Abstract
Aim: To compare outcomes of patients with primary hepatocellular carcinoma (HCC) and portal hypertension (PHT) who underwent laparoscopic liver resection (LLR), open liver resection (OLR) or radiofrequency ablation (RFA). Patients and Methods: We retrospectively reviewed 88 patients with primary HCC and PHT who underwent LLR (n=20), OLR (n=48) or RFA (n=20) and analyzed their outcomes by treatment group. To reduce selection bias, covariate distributions in groups were adjusted using inverse probability treatment weighting (IPTW). Results: Five-year recurrence-free survival (RFS) was significantly better in the LLR and OLR than in the RFA group both before and after IPTW adjustment. The OLR group had significantly more postoperative complications than the RFA group; however, there was no significant difference in the postoperative complication rate between LLR and RFA groups. Conclusion: LLR may be a feasible treatment for patients with Barcelona Clinic Liver Cancer (BCLC) stage 0 or advanced hepatocellular carcinoma (A HCC) and PHT.
Since the first international consensus conference in 2008 (1), laparoscopic liver resection (LLR) has been increasingly performed, with the technique continuing to evolve worldwide (2-6). However, clinical evidence is lacking in some areas. For example, it has not yet been determined whether LLR is indicated in patients with hepatocellular carcinoma (HCC) and portal hypertension (PHT) (7, 8).
The current guidelines of the Barcelona Clinic Liver Cancer (BCLC) Group (9) and the American Association for the Study of Liver Disease (AASLD) (10) advise that patients with HCC and clinically significant PHT are candidates for radiofrequency ablation (RFA) or liver transplantation and not for open liver resection (OLR). However, this recommendation is questionable because there are no sound data comparing the outcomes of RFA, OLR and LLR or reporting outcomes of LLR in patients with cirrhosis, HCC and PHT.
Clinically significant PHT, defined as a hepatic venous pressure gradient (HVPG) ≥10 mmHg, is reportedly the most powerful predictor of postoperative liver failure and poor long-term survival in patients with Child–Pugh A liver function (11). However, this conclusion was based on a retrospective cohort study of only 77 patients who had undergone open liver resection (OLR). Since preoperative HVPG measurement is invasive, it is not performed routinely in most liver centres. Rather, indirect clinical characteristics, including presence of oesophageal varices (EVs) and/or splenomegaly associated with thrombocytopenia, are used to diagnose PHT (12). Conversely, many studies have reported that OLR improves survival in patients with PHT and that PHT is not an absolute contraindication to hepatectomy in patients with cirrhosis (13, 14). Given the very limited availability of liver transplant donor organs (15), surgical resection may be a better treatment option for some patients with HCC and clinical PHT (16). Moreover, LLR may be safer than OLR in patients with HCC and cirrhosis because postoperative ascites and other complications occur less frequently after the former procedure (17-19).
In this study, we compared the outcomes of patients with BCLC stages 0 and advanced hepatocellular carcinoma (A HCC) and PHT who had undergone LLR, LR or RFA. We used inverse probability treatment weighting (IPTW) to reduce patient selection bias. IPTW enhances the validity of comparing non-matched cases by using propensity scores to weight the samples. This minimises the confounding that otherwise frequently occurs in cohort studies of the effects of treatment on outcome. In addition, IPTW enables estimation of marginal or population-average treatment effects (20).
Characteristics of patients with BCLC stage 0 and A HCC and portal hypertension according to treatment group.
Patients and Methods
Patients and inclusion criteria. This study included 88 consecutively enrolled patients with primary HCC who had undergone LLR, OLR or RFA at the Hiroshima Red Cross and Atomic Bomb Survivors Hospitals between January 2008 and September 2015. All patients had a maximum of three tumours each measuring <3 cm or a solitary tumour <5 cm and had been diagnosed as having PHT, defined as the presence of EVs and/or a platelet count of <100,000/μl in association with splenomegaly. Presence of EVs was determined preoperatively based on upper gastrointestinal endoscopic findings (21). Splenomegaly was diagnosed as present when spleen length exceeded 10 cm on preoperative computed tomography (CT) (12, 22).
