Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is one of the most common malignancies in the world with more than 1 million cases occurring each year worldwide (see Chapter 80; Nordenstedt et al, 2010). It is well known that HCC is the most common malignancy in Southeast Asia and in sub-Saharan Africa owing to endemic viral hepatitis, especially hepatitis B. Since the 1990s the incidence of HCC has increased significantly in the United States and other Western countries as hepatitis C rates have risen. In the United States, the mortality rate for HCC has improved over the past 2 decades, as screening programs have identified patients with tumors at earlier stages. Of patients presenting with HCC, only 15% to 25% are operable, and less than 10% are resectable with curative intent (Bolondi et al, 2001; Farmer et al, 1994); however, most patients with unresectable disease have tumor confined to the liver. At a regional referral center for liver cancer, curative treatments can be applied to 30% to 40% of patients (Bruix & Llovet, 2002). More than half of patients presenting with primary liver cancer are potential candidates for hepatic regional therapy, of which a significant number may be amenable to ablative techniques.

Metastatic Disease

The liver is one of the most common sites of metastatic spread, particularly from gastrointestinal malignancies (see Chapter 81A). Most malignancies that recur in the liver do so in combination with extrahepatic metastases, making resection or ablative regional therapies unsuitable management strategies. In contrast, recurrences from colorectal primary tumors are localized to the liver in approximately one third of patients, allowing use of potentially curative regional interventions. Only a small proportion of patients with metastases confined to the liver are operable, and an even smaller fraction are resectable, owing to technical and prognostic factors. By evaluating all patients with hepatic colorectal metastases at a single center, Scheele and colleagues (1995) found 21% of patients with synchronous disease and 51% of patients with metachronous disease to be potentially resectable.

Most patients with metastatic disease confined to the liver are not candidates for resection therapies but are potentially eligible for regional ablative treatments. Systemic and regional chemotherapy for metastatic colorectal cancer has improved dramatically, with median survivals of greater than 20 months now commonly reported (see Chapters 86 and 87). Additionally, response rates for systemic chemotherapy are generally greater than 50%, and combined with regional chemotherapy, they approach 90%. Survival at 5 years for patients after resection of colorectal liver metastases in modern series ranges from 46% to 58% (Karanjia et al, 2009; Pawlik et al, 2005). Resectional and ablative therapies now must be interpreted in the context of effective chemotherapy, opening the door for downstaging of unresectable tumors and adjuvant therapy for borderline resectable tumors.

Cooling Rate

The rate at which tissue cools affects the proportion of cells killed by a single freeze cycle. Maximal cell death is achieved at slow and rapid cooling rates, although from different mechanisms, whereas greatest cell survival is seen with intermediate cooling rates. Cellular dehydration causes lethal injury at slow cooling rates, whereas rapidly cooled cells are destroyed by the mechanical action of ice crystallization and expansion.

Depth of Hypothermia

Independent of the cooling rate, extensive tissue injury occurs at temperatures below −20° C, and temperatures below −40° C are lethal for almost all tumor cells. The mechanism of cell injury at these temperatures is the formation of intracellular ice. At higher temperatures, the rate of cooling influences the rate of ice formation and the mechanism of cell death; however, when the temperature is below −40° C, almost all of the water within the cell is frozen, ensuring complete tissue ablation. For cryotherapy to be reliable, all targeted tissue must reach these levels of hypothermia to achieve reproducible and certain cell death, the theoretic goal of cryotherapy. This level of hypothermia is not consistently reached at the periphery of the ice ball, where temperatures may be 0° C.

The sensitivity of different tissues to hypothermia varies considerably. Most normal hepatocytes die at −15° C to −20° C, whereas at −10° C most hepatocytes survive. Bile ducts, connective tissue, and vascular structures tolerate slightly lower temperatures than hepatocytes. In contrast, liver tumors tend to require deeper hypothermia to −40° C for complete and reliable cell death. As a general rule, the −40° C isotherm is located approximately three quarters of the distance from the probe to the edge of the ice ball as seen at intraoperative sonography. To reach this level of hypothermia and obtain reliable ablation at the tumor margin, the ice ball is extended 1 cm beyond the peripheral edge of the tumor.

