Treatment optimization for recurrent hepatocellular carcinoma: Repeat hepatic resection versus radiofrequency ablation

Abstract Background and aims The optimal treatment strategy for recurrent hepatocellular carcinoma (HCC) remains unclear. Therefore, we aimed to compare the outcomes of repeat hepatic resection (RHR) and radiofrequency ablation (RFA) for recurrent HCC. Method From December 2004 to December 2015, 138 patients who underwent RHR and 194 patients who underwent RFA were enrolled. Propensity score matching (PSM) was performed to establish 1:1 RHR‐RFA group matching. Clinical outcomes were compared before and after matching. Results Before matching, the 1‐, 3‐, and 5‐year postrecurrence survival (PRS) rates were 91.8%, 82.0%, and 72.9% for the RHR group (n = 138) and 94.4%, 75.4%, and 61.7% for the RFA group (n = 194), respectively (P = .380). After matching, the PRS rates at 1, 3, and 5 years were 90.5%, 81.5%, and 71.8% for the RHR group (n = 120) and 91.0%, 61.0%, and 41.7% for the RFA group (n = 120), respectively (P = .002). In the subgroup analysis, the PRS rates for the RHR group were better than those for the RFA group for patients who relapsed within 2 years (P = .004) or patients with primary tumor burden beyond the Milan criteria (P = .004). Multivariate analysis showed that treatment allocation was identified as an independent prognostic factor for PRS. Conclusion Compared with RFA, RHR provided a survival advantage for recurrent HCC, especially for patients who relapsed within 2 years and those with primary tumor burden beyond the Milan criteria.


| INTRODUCTION
Hepatocellular carcinoma (HCC) is among the most common cancers and is the leading cause of cancer-related mortality worldwide. 1 Although hepatic resection is commonly chosen as a curative treatment for HCC, long-term outcomes after hepatic resection are not yet satisfactory, as tumor recurrence within the liver occurs in up to 60%-80% within 5 years. [2][3][4] Moreover, guidelines for the management of recurrent HCC remain controversial and poorly defined.
Intrahepatic recurrence is the main cause of postoperative death for HCC in clinical practice, and the postrecurrence survival (PRS) of HCC patients is greatly impacted by the characteristics of the recurrent tumor and the corresponding treatment modalities. [5][6][7] Currently, the available treatment options for recurrent HCC are not particularly different from those for primary HCC. Repeat hepatic resection (RHR) continues to be the conventional option for recurrent HCC with preserved liver function and residual liver volume, [8][9][10][11] and advances in surgical techniques and perioperative care have resulted in improved safety in RHR. In addition, as a minimally invasive option, radiofrequency ablation (RFA) has emerged as another alternative treatment modality for small HCC. 12 Several studies have previously recommended RHR when possible in the treatment of recurrent HCC. [9][10][11] However, conflicting data have also shown that outcomes following RFA are similar to those following RHR for the management of recurrent HCC. [13][14][15][16] Thus, the outcomes of RHR and RFA for the treatment of recurrent HCC remain unclear and controversial.
In this study, we compared the outcomes of recurrent HCC treated by RHR or RFA using a large cohort. Propensity score matching (PSM) analysis was carried out to minimize the bias that arises from different patient backgrounds.

| Patients
From December 2004 to December 2015, a total of 1186 HCC patients developed recurrence after initial resection. Among these patients, 144 (12.1%) were amenable to RHR, 203 (17.1%) received RFA, 678 (57.2%) underwent chemoembolization, 76 (6.4%) received sorafenib, and 85 (7.2%) received only supportive treatment. Recurrent HCC was defined as the appearance of a new lesion with radiologic features typical of HCC, as confirmed by two or more imaging modalities. The following criteria were used for patient selection: (a) age between 18 and 75 years; (b) a diagnosis of intrahepatic recurrent HCC after initial curative hepatic resection; (c) no radiologic evidence of macroscopic vascular invasion or extrahepatic metastasis; (d) RHR or RFA performed as the initial treatment for recurrent HCC; and (e) Child-Pugh class A or B. Thus, six patients in the RHR cohort were excluded, including four patients who received palliative repeated resection and two patients with other malignancies. Nine patients in the RFA cohort were excluded, including five patients older than 75 years and four patients with extrahepatic metastasis. In total, 342 consecutive patients who underwent RHR (n = 138) or RFA (n = 194) were ultimately enrolled. The correction of potential confounding factors that may affect the outcome of these groups was performed with 1:1 PSM. Finally, 120 patients who underwent RHR and 120 patients who underwent RFA as the first treatment for recurrent HCC were enrolled. This study was approved by the Institutional Review Board of our center and conducted according to the ethical guidelines of the 1975 Declaration of Helsinki.

