Hepatocellular carcinoma (HCC), the most common type of primary liver cancer and the most common pathohistological type, has been reported to account for ~90% of cases (1). Currently, HCC treatment involves a multidisciplinary approach, multiple therapeutic approaches, and individualized treatment. Common treatments include surgery, ablation, transcatheter arterial chemoembolization, radiation therapy, and systemic antitumor therapy.

Patients with early HCC recurrence typically have a poorer prognosis than those with late recurrence. Kim et al. (2) noted that the risk rate of recurrence following HCC treatment peaked at one year (21.7%) and gradually declined after five years. Surgical resection modalities for HCC are divided into anatomical liver resection (AR) and non-anatomical liver resection (NAR), with NAR involving a local resection 1–2 cm from the tumor margins without regard to Couinaud liver segments, which can facilitate the preservation of hepatic parenchyma. AR was first proposed by Makuuchi et al. (3) as a systematic resection of liver segments limited by tumor portal vein branches to reduce the incidence of liver tumor cells invading the hepatic vascular system and transferring along the blood vessels. The effect of AR and NAR on the early prognosis of patients remains unclear, with some studies suggesting that AR is more likely to improve oncologic outcomes. In the retrospective study of 362 patients by Zhong et al., after propensity score matching (PSM), one-, three-, and five-year disease-free survival (DFS) rates were reported to be 51.1%, 44.7%, and 42.0%, respectively, in the AR group, and 44.9%, 34.3%, and 26.4%, respectively, in the NAR group (P = 0.039). Using multifactorial regression analysis, AR was reported to be an independent favorable prognostic factor in patients with HCC in combination with microvascular invasion (MVI) (HR 1.054, 95% CI 1.105–2.045, P = 0.009) (4). However, it has also been suggested that there is no significant difference in the effects of the two surgical procedures on patient DFS rates. In a retrospective study by Elvan et al., that included 94 samples (5), one-, three-, and five-year DFS rates in the NAR group were 73.6%, 39.1%, and 32.8%, respectively, and 48.8%, 22.7%, and 22.7%, respectively, in the AR group (P = 0.085). However, most of the above studies utilized a single baseline characteristic (for example, whether the tumor diameter exceeded 5 cm or whether MVI accompanied it) for survival analysis, with less consideration for each patient’s comprehensive situation. As postoperative recurrence in patients with HCC is a highly heterogeneous outcome involving the interaction of many potential factors, our stratified randomized study aimed to develop more comprehensive baseline characteristics using a self-developed recurrence prediction model to group our study population accordingly and to investigate the effect of each surgical modality on early recurrence of HCC.

Currently, HCC is mainly treated via surgical resection using a multidisciplinary team approach (MDT, medical oncology, interventional medicine, and gastroenterology) and supplemented with local therapy, hepatic arterial cannulation for continuous chemotherapy infusion, and adjuvant therapy according to each patient’s physical condition and tumor stage. With advances in surgical techniques and increased usage of large medical centers, combined with the oncological characteristics of HCC, the use of AR has gradually become more prominent in hepatectomy, with many studies reporting a better prognosis with AR than with NAR (6–8).

The tumor-node-metastasis (TNM) staging system has been widely used to predict the survival rate for patients with tumors; however, in recent years, There are already many studies that have predicted prognoses using prediction models with higher accuracy and precision than TNM staging (9–11). Our model showed better performance than the AJCC, BCLC, and CNLC staging systems (0.813, 0.721, 0.743, and 0.763, respectively; P < 0.001). This may be because, in addition to the inclusion of assessment factors (tumor diameter, tumor number, and combined vascular invasion) that are widely used in these staging systems, we included other clinicopathological patient features, such as combined MVI, elevated AFP, and elevated ALP, which have been shown to significantly affect the prognosis of patients with HCC (12–14).

Previous studies have shown that the morphological characteristics of primary tumors and the early recurrence of postoperative are closely related to long-term prognosis in patients with HCC. The risk factors for HCC recurrence are largely related to the morphological characteristics of a patient’s tumor, e.g., tumor diameter ≥5 cm and portal vein invasion are indicative of increased invasiveness of the tumor and a higher risk of recurrence (15, 16). Moreover, the occurrence of MVI is closely related to a patient’s tumor load, differentiation level, AFP level, and HBV activity and replication (17), while the presence of satellite nodules in patients with MVI is suggestive of tumor progression (18).

Serum biomarker testing is also often used to predict HCC recurrence. ALP is a common liver function test, and an elevated ALP level is often in response to liver injury. Elevated serum ALP can also be used to evaluate the prognosis of bone diseases such as osteosarcoma, as well as tumors accompanied by bone metastases (19). Similarly, our study found that elevated ALP levels were positively correlated with the risk of early postoperative recurrence in patients (HR 2.803, 95% CI 1.617–4.860), which may be attributed to higher ALP activity in the nucleus of cancer cells and to the dynamic changes that occur during the cell cycle, suggesting that ALP may contribute to tumor formation through altering cell cycle regulation and proliferation (20). ALP can also be elevated in a number of inflammation-related diseases (e.g., hepatitis, choledocholithiasis, cholangitis, and pancreatitis), and the inflammatory milieu plays an important role in driving the development and progression of cancer (21); therefore, elevated serum ALP levels may reflect the fact that patients with HCC have more severe inflammation with a poorer prognosis. Alpha-fetoprotein (AFP) is a conventional tumor marker for predicting HCC recurrence, but its sensitivity is relatively low. des-γ-carboxy-prothrombin (DCP) was recently reported to be superior to AFP in detecting HCC recurrence (22, 23). Furthermore, biomarkers such as circulating tumor DNA assays and circulating tumor cells have also been noted in the improvement of AFP prediction performance (24). However, research concerning biomarkers or combinations of biomarkers that are highly accurate and easy to detect is ongoing.

