogram we established can accurately predict the OS and CSS of the patients with small HCC to assist the surgeons for clinical decision-making.