Diagnoses of HCC were made by pathological examination in the LLR and OLR groups and by pre-treatment imaging studies, including abdominal contrast-enhanced dynamic CT and gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) enhanced magnetic resonance imaging (MRI), in the RFA group (23, 24). Serious postoperative complications were defined as Clavien-Dindo grade III or higher (25, 26).
Surgical and RFA procedures and basis of treatment decisions. The indications for open and laparoscopic surgery were based on the patient's daily living activities, age and fitness, tumour location, degree of tumour invasion, extent of resection and remnant liver function (27-31), with the choice of these procedures mainly being based on Child-Pugh class and results of indocyanine green retention test at 15 min (ICGR15). If their liver function was Child-Pugh class A or B and ICGR15 ≤45%, they underwent LLR or OLR. If their ICGR15 was greater than 45%, 99mTc-GSA scintigraphy was performed. If this showed an ICGR15 of less than 45%, they underwent partial resection (32). Selection criteria for LLR depended on tumour location (mainly liver wedge and surface) and specific requests from the patient for laparoscopic surgery. The presence of ascites was considered an absolute contraindication to resection. The open hepatectomy procedures performed have been described elsewhere (28, 29). For laparoscopic liver resection, patients were placed in a supine position with the primary surgeon positioned on the right side of the patients and scopist positioned between legs. When lesions were located in the right lateral sectors, patients were placed in the left-lateral decubitus position. Trocars were inserted using an open technique. Until 2011, LLR procedures involved inducing a carbon dioxide (CO2) pneumoperitoneum and performing parenchymal division with a Cavitron Ultrasonic Surgical Aspirator (CUSA) system (Valleylab™, Boulder, CO, USA) under direct vision through a small laparotomy incision after minimal mobilization of the liver (33). From 2012, pure LLR was introduced in our institution.
Prognosis of 88 patients with Barcelona Clinic Liver Cancer (BCLC) stage 0, advanced hepatocellular carcinoma (A HCC) and portal hypertension (PHT) who underwent laparoscopic liver resection (LLR group; n=20), open liver resection (OLR group; n=48) or percutaneous radiofrequency ablation (RFA group; n=20) therapy. a: Unweighted (crude) Kaplan-Meier curve showing the percent overall survival (OS) after treatment. The OS of the LLR group (black line) was comparable to those of OLR (dotted line) and RFA group (small dotted line). b: Unweighted (crude) Kaplan-Meier curve of recurrence-free survival (RFS) after treatment. Five-year RFS was significantly better in the LLR (black line; p=0.0007) and OLR (dotted line; 0.0002) groups than in the RFA group (small dotted line).
All patients scheduled for RFA underwent outpatient abdominal ultrasonography (US) to assess the feasibility of US-guided percutaneous RFA (34). The technique and therapeutic strategy for RFA were the same as previously described (35). RFA treatment was selected for patients with Child-Pugh class A or B, contraindications to surgery or general anaesthesia, and those who requested RFA. Exclusion criteria for RFA comprised segment I tumours, tumours overhanging the liver margin and in close proximity to adjacent organs, such as the gallbladder, stomach or colon, and tumours located near major intrahepatic vessels. Written informed consent was obtained for surgical treatment or RFA therapy from all patients, in accordance with our hospital's guidelines. Decisions regarding treatment modalities were made at a meeting attended by all participating surgeons and gastroenterologists. Our study protocol conforms to the updated ethical guidelines of the 2013 Declaration of Helsinki and was approved by our Institutional Review Board.
Survival and recurrence. Patients underwent blood tests and CT scans every 3 months after LLR, OLR or RFA. Recurrence was diagnosed based on imaging findings. Patients with intrahepatic distant or local recurrence were variously managed with ablative therapy, transcatheter arterial chemoembolization, surgery, laparoscopic microwave coagulation therapy, radiation or best supportive care. In case of death, survival time after surgery and cause of death were recorded.
Location in liver, surgical procedure, and resection weight of HCC tumors according to treatment group.
Univariate and multivariate analyses of overall survival in all HCC patients with portal hypertension.
