Construction of a predictive nomogram and bioinformatic investigation of the potential mechanism of postoperative early recurrence of hepatocellular carcinoma meeting the Milan criteria

BACKGROUND: Hepatectomy is the most common treatment for hepatocellular carcinoma (HCC) meeting the Milan criteria; however, postoperative early recurrence (PER) compromises the survival time. This study aimed to construct a predictive nomogram for PER of HCC patients within the Milan criteria. And the underlying mechanism related to PER may associate with the independent risk factors used to construct the nomogram, therefore, we preliminarily investigated the potential mechanism of PER using The Cancer Genome Atlas (TCGA) database to provide an idea for preventing PER. METHODS: Patients with HCC meeting the Milan criteria receiving hepatectomy in our center between 2009 and 2015 were enrolled. The clinical and histological data of all participants were collected. Follow-up was performed at outpatient and PER was defined as recurrence within 2 years after resection. All participants were randomly assigned to the training or validation cohort at a 4:1 ratio. A nomogram was constructed based on the independent risk factors in the training cohort. The accuracy and clinical utility of this nomogram were evaluated using the C-index, calibration plot, and decision curve analysis (DCA). The differentially-expressed genes (DEGs) between early-stage HCC patients with and without PER in TCGA database were identified. Enrichment analysis was performed to determine the potential relapse-related mechanism. RESULTS: The independent risk factors were alpha-fetoprotein (AFP) ≥400 ng/mL, gamma-glutamyl transpeptidase (GGT) ≥60 U/L, Glisson’s capsule invasion, microvascular invasion (MVI), and satellite lesions. The C-index value of the nomogram was 0.693 [95% confidence interval (CI): 0.632–0.754; P<0.001] in the training cohort and 0.658 (95% CI: 0.529–0.787; P=0.016) in the validation cohort. The calibration and decision curves demonstrated good accuracy and clinical utility of this nomogram respectively. 133 DEGs were identified and enrichment analysis showed the bile secretion pathway related to PER and two bile secretion pathway-related genes {ATP1A2 [P=0.027; hazard ratio (HR) =2.086, 95% CI: 0.916–4.749] and SLC5A1 (P=0.0016; HR =0.361, 95% CI: 0.145–0.898)} were significantly associated with disease free survival (DFS). CONCLUSIONS: Our nomogram has satisfactory accuracy and clinical utility in predicting the PER of patients with HCC meeting the Milan criteria. Aberrant bile secretion may be an important mechanism of PER.