Primary and secondary malignancies in the liver present one of the most challenging problems in clinical oncology.

Colorectal cancer is the second leading cause of cancer-related death in the United States, with liver metastases accounting for about half of these deaths. Other tumors that frequently develop fatal hepatic metastases-despite a resectable primary tumor-include ocular melanoma, neuroendocrine tumors, and gastrointestinal sarcomas. Hepatocellular carcinoma (hepatoma) is the most common fatal malignancy worldwide, killing 1.25 million people annually.

Less than half of patients with metastatic colorectal cancer respond to systemic chemotherapy, and the survival benefit is no more than several months. Hepatoma has very little response to systemic therapy. External beam radiotherapy is limited by the radiosensitivity of the liver. Novel approaches for treating hepatic malignancies include direct infusion of chemotherapeutic drugs into the hepatic artery or portal vein, em-bolization, percutaneous ablation with various physical and chemical agents, and interstitial or intravascular injection of radioactive agents. These techniques have in common a regional or local approach to liver disease.

Chemoembolization

Chemoembolization combines hepatic artery embolization with simultaneous infusion of a concentrated dose of chemotherapeutic drugs. There are several theoretical advantages to this technique. First, embolization renders the tumor ischemic, depriving it of nutrients and oxygen. Second, tumor drug concentrations are 1 to 2 orders of magnitude higher than are achieved by infusion alone.1,2 Third, because blood flow is arrested, the dwell time of the chemotherapy is markedly prolonged, with measurable drug levels present as long as a month later.3,4 Fourth, the ischemia induced by embolization helps to overcome drug resistance by causing metabolically active cell membrane pumps to fail, thereby increasing intracellular retention of the chemotherapeutic drug.5 Finally, because most of the drug is retained in the liver, systemic toxicity is reduced.6

Patient selection for chemoembolization

Patients selected for regional therapy must have their tumor confined to the liver. Patients with minimal or indolent extrahepatic disease may be candidates if the liver disease is considered the dominant source of morbidity and mortality for that individual.

Tolerance of hepatic artery occlusion is dependent upon the presence of portal vein inflow. Compromise in portal venous blood flow is a relative contraindication to hepatic embolization. Chemoembolization can be performed safely despite portal vein tumor thrombus if hepatopetal collateral flow is present.7 In this setting, a smaller volume of the liver should be embolized at any one time.

When the parenchyma is diseased, the liver becomes more dependent on the hepatic artery for its blood supply. A subgroup of patients has been identified who are at high risk of acute hepatic failure following hepatic artery embolization. They have a constellation of >50% of the liver volume replaced by tumor, lactate dehydrogenase (LDH) >425 IU/L, aspartate aminotransferase (AST) >100 IU/L, and total bilirubin 32 mg/dl.8 The presence of hepatic encephalopathy or jaundice are absolute contraindications to embolization.

Biliary obstruction is also a contraindication. Even with a normal serum bilirubin, the presence of dilated intrahepatic bile ducts places the patient at high risk for biliary necrosis of the obstructed segment(s) of the liver.

Procedure and peri-procedural care

Pretreatment assessment-Preoperative evaluation for chemoembolization includes a tissue diagnosis; cross-sectional imaging of the liver; exclusion of extrahepatic disease; and laboratory studies, including complete blood count, prothrombin time, partial thromboplastin time, creatinine level, liver function tests, and tumor markers.

Patient education-Before em-barking on this fairly arduous palliative regimen, patients should be thoroughly informed of the side effects and risks. Up to 80% to 90% of patients suffer from postembolization syndrome, characterized by pain, fever, and nausea and vomiting. The severity of these symptoms varies tremendously from patient to patient, and can last from a few hours to several days. Other significant toxicities are rare. Serious complications occur after 5% to 7% of procedures (see below). Given the significant discomforts, hazards, and expense of this treatment, its palliative role should be clearly understood.

Procedure-Patients must fast overnight, and are admitted to the hospital the morning of the procedure. A Foley catheter is inserted, and vigorous hydration is initiated (normal saline solution [NSS] at 200 to 300 mL/hr). Prophylactic antibiotics (cephazolin 1 gram, metronidazole 500 mg) and antiemetics (odansetron 24 mg, decadron 10 mg, diphenhydramine 50 mg) are administered intravenously. The benefit of antibiotic prophylaxis is unproven.

Diagnostic visceral arteriography is performed to determine the arterial supply to the liver and confirm patency of the portal vein. The origins of vessels supplying the gut, particularly the right gastric and supraduodenal arteries, are carefully noted in order to avoid embolization of the stomach or small bowel.

