Background: Early relapse after hepatectomy presents a significant challenge in the treatment of hepatocellular carcinoma (HCC). The aim of this study was to construct and validate a novel nomogram model for predicting early relapse and survival after hepatectomy for HCC.
Patients and methods: We conducted a large-scale, multicenter retrospective analysis of 1,505 patients with surgically treated HCC from 4 medical centers. All patients were randomly divided into either the training cohort (n=1,053) or the validation cohort (n=452) in a 7:3 ratio. A machine learning-based nomogram model for prediction of HCC was established by integrating multiple risk factors that influence early relapse and survival, which were identified from preoperative clinical data and postoperative pathologic characteristics of the patients.
Results: The median time to early relapse was 7 months, whereas the median time from early relapse to death was only 19 months. The concordance indexes of the postoperative nomogram for predicting disease-free survival and overall survival were 0.741 and 0.739, respectively, with well-calibrated curves demonstrating good consistency between predicted and observed outcomes. Moreover, the accuracy and predictive performance of the postoperative nomograms were significantly superior to those of the preoperative nomogram and the other 7 HCC staging systems. The patients in the intermediate- and high-risk groups of the model had significantly higher probabilities of early and critical recurrence (P<.001), whereas those in the low-risk group had higher probabilities of late and local recurrence (P<.001).
Conclusions: This postoperative nomogram model can better predict early recurrence and survival and can serve as a useful tool to guide clinical treatment decisions for patients with HCC.
Keywords: Early relapse; Hepatectomy; Hepatocellular carcinoma; Nomogram; Postoperative.