Thawing Process

Further damage to tissue occurs during the thawing process and varies according to the rate of the thaw. Rapid thawing of frozen tissue tends to increase cell survival, whereas slow thawing is more destructive than either rapid or slow cooling. In slowly thawed tissue, the extracellular ice melts before the intracellular ice, briefly making the extracellular fluid relatively hypoosmolar compared with the intracellular fluid. Free water flows down this osmotic gradient into the cells, causing them to swell and ultimately burst. Simultaneously, the ice within the cell undergoes recrystallization, especially in the temperature range of −20° C to −25° C. Recrystallization is a process by which ice crystals reform, coalesce, and enlarge, mechanically disrupting the cellular membranes. The effects of thawing are potentiated by allowing the entire lesion to reach ambient temperature slowly and passively.

Microvascular Effects

In addition to the acute physicochemical and structural damage to the cell and its environment, the lethal effects of cryoablation are potentiated by disruption of vascular structures, which causes delayed hypoxia and necrosis (Rubinsky et al, 1990). The vascular effects are acute and chronic. Freezing at slow cooling rates causes the radius of the sinusoids to increase by a factor of 2, which is equivalent to increasing the volume in the intravascular space by a factor of 4. The expansion of the vascular space tears the endothelium, exposing and disrupting the underlying basement membrane. Platelet thrombi develop, and permeability increases, leading to swelling and microcirculatory failure. The tissues supplied by these damaged vessels become ischemic and necrotic. This mechanism of injury is more important in the intermediate and slow cooling zones, where direct cellular injury by intracellular ice formation or dehydration is not reliable. For maximal benefit, cryosurgery should be performed so that there is rapid freeze, slow thaw, and repeated freeze-thaw cycles; however, experimental evidence has shown that high-flow venules at the periphery of an ice ball are crucial in shutting down the microcirculation of the tumor, and repeated cycles do not seem to improve this effect (Richter et al, 2005). It also has been shown in experimental models that the serine protease inhibitor aprotinin decreases platelet trapping and improves tissue destruction in cryolesions (Kollmar et al, 2004). The effects of cryosurgery on tissues are a result of direct cellular damage secondary to physicochemical effects and indirect cell damage secondary to loss of cellular integrity and destruction of vascular channels.

Morphologic and Histologic Changes After Cryotherapy

Immediately after thawing, the area of liver treated with cryoablative techniques appears dark red, swollen, and usually well demarcated. Histologic examination reveals congestion of the hepatic sinusoids associated with hemorrhage, decreased levels of cellular glycogen, and an absolute reduction in the number of mitochondria. With time, the lesion takes on a light gray color and slowly resorbs over the ensuing weeks. Microscopically, this period correlates with polymorphonuclear cell and macrophage infiltration of the tissue. The cryolesion is progressively replaced with fibrous tissue, usually with a persistent area of necrosis and fibrosis, but occasionally the lesion is completely resorbed.

Immunologic Effects of Cryoablation

In early studies of cryotherapy, anecdotal reports of regression of untreated tumor sites suggested a generalized, tumor-specific immunologic response to the cryoablated lesion (Faraci et al, 1975). These effects have not been substantiated in animal or human models, although it is unlikely that an antitumor immune response enhances the effect of cryotherapy.

Preoperative Preparation

The extent, location, and proximity of the liver tumor to the hepatic vascular and biliary structures should be defined by careful review of preoperative imaging studies. Patients with more than 40% of the liver replaced with tumor are not candidates for cryoablative techniques, because it is difficult to ablate this volume of tumor, and the perioperative morbidity is prohibitive. In general, the same preoperative preparations are required before cryosurgery as for liver resection, because a small percentage of patients may be converted to resection owing to unexpected resectability or intraoperative complications.

Two devastating complications associated with hepatic cryotherapy are the cryoshock syndrome, a complex of coagulopathy and renal and pulmonary failure, and intraoperative hypothermia (Seifert & Morris 1998); preoperative hydration is important to decrease the incidence of renal dysfunction (Bagia et al, 1998). Hypothermia is another significant complication that can be alleviated using heated fluid and airway circuits and a Bair Hugger (Arizant Healthcare, Eden Prairie, MN) warming system (Onik et al, 1993). Blood and blood products should be made available for the infrequent cases of intraoperative bleeding.

Cryotherapy at Laparotomy: Open Cryotherapy

The abdomen is approached through a right subcostal or bilateral subcostal incision with vertical extension in the midline. The peritoneal cavity is explored thoroughly to detect extrahepatic metastatic disease. Enlarged lymph nodes, particularly those in the porta hepatis, are evaluated by frozen section. The liver is mobilized from its ligamentous attachments and examined bimanually. Using a 5-mHz intraoperative ultrasound (US) transducer, each Couinaud segment of the liver is systematically scanned to determine the extent of disease and its relationship to the hepatic vascular and biliary structures. Suspicious lesions undergo US-directed core biopsies for histologic confirmation, but all lesions seen on preoperative studies should be confirmed by intraoperative US, biopsy, or both.

