Radiofrequency Ablation for Colon and Rectal Carcinoma Liver Metastases: What’s Missing?

Abstract

Radiofrequency ablation (RFA) of liver tumors has quickly gained acceptance since its introduction in the early 1990s. This technique is increasingly being used to treat liver metastases from colorectal (CRC) cancers. Most of the published trials of RFA in CRC patients are retrospective analyses; no prospective comparisons of RFA vs. resection or chemotherapy are yet available. Nevertheless, conclusions that can be safely drawn from available data include that RFA effectively destroys very small tumors, it is minimally invasive, is better able to spare uninvolved liver than is liver resection, and is associated with lower treatment costs than surgical resection. The major limitation of RFA is the high rate of local recurrence in treated patients. Potential contributors to the high local recurrence rate are discussed, including the current RFA technique, large blood vessels serving as “heat sinks,” potential islands of viable tumor cells within an ablated zone, and the need for improved methods of real-time feedback. Improved multidisciplinary evaluation and treatment planning are needed for better management of patient candidates for RFA. Furthermore, prospective studies with defined end points and detailed documentation of the RFA technique are needed.

Since the time of its introduction by Italians in the early 1990s, radiofrequency ablation (RFA) of liver tumors has quickly gained acceptance, and its use has grown rapidly in the past 5 years. The first use of RFA for treatment of a human liver tumor in the United States was in 1996, representing a collaborative effort between one of the authors (K.K.T.) and a radiologist (Dr. Nahum Goldberg) at the Massachusetts General Hospital. Since then, RFA has gradually replaced percutaneous ethanol injection as the preferred method for local ablation of hepatocellular carcinoma worldwide. It has also replaced cryosurgery as the preferred method of ablation of colon and rectal carcinoma liver metastases. While it was initially tested in hepatocellular carcinoma, RFA is now being used in colorectal carcinoma (CRC) liver metastases and other secondary liver tumors, primary lung tumors, lung metastases, renal cell carcinoma, and bone metastases.^1–5

PUBLISHED RFA STUDIES AND THEIR POTENTIAL LIMITATIONS

The list of peer-reviewed publications touting RFA efficacy in patients with CRC is rapidly increasing. A majority of the published literature on clinical outcomes describes retrospective analyses of RFA-treated patients. Conclusions regarding RFA efficacy that can be safely drawn from these publications are limited by the shortcomings of most reports. There are no prospective randomized clinical trials comparing RFA to resection or chemotherapy for treatment of CRC liver metastases. Thus, the most significant limitation of the existing reports is the absence of appropriate control groups. Unfortunately, too often, the selected control group has greater tumor burden than the RFA-treated group (which is why they were not treated with RFA). Another shortcoming of most reports is the absence of a robust definition of local recurrence, which is an important end point. In addition, studies frequently include patients who have been at risk for less than 6 months following RFA. A vast majority of the reports are from a single center, and since each center performs RFA a bit differently, extrapolation of results to a broad set of providers may not be reliable. Moreover, comparison of results between centers is precluded by these differences in technique.

A study of percutaneous RFA conducted by a very experienced Italian group illustrates the challenges in interpreting published reports.⁶ A total of 117 patients with 179 metastases ranging from 0.9 to 9.6 cm in diameter were treated by percutaneous RFA using a cooled tip electrode under ultrasound guidance. Based on the observation that estimated 1-, 2-, and 3-year survival rates of 93%, 69%, and 46% were similar to those in published reports of surgical resection, the authors concluded that RFA is an effective method to treat hepatic metastases from CRC. However, analysis of local recurrence rates, stratified by lesion size, reveals the chief deficiency of RFA technology: the overall local recurrence rate was 39%, and was even higher (68%) for tumors larger than 4.1 cm. Notably, survival data of this single-arm study reveal more about eligibility criteria, whereas the local recurrence rates reflect the efficacy of RFA as a technology. In other words, survival statistics from reports of uncontrolled clinical studies—which represent the vast majority of the RFA literature—are not reliable indicators of RFA efficacy. Rather, survival statistics reflect what types of patients were included in the study. On the other hand, reports of recurrence patterns provide valuable information regarding efficacy of the modality.

