This study recruited HCC patients who underwent preoperative CTC testing at the Department of General Surgery I of Zhongshan People’s Hospital from January 2010 to December 2017, with the following inclusion criteria: (1) patients diagnosed with HCC by postoperative pathology, (2) no treatment other than TACE, (3) radical tumor resection (R0) that is, negative macroscopic and microscopic tumor resection margins, (4) complete serological and imaging data, (5) tumor diameter not less than 5 cm, and (6) single tumor. Exclusion criteria were (1) patients younger than 18 years of age, (2) presence of vascular tumor thrombus and distant metastasis, (3) patients with Child-Pugh grade C liver function, (4) no severe vital organ dysfunction, (5) patients who underwent palliative resection, and (6) loss of postoperative follow-up data. The Ethics Committee of Zhongshan People’s Hospital has approved the retrospective study, all patients have signed an informed consent form.

All patients underwent abdominal enhanced CT or MRI, chest CT or X-ray scan. Laboratory tests include blood routine, liver and kidney function, coagulation function, hepatitis B surface antigen, hepatitis C antibody, AFP, and other examinations. The basic data of the included study population, such as gender, age, hepatitis B surface antigen, hepatitis C antibody, AFP level, alanine aminotransferase (ALT), aspartate aminotransferase (AST), γ-glutamyl aminotransferase (GGT), alkaline phosphatase (ALP), creatinine (Cr), albumin (Alb), total bilirubin (TIBL), direct bilirubin (DIBL), international normalized ratio (INR), platelet count, presence of cirrhosis, Child-Pugh grade, presence of microvascular infiltration (MVI), maximum tumor diameter, pathological classification, extent of liver resection, type of liver resection and other data were collected. Minor liver resection was defined as resection of fewer than three Couinaud liver segments, while major liver resection was defined as resection of three or more liver segments. Non-anatomical liver resection included a limited resection or wedge resection; anatomical resections were defined by the Brisbane 2000 system. The continuous variables are transformed into binary variables, and the cut-off value is the upper and lower lines of the recognized normal value.

Considering that this was a retrospective study, the decision to use TACE prior to surgery was left to the discretion of the treating surgeon and the patient at that time. The patient was placed supine, locally disinfected, draped, and given local anesthetized. The puncture site was chosen to be 2 cm below the inguinal ligament, and the catheter sheath was placed into the femoral artery using the Seldinger technique. Firstly, the DSA technique helps with abdominal trunk and standard hepatic artery angiography to determine the tumor’s location, size, and condition of the tumor. Once the tumor is understood, the catheter sheath is continued deeper into the left or right hepatic artery or the vessel that feeds the tumor, 5-fluorouracil (500 mg/m2) or oxaliplatin (100 mg/m2) was injected into the proper hepatic artery, and embolization was performed using different embolization materials. Patients were asked to return to the hospital 4-6 weeks after embolization for follow-up serology, including blood routine, liver and kidney function, coagulation function, AFP, and imaging included abdominal enhanced CT or MRI, chest X-ray scan, etc. All of the above procedures were performed by highly qualified attending physicians who received relevant interventional medicine.

The Cyttel method is used to detect CTCs, and its main principles include the negative immunomagnetic particle assay and immunofluorescence in situ hybridization (im-FISH). Jiangsu Lyle Biomedical Technology Co manufactures the kit. For patients with preoperative TACE, samples were obtained within three days before TACE, while for patients without preoperative TACE, the sample extraction must also be completed within three days before surgery. Generally, we draw 5ml peripheral blood, and process the samples strictly according to the manufacturer’s instructions. Firstly, the samples was treated with negative immunomagnetic powder method to remove leukocytes from the peripheral blood, and isolate rare cells in the blood, and finally obtain CTCs. Then, the im-FISH technique was used to fix and dehydrate the samples, then hybridization with chromosome centromeres 1 and 8, followed by sealing with 4-diamidine-2-phenylindole (DAPI) staining solution, and then observation and counting under a fluorescence microscope (29–32). It defined CTC count ≥1 as CTC-positive (32).

Each follow-up visit for all patients include AFP, routine blood tests, liver and kidney function tests, coagulation function tests. Enhanced CT or MRI of the abdomen, chest CT, and the bone scan will be performed if tumor residue and signs of tumor recurrence are suspected. The first postoperative follow-up visit would performed one month after the operation. The follow-up frequency was once every 2-3 months within six months after the operation, once every 3-4 months within 6-24 months after the operation, and once every 4-6 months after 24 months after the operation. After a recurrence of HCC, treatment options are chosen according to the recurrence and the patient’s general condition. Treatment options include surgical re-resection, radiofrequency ablation (RFA), percutaneous ethanol injection (PEI), TACE, taking targeted drugs, immune drug therapy, and even liver transplantation. OS was defined as the date of surgery until patient death or last follow-up, and RFS was defined as the date of surgery until patient signs of recurrence or last follow-up. Recurrence was classified as early recurrence and late recurrence using a cut-off value of 24 months.