When an operative strategy is defined, the porta hepatis is encircled for the possibility of needing a Pringle maneuver during the procedure. The cryoprobe is introduced into the center of each lesion under US guidance. The placement is confirmed sonographically by viewing the probe in two or three perpendicular planes. In general, it is preferable to introduce the cryoprobe through the anterior surface of the liver, while monitoring the introduction of the probe and formation of the ice ball with the US transducer placed on the posterior side of the liver. Intraoperative US allows safe placement of the probe, avoiding injury to major intrahepatic vascular and biliary structures. To minimize tumor spillage and to avoid “cracking” the surface of the tumor, the cryoprobe is optimally inserted through normal liver before entering the tumor.

Depending on the depth of the lesion, direct introduction of the probe into the tumor or wire-guided, Seldinger-style localization may be used. The Seldinger technique is best used for deep intraparenchymal and nonpalpable tumors. First, the tumor is identified by intraoperative US. An echogenic needle is passed into the center of the lesion under US guidance and subsequently is exchanged for a J-wire (Fig. 85B.1A). When the J-wire is in place, the tract of the wire is dilated with a coaxial dilator and peel-away sheath. The dilator is withdrawn, and the cryoprobe is inserted through the sheath into the center of the tumor. The position of the cryoprobe is evaluated again using US in two or three different axes. The probe must be placed through the center of the tumor with the tip near the opposite margin to encompass the tumor completely within the cryolesion.

FIGURE 85B.1 A, Ultrasound (US) of an echogenic wire-guided localization of an intraparenchymal hepatic metastasis. Large arrows indicate the J-wire. Small arrows indicate the margins of the tumor. B, Intraoperative US of the ice ball. Arrows indicate the limits of the freeze front. C, US of the thawed ice ball showing the halo caused by edema around the ablated tumor. Arrows indicate margins of edema.

It is crucial to tailor the type of cryoprobe and freezing technique to the size and location of the tumor. The temperature gradient within the ice ball significantly affects the efficacy of the procedure. The temperature near the probe approximates that of liquid nitrogen, whereas at the periphery of the ice ball, the tissue may be only a few degrees below 0° C. In addition, the rate of cooling is determined by the size of the probe, the flux of refrigerant, and the thermoconductivity of the uninsulated portion of the probe. To achieve maximum tumor ablation, the operator must understand the physical properties of the cryotherapy system. The system we use employs vacuum insulated probes and supercooled liquid nitrogen refrigerant. It permits the use of four simultaneous and completely independent probe placements. Two probes are available with this system, accommodating different size tumors. The 3-mm blunted probe creates a cryolesion 4 cm in diameter, whereas the 8-mm trocar point probe creates a 6-cm freeze zone. Two 8-mm probes, used in tandem, create a 10-cm cryolesion (Ravikumar et al, 1991b). Other systems have flat-faced probes that can be applied to a flat surface to create a hemispheric cryolesion of 3 to 4 cm, which can be helpful to treat close margins (Gruenberger et al, 2001).

When the trajectory is chosen, and appropriate placement of the probes is accomplished, the liver is insulated from surrounding tissues using laparotomy pads, towels, or rubber mats. After the probe is placed, and the surrounding tissue is protected, the freezing process is started. At −100° C, the probe sticks to the tissue, allowing placement of additional probes safely without dislodging the first probe; two or three probes can be placed in the liver at the same time, but we do not recommend treating more than three lesions simultaneously, because it is difficult to monitor concurrent cryoablations adequately, and the incidence of intraoperative hypothermia is higher.

Maximal cooling is started, and real-time US monitoring of the freezing process is carried out. The freeze front is seen as a hyperechoic rim with posterior acoustic shadowing (Fig. 85B.1B). Cooling continues until the freeze front extends 1 cm beyond the US margin of the tumor. Typically, the freezing process takes 8 to 15 minutes to complete, depending on the efficiency of the cryotherapy system and the size of the tumor. The freeze zone shown by intraoperative US closely approximates the pathologic and histologic volume of cryonecrosis.