Another example of the difficulties associated with interpretation of retrospective RFA studies comes from a report of intraoperative RFA in 418 patients.⁷ The study cohort comprised patients who underwent laparotomy for isolated liver metastases and were found to have extent of disease that could be treated by resection alone, RFA alone, resection combined with RFA, or that was unresectable. The overall survival of the two RFA groups was poorer than in the group treated by resection alone. Furthermore, survival of patients treated by RFA alone or resection combined with RFA was better than that of patients treated only with chemotherapy. Consequently, the observed survival differences more likely resulted from differences in extent of disease in patients in each group, rather than from differences in efficacy of the various treatment modalities. Nonetheless, local recurrence rates in this type of study are informative regarding RFA efficacy; the 9% local recurrence rate in RFA-treated patients in this study demonstrates the potential efficacy of this treatment modality.

Despite the limitations of the published studies, several conclusions regarding RFA for CRC liver metastases may be safely drawn from these studies. First, RFA effectively destroys tumors when they are very small, and some patients are “cured” with RFA alone (although most of the cured patients are those with hepatocellular carcinoma rather than CRC metastases). Second, despite the relatively rare occurrence of peri-procedural complications, RFA is minimally invasive. Third, RFA is better able to spare uninvolved liver than is liver resection. Finally, RFA treatment cost is much lower than that of surgical resection; however, frequent overuse of RFA (eg, when not indicated) increases its overall cost.

RFA AND LOCAL RECURRENCE RATES — WHAT IS MISSING?

Treatment Planning and Technique

Recognizing these general advances in the field, we address the question, “What is missing?” Undoubtedly, the biggest limitation of RFA is the relatively high rate of local recurrence in treated patients. Perhaps the most frequent cause of local recurrences is use of overlapping zones of ablation that do not completely overlap. For example, to ablate a 3-cm tumor with 1-cm margins around the tumor, a 5-cm zone of ablation is required. One of the most commonly used and reliable RFA probes creates a 3.0- to 3.5-cm zone of ablation. Mathematical calculations reveal that 14 overlapping ablation spheres, each with a 3.0-cm diameter, are required to completely destroy the 5.0-cm sphere.⁸ In addition, the 14 zones need to be positioned perfectly relative to the 3-cm diameter target tumor—a daunting task.

Thus, the “pullback” technique was developed, whereby, with a single electrode position, a spherical ablation shape is first created, and then the electrode is simply pulled out slightly and the process is repeated. Several repetitions of this process create an ablation volume that is cylindrical in shape. By overlapping these cylinders, one can destroy a large tumor. However, while one envisions placement of electrodes parallel to each other to precisely overlap the cylinders, the angle of an electrode position may easily be placed off-target by a few degrees, especially when long distances are traversed to reach the tumor; and the absence of overlap of the cylindrically shaped ablation volumes leads to local recurrences. An additional challenge is that not all tumors are shaped alike; some have a spherical shape, whereas others are shaped like a dumb-bell, egg, or disk.

These difficulties in accurate electrode placement result in imperfect treatment planning. RFA treatment planning is now at the stage where radiation oncology was in its infancy. Somewhat remarkably, the current methodology of tumor ablation involves mentally envisioning the desired location of the ablation zone, and then use of the right side of the brain to integrate the images and guide the electrode into the proper position. While imaging (eg, ultrasound or computed tomography [CT]) can confirm the accuracy of electrode positioning to some degree, the inability to judge the extent of overlapping ablation zones with these radiology tools still results in much of the treatment planning being performed in the operator’s mind. In many ways, this current state of RFA treatment planning seems akin to simply using hand-eye coordination to point a radiation source for radiotherapy. There is substantial room for improvement in this particular area.

Treatment planning is further hampered by zones of ablation that are not reproducible in size or shape from patient to patient. Even if one uses the same electrode and energy in patient after patient, the size of the ablation zone may differ by up to 30%; even with smaller variances (eg, 10%) between patients in the size of an ablation zone, this difference could understandably lead to local recurrences.