Statistical analysis. For continuous variables, the Wilcoxon rank-sum test was used for non-parametric analyses. Categorical variables were compared using the χ2 test. The Kaplan-Meier method was used to construct overall survival (OS) and recurrence-free survival (RFS) curves. All survival curves were compared using log-rank tests. To identify independent predictors of OS and RFS, factors with p-values <0.10 on univariate analysis were subjected to multivariate analysis using a Cox proportional hazards model.
IPTW analysis was used to overcome possible patient selection bias. Using propensity scores from the IPTW analysis, a logistic regression model was used to calculate the probability of each patient receiving LLR, OLR or RFA on the basis of clinicopathological variables. After IPTW balancing of the three groups, differences in OS and RFS rates were tested by Cox regression analyses. Statistical significance was defined as a p-value <0.05. Statistical analyses were performed using the R statistical programming environment and JMP 9.0 software (SAS Institute, Cary, NC, USA) (20).
Results
Clinicopathological characteristics of patients with BCLC stage 0 and A HCC and PHT according to treatment group. The preoperative background characteristics of the study patients according to treatment group are summarized in Table I. Compared to the OLR group, the LLR group had a greater mean age (74 vs. 70 years, p=0.007), mean serum albumin concentration (4.2 vs. 3.8 mg/dl, p=0.027) and mean platelet count (12 vs. 9×104/μl, p=0.003), lower Child-Pugh score (5.2 vs. 5.6 points, p=0.03), shorter mean operation time (185 vs. 232 minutes, p=0.03) and less mean blood loss (241 vs. 412 ml, p=0.02). Compared to the RFA group, the OLR group had a lower mean age (70 vs. 73 years, p=0.02). Compared to the RFA group, the LLR group had a greater mean serum albumin (4.2 vs. 3.7 mg/dl, p=0.003) and greater mean platelet count (12 vs. 8×104/μl, p=0.003). Table II shows tumour numbers, locations and weights, as well as surgical procedures performed according to treatment group. The mean resection weights were 53 and 93 g in the LLR and OLR groups, respectively (p=0.08, Table II).
Univariate and multivariate analyses of recurrence-free survival in all HCC patients with portal hypertension.
Crude OS and RFS (before IPTW adjustment). The median duration of follow-up after surgery was 29.3 months (range=0.3-89.2). There were no significant differences in OS between the LLR, OLR and RFA groups (Figure 1a). By contrast, RFS at 1, 3 and 5 years was significantly better in the LLR than in the RFA group (p<0.0007): 81.3%, 46.3% and 30.9% vs. 36.8%, 9.2% and 0%, respectively (Figure 1b). RFS at 1, 3 and 5 years was also significantly better in the OLR than in the RFA group (p<0.0002): 76.4%, 53.2% and 44.2% vs. 36.8%, 9.2% and 0%, respectively (Figure 1b).
The clinicopathological factors were evaluated to identify those associated with worse OS (Table III). In univariate analyses, p-values for treatment with LLR or OLR, presence of oesophageal varices, platelet count, ICGR15, Child-Pugh score and α-fetoprotein (AFP) were all <0.10. The inclusion of these seven variables in a multivariate Cox proportional hazards analysis showed that platelet count (p=0.046; hazard ratio (HR)=0.83; 95% confidence interval (CI)=0.64-0.99) was the only independent predictor of worse OS.
The clinicopathological factors were also evaluated to identify those associated with worse RFS (Table IV). In univariate analyses, p-values for treatment with RFA or OLR, serum total bilirubin and platelet count were all <0.10. The inclusion of these four variables in a multivariate Cox proportional hazards analysis showed that RFA (p=0.02; HR=3.02; 95% CI=1.53-4.29) was the only independent predictor of worse RFS.
Postoperative complications, hospital stay and recurrence pattern. In the OLR group, one in-hospital death secondary to sepsis occurred. This group had a significantly higher rate of postoperative complications than the RFA group (47.9 vs. 20%; p=0.032, Table V). There were no significant differences in overall postoperative or serious postoperative complications between the LLR and RFA groups. The OLR group had a significantly higher rate of postoperative ascites and pleural effusion than the LLR group (18.8 vs. 0%; p=0.038).