After completion of the first freeze cycle, the tissue is allowed to thaw passively. Active rewarming is unnecessary and may adversely influence results. Complete thawing of the tumor may take 20 to 30 minutes, extending the overall operative time. Because cryotherapy failure usually occurs at the periphery of the ice ball, it is necessary only to thaw the most peripheral centimeter of the ice ball before initiating the second freeze cycle; this reduces operative time without compromising clinical efficacy. The second freeze-thaw cycle is accomplished in a similar fashion.

After completion of the second freeze-thaw cycle, the probe is actively rewarmed, allowing it to disengage before the tumor completely thaws. The probe tract is packed with absorbable knitted fabric, such as Surgicel or another hemostatic material, and gentle pressure is applied to the liver, preventing delayed hemorrhage from the probe tract. Rough handling of the thawing ice ball may cause a cleavage plane to develop at the interface of the ice ball and the normal liver parenchyma, which can cause massive bleeding. Generally, the probe insertion tract stops bleeding promptly, as the coagulation cascade activates at body temperature. After the two freeze-thaw cycles, the treated volume usually remains hyperechoic on US, whereas the surrounding normal liver becomes hypoechoic owing to edema; this gives the treated tumor a “halo” appearance on imaging studies (Fig. 85B.1C).

Inflow Occlusion

The flow of warm blood through vessels adjacent to the tumor may act as a heat sink that changes the conformation of the ice ball and alter tumor ablation near the vessel. Some groups have used inflow occlusion with a Pringle maneuver to diminish these effects. Experimental studies have not confirmed these assertions in the porcine model, in which clamping the hepatoduodenal ligament did not alter the size of the ice ball that formed, the amount of liver necrosis, or the incidence of large vessel infarction (Kahlenberg et al, 1998). The relative contraindication to cryoablation near vessels may be only theoretic, and further evaluation of this problem is necessary.

Monitoring the Ice Ball

Most investigators use US to monitor the ice ball, because the pathologic size of liver necrosis closely correlates with the US size of the ice ball. Others have used thermocouples to monitor the temperature at the margin of the ice ball; however, this is an unnecessarily invasive procedure that provides no important additional information. Electrical impedance through the cryolesion is not as accurate as US in defining adequate ablation; but MRI is a promising technology, because it can accurately evaluate temperature gradients within the ice ball in real time. Spectroscopic information obtained by MRI also may provide accurate information on cellular necrosis. MRI is generally unavailable in most operating and interventional radiology suites and remains an investigational tool.

Bile Duct Warming

To protect the major bile ducts from the adverse effects of cryotherapy, some surgeons cannulate the bile ducts to circulate warmed fluids during the cryoablation. At present, there is not enough evidence of benefit to support this invasive procedure.

Generalized Hypothermia

Generalized hypothermia, a potentially serious adverse event, is generally rare, but it depends on the number of lesions treated and their proximity to blood vessels, especially the vena cava and hepatic veins. The consequences of intraoperative hypothermia are multiple and include cardiac depression, arrhythmias, and coagulopathy. These effects can be ameliorated by warming parenteral fluids and using closed-circulation body warmers and heating blankets. Perhaps most importantly, treatment of multiple large lesions simultaneously should be avoided.

Hemorrhage

After removal of the cryoprobe, hemorrhage through the probe tract is common. The blood does not clot, because the activity of the enzymes that activate the coagulation factors is depressed by the localized hypothermia. Pressure and hemostatic materials usually control the probe tract bleeding. More significantly, the tumor can crack at the interface with the normal uncryoablated tissue, and massive hemorrhage can ensue that requires transfusion, packing, or even conversion to resection. Careful handling of the frozen tissue and gentle extraction of the cryoprobe can minimize this complication, which has been reported in 0% to 25% of treated patients. Diligent use of intraoperative US to avoid moderate-sized vessels during probe placement minimizes this problem.

Bile Fistula and Bile Collections

Bile collections are reported after cryotherapy, affecting approximately 3% of patients in collected series (Seifert & Morris, 1998). Bile collections and fistulae are most common when superficial lesions are treated. Late strictures of the major bile ducts owing to injury during the freezing process may occur and predispose patients to cholangitis. The risk is increased for cryoablated tumors located near the hepatic hilum or the bifurcation of the major biliary ducts, although no long-term studies have confirmed this risk of bile duct injury and late stricture (see Chapters 42A and 42B).