| Treatment strategy
Treatment selection for recurrent HCC was decided by our multidisciplinary team. RHR was assigned when there was the possibility for the complete removal of all tumors while retaining a sufficient liver remnant with an expected remnant liver volume of no less than 250 ml/m 2 , as assessed by our multidisciplinary team. Resection was avoided if patients had gross ascites, severe portal hypertension, or inadequate liver remnant. Reasons for assigning RFA instead of RHR included psychological resistance to invasive treatment, refusal of general anesthesia, and an insufficient liver remnant.

| RHR procedure
Repeat hepatic resection was performed using previously described techniques. 17 Intraoperative ultrasound was routinely performed to assess the tumor burden, the liver remnant, and the possibility of a negative resection margin. Pringle's maneuver, with a clamp/unclamp time of 10 min/5 min, was performed if necessary. Anatomic resection was our preferred surgical method. If the liver remnant was inadequate, nonanatomic resection was performed with a negative resection margin.

| RFA procedure
Radiofrequency ablation was performed with the use of conscious analgesic sedation and local anesthesia. All procedures were performed percutaneously under real-time ultrasound guidance. 18,19 A Cool-tip RFA system (Radionics) with a 3 cm active tip length was used for ablation. The numbers of overlapping ablations and ablation points were determined by the number and diameter of the tumor with the aim of achieving an ablative margin of at least 0.5 cm in the normal tissue surrounding the tumor. At the end of the procedure, the needle tract was ablated to prevent bleeding and tumor seeding.

| Follow-up
In both groups, dynamic enhanced computed tomography (CT) was conducted 4 weeks after the treatment to evaluate the efficacy of the technique. Thereafter, the patients were followed up once every 3 months for the first 2 years and then once every 6 months after 2 years. At each follow-up session, blood tests, including serum liver function tests and AFP tests, as well as abdominal contrast material-enhanced three-phase dynamic spiral CT or magnetic resonance imaging, if necessary, were carried out. All patients with hepatitis-related HCC who were prepared for the initial resection of their HCC in our hospital were counseled by a hepatologist for antiviral therapy regardless of the serum HBV DNA result. 20,21

| Statistical analysis
The main end point of this study was PRS, which was defined as the date of recurrence until death or the end of the follow-up period. PSM was performed to reduce the patient selection bias and confounding variables between the RHR and RFA groups according to age, tumor size at recurrence, tumor number at recurrence, and Child-Pugh score. A oneto-one nearest neighbor matching algorithm was performed with a caliper of 0.1. Comparisons of the RHR and RFA groups were performed using t tests for continuous variables and chi-square or Fisher's exact tests for categorical variables before and after PSM. The PRS curves were calculated by the Kaplan-Meier method and analyzed with a log-rank test. All variables that yielded a P value of <.05 in the univariate analysis were subjected to a multivariate analysis using Cox proportional hazards models. All tests were two-sided, and a significant difference was considered when P < .05. Statistical analysis was performed using SPSS 24.0 (SPSS Inc).

| RESULTS
The baseline characteristics before and after PSM are summarized in Tables 1 and 2. Before matching, the median follow-up period after recurrence was 37.6 months for the RHR group and 41.6 months for the RFA group. Compared with those in the RHR group, patients in the RFA group were older (52.9 ± 11.8 years vs 50.1 ± 10.9 years, P = .028), more patients relapsed within 2 years after initial hepatic resection (65.5% vs 39.1%, P < .001), and exhibited smaller recurrent tumor sizes (1.9 ± 0.9 cm vs 2.8 ± 1.9 cm, P < .001). After matching, there was no significant difference between the RHR and RFA groups (Tables 1 and 2). For patients with recurrent HCC who received RHR or RFA, repeated recurrence was observed in 128 patients, including 69 of 120 (57.5%) in the RFA cohort and 59 of 120 (49.2%) patients in the RHR cohort. Among the 128 patients with repeated recurrence, 52 of 69 (75.4%) patients in the RFA cohort and 31 of 59 (52.5%) patients in the RHR cohort developed only intrahepatic recurrence.

| Subgroup analysis of PRS
For 47 patients in the RHR group and 62 patients in the RFA group who presented tumor recurrence within 2 years after the initial hepatic resection, the respective 1-, 3-, and 5-year PRS rates were 83.5%, 69.7%, and 65.4% for the RHR group and 84.8%, 45.9%, and 22.7%, respectively, for the RFA group (P = .004) (Figure 2A). For patients with an interval of tumor recurrence >2 years, there was no significant difference in the PRS rates between patients treated by RHR or RFA (P = .718) ( Figure 2B).
For 45 patients in the RHR group and 54 patients in the RFA group who had an initial tumor burden beyond the Milan criteria, the respective 1-, 3-, and 5-year PRS rates were 90.4%, 84.5%, and 64% in the RHR group and 88.2%, 53.6%, and 29.8%, respectively, in the RFA group (P = .004) ( Figure 2D). There was no significant difference in PRS rates between patients with an initial tumor burden within the Milan criteria (P = .237) ( Figure 2C).

