Minimizing Complications With Radiofrequency Ablation for Liver Cancer

Surgical resection has long been considered to be the only treatment to offer prolonged survival to patients with primary or metastatic cancer confined to the liver. Unfortunately, most of patients with hepatic malignancy have disease that is not amenable to surgical resection because of large tumor size, multifocal disease, inadequate predicted liver reserve, or comorbid medical conditions precluding safe operation. These facts led to the recent development of multiple nonresectional modalities for ablating liver tumors, one of which is radiofrequency ablation (RFA).¹ When used properly in appropriately selected patients, it is hoped that RFA will improve survival for some patients with hepatic cancer who cannot undergo liver resection.

The safety profile of RFA for liver tumors was initially established through uncontrolled series of heterogenous groups of patients treated with various techniques; overall complication rates appear to be less than 10% and procedure-related mortality is less than 1% in experienced hands.^1,2 This safety profile led the Food and Drug Administration to approve RFA devices for clinical use in The United States. It is important to note, however, that this approval was granted for the ability to destroy tissue and not for the ability to completely destroy or cure cancer. There has been no controlled prospective clinical trial published demonstrating that RFA extends the survival of patients with any type of cancer of the liver.² In addition, a recent technology assessment on RFA for the treatment of liver cancer emphasized the substandard quality of data on the subject.² Despite these issues, aggressive marketing programs by the manufacturers of RFA devices led to the rapid diffusion and worldwide use of RFA for metastatic and primary liver cancer. Even more concerning is the fact that RFA is widely perceived as a safe and “minimally invasive” treatment modality for unresectable liver cancer although a clear understanding of its limitations and risks are not clearly delineated.

Strasberg and Ludbrook recently published an eye-opening and insightful work in which they discussed the need for regulatory oversight to monitor the implementation of new surgical technologies and innovative practice.³ They argued that the clinical use of new technologies, such as RFA for the ablation of liver cancer, whose clinical efficacy or risks are not yet thoroughly assessed,² be restricted to clinical trials so that patient safety is monitored under the purview of human subjects research. Unfortunately, few studies exploring the utility of RFA for ablating liver cancers have been carried out and published under the auspices of a closely monitored clinical trial.

In this issue of the Annals of Surgery, 2 groups of investigators present the results of their ongoing clinical experience with RFA for the treatment of patients with primary and secondary liver cancer. Curley and colleagues from MD Anderson Hospital in Houston, Texas, and clinicians from The Pascale National Cancer Institute in Naples, Italy, discuss early (<30 days) and late (>30 days) complications after RFA of liver tumors in 608 consecutively treated patients. Poon and colleagues from The University of Hong Kong Medical Centre, Queen Mary Hospital, present the complication rates associated with RFA in their first 100 patients. The focus of the latter study was to emphasize the importance of the learning curve the authors acquired that led to a reduction in complications after RFA. In both studies, the investigators monitored the number of patient-, tumor-, and treatment-related variables and performed univariate and multivariate analysis in an effort to identify factors associated with RFA-induced complications. The principal investigators of these studies are to be congratulated for treating all patients with RFA in their institutions only after approval of their protocols by their institutional medical review boards.

Although the overall percentage of patients experiencing complications was low (<10%) in both reports, the incidence of complications in particular subgroups of patients was substantially higher (eg, 11%-15% in those undergoing open operation and 26% in those with an elevated bilirubin). In addition, the morbidity associated with many complications was high and required substantial additional interventions and hospitalization at the expense of patients’ quality of life. Patients with advanced cancer who experienced a serious complication did not appear to be well served by RFA. The unknown risks of subjecting patients with terminal cancer to serious complications, especially if they are asymptomatic, by a therapy that has yet to demonstrate an improvement in survival would be morally and ethically unjust outside of a clinical trial.

Strasberg argues there is a moral and ethical obligation on the part of treating physicians to explore the limits of new technologies, such as RFA, under the protection of human clinical research, and this is exactly what the investigators in Hong Kong, Houston, and Naples did. One of the most important aspects of these trials is related to the fact that there was 100% follow-up and that patients were closely and thoroughly monitored for the development of any complication or untoward events. For these reasons, these studies are of value to the RFA literature. However, there are substantial differences in design between these 2 studies, in the type of RFA probes used and in RFA technique, all of which prevent reliable comparisons between their respective outcomes. In addition, the heterogenous mixture of patients, changes in RFA technique over time and failure to report the outcomes according to tumor stage will prevent the data from being meaningfully combined with that from other studies. Nevertheless, some of the observations made by these investigators are worthy of discussion.