The recurrence pattern of HCC has been extensively studied and mostly involves intrahepatic recurrence at the hepatic end of the residual liver (90.1%), with most recurrences occurring within 1–2 years postoperatively (54.5%) (18, 53). In 1998, Ueda et al. used CT hepatic angiography to determine the hemodynamics of HCC during carcinogenesis (25) and confirmed, for the first time, the theory of tumor blood flow (TBF). The TBF drainage area is considered to be the peritumor area where tumor blood drains and may contain more micrometastases than other areas. Therefore, to completely resect the peritumor area, the optimal surgical area for HCC should be determined based on the patient’s tumor hemodynamics, that is, the TBF drainage area. Based on this understanding, in 2000, Sakon et al. proposed the concept of three modes of intrahepatic recurrence of HCC, namely, (i) local intrahepatic metastasis, that is, recurrence of HCC spreading directly to the periphery of the tumor through portal venous blood flow or venous drainage; (ii) systemic intrahepatic metastasis, that is, HCC recurrence due to circulating tumor cells (CTCs), and (iii) multicenter HCC recurrence due to HCC redevelopment (26). Based on recurrence patterns after TBF hepatectomy, AR can be used to appropriately resect the drainage area of the TBF and effectively avoid the recurrence of HCC intrahepatic metastasis.

AR was first proposed by Makuuchi et al. with the aim of completely resecting the tumor-carrying portal branches supplied by the portal vein and the branches of the hepatic artery, thereby improving the surgical prognosis of patients with HCC. Owing to HCC’s tendency to progress to intrahepatic vascular structures (27, 28), AR allows complete resection of the tumor, peritumoral liver tissue, and possible micrometastases around the tumor. This procedure is preferred by most surgeons. However, NAR is typically used for patients with HCC with relatively deteriorated liver function reserves because of chronic hepatitis or cirrhosis. However, when comparing AR and NAR, it is also necessary to consider the difficulty of performing both surgical procedures as well as postoperative differences in complications among patients, such as residual liver ischemia and liver failure. Previous studies have shown that AR has a longer operative time and a higher likelihood of postoperative residual hepatic ischemia than NAR and that residual hepatic ischemia is an independent risk factor for postoperative recurrence in patients (29). In recent years, the gap between AR and NAR in terms of operative time, intraoperative bleeding, and the occurrence of serious postoperative complications has narrowed with the advancement of intraoperative aids and surgical techniques (30). Our study shows that in the low-risk recurrence group, no statistically significant difference was observed between the two surgical techniques in terms of operation time, average hospitalization time, intraoperative bleeding, postoperative liver failure, and related complications. In the high-risk recurrence group, patients in the AR group had less bleeding than those in the NAR group (200 mL vs. 400 mL, respectively; P = 0.008). This may be because, as surgical techniques have developed, the operative times for AR and NAR have become lesser. However, AR following hepatic vascular anatomy is more effective in avoiding intraoperative bleeding. DFS was superior to NAR in the low-risk group for recurrence, and this advantage was more pronounced in the subgroup of high-risk patients.

Previous studies have shown that the effect of surgical resection modality on the prognosis of patients with HCC, especially those at a relatively high risk of recurrence (4, 5, 31), was not statistically significant, whereas our study found that patients in the high-risk recurrence group who underwent AR had a significantly better one-year RFS rate than those who underwent NAR. This could be because some high-risk patients have CTCs that have not begun to spread through the bloodstream, and AR resection may be able to remove some CTCs that have not yet spread via blood metastases. However, the clinical application of CTCs remains challenging. The most important difficulty is that the earlier the cancer stage is, the fewer the CTCs. Furthermore, there are conflicting definitions of CTCs (32). Therefore, for patients with late tumor staging, we consider that risks and benefits should be balanced with consideration of the possible risk of high CTCs, the difficulty of intraoperative resection, and postoperative hepatic function compensation, and that AR should be selected to treat patients with surgical indications.

This single-center retrospective study had some limitations. We intend to plan a large multicenter large-sample study to explore a preoperative CTC model and enhance the validity of the model through internal and external validation, which may allow for a more in-depth investigation of changes in relation to the benefits of varying surgical modalities for different patient populations.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Biomedical ethics committee of Mianyang Centre Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

RS: Data curation, Writing – original draft, Writing – review & editing. JW: Data curation, Writing – original draft, Writing – review & editing. XZ: Methodology, Supervision, Writing – review & editing. HL: Supervision, Writing – review & editing. XY: Data curation, Investigation, Writing – review & editing. YG: Data curation, Investigation, Writing – review & editing. LY: Data curation, Investigation, Writing – review & editing. HD: Data curation, Software, Writing – review & editing. PY: Supervision, Writing – review & editing.