Other Contributors to Local Disease Recurrence

Another cause of local recurrences is thermal protection of tissues by large blood vessels. Blood flow in vessels carries heat away from tissues directly adjacent to the vessel, and thereby maintains the tissue temperature at 37° C. Thus, larger blood vessels are referred to as “heat sinks,” and tumor cells on these vessels are extremely difficult to heat via RFA (Figure 1). Smaller blood vessels are commonly thrombosed during treatment, and thus do not serve as heat sinks. One approach to this problem is to temporarily interrupt blood flow during RFA. Ablation volumes are clearly larger when blood flow is temporarily stopped (Figure 2).⁹ Moreover, tumor cells adjacent to blood vessels in which flow is stopped are no longer thermally protected by the heat sink effect. Of note, it is not feasible to stop blood flow in the vena cava or large hepatic veins during RFA.

Figure 1.

Computed tomography (CT) images of local recurrence following radiofrequency ablation (RFA) of colorectal carcinoma liver metastases. The metastases are shown prior to ablation (left panel) and immediately following ablation (center panel). Close inspection of the contour of the ablation zone reveals that it is sculpted by a small blood vessel, suggesting thermal protection of the tumor cells adjacent to the vessel. The right panel shows extensive local recurrence 6 months after RFA.

Figure 2.

Computed tomography (CT) image shows two colorectal carcinoma liver metastases of equivalent size. Although both metastases were treated by radiofrequency ablation (RFA) using an identical amount of energy, one of the ablation zones is significantly larger than the other. The larger, anterior ablation zone was created by temporarily interrupting hepatic arterial and portal venous blood flow during the ablation. This maneuver reduces the amount of heat drawn away from the area by blood flow. This maneuver can also be used to increase the likelihood of destruction of tumor cells adjacent to large blood vessels.

Islands of viable tumor cells within the ablation zone are another potential source of local recurrence. RFA ablation electrodes are designed and tested using normal animal livers, which have little heterogeneity in their consistency. Unlike normal animal livers, however, each human liver tumor is unique, and many are heterogeneous in makeup. We and others have observed nests of tumor cells within an ablation zone. While the RFA-treated cells appear morphologically bizarre and their viability is admittedly difficult to determine, some of the cells may be viable such that they give rise to local recurrences.⁹

Finally, local recurrence rates after RFA are most likely influenced by operator experience, commitment, and patience, although there are no published studies on these factors in relation to treatment outcome.

WHAT IS MSSING IN RFA TECHNOLOGY?

Many potential future developments in RFA technology would benefit the field. Most important is the need to be able to create reproducible, larger ablation zones with a single electrode position: perhaps “reproducible” deserves more emphasis than “larger.” Numerous strategies are being examined for creating larger zones of ablation (Table 1).

Table 1.

Strategies to increase the size of radiofrequency ablation (RFA) zones created from a single electrode position.

Active cooling of electrode Vascular inflow occlusion Multi-electrode arrays Larger electrode arrays Pulsed energy delivery Increasing tissue conductivity Combine with other modalities (eg, embolization) Injection of hypertonic saline prior to RFA Enhancing tissue sensitivity to thermal injury

Improved methods of real-time feedback are needed for simultaneous monitoring of the tumor, the boundary between tumor and normal tissues, electrode position, and the evolving ablation zone; and this needs to be achieved in three—not just two—dimensions. Presently, ultrasound is the most commonly used imaging modality for RFA treatment. Ultrasound is two-dimensional, and while it is dynamic and real-time, it is difficult to visualize the ablation zone because ablation creates echogenicity (Figure 3). CT and magnetic resonance imaging (MRI) have some advantages, but are generally not portable to the operating room.

Figure 3.

Ultrasound images of metastasis prior to radiofrequency ablation (RFA) (left) and during RFA (right). The ultrasound echogenicity of the RFA zone obscures views of the deep margin.

The optimal technique to perform RFA also has not been defined, and differs at various centers. RFA may be performed during open laparotomy, via laparoscopy, or via a percutaneous approach. Each of these approaches has its advantages, as shown on Table 2. Comparisons of results between published studies should control for differences in these approaches. Other unanswered issues include that the requisite ablation margin remains unknown; the best method to follow lesions radiographically after treatment lacks consensus; and one of the most important unresolved issues is which patients to treat.

Table 2.