Curley's study identified 2 factors as being predictive of having an RFA-related complication: open operation and cirrhosis. Poon's study identified 4 factors that were significantly associated with having an RFA-related complication: elevated serum bili (> 20 μm/L), ablation of more than 1 tumor, open operation, and early experience (fewer than 50 cases). Both studies identified open operation as being associated with higher complications compared with the percutaneous route.

The significance of this finding is uncertain and confounded by the fact that patients treated by open procedure were different from those treated by the percutaneous route and were subjected to variations in RFA technique. In both studies, patients treated open had more extensive tumor burden. In the Curley study, all patients undergoing open operation had hepatic arterial and portal inflow occlusion for 2–3 minutes and many patients had a concomitant liver resection. Therefore, it is possible that complications sustained in patients undergoing open surgery were related to tumor factors or a Pringle maneuver and not to the “route of RFA therapy.” The volume of tissue necrosis and total delivery of current on complications are additional factors that could increase complications in patients treated by laparotomy and the importance of these factors could be tested in future clinical trials.

Even though tumor size or central location were not predictive for early or late complications in the Curley article, the potential negative impact of these factors cannot be dismissed. It is possible that some large, central tumors were incompletely ablated because of inadequate current deposition or a heat sink phenomenon whereas those that were completely ablated were the ones that resulted in biliary injury or abscess. Without knowledge about variables, such as total current delivery/volume of tissue, peak temperatures reached at the margins of ablated tumors, extent and thoroughness of the zone of necrosis, or type of tumor treated, it is unwise to discount the potential danger of ablating a large central tumor located in the posterior medial aspects of segments V, VIII, or IVb. The Hong Kong group, in fact, reported that their only biliary injury occurred in a patient whose tumor at the base of segment IVb was ablated.

Two-thirds of cirrhotics in the Curley study were Child Class B or C whereas the majority of patients in the Poon study were Child Class A. This difference may explain why Curley identified cirrhosis as a predictor of complication whereas Poon did not. Unfortunately, neither group stratified their complication rate by Child class, and so it remains unknown whether complications in these studies were related to the severity of cirrhosis. The observation by Poon that an elevated serum bilirubin above 20 μm/L was significantly associated with having a complication strongly suggest that the Child class, CLIP score, or other clinical staging system could be independent predictors of RFA-induced complications in patients with HCC and cirrhosis.

The Hong Kong group implemented RFA in 2001, well after the basic techniques and complications associated with RFA were defined and widely published. Hence, one would expect these clinicians to start higher up on the learning curve of RFA when compared with others, such as Curley, who acquired a very large earlier experience. The Hong Kong group also consists of world-class hepatobiliary surgeons with extensive experience in liver surgery and imaging. Despite these facts, Poon and his group demonstrated that the complications associated with RFA were reduced with increasing experience beyond their first 50 cases. For this reason, Poon suggests that future publications on hepatic RFA include a statement by the authors on their experience so that the results have more meaning and comparisons to other groups can be made legitimately.

The early complication rate reported by Curley's group in their first publication after treating their first 123 patients was 2.4%.⁴ This is in contrast to a short term complication rate of 8% (40 out of their last 483 patients treated) in the current publication. This represents a 300% increase in complication reported in the second publication despite the fact that the clinicians at both MD Anderson and the Pascale Cancer Institute are well beyond the learning curve of RFA. These results are in contrast to the Hong Kong group's experience, who reported a 400% reduction in RFA-associated morbidity over time as their experience increased beyond the first 50 patients.

A number of possibilities can be proposed to explain the reduction in morbidity experienced by the Hong Group and the increase in complication rate reported by the MD Anderson/Pascale Group. The Hong Kong Group refined their application of RFA over time through better targeting and by customizing their total current delivery and limiting the extent of tissue ablation. They also tended to avoid ablating central, larger tumors and they did not perform concomitant liver resection, as did the MD Anderson group.

In Curley's initial report,⁴ Child class C patients were excluded from treatment though in the current report, 2/3 of patients with HCC had Child class B or C cirrhosis. This indicates that the criteria for enrollment and patient selection changed over time in the MD Anderson/Italian study and could explain a higher complication rate in the patients treated over the past 4 years. In the MD Anderson/Pascale clinical trial, the RF generator was upgraded from a 100 W to 200 W system and the delivery of RF current was always dictated by the manufacturer's recommendations⁴ as opposed to being custom tailored to individual tumors as was the case in Hong Kong. One wonders whether the advances in generator technology and more inclusive patient selection are evidence of the authors’ willingness to treat patients with more advanced stages of cancer with more aggressive RFA.

