Radiofrequency ablation versus resection for technically resectable colorectal liver metastasis: a propensity score analysis

This retrospective study evaluated collected data from 428 consecutive patients who underwent RFA or resection for CRLM at the Peking University Cancer Hospital between November 2010 and December 2015. The study was approved by the Clinical Research Ethics Committee of the same hospital and was performed in compliance with the Helsinki Declaration. Written informed consent was obtained from all patients.

Inclusion criteria was patients with ≤ 3 tumors, well-located tumor size of ≤ 5 cm, and absence of uncontrolled extrahepatic disease. The exclusion criteria were patients with recurrent CRLM after previous resection or RFA, or who underwent both RFA and resection in one session, and those who received palliative treatment. The patients’ preoperative images were retrospectively viewed to confirm the technically resectable disease CRLM which was feasibility of complete macroscopic resection to maintain at least 30% future liver remnant [24]. Based on these criteria, we included 50 patients who received RFA and 160 patients who underwent resection with curative intent.

Baseline data included sex, age, timing of metastasis, location of primary cancer, T stage and N stage, number and diameter of hepatic metastases, carcinoembryonic antigen (CEA) level, and neoadjuvant chemotherapy in the two groups. Disease-free survival and overall survival was determined in both the groups. Variables between the two groups and those included in clinical risk score that could have impacted on survival were identified.

The propensity scores were estimated using a logistic regression model that included the following five covariates primary lymph node status, synchronicity, number of metastases, size of the largest metastasis, and preoperative CEA levels. A 1:2 “nearest neighbor” match paradigm was used. Patients were matched using a caliper of 0.15 in each group (Fig. 1).

The liver was examined, and intraoperative ultrasonography was performed to identify the number and locations of metastases. The extent of hepatic resection was determined by the number, diameter, and locations of the tumors, and lobectomy, segmentectomy, or limited resection was adopted. Parenchymal dissections were performed using the clamp method with Peng’s multifunctional operative dissector (Hangzhou Shuyou Medical Instrument Co., Ltd., PR China; FDA 510[K] number K040780). An intermittent Pringle’s maneuver with clamping of the hepatoduodenal ligament was occasionally performed during parenchymal transection for vascular occlusion. The preserved margin during parenchymal dissection was ≥ 5 mm.

The indications for RFA were complete necrosis achieved based on the tumor size and its position, patients’ comorbidities that precluded general anesthesia or surgery, and patient choice. RFA was more often used in deeply situated tumors that would have required excessive sacrifice of the normal parenchyma in resection. Ablation of tumors next to major bile ducts (common bile duct, common hepatic, right and left hepatic ducts) within 1 cm, in contact with larger blood vessels (portal vein and hepatic vein), or in close proximity to vulnerable structures (colon, gallbladder etc.) were relatively restricted. All RFA procedures were performed using the Celon system (Teltow, Germany) by radiologists with > 5 years of interventional experience. The bipolar electrode needles were 16G, and scanning/guidance ultrasonography was performed using the Aloka α-10 (Tokyo, Japan) and GE Logiq E9 (Connecticut, USA) devices. Electrodes were inserted into the tumor under ultrasonographic guidance, and overlapping ablations were used for > 3-cm tumors. The ablation end-point was determined based on the impendence and output power, as well as coverage of the safety margins. Track ablation was performed after the treatment. The ablative area appeared hyperechoic on ultrasound during RFA procedure, which should cover the tumor area. For cases of difficult to assess, contrast-enhanced ultrasound was performed immediately after RFA. If tumor residual occurred, additional RFA session was performed.

Patients were evaluated by contrast-enhanced computed tomography (CECT) or magnetic resonance imaging (MRI) at 1 month after resection or RFA procedure. Then, CEA test, MRI of the abdomen, CT of the chest, and MRI or CT of the pelvis were repeated every 3 months for 2 years and every 6 months thereafter. Recurrences were typically identified radiologically.