Patients in the OLR group also had a significantly longer postoperative hospital stay (median=15 days; range=7-66) than the RFA group (median=8 days; range=2-14, p<0.0001). There was no significant difference in postoperative hospital stay between the LLR (median=9 days; range=7-14) and RFA groups.
Complications and mortality of HCC patients with portal hypertension according to treatment group
At the time of data collection, tumour recurrence had developed in 8/20 patients in the LLR group, all of which (100%) were intrahepatic distant recurrences. In the OLR group, tumour recurrence had developed in 21/48 patients: 20 (95.2%) being intrahepatic distant and one (4.8%) local recurrence, whereas in the RFA group in 13/20 patients had developed recurrences: 9 (69.2%) intrahepatic distant and four (30.8%) local recurrences. The frequencies of intrahepatic distant and local recurrence differed significantly between the OLR and RFA groups (p=0.038).
OS and RFS after IPTW adjustment. Among the 20 clinicopathological variables (age, sex, hepatitis B surface antigen (HBsAg)-positive, anti-hepatitis C virus (HCV) antibody, oesophageal varices, splenomegaly, ASA status, albumin, total bilirubin, aspartate transaminase, alanine transaminase, creatinine, prothrombin activity, platelet count, ICGR15, Child-Pugh score, AFP, maximum tumour diameter, number of tumours and blood transfusion), the distributive covariates that differed between the LLR and RFA groups were serum albumin concentration and platelet count (Table I). After IPTW adjustment of these two covariates, the weighted RFS of the LLR group remained significantly higher than that of the RFA group (p=0.049; adjusted HR= 0.39; 95% CI=0.12-1.24; Figure 2a). However, there was no significant difference between the weighted OS of the LLR and RFA groups (p=0.70, adjusted HR=0.61; 95% CI=0.05-7.63).
The distributive covariates that differed between the OLR and RFA groups were age, maximum tumour diameter and blood transfusion (Table I). After IPTW adjustment of these covariates, the weighted RFS of the OLR group remained significantly higher than that of the RFA group (p=0.004, adjusted HR=0.24; 95% CI=0.12-0.47; Figure 2b), whereas there was no significant difference between the weighted OS of the OLR and RFA groups (p=0.60, adjusted HR=1.34; 95% CI=0.39-4.64).
By contrast, there was no significant difference between the weighted OS and RFS of the LLR and OLR groups (p=0.11, adjusted HR=0.30; 95% CI=0.04-2.16 and p=0.88, adjusted HR=1.07; 95% CI=0.43-2.68, respectively; Figure 2c), considering the distributive covariates that differed between the LLR and OLR groups were age, serum albumin concentration, Child-Pugh score, operation time and blood loss (Table I).
Discussion
Treatment of patients with cirrhosis, HCC and PHT is currently hampered by a shortage of liver graft donors (15). According to the BCLC and AASLD guidelines, RFA is a valid option for the treatment of such patients; however, RFA has high local recurrence rates and there are some technical problems, such as locating tumours (19, 35, 36). We have reported that OLR may be an appropriate treatment option for patients with BCLC stage 0 or A HCC and PHT (36). However, in that study, RFA treatment was associated with fewer postoperative complications than OLR (36). We, therefore, decided to investigate the usefulness of treatment with the minimally invasive procedure of LLR in patients with cirrhosis, HCC and PHT postulating that this might further reduce the incidence of postoperative complications. In this study, we demonstrated that, in patients with primary HCC of BCLC stages 0 and A, the outcomes of the LLR group were better than those of the RFA group. Additionally, there was no significant difference in rate of postoperative complications between the LLR and RFA groups. To the best of our knowledge, this is the first study to report the feasibility of LLR treatment in cirrhotic patients with HCC and PHT.
Being a minimally invasive procedure, LLR results in less disturbance of collateral blood and lymphatic flow and, therefore, less bleeding, postoperative ascites and mesenchymal injury caused by compression of the liver than does laparotomy, which involves liver mobilization (37). During laparotomy, mobilization and detachment of the liver can result in massive postoperative ascites, pleural effusion and bleeding, especially in patients with PHT because collateral veins and lymphatic vessels have developed around the liver and peritoneum. Such complications, which can lead to severe postoperative liver failure, are less frequent and severe after LLR than after OLR, partly because the small incision of LLR limits evacuation of ascites, thereby reducing the risk of infection with the dual benefits of less need for fluid infusion and improved reabsorption of ascites.