Poon points out that when using a new technology like RFA, clinicians are obligated to audit their own performance in an effort to improve treatment results and reduce complications as they accumulate experience. I agree with his philosophy. However, I also believe that there is an even higher level of responsibility on the part of physicians planning to use RFA for the treatment of liver tumors and on the regulatory process of granting privileges for performing RFA. Physicians have a moral and ethical obligation to acquire a thorough understanding of the physics, technical limitations and application of RFA prior to using it in patients. Training and experience in tumor targeting should also be acquired using phantom models in simulated environments and not on patients.

RFA device-related technical issues and application factors are 2 areas that could potentially increase the risk for developing an RFA related complication and these factors were not investigated by either the MD Anderson or Hong Kong groups. Table 1 lists some but not all of the technique-related variables that could potentially influence complications following RFA of liver cancer. This list illustrates the complexity of the procedure and also emphasizes the need to describe the technique in detail in publications so that meaningful comparisons of outcomes between studies can be compared and data can be pooled.

TABLE 1. Technical Factors Potentially Affecting RFA-Induced Complications

The technical and clinical limits of RFA and potential pitfalls and problems related to its use were recently described in an article by Rhim and other leading experts in the field of RFA technology.⁵ An understanding of these limits is essential for any clinician using RFA for the treatment of liver tumors. Goldberg and other leaders in the field of RFA technology have more recently pleaded for the need to standardize the nomenclature and methods for reporting the results of RFA to better refine the technology and understand its clinical efficacy and limitations.⁶ The fact that these type of articles are being published by pioneers who developed RFA technology is testimony to the fact that RFA is a medical technology whose limits and application are not well understood at present and for which there is tremendous need for additional study and refinement in reporting.

An important question yet to be addressed by any clinical RFA study is whether or not residual microscopic disease in an RF-ablated liver cancer has any impact on patient survival. Whereas it is well established that incompletely resected metastatic or primary liver cancer does not confer additional survival, maybe this is not the case for patients undergoing incomplete liver tumor ablation. Should a small amount of residual tumor be subjected to repeat RFA? Will repeat ablation of a residual hepatoma make a difference in patient survival when 20% of patients will develop additional tumors elsewhere in the liver every year? At what expense and risk should patients be subjected to repeat or more extensive ablation to achieve 100% tumor necrosis? Complication rates following RFA will be increased if efforts are taken to achieve 100% necrosis for large and centrally located cancers. Certainly, a tumor-free margin and 100% tumor necrosis are preferred endpoints to strive for but these goals must also be weighed against the development of complications that impair patients’ quality of life. No clinical trial has addressed these questions and yet the answers are critically important for how RFA is applied clinically in the future.

Whether recent advances in RF technology or the manner in which it is used will impact patient outcomes is currently unclear and requires additional meticulous study. For example, larger multitined probe arrays, more powerful RF generators, the use of hypertonic saline,⁷ controlled hepatic hypoperfusion, and pre- RFA chemotherapy⁸ will enable clinicians to ablate larger liver tumors but this may result in higher complication rates. The expanded use of recently developed image guidance systems may improve tumor targeting and thereby reduce complications by avoiding critical vascular and biliary structures in the liver.⁹

Until there is evidence from well-controlled prospective clinical trials that RFA of liver cancer results in improved patient survival, it is essential that physicians employing RFA technology do everything possible to minimize the risk associated with this therapy. Both Curley and Poon's groups provide valuable information on the complications associated with RFA of liver cancer. Their observations were possible because of close and comprehensive follow-up of their patients. Their critical observations provide some guidance for clinicians and investigators using RFA in the management of patients with liver cancer with the hope that major complications such as hepatic abscesses, biliary complications, delayed tumor bleeding, and tachyarrhythmias can be avoided in future patients. However, to better understand the limits and further reduce the risks of RFA, additional study is necessary in which larger numbers of patients are enrolled in even more strictly controlled prospective clinical trials than those carried out by Curley and Poon.

To date, standardized means of reporting have not been established or used to describe the outcomes of patients undergoing RFA.² To rectify this problem, it is essential that investigators provide detailed descriptions of individual patients and their tumors along with detailed descriptions of the technical aspects of RFA using standardized nomenclature and reporting systems.⁶ In addition, to understand how RFA influences the survival of patients with liver cancer it will be necessary to stratify the results using internationally recognized tumor staging systems.¹¹ Finally, the temptation to use RFA as an alternative to surgical resection for patients with resectable liver cancer must be discouraged until prospective randomized clinical trials establish them as being equally or more effective as surgical resection.¹²