Continuous variables were expressed as median ± square difference (Median ± SD), and categorical variables were expressed as number (n) or percentage (%) of patients. The t-test or Mann-Whitney test was used to compare two groups of continuous variables, and the χ2 or Fisher’s exact test was used to compare two groups of categorical variables. The survival curves of OS and RFS of the patients were plotted using the Kaplan-Meier method, and the OS and RFS of the patients in the preoperative TACE group and the two groups without preoperative TACE were compared using the log-rank. We also used the Landmark analysis method to analyze the results of assessing early recurrence (recurrence 24 months after surgery) and late recurrence. Univariate and multivariate Cox regression models were used to analyze the independent risk factors of each factor on patients’ RFS and OS. All statistics and graphs for this study were completed in R (version 3.62). P values < 0.05 were considered statistically significant.

Baseline characteristics of the total population of HCC are listed in Table 1 . The population was divided into positive and negative subgroups based on the preoperative CTCs count. The clinical baseline of each subgroup is shown in Table 2 . In this study, a total of 361 patients with HCC were enrolled in this study, including 211 patients of CTC-positive (58.4%) and 103 patients of preoperative TACE (28.5%). The median follow-up time of the CTC-positive group was 38.0 months, while the median follow-up time of the CTC-negative group was 44.5 months. The median follow-up time of HCC patients with preoperative TACE was 41.0 months, while patients without TACE were 36.5 months. During follow-up, 134 patients died, and 275 patients developed tumor recurrence. In the CTC-positive and CTC-negative subgroups, the clinicopathological variables were similar, comparable and not statistically significant between patients who underwent preoperative TACE and those who did not (P > 0.05; Table 2 ). In the overall population, RFS and OS were similar of patients with and without preoperative TACE; Preoperative TACE did not improve the prognosis of HCC (P > 0.05; Figures 1A, B ).

Using survival curves drawn by the Kaplan Meier method, we found that OS (median 39 months vs. 47 months, P < 0.05, Supplementary Figure 1A ) and RFS (median 17.0 months vs. 24 months, P < 0.05 Supplementary Figure 1B ) in CTC-positive group were worse than those in CTC-negative group. We also analyzed the effect of CTCs status on postoperative recurrence patterns using the landmark method. Using a 24-month cut-off, postoperative recurrence was divided into an early recurrence and late recurrence. We found that CTC positive was associated with postoperative early recurrence (P < 0.05; Supplementary Figure 2 ) but not with late recurrence (P > 0.05; Supplementary Figure 2 ).

To determine whether CTCs status affects the clinical efficacy of TACE, we stratified patients’ CTCs status of and compared the OS and RFS between patients with and without preoperative TACE at different CTCs status. In the CTC-positive group, preoperative TACE prolonged OS and RFS in HCC patients; the difference was statistically significant (P < 0.05; Figures 2A, B ). In CTC-negative group, preoperative TACE could not improve RFS and OS, and the difference was not statistically significant (P > 0.05; Figures 3A, B ).

Univariate and multivariate Cox regression analysis also showed that in CTC-positive group, non-preoperative TACE was an independent risk factor for OS (hazard ratio [HR]= 2.330, 95% confidence interval [CI], 1.318-4.120, P <0.05; Table 3 ) and RFS(hazard ratio [HR]= 2.332, 95% confidence interval [CI], 1.584-3.432, P <0.05; Table 4 ) of HCC patients, while in CTC-negative group, preoperative TACE had no effect on OS (hazard ratio [HR]= 0.655, 95% confidence interval [CI], 0.338-1.267, P >0.05; Table 5 )and RFS (hazard ratio [HR]= 0.805, 95% confidence interval [CI], 0.511-1.269,P >0.05; Table 6 ) of HCC patients.

Using landmark analysis and taking 24 months as the cutoff value, we found that preoperative TACE could reduce the early recurrence of patients in CTC-positive group (P < 0.05, Figure 4A ), but could not improve the late recurrence rate of patients (P > 0.05, Figure 4A ). In the CTC-negative group, preoperative TACE could not improve the early and late recurrence (P > 0.05, Figure 4B ).

The comparison of clinicopathological variables between the CTC-positive group and CTC-negative group is shown in Table 7 ; the proportion of patients with tumor diameter ≥10cm (48.3.1% vs. 34.7%, P < 0.05; Table 7 ) and positive rate of MVI (65.9% vs. 54.0%, P < 0.05; Table 7 ) in CTC-positive group was higher than that in CTC-negative group. At the same time, other clinicopathological indicators such as age, sex, HBV, cirrhosis, Child-Pugh, Edmondson stage, and AFP≥400ng/ml were not significantly different(P > 0.05; Table 7 ).

We compared the effects of preoperative TACE on perioperative complications and mortality. We found that preoperative TACE did not increase perioperative mortality, liver failure, bile leakage, ascites, wound infection, and other complications compared to patients without preoperative TACE (P > 0.05; Table 8 ).