Cryoshock

Cryoshock is a complex of multisystem organ failure, renal failure, and DIC after cryotherapy, and it is potentially lethal. The cause of this symptom complex is unknown, but it is responsible for 18% of deaths after hepatic cryotherapy. In a survey of all groups using cryotherapy, cryoshock was reported in only 21 of 2173 patients undergoing hepatic cryotherapy for tumor (Seifert & Morris, 1998). The cause of this phenomenon is likely related to systemic release of proinflammatory cytokines, such as tumor necrosis factor, interleukin (IL)-1, IL-2, and IL-6. Experimental studies have correlated the risk of this syndrome with larger ablations and with higher systemic levels of these cytokines (Ng et al, 2004; Seifert et al, 2002).

Mortality

The overall early mortality rate for cryotherapy is low at 1.5% (range, 0% to 8%) (Seifert & Junginger, 2004). In expert hands, the operative mortality for major hepatic resection is similar. The most common causes of death after hepatic cryotherapy are not specific to the cryotherapy but rather to cardiac and pulmonary comorbidities. The more common treatment-specific causes of death include cryoshock, liver failure, and hemorrhage.

Cryoablation of Hepatocellular Carcinoma

Cryoablation of primary liver cancer has been reported in more than 200 patients. The outcome of these reported series is difficult to interpret, however, because many patients were treated with cryotherapy in combination with resection, chemotherapy, or hepatic artery ligation. There is about a 60% reduction (range, 40% to 82%) in AFP levels after cryotherapy in patients in whom AFP was elevated preoperatively. Reduction of AFP serum levels after cryotherapy does not clearly translate into a survival benefit.

For primary liver cancer patients treated with cryotherapy alone, the 5-year survival is approximately 30%. As expected, survival correlates well with the biologic characteristics of the tumor, of which size seems to be the most important factor. Zhou and colleagues (1996) showed that 5-year survival was 48% for tumors less than 5 cm, whereas it was only 25% for larger tumors (Zhou et al, 1996). The pattern of recurrence is predominantly in the liver after cryotherapy, but the reports typically do not describe whether the cryotherapy was complete, or what the intended margin was. In addition, it is often not reported whether the tumor recurred in the treated area, or whether these recurrences were a separate, intrahepatic, distant tumor resulting from intrahepatic metastases or new primaries. Few studies contain longitudinal survival information, and these results must be corroborated by larger multicenter studies controlling for tumor size, number of lesions treated, and adequacy of treatment.

It is difficult to compare the results of resection with cryotherapy, because the reported series generally are not comparable in terms of tumor characteristics. The operative mortality for resection of HCC historically has been 6% to 8% in most series, but now with better selection criteria and improved techniques, mortality is generally less than 5%. The median survival is approximately 30 months, with 5-year survivals of 30% to 40%. Similar survivals are reported in selected patients who undergo percutaneous ethanol injection (PEI); however, these are performed in highly selected patients (see Chapter 85A). Livraghi and colleagues (1995) reported a multicenter trial of PEI for HCC that included 746 patients (Livraghi et al, 1995). Survival was related to the size of the lesion, number of tumors, and Child-Turcotte-Pugh (CTP) class of the associated cirrhosis. For unifocal tumors less than 5 cm, the 5-year survival rates were 47%, 29%, and 0% for class A, B, and C cirrhosis, respectively. Even in the presence of multifocal disease, 36% of patients with class A cirrhosis survived 5 years after PEI. Similar results have been reported by Castells and associates (1993) in a comparative study of resection and PEI for HCC less than 4 cm. At 4 years, survival was equivalent for the two modalities.

Hepatic artery infusion chemotherapy, hepatic artery ligation, and systemic chemotherapy all have 25% response rates for HCC. These responses have not led to improved survival, however, and these approaches remain investigational.

Because of patient selection and the combination of therapies used with cryotherapy, for the most part, comparisons with other modalities are meaningless. The development of RFA has generally replaced cryotherapy in the treatment of HCC. Future directions may include hepatic artery infusional chemotherapy after ablative therapies for unresectable tumors.

Cryoablation of Liver Metastases

Liver metastases arising from colorectal, lung, pancreas, and stomach primary tumors constitute the most common malignant tumors of the liver (see Chapter 81A). Most cancers that metastasize to the liver do so in combination with extrahepatic dissemination. The natural history and pattern of recurrence for colorectal cancer is different. Of the approximately 160,000 new cases each year, half develop liver metastases within 5 years of diagnosis. Approximately 20% of these patients (16,000) develop metastatic disease confined to the liver. Only a quarter (4000 to 5000 patients) are amenable to hepatic resection with curative intent; the remaining 12,000 patients with liver-only colorectal metastases are candidates for regional therapies to the liver. The number of patients being considered for liver resection is now increasing because of improved resection techniques and dramatic improvements in adjuvant chemotherapy.

