The comprehensive treatment based on radical surgery is currently considered as the preferred and most effective method for HCC treatment[15, 16]. By eradicating the microscopic metastatic lesions around tumor to a certain extent, the promotion of the concept of anatomic hepatectomy has brought a very ideal surgical mean for the radical surgical treatment of HCC[17, 18]. Nevertheless, for the MVI positive patients, even after anatomic hepatectomy, the 5-year recurrence rate is still as high as 60–70%. Therefore, preoperative prediction of MVI is particularly important. In our study, first, the clinical outcomes and prognosis of clinical HCC patients with or without MVI were compared and analyzed; then, based on preoperative AFP, tumor diameter and TNM stage, a MVI prediction model with superior predictive efficacy was constructed and validated, which provided a new therapeutic strategy for clinical comprehensive and standardized treatment of HCC patients.
AFP is an important indicator of HCC screening, which can reflect the occurrence and development of HCC, which is positively correlated with the pathological HCC process[20, 21]. AFP was proved to be an independent risk factor for early recurrence and poor overall survival of HCC patients after hepatectomy, the association between MVI and AFP was always been concerned by researchers. Patients with AFP ≥ 400 kU/L were usually considered the possibility of diagnosis of HCC, and it was also an important risk factor for the early occurrence of HCC[23, 24]. However, A single haematological index is generally not highly specific and its clinical application is limited. Several studies have confirmed that in the growth process of tumors, the poorly differentiated areas will gradually replace the well-differentiated areas, thus increasing the degree of malignancy and invasiveness of tumors. Furthermore, large HCC tumor size stimulates invasive behavior, such as local, lymphatic and distant metastases. Anatomic hepatectomy has been shown to be significantly beneficial for long-term survival in patients with tumors > 5 cm in diameter, which also suggested that the occurrence of MVI was higher in tumors with larger diameter[27, 28]. The tumor diameter and TNM stage based on the imageological data also were independent risk factors of for postoperative MVI[29, 30]. The greater tumor size might be associated with capsular invasion, satellite nodules, and tumor thrombus, and the poor TNM stage might be relevant to the degree of aggressiveness and malignancy of the tumor[31, 32]. In our MVI prediction model, consist of the preoperative AFP ≥ 400 kU/L, TD ≥ 5cm and III-IV TNM stages, demonstrates reasonable composition structure.
Nowadays, radiometrics technology is used to extract predictive MVI model from CT images and preoperative MVI prediction of hematological indicators[33, 34]. Although some studies have found that special features of MRI can be used as typical features of MVI imaging diagnosis, such as image cystic insufficiency, coronary enhancement in arterial phase, and peritumoral low signal, etc[35, 36]. However, the process is quite complex, because radiomic feature extraction requires algorithms developed by scientists and engineers, and many algorithms are difficult to be recognized and put into clinical practice due to their overlapping or inadequate imaging. MVI status can also be reflected by specific clinical hematologic indicators including des-gamma carboxy prothrombin (DCP) and peripheral neutrophil to lymphocyte ratio (NLR), etc[37, 38]. Unfortunately, the predicted effect of preoperative hematological indexes is poor. In our prediction model, based on the risk factors of haematological and imageological data, with a high AUC of 0.7997 and a good H-L goodness of fit (P = 0.231), by successfully predicting the poor prognosis in high-risk patients, the prediction model can comprehensively screen tumor occurrence and invasiveness, so as to effectively preoperatively predict MVI for HCC patients.
In addition, our study proved that the HCC patients with poor differentiation resulted in a higher incidence of MVI compared with the patients with high/moderate differentiation, also leaded to a poor prognosis. Consistent with previous studies, poorly differentiated tumors were more aggressive than well-differentiated ones, so the prognosis of patients with poorly differentiated HCC after resection was worse than that of patients with highly differentiated, and the possibility of postoperative recurrence was increased[39, 40]. Although MVI positive has shown no significant difference in intraoperative and postoperative clinical indicators, while the difference of clinical prognosis can effectively guide the further comprehensive treatment strategy. MVI was an independent risk factor of HCC for poor prognosis after radical resection[41, 42]. Therefore, through accurate prediction of preoperative MVI, more reasonable and timely treatment options can be selected clinically, including radiofrequency ablation, anatomic hepatectomy or liver transplantation, and even the choice between neoadjuvant and adjuvant therapy can be extended. In addition, although targeted therapy for MVI is the first choice in the traditional sense, targeted therapy should also be an important systematic treatment method for MVI, so as to truly achieve personalized treatment for patients based on tumor biological behavior[43–45].
This study also has some limitations, some studies have found that the margin of the tumor to the surgical margin plane can significantly affect postoperative outcomes, but this study was not further discussed in the survival analysis. Although the sample size included in this study reached the standard of statistical analysis, it was necessary to carried out a randomized controlled study with large sample for analysis and research on the complex etiology of HCC recurrence. In addition, the follow-up time of this study was 36 months, so it is necessary to carry out longer follow-up of patients and analyze the mortality and recurrence rates at different stages, so as to further study the relationship between MVI and the prognosis of HCC patients.