Pretreatment risk management of a novel nomogram model for prediction of thoracoabdominal extrahepatic metastasis in primary hepatic carcinoma

• Published in: Journal of translational medicine Vol. 17; no. 1; p. 117
• Main Authors: Hu, Jia, Wang, Ting, Zhang, Kun-He, Jiang, Yi-Ping, Xu, Song, Chen, Si-Hai, He, Yu-Ting, Yuan, Hai-Liang, Wang, Yu-Qi
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 08.04.2019; BioMed Central; BMC
• Subjects: Abdomen; Alcohol; Algorithms; Antigens; Bacterial infections; Biomarkers; Blood; Calibration; Cancer; Cancer metastasis; Cancer therapies; Carbohydrates; Carcinoma; Care and treatment; Chinese medicine; Clinical Decision-Making; Colorectal cancer; Decision making; Female; Hepatitis; Humans; Individualized clinical decision-making; Infections; Liver; Liver cancer; Liver Neoplasms - pathology; Male; Medical imaging; Medical prognosis; Metastases; Metastasis; Methods; Middle Aged; Models, Biological; Neoplasm Metastasis; NMR; Nomogram; Nomograms; Nomography (Mathematics); Novels; Nuclear magnetic resonance; Patients; Population; Portal vein; Pretreatment risk management; Primary hepatic carcinoma; Prognosis; Risk Factors; Risk Management; ROC Curve; Sepsis; Thoracoabdominal extrahepatic metastasis; Thrombosis; Tumors; Weighing; Primary hepatic carcinoma; Thoracoabdominal extrahepatic metastasis; Pretreatment risk management; Nomogram; Individualized clinical decision-making
• ISSN: 1479-5876; 1479-5876
• DOI: 10.1186/s12967-019-1861-z

Extrahepatic metastasis is the independent risk factor of poor survival of primary hepatic carcinoma (PHC), and most occurs in the chest and abdomen. Currently, there is still no available method to predict thoracoabdominal extrahepatic metastasis in PHC. In this study, a novel nomogram model was developed and validated for prediction of thoracoabdominal extrahepatic metastasis in PHC, thereby conducted individualized risk management for pretreatment different risk population. The nomogram model was developed in a primary study that consisted of 330 consecutive pretreatment patients with PHC. Large-scale datasets were extracted from clinical practice. The nomogram was based on the predictors optimized by data dimension reduction through Lasso regression. The prediction performance was measured by the area under the receiver operating characteristic (AUROC), and calibrated to decrease the overfit bias. Individualized risk management was conducted by weighing the net benefit of different risk population via decision curve analysis. The prediction performance was internally and independently validated, respectively. An independent-validation study using a separate set of 107 consecutive patients. Four predictors from 55 high-dimensional clinical datasets, including size, portal vein tumor thrombus, infection, and carbohydrate antigen 125, were incorporated to develop a nomogram model. The nomogram demonstrated valuable prediction performance with AUROC of 0.830 (0.803 in internal-validation, and 0.773 in independent-validation, respectively), and fine calibration. Individual risk probability was visually scored. Weighing the net benefit, threshold probability was classified for three-independent risk population, which was < 19.9%, 19.9-71.8% and > 71.8%, respectively. According to this classification, pretreatment risk management was based on a treatment-flowchart for individualized clinical decision-making. The proposed nomogram is a useful tool for pretreatment risk management of thoracoabdominal extrahepatic metastasis in PHC for the first time, and may handily facilitate timely individualized clinical decision-making for different risk population.