Comparisons of recurrence-free survival and overall survival between microwave versus radiofrequency ablation treatment for hepatocellular carcinoma: A multiple centers retrospective cohort study with propensity score matching

Both microwave (MW) ablation and radiofrequency (RF) ablation are widely used for hepatocellular carcinoma (HCC) treatments in clinic. However, it is still unclear if ablative methods could influence the recurrence-free survival (RFS) and overall survival (OS) of HCC patients. Therefore, we carried out this multi-center retrospective cohort study to investigate the differences of recurrence-free survival (RFS) and overall survival (OS) between MW ablation and RF ablation by survival analysis. From January 2014 to December 2016, patients who received thermal ablation surgery for HCC treatment were screened. Finally, 452 patients met the eligibility criteria and finished the follow-up. Univariable and multivariable regression analyses were used to identify independent predictive factors of the RFS and OS. Also, propensity score matching (PSM) was used to balance the bias between two groups. Finally, we found that before the PSM, the univariable and multivariable regression analyses revealed that there were no significant differences on the RFS between two groups. Same results were obtained for the OS. After PSM, 115 pairs of patients were created, and both the univariable and multivariable regression analyses suggested that there were still no significant differences on the RFS between two groups. Same results were obtained for the OS. In conclusion, our present study showed that there were no significant differences between MW ablation and RF ablation for HCC patients on the RFS or OS.

Introduction and background confirmed by postoperative histopathology. Severe organ failure mainly includes liver failure (Child-Pugh C degree), heart failure (NYHA Ⅲ-Ⅳ), and renal failure (serum creatinine >442 μmol/L). Severe immune system disease mainly includes immunodeficiency disease, systemic lupus erythematosus, rheumatoid arthritis and ankylosing spondylitis.

Follow-up
All patients were reexamined using serum alpha fetoprotein (AFP), ultrasound or CT, and chest X-ray at 1 month after surgery. Then, patients were followed-up at a 2-monthly interval for the first 6 months and at a 3-monthly interval thereafter. Tumor recurrence was defined as new appearance of intra-or extrahepatic tumor nodule. The clinical practice guidelines of EASL-EORTC was used for the diagnosis of tumor recurrence [16]. Patients with tumor recurrence were actively treated with percutaneous ablative, hepatic resection, transcatheter arterial chemoembolization (TACE), radiotherapy or conservative treatment.
The follow-up began from January, 2014 and ended at March, 2018. All data used for analysis was collected from digital medical history or paper medical records, and all data were fully anonymized before access. In every center, two trained researchers were in charge of follow-up and all data were entered using "Epidata".

Variables and outcomes
In this study, 22 variables were collected and analyzed. All variables could be divided as patient characteristics (age, sex, ASA score, hypertension, diabetes and cardiopathy), liver function variables (cirrhosis, HBV/HCV infection, child-pugh stage, AFP, TBiL, ALB, ALT, and AST), operative variables (tumor number, tumor size, anesthetic methods and adjuvant chemoradiotherapy) and follow-up information (dates of operation, inpatient days, dates of tumor recurrence and dates of death). All data were collected and entered with the same way as the follow-up data. In every center, two trained researchers were in charge of collecting data.
The main outcome of this study was RFS and the second outcome was OS. The RFS was defined from the date of the surgery to the date of first recurrence. If the recurrence of tumor was not recorded, the RFS was defined as the time between the date of surgery and the date of last follow-up. The OS was calculated from the date of the surgery to either the date of death or the date of the last follow-up visit.

Propensity score matching
Because patients were not randomly allocated to the MW ablation group and RF ablation group, and variables in two groups were imbalanced. We decided to use the propensity score matching as described before [17] to eliminate the imbalance between two groups. This method consisted of ordering the case and control subjects, then selecting the first case subject and finding the control subject with the closest propensity score [18]. A logistic regression model was built given the covariates of tumor number, cirrhosis, HBV/HCV infection, Child-Pugh stage, AFP, ALB, TBIL, anesthetic methods, tumor size, adjuvant chemoradiotherapy and hypertension, and the dependent variable of ablation methods. We applied 1:1 nearest neighbor matching without replacement to ensure that conditional bias was minimized. For each patient having MW ablation, a patient having RF ablation with a minimum in distance of propensity scores was matched. The caliper width was 0.05 for propensity score matching. Propensity score matching was carried out using IBM SPSS Statistics 23.0 version.