Local recurrence was defined as tumor growth at the treatment site. Intrahepatic recurrence was defined as new liver lesions emerging at a non-treatment site. Systemic recurrence was defined as tumors at both hepatic and extrahepatic sites, including recurrence at the site of the primary tumor.

Continuous variables were reported as median and interquartile range. Inter-group differences were analyzed using the chi-square test, Fisher’s exact test, or Student’s t test, as appropriate. Survival data were analyzed using the Kaplan-Meier method and the log-rank test. Variables with a univariate p value of < 0.1 were entered into the Cox regression model for multivariate analysis. A p value of < 0.05 was considered statistically significant.

The resection group included 92 patients (58 males, 34 females) with a median age of 63 years (interquartile range 51.0–65.8), and the RFA group included 46 patients (29 males, 17 females) with a median age of 63 years (interquartile range 50.8–67.0). The patients’ clinicopathological characteristics are shown in Table 1. After matching according to the propensity score, there was no significant difference between the two groups although differences were originally observed for preoperative CEA levels and the number, size, and location of the liver metastases. The 46 patients in the RFA group underwent treatment for 55 lesions (1.2 ± 0.5 lesions/patient), and the 92 patients in the resection group underwent treatment for 114 lesions (1.2 ± 0.4 lesions/patient). The median diameter in the RFA group was 2.3 cm (range, 1.7–3.6 cm), compared to 3 cm (range, 1.9–3.6 cm) in the resection group. Thirty-four patients (37.0%) in resection group and 22 patients (47.8%) in the RFA group received neoadjuvant chemotherapy. Patients in the two groups received regular systemic chemotherapy regimens, such as FOLFOX, CAPEOX, or FOLFIRI, combining biologic-targeted agents (bevacizumab or cetuximab) which were selectively used in high risk of recurrence patients only. After treatment, 45 (48.9%) patients in the resection group and 16 (34.8%) patients in the RFA group received adjuvant chemotherapy according to preoperative chemotherapy response, Fong’s score, and postoperative recovery condition, and the difference was not statistically significant (P = 0.115).

All follow-ups ended in July 2018, and the median follow-up was 44 months (range, 6–96 months). The overall survival (OS) rates were similar in the resection and RFA groups at 1 year (97.8% vs. 95.7%), 2 years (83.6% vs. 91.3%), and 3 years (66.8% vs. 71.6%). Based on the Kaplan-Meier analyses, the median OS was 74 months in the resection group and was 59 months in the RFA group (P = 0.484, Fig. 2a). The median disease-free survivals (DFS) were 22 months after resection and 14 months after RFA (P = 0.032, Fig. 2b). However, the DFS for resection and RFA were similar among patients with a tumor size of ≤ 3 cm (24 months vs. 21 months, P = 0.41).

The first sites of disease progression after treatment are shown in Table 2. Intrahepatic recurrence was significantly common (36.9% vs. 11.9%, P = 0.001), and local recurrence was more common in the RFA group (15.2% vs. 6.5%, P = 0.099) (Table 2). The systemic recurrence rates were similar in both groups (26.1% vs. 39.1%, P = 0.129). Hepatic recurrence was more common after RFA compared to resection (69.6% vs. 32.6%, P < 0.001) (Table 2).

The time to local, intrahepatic, and systemic recurrences are shown in Fig. 3. The RFA group had a significantly shorter time to intrahepatic recurrence, compared to the resection group (P < 0.001). No significant differences were observed between the two groups for the times to local recurrence (P = 0.083) or systemic recurrence (P = 0.478). Additional treatments with curative intent (resection, RFA, radiotherapy, or combination therapy) were performed after recurrence for 18 patients (50.0%) in the RFA group and 17 patients (37.0%) in the resection group (P = 0.089).

Cox multivariate analyses were used to evaluate DFS, and the results revealed that poorer DFS was independently associated with RFA, T4 status, lymph node positivity, and tumor diameter > 3 cm (Table 3). OS was independently associated with tumor diameter > 3 cm and T4 stage, but was not significantly associated with RFA or resection as first-line treatment.