Once the arterial anatomy is understood clearly, a catheter is advanced superselectively into the right or left hepatic artery, depending upon which lobe holds the most tumor (figure 1). The chemoembolic mixture is injected until nearly complete stasis of blood flow is achieved (figure 2). In our institution, we use 100 to 150 mg cisplatin, 50 mg doxorubicin, and 10 mg mitomycin-C dissolved in 10 mL of radiographic contrast and emulsified with 10 mL of iodized oil and 150 to 250 µm polyvinyl alcohol (PVA)

particles. The patient receives intra-arterial lidocaine (30 mg boluses up to 200 mg total) and intravenous narcotics to alleviate pain during the embolization. After the procedure, vigorous hydration (NSS 3L/24 hr), intravenous antibiotics, and antiemetic therapy (odansetron and decadron) are continued. Narcotics, perchlorpromazine, and acetominophen are supplied liberally for control of pain, nausea, and fever. The patient is discharged as soon as oral intake is adequate and parenteral narcotics are not required for pain control. About half of patients are discharged in 1 day, most in 2 days. Oral antibiotics are continued for another 5 days, as well as antiemetics and oral narcotics if needed. The patient returns for a second procedure directed at the other lobe of the liver 3 to 4 weeks later. Depending upon the arterial anatomy, two to four procedures are required to treat the entire liver, after which response is assessed by repeat imaging studies and tumor markers.

Complications-Major complications of hepatic embolization include hepatic insufficiency or infarction, hepatic abscess, tumor rupture, surgical cholecystitis, and non-target em-bolization to the gut. With careful patient selection and scrupulous technique, the collective incidence of these serious events is 3% to 4%. Other complications include periprocedural cardiac events, renal insufficiency, and anemia requiring transfusion, with incidences of <1% each. Thirty-day mortality ranges from 1% to 4%.

Results in specific diseases

Hepatoma-Among combined series of 800 patients with unresectable hepatocellular carcinoma treated palliatively with chemoembolization in Asia, Europe, and in the United States, response rates (as measured by decreased tumor volume and decreased serum alpha-fetoprotein levels) were 60% to 83%.3,9-14 Cumulative probability of survival ranged from 54% to 88% at 1 year, 33% to 64% at 2 years, and 18% to 51% at 3 years, with the best results obtained by repeated embolizations with a combination of iodized oil, Gelfoam (Upjohn, Kalamazoo, MI), and chemotherapeutic drugs. Survival varies directly with oil uptake and retention, and inversely with tumor volume, stage, and Childs class (figure 3).

Despite the extent of single-institution experience with chemoembolization of hepatoma published over the past decade, few controlled trials have been reported. A multicenter European trial comparing cisplatinum/

lipiodol/Gelfoam chemoembolization to no therapy in 100 patients with relatively small tumor burdens (90% stage I) found 1-year survivals of 62% and 43% respectively, and 2-year survivals of 38% and 26%.15 A French multicenter trial of 127 patients with more advanced disease (62% stage II or III) showed almost identical survival rates in the chemoembolization arm (64% and 38% at 1 and 2 years), with survival in the control arm of only 18% and 6%, respectively (p <0.0001).16

Colorectal metastases-Phase II studies of chemoembolization for metastatic colorectal cancer have been reported by several centers in the United States. Patients enrolled in these trials are usually individuals who have failed systemic and/or intra-arterial infusion chemotherapy. Lang17 used a combination of superselective segmental and selective lobar injections of a doxorubicin-iodized oil emulsion on 46 patients; 59% achieved stabilization or regression of disease with 17% complete responses. Actuarial survival was 68% at 1 year and 37% at 2 years.

At the Boston Center for Liver Cancer, 40 patients were chemoembolized with 5-FU, mitomycin-C, oil, and gelatin sponge.18 Sixty-three percent had partial or minor morphologic responses and 62% had a > 50% drop in CEA level. Median survival from first chemoembolization was 10 months. A number of prognostic factors were identified. Patients with an ECOG performance status of 0 to 1 had a median survival of 24 months, versus 3 months for those with a performance status of 2. Patients with extrahepatic disease had a median survival of 3 months versus 14 months for those without. Among patients with good performance status and no extrahepatic disease, actuarial survival was 73% at 1 year and 61% at 2 years. Elevation of the alkaline phosphatase or LDH levels to above 3-times normal, or elevation of the AST above normal, all predicted worse survival.