| Treatment complications
Treatment-related complications are summarized in Table  4. Complications were reported according to the Clavien-Dindo grade. 22 Major complications were those classified as Clavien-Dindo grade III or higher. No treatment-related mortality was reported in this study, and the incidence of major complications did not differ significantly between the RHR and RFA groups (Table 4). Note: Variables are expressed as the mean ± SD or n (%).

| DISCUSSION
This study demonstrated that RHR was more effective than RFA for extending the postoperative recurrence survival time in HCC patients after the initial recurrence, especially for patients who relapsed within 2 years or those who had an initial tumor burden beyond the Milan criteria. The Barcelona clinic liver cancer (BCLC) staging system was initially established as a link between staging and treatment indications for primary HCC. 23 For naïve HCC Major liver resection: resection of three or more segments; minor liver resection: resection of fewer than three segments.aaa patients, it is generally accepted that the survival outcome of RFA is comparable to hepatic resection for small HCC. [24][25][26] However, the treatment of recurrent HCC does not strictly follow the BCLC staging system, as expected differences should be considered when determining treatment options for recurrent tumors. First, the size of the recurrent tumor may be smaller than the initial tumor due to more frequent surveillance. Because all patients enrolled in this study were diagnosed with intrahepatic recurrent HCC after the initial curative hepatic resection, the finding that the median tumor size of the primary HCC summarized in the table was >5 cm is reliable, since RFA may be inferior for local tumor control. However, due to more frequent surveillance after initial resection, the median size of recurrent HCC in both cohorts in this study was mostly <3 cm, with no significant difference observed, as summarized in Table 1 (RHR vs RFA: 2.4 ± 1.1 cm vs 2.2 ± 1.0 cm, P = .091). In addition, the limitations in liver function reserve cannot be ignored after the initial resection. Several studies reported generally comparable outcomes between RHR and RFA, [13][14][15][16] although a tendency toward longer survival time was observed in the RHR group than in the RFA group. Conversely, others have argued that RHR should be considered the most effective treatment for intrahepatic recurrence HCC. 7,9,11 In summary, the study limitations, including the small sample size and short follow-up period, may have led to our failure to identify an association between PRS and risk factors. In addition to the type of treatment, our study identified the presence of hepatitis and serum AFP level at recurrence as independent factors associated with PRS in the multivariate analysis. Among patients who relapse within 2 years or who have an initial tumor burden beyond the Milan criteria, RHR provided a survival advantage for recurrent HCC compared with RFA. As HCC is more likely to microdisseminate into the tributaries of the portal branches and shed tumor

T A B L E 3 Univariate and multivariate
analyses of prognostic factors for PRS emboli into the neighboring branches of the same liver segment, 27 RHR may offer a better chance to eradicate intrahepatic micrometastases generated from the original tumor than RFA. Another explanation may be that RFA was reported to be inferior to surgery for local tumor control. 24,25 Relapse beyond 2 years after initial resection is now generally believed to be due to multicentric carcinogenesis because of hepatitis or cirrhosis, 28,29 and our study showed that there was no significant difference in the long-term outcomes of patients who received RHR or RFA, consistent with previous studies. [30][31][32] In fact, the rate of repeated resection in clinical practice was less than 30%. 33 In addition to the limited liver remnant, another possible obstacle to RHR may be the assumption that RHR is more difficult to perform due to postoperative adhesion and anatomic changes. However, a systematic review reported that the 5-year patient survival rate after recurrence who underwent repeated liver resection ranged from 22.0% to 84.0%, 34 and the data of Yoh et al showed that the 5-year survival rate after recurrence was 73.8% for HCC patients with intrahepatic recurrence who underwent repeat resection, 35 which was consistent with our results. Thus, the results of our study indicated that RHR should not be ignored but performed carefully when surgical resection is considered feasible in clinical practice. Although RFA is generally regarded as safer and less invasive than RHR, both treatment groups suffered similar treatment complication morbidity rates. This may be related to the introduction of PSM and the improvement in the perioperative management of modern surgery.
The limitations of this study should also be noted. First, deviation may be unavoidable due to the retrospective design of our study. Second, treatment selection for recurrent HCC was decided by our multidisciplinary team and not randomly assigned. For HCC patients who had major resection, when recurrence was present and located deep within the liver or in a patient with insufficient liver function reserve, RHR should be carefully considered to determine whether it would offer any benefit to the patient. Thus, a prospective, randomized trial is needed to compare the efficacy of RHR and RFA for the treatment of recurrent HCC.
In conclusion, the results of our study showed that RHR is relatively safe and yielded better PRS rates than RFA for recurrent HCC, especially in patients who relapse within 2 years or who have a primary tumor burden beyond the Milan criteria. Note: Major complications were defined as those with a Clavien-Dindo classification of grade III or higher.