Relative advantages of different approaches to radiofrequency ablation (RFA) of liver tumors.

Laparotomy	Laparoscopy	Percutaneous
	Minimally invasive	Least invasive
Visual inspection for extrahepatic metastases	Visual inspection for extrahepatic metastases	CT or MRI guidance may be used
Thorough ultrasound examination of liver	Relatively complete ultrasound examination of liver	Least costly
May be combined with visceral resection (colon or liver)	May be combined with visceral resection (colon or liver)
Readily access all areas of liver	Readily access most areas of liver
Readily combined with portal inflow occlusion	May be combined with portal inflow occlusion
May dissect away organs adherent to tumor	May dissect away organs adherent to tumor

Abbreviations: CT = computed tomography scan; MRI = magnetic resonance imaging.

RFA FOR RESECTABLE CRC LIVER METASTASES

Should RFA be used for resectable CRC metastases? Most experts concur that RFA should not be used in lieu of resection, when resection is feasible. The challenge, however, is in defining what is unresectable. Based on many advances in liver surgery, extensive liver resections are being performed more commonly and more safely. Strategies such as preoperative portal vein embolization,^10,11 neoadjuvant chemotherapy,^12,13 and staged, sequential liver resections allow for resection of liver metastases in many patients who were previously classified as unresectable.

Equally important is the question of who determines which tumors are unresectable. The concept of unresectable is different to a radiologist than to a surgeon. Furthermore, a general surgeon and a liver surgeon may have different ideas about which lesions are resectable. More to the point, who decides which patients should undergo tumor ablation? These questions highlight an important entry on the list of “what’s missing,” namely, multidisciplinary evaluation and treatment planning. Unfortunately, it is not uncommon to observe that some radiologists unilaterally recommend percutaneous ablation, non-liver surgeons unilaterally recommend operative RFA rather than referral to a liver surgeon who can appropriately assess resectability, and liver surgeons unilaterally recommend RFA “debulking” added to chemotherapy with hopes that this approach is superior to chemotherapy alone.

WHAT IS MISSING IN CURRENT DATA SETS?

Shortcomings in data sets are another area for improvement in determining optimal clinical use of RFA. First, a common terminology and set of definitions are absent, including definitions of important end points such as local recurrence. Second, techniques are often not documented with meticulous attention to detail, recognizing that every center performs RFA differently. In addition, the RFA technology is evolving rapidly, such that in many published studies, not all patients were treated with the same RFA technique. Some data sets do not sufficiently distinguish between various tumor histologies. In many data sets, patient follow-up is censured at first recurrence, which artificially lowers reported local recurrence rates for each treated lesion. Finally, the vast majority of data sets lack the appropriate control group needed to draw inferences regarding RFA efficacy.

With these limitations in data sets, it is still not possible to answer the following questions with a high degree of statistical certainty: First, is RFA of a limited number (eg, 1 to 2) of small metastases (eg, < 3.0 cm) as effective as surgical resection? Second, does RFA of unresectable CRC liver metastases provide a measurable prolongation of survival? Third, in patients with unresectable CRC liver metastases, is chemotherapy combined with RFA superior to chemotherapy alone? Some of these questions are being addressed by the ongoing European CLOCC (Chemotherapy + Local Ablation vs. Chemotherapy) trial, in which patients with unresectable liver metastases are randomly assigned to chemotherapy alone, or chemotherapy combined with RFA and surgery for their liver metastases.¹⁴ This is an important study, which, unfortunately, has been hampered by slow accrual (target accrual is 152 patients).

CONCLUSION

In the future, RFA devices that create larger ablation zones with a single electrode position will be required. While significant progress has already been made on this front through use of larger probes and more powerful generators, perhaps alternative energy sources (eg, microwave) will be useful. Strategies to increase tumor cell susceptibility to lower temperatures remain ripe for research and development. Technology is also needed to significantly enhance real-time feedback regarding ablation zone relative to location of tumor, and three-dimensional treatment planning. More important than technologic improvements is the need to enhance multidisciplinary treatment planning in all hospitals and centers involved in RFA. Finally, it is necessary to accrue patients to clinical trials to provide appropriate control groups in studies of RFA.