At Northwestern Memorial Hospital, 30 patients were chemoembolized with cisplatin, doxorubicin, mitomycin-C (CAM), and bovine collagen.19 They reported 95% of patients had a 3 25% drop in CEA; 63% had a "radiologic" response, which was defined as tumor necrosis or 3 25% decrease in size. Median survival was 8.6 months from first chemoembolization and 29 months from diagnosis.

At the University of Pennsylvania, 51 patients were chemoembolized with CAM, iodized oil, and PVA.20 Morphologic stabilization or regression occurred in 72% of patients; CEA stabilized or regressed in 90%; median duration of response was 12 months (figure 4). Actuarial survival from diagnosis with liver metastases was 86% at 1 year, 55% at 2 years, and 23% at 3 years, with a median survival of 24 months.

The results in these extensively pretreated patients are promising, but high early response rates do not necessarily lead to improved survival. Phase III trials of chemoembolization are needed to determine if a survival benefit exists. The American College of Radiology Imaging Network (ACRIN) is currently funding a multicenter randomized trial of systemic chemotherapy with or without chemoembolization for colorectal metastases to liver.

Ocular melanoma-Patients with ocular melanoma frequently develop rapidly progressive, fatal hepatic metastases, with median survivals of 2 to 6 months. The MD Anderson Cancer Center reported results of 30 patients treated by serial chemoembolizations with cisplatin and PVA.21 There was one complete response; 46% of patients had a >50% morphologic response (figure 5). Median survival was 11 months (14 months for responders versus 6 months for non-responders), with an actuarial survival of about 33% at 1 year. There are no other large series for this tumor. Vogelzang22 reported dismal results using cisplatinum, doxorubicin, mitomycin-C, and bovine collagen with only one long-term survivor among eight patients, and a median survival of 2 to 3 months.

Neuroendocrine tumors-Embolization has an established role in the palliation of these hypervascular tumors, typically producing symptom-free intervals of 5 to 10 months in

> 90% of patients. Three reports have looked at chemoembolization of neuroendocrine tumors23-25 (figure 6). Response rates and survival are similar for embolization and chemoembolization, but the duration of response after chemoembolization averages close to 2 years. Other investigators have warned of an increased complication rate after chemoembolization of carcinoid tumors. A prospective randomized trial is necessary to determine if there is enough additive benefit from chemoembolization to justify any additional risk and cost.

Sarcomas-Mavligit26 reported major regression of metastatic leiomyosarcoma in 10 of 14 patients with cisplatinum/gelfoam chemoembolization, followed by a 2-hour vinblastine infusion into the hepatic artery, with a median duration of response of 1 year. Ten of 16 patients treated with CAM/oil/PVA at the University of Pennsylvania responded, with extensive tumor necrosis demonstrated on CT in all cases.27 Three became resectable. Median duration of response was 10 months. Since systemic chemotherapy and radiotherapy are ineffective against metastatic sarcomas in the liver, chemoembolization is worthy of investigation.

Other metastases-Liver metastases from lung, breast, pancreas, stomach, small bowel, kidney, bladder, thymus, cholangiocarcinoma, ovary, thyroid, and unknown primary tumors have been treated with chemoembolization. Published reports group patients with different tumor types together, making interpretation of the results difficult. Overall, mixed metastatic lesions have a 60% to 75% objective response rate and median survivals of 8 to 11 months.

Percutaneous tumor ablation

A variety of chemical and physical agents have been used to destroy cancerous tissue in the liver directly.28 Chemical ablation, typically performed with absolute alcohol or acetic acid, depends upon direct contact between the liquid sclerosant and the tumor cells. Cell death occurs due to dehydration, protein denaturation, coagulative necrosis, and small vessel thrombosis. Areas of septation, necrosis, or other tissue inhomogeneity cause incomplete diffusion of liquid agents through the tumor, which limits the effectiveness of this technique. Complete necrosis is achieved in 70% to 75% of small hepatomas, where the surrounding cirrhotic liver helps to contain the alcohol within the tumor.

Alcohol is relatively ineffective for metastatic lesions, which tend to be harder than the surrounding normal liver. Metastatic disease is better treated using thermal ablation techniques, such as cryotherapy or heating with radiofrequency, microwave, laser, or focused ultrasound energy. These energized sources cause thermal death within a predictable volume of tissue around the source probe, irrespective of the nature of the tissue. The major limitation of this technique is the heat sink caused by the substantial blood flow within the liver, which limits the size of a single thermal lesion to 2 to 3 cm with most devices.

