Patient characteristics
A total of 124 patients were included in this study, the age range was 18-72 years, and the median age was 53 years, of which 72 (58.06%) patients were detected with MVI. There was significant difference between tumor size, T stage and MVI (p< 0.05). There was a higher proportion of MVI in HCC patients with tumor size ≥5 cm (74.13% vs. 43.93%, p=0.001) and T3-4 (72.85% vs. 38.89%, p<0.001). There was no significant difference between other clinicopathological features and MVI (p> 0.05) (Table 1).
18F-FDG PET/CT metabolic parameters
There were significant differences between SUVmax, SUVmin, SUVavg, SUVpeak, SUVratio, TLG, MTV and MVI. The HCC patients with high SUVmax (5.67 vs. 4.05, p<0.001), low SUVmin (0.45 vs. 0.72, p=0.031), high SUVmean (3.08 vs. 2.51, p=0.001), high SUVpeak (4.61 vs. 3.36, p=0.001), high SUVratio (1.70 vs. 1.29, p<0.001), high TLG (368.03 vs. 115.65, p=0.006), and high MTV (95.94 vs. 41.72, p=0.011) had a higher proportion of MVI (Table 2).
Multivariate analysis
The tumor size, T stage, SUVmax, SUVmin, SUVmean, SUVpeak, SUVratio, TLG and MTV were statistically analyzed. The results showed that tumor size (OR 0.375, p=0.018), T stage (OR 2.876, p=0.017), TLG (OR 0.998, p=0.023) and MTV (OR 1.009, p=0.015) were independent predictors of MVI. Meanwhile, SUVmax (OR 1.785, p=0.171), SUVmin (OR 0.640, p=0.356), SUVmean (OR 0.627, p=0.440), SUVpeak (OR 0.890, p=0.792), and SUVratio (OR 1.105, p=0.867) were not independent predictors of MVI (Table 3).
ROC curve
In the ROC curve for predicting MVI based on relevant parameters, the maximum Jordan index was taken as the cutoff point. The cutoff value of 18F-FDG PET/CT metabolic parameter TLG for predicting MVI was 0.332, and the sensitivity and specificity were 48.61% and 84.62%, respectively. The cutoff value of MTV for predicting MVI was 0.292, and the sensitivity and specificity were 54.17% and 75.00%, respectively. The incidence of MVI was higher in patients with TLG >400.67 and MTV >80.58 (Table 4, Figure 1 ).
Discuss
In our study involving 124 hepatocellular carcinoma (HCC) patients, we identified that 58.06% exhibited microvascular invasion (MVI), a critical prognostic factor for HCC and a significant predictor of post-transplant tumor recurrence. Univariate analysis revealed significant correlations between MVI and various parameters, including tumor size, T stage, SUVmax, SUVmin, SUVmean, SUVpeak, SUVratio, TLG, and MTV. Further, multivariate analysis confirmed that tumor size, T stage, TLG, and MTV are independent predictors of MVI. Our findings align with previous studies, which also noted the association between FDG uptake in HCC lesions and tumor aggressiveness, suggesting the potential of 18F-FDG PET/CT as a non-invasive method for detecting malignant tumors and assessing tumor histological features. Notably, higher TLG and MTV values were linked to a higher incidence of MVI, indicating their potential as predictive markers. Despite the retrospective design of our study, which may introduce selection bias, and the lack of a survival analysis due to a short follow-up duration, our results underscore the value of 18F-FDG PET/CT in predicting MVI before liver transplantation. We plan to continue with patient follow-ups, expand the sample size, and delve into deeper deep-learning studies to enhance the clinical management of HCC patients.