The ultimate goal of regional treatment of liver metastases is increased survival, but long-term follow-up of patients after cryotherapy for liver metastases is inadequate. Most published studies do not have follow-up exceeding 2 years, and few groups have treated enough patients to draw valid conclusions regarding the efficacy of cryotherapy in this setting. Series with adequate follow-up reporting 3- and 5-year survival have been published, albeit with relatively small numbers. In addition, cryotherapy has been used primarily as a salvage procedure or in combination with other treatment modalities, including resection or hepatic artery ligation or infusion, which obscures the results for cryotherapy alone. Ravikumar and associates (1991a) reported overall and disease-free survivals of 78% and 39% at 5 years for 18 patients undergoing complete cryoablation of liver metastases. These encouraging results have not been duplicated in more recent studies, probably because of less stringent selection criteria. Most series (Table 85B.2) report median overall survival of about 2 years after cryoablation of liver metastases. In 24 patients treated with cryotherapy, regardless of whether the ablation was complete or not, Ravikumar and associates (1991b) reported disease-free and overall survival rates of 24% and 63%, respectively. Similarly, Weaver and colleagues (1995) found that 11% of patients were disease-free at a mean follow-up of 30 months, and 62% of treated patients were alive 24 months after cryoablation. Seifert and Morris (1998), in a series of 116 patients with colorectal metastases, showed a median survival of 26 months, and 13% of patients were alive at 5 years. More recent publications have reported 4- and 5-year actuarial survival ranging from 22% to 36% (Kerkar et al, 2004; Rivoire et al, 2002; Seifert & Junginger, 2004). Despite the wide variation in outcome with cryotherapy, some patients in each of the series achieved durable survival with this treatment. Actual follow-up is greater than 5 years for individual patients, suggesting that cryotherapy can be curative in some cases. Based on the worldwide data, it seems that durable local control is achieved in about 20% of patients treated with cryotherapy.

Because the series reporting survival are generally small and include patients treated with cryotherapy, in combination with resection for control of inadequate margins, and patients receiving hepatic artery infusion chemotherapy (see Chapter 86), the results cannot be combined for a proper meta-analysis. Because there are no randomized trials of cryotherapy, Tandan and coworkers (1997) undertook a critical comparative review of resection and cryoablation of colorectal liver metastases. This analysis reviewed 178 studies published from 1973 to 1995. Only studies with survival data exceeding 2 years and data for liver resection if the series included more than 60 patients were included; four cryotherapy and nine resection studies met these minimum inclusion criteria. The median follow-up times for the two procedures were 12 to 28.8 months for cryotherapy and 21 to 69 months for liver resection. The data on liver resection were found to be more valid and consistent, but this review supported a role for cryotherapy in the management of liver metastases. Based on this evaluation, the authors concluded that cryosurgery offers potential advantages in the treatment of liver metastases, it should not be used in cases of resectable disease outside of a clinical trial, and randomized trials comparing the two modalities are needed. Survival seems to be related to the size of the lesions, the volume of liver replaced by tumor, a low pretreatment serum CEA level, absence of extrahepatic metastases, and whether the lesion underwent complete ablation. Normalization of an elevated precryotherapy CEA has been a favorable prognostic factor. Failure to achieve this normalization of CEA probably results from residual unrecognized micrometastatic disease.

Complete response at the treated site ranges from 60% to 90%, and most recurrences are at sites away from the cryolesion. The pattern of recurrence in these patients is primarily in the liver, with or without synchronous extrahepatic disease. This finding underscores the need for additional therapy to further reduce the incidence of hepatic recurrence. With the improvements in regional and systemic chemotherapy, the utility of cryotherapy combined with chemotherapy has been shown even in grossly unresectable patients (Kemeny et al, 2001). An additional use for cryoablation is in the treatment of close or inadequate margins. Edge cryotherapy in these situations has resulted in good local tumor control (Gruenberger et al, 2001; Hou et al, 2007; Yan et al, 2006). Survival after hepatic resection of liver metastases is substantially better (20% to 58%) than after cryotherapy, but the patient populations are not comparable. Cryotherapy generally has been reserved for patients who are unable to undergo resection and who might have a poorer prognosis.