Patient selection for

percutaneous ablation

Similar to regional catheter-directed therapy, patients undergoing percutaneous tumor ablation should have unresectable disease confined to the liver. In addition, the tumor must be visible and accessible to needle puncture under cross-sectional imaging guidance (ultrasound, CT, or MR). Practical limitations exist on the size and number of lesions that are likely to benefit from this form of therapy. For tumors larger than 3 cm, multiple overlapping chemical injections or thermal lesions must be performed to completely treat the tumor volume and a 5- to 10-mm margin of liver around the tumor. This becomes impractical once tumors are larger than 5 cm, though some physicians aggressively retreat such larger lesions in an effort to eliminate any residual viable tumor. Treatment of more than 3 to 4 individual lesions not only has practical limitations, but such patients are likely to harbor additional sites of disease and, therefore, tend to progress rapidly.

Technique of chemical ablation

Intravenous access is obtained for administration of analgesics. The tumor is localized in the usual fashion as for a biopsy, and the skin and liver capsule anesthetized. Specialized 20- to 22-gauge needles for alcohol ablation have a closed diamond tip and multiple sideholes to facilitate distribution of the alcohol through the tumor. Under imaging guidance, the needle is placed to the far wall of the tumor, then alcohol is injected and the needle gradually withdrawn toward the proximal margin. Ultrasound guidance is preferable since it permits realtime observation of the diffusion of the alcohol through the tumor. If the alcohol is observed to track into vessels, bile ducts, the surrounding liver, or back along the access tract, injection is halted and the needle repositioned. Alcohol is also quite conspicuous on CT because of its low density (-200 HU), but the injections cannot be monitored in realtime. Once the injection is complete, the needle is left in place for 1 to 2 minutes to allow the liquid to diffuse, then the needle is withdrawn while aspirating, to avoid reflux of any residual alcohol into the peritoneum. Additional passes may be required to treat the entire tumor. The volume of alcohol required is estimated using the formula for the volume of a sphere, adding 0.5 cm to the tumor radius to allow for ablation of a margin of liver. Lesions up to 3 cm in diameter can be treated in a single session. A 4-cm tumor requires 65 mL of alcohol, which is an intoxicating and potentially toxic dose. Therefore tumors larger than 3 cm, or multiple tumors, need to be treated in multiple separate sessions.

Pain, nausea, and vasovagal reactions are common during percutaneous alcohol injection. Fever and elevation of liver function tests occur during the ensuing few days. Complications such as bleeding, pneumothorax, pleural effusion, liver abscess, biliary stricture, hepatic infarct, and tumor seeding occur with a frequency of <1% each.

Results of chemical ablation

Histologic examinations of resected hepatocellular carcinoma nodules after alcohol ablation indicate complete tumor necrosis in 70% to 75% of tumors. For the usual patient not undergoing resection, follow-up imaging is performed using triple-phase scanning to examine the tumor during the arterial phase of contrast enhancement. The overall size of the lesion after successful ablation is larger due to ablation of a margin of normal liver, so static imaging alone is not sufficient to assess response. Peripheral rim enhancement is typical, and reflects the inflammatory reaction in the surrounding liver. Any intratumoral enhancement suggests residual viable tumor, which may be treated with repeat ablation. Local recurrence rates at the ablation site are 2% to 17%.29,30 As expected, almost all patients develop new sites of disease elsewhere in the liver within 5 years, so continuous surveillance is critical.

Survival after alcohol ablation of hepatocellular carcinoma is reported to be 80% to 98% at 1 year and 55% to 70% at 3 years. This, in part, reflects the early stage of disease in patients who are candidates for local ablation, in contrast to patients treated with chemoembolization, who have heavier tumor burdens at the time of diagnosis.

Results of attempted alcohol ablation of metastatic liver lesions are dismal, and this approach has largely been abandoned in favor of thermal ablation.

Technique of thermal ablation

Currently, there are three commercially available radiofrequency (RF) ablation devices in the United States for treatment of soft tissue (figure 7). Each uses 15- to 18-gauge needles with either multiple parallel needles or a radial array to disperse the energy over a larger volume. The needle is placed into the lesion under imaging guidance. Both CT and ultrasound work well for this technique (figure 8). The exact protocol for needle placement and delivery of the RF current varies by vendor. The endpoint for treatment also depends upon the device, with varying use of time, temperature, or electrical impedance as endpoints. This technology is still relatively unstable, in that the vendors regularly introduce more powerful generators and new probe configurations in an effort to increase the treatment volume.

Using current technology, 3- to 4-cm tumors can usually be ablated using overlapping burns (figure 9). Ablation of lesions >5 cm is impractical, because it is difficult to achieve complete treatment