Kaplan–Meier curve after inverse probability treatment weighting (IPTW) showing recurrence-free survival (RFS) after treatment. a: The RFS (%) of the laparoscopic liver resection (LLR) group (black line) was significantly better than that of the RFA group (dotted line, p=0.049). b: The RFS of the open liver resection (OLR) group (black line) was significantly better than that of the RFA group (dotted line, p=0.004). c: The RFS of the LLR (black line) and OLR (dotted line) groups were comparable (p=0.88).
In our study, the reasons for better RFS of the LLR and OLR groups than the RFA group are not clear; however, at least in theory, LLR and OLR have the advantage of offering better local control of HCC because of complete resection of HCCs, whereas RFA carries a potential risk of local recurrence associated with insufficient ablation (38).
Moreover, despite significant differences in recurrence rates, survival rates were similar in the LLR, OLR and RFA groups. Patients were followed up every 3 or 4 months after RFA treatment, enabling early detection of recurrence and more rapid and appropriate treatment with surgical or interventional methods, which may explain why the high HCC recurrence rate in the RFA group was not associated with a detectably higher death rate.
There are several technical aspects to performing LLR in patients with cirrhosis, HCC and PHT. When the five or six surgical trocars are inserted into the abdominal wall, care must be taken to avoid injuring the collateral veins that have developed in the abdominal wall. Since these patients often have thrombocytopenia, injury to vessels in the abdominal wall by insertion of trocars is one of the major causes of postoperative bleeding. In this study, none of the tumours were located in Segment I in the liver; however, we believe that LLR is indicated for all HCCs, regardless of their locations in the liver. In general, it is technically difficult to resect tumours located in Segment VII and VIII by LLR; however, resection of tumours in these locations can be facilitated by placing the patient in a semi-prone position and using a transthoracic trocar (39). On the other hand, since this approach requires only a small incision, LLR has a technical advantage over OLR when resecting tumours located on the edge and surface of liver in Segment II-VI. However, central lesions in patients with cirrhosis, HCC and PHT that would require massive hepatectomy may be better treated by RFA to minimize the risk of postoperative liver failure. To prevent intraoperative bleeding during hepatectomy by LLR, it may be necessary to implement both Pringle's manoeuvre (40) and precoagulation methods (41) to achieve hemostasis during liver parenchyma transection before hepatectomy. These techniques may make the LLR in patients with cirrhosis, HCC and PHT more feasible and effective in the future.
Our study had several limitations. First, the sample was relatively small, which made it difficult to evaluate the outcomes of HCC patients with BCLC stage 0 and A disease and PHT who underwent LLR, OLR or RFA. However, no previous study has compared these three treatment groups. Second, our findings may not be applicable in other centres because this was a single-centre retrospective study and not a randomized controlled trial, which is why we used IPTW to adjust for selection bias. Additionally, LLR in patients with PHT is a technically challenging procedure. Technological advances and accumulation of experience may make LLR in such patients safer and more effective and, thus, achieve better OS than OLR and RFA. In the future, more favourable outcomes may be achieved in carefully selected patients in high volume centres.
In conclusion, our findings that the RFS of the LLR group was significantly better than that of the RFA group and also the fact that there was no significant difference in incidence of postoperative complications between the LLR and RFA groups indicate that LLR may be a feasible treatment option for patients with BCLC stage 0 or A HCC and PHT. Good candidates for LLR include patients with tumours on the liver edge or surface and in whom ablation therapy would be difficult.
Acknowledgements
The Authors thank Dr. M. Fujiwara at the Hiroshima Red Cross Hospital and Atomic Bomb Survivor Hospital for diagnosing the pathological characteristics of HCC.
Footnotes
Conflicts of Interest
The Authors declare that they have no conflict of interest.
- Received April 23, 2016.
- Revision received May 28, 2016.
- Accepted May 30, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