Statistical analysis
Statistical analyses were carried out using the IBM SPSS Statistics 23.0 (SPSS Inc., Armonk, NY, USA). Categorical variables were reported as number (n) or proportion (%) and continuous variables were expressed as mean ± standard deviation (SD) or median (range). The Student's t test was used for comparisons of continuous variables. Otherwise, the Mann-Whitney U test was applied. Categorical variables were compared with the χ 2 test with the Yates correction or the Fisher's exact test, as appropriate. To identify independent predictive factors of prognosis, univariable and multivariable regression analyses were used. The RFS and OS rates were compared between the MW ablation and RF ablation groups before and after propensity matching using the Kaplan-Meier regression analyses or univariable Cox regression analyses. Multivariable Cox proportional hazard regression analyses were then performed to adjust for other prognostic factors which were associated with OS and RFS [19]. All statistical tests were 2 sided, and P values <0.05 were considered statistically significant.
Result 532 patients who underwent MW or RF ablation met the eligibility criteria, and finally 452 patients finished follow-up. Patients were divided into two groups: the MW ablation group (N = 218, 48.2%), and the RF ablation group (N = 234, 51.8%). The comparisons of patients' characteristics and other variables between two groups in the entire cohort are illustrated in Table 2. Patients' characteristics including Hypertension, Tumor size and Anesthetic methods are significantly different between two groups (P<0.05). The follow-up time was at a range of 1.25-to 4.25-year, and the average follow-up time was 2.34-year.
In our retrospective study, all patients received ablation surgery by percutaneous approach. Data regarding the complete response, differentiation between local and distant recurrence, need for repeat ablations, postoperative complications and Edmandson grade were showed in Table 3. No statistical differences were observed between two groups.
First, we used Kaplan-Meier analysis or univariable Cox regression model analysis to screen variables which had significant influence on the RFS and OS. From the Table 4 showed that ASA score, Hypertension, Tumor number, Cirrhosis, Adjuvant chemoradiotherapy, AFP, Tumor size and anesthetic methods would significantly influence patients' RFS (P<0.05). However, we found that the Ablation methods had no observable influence on the RFS by logrank test (Fig 2A, P = 0.089). Also, results suggested that Age, ALB, Child-Pugh stage, Tumor size and anesthetic methods could remarkably influence the OS. But the ablation methods had no significant influence on the OS by log-rank test, too (Fig 2B, P = 0.160).
Significant variables (P<0.05) as shown in Table 4 were entered into multivariable Cox regression model analysis. As the Table 5 showed, the ablation methods had no significant influence on the RFS and OS (both P>0.05). It suggested that different ablation methods would not influence the RFS or OS for HCC patients.
PSM analysis was carried out as illustrated above and finally created 115 pairs of patients. After the PSM, there no significant differences of variables and PS score between two groups ( Table 2). Comparisons of patients' RFS and OS between two groups after PSM were shown in Table 6. It suggested again that different ablation methods would not influence the RFS (Fig  3A, P = 0.162) or OS of HCC patients (Fig 3B, P = 0.726). Results in multivariable Cox regression model analysis showed that there were no differences on the RFS and OS after PSM ( Table 7).

Discussion
According to this multi-center retrospective cohort study, we found that there were no significant differences between MW ablation and RF ablation for HCC patients on the RFS or OS. Though liver resection is the first-line curative treatment for patients with HCC, several studies had verified that hepatic resection contributed to a higher rate of complications and surgical mortality [20,21]. At the same time, more studies also found that ablation surgery was equivalent to surgical resection for overall survival [22,23]. In conclusion, ablation surgery is a kind of effective and less invasive method for tumor treatments. However, with the development of MW ablation and RF ablation, researchers focus on the differences between two methods. Some studies [11,24,25] reported that MW ablation seemed to have a lower rate of local recurrence of tumor. It could be explained that MW ablation has an improved convection profile, higher intratumoral temperatures, faster ablation time, larger ablation volume, and less susceptibility to heat-sink effect [26,27]. But still other studies found that no significant differences on the RFS or OS were observed between MW ablation group and RF ablation group [28,29]. In our study, we further verified that there were similar RFS In this study, more than 20 related variables were collected and analyzed. After PSM, results of multiple Cox regression analysis showed that AFP, Tumor size and Tumor number were independent risk factors for the RFS, and the Anesthetic methods and Age were independent risk factors for the OS. In this study, we found that the local anesthesia was better than general anesthesia, and the OS in local anesthesia group was longer than the general anesthesia group. This finding could be verified by related studies [30,31].
This study has several limitations. First, it is a retrospective cohort study rather a randomized controlled trail. But we had used the PSM and multiple Cox regression analyses to minimize the bias between two groups. Second, the follow-up time could be longer. In the further diverse medical record software, some data was missing during the replacement of software.
And some data about the dosage of anesthetic drugs and adjuvant chemotherapeutic drugs were not recorded with details, so we could not collect relevant data intactly. In summary, after using PSM analyses and multivariable Cox regression analyses, our present study showed that there were no significant differences between MW ablation and RF ablation for HCC patients on the RFS or OS. Both MW ablation and RF ablation were effective and safe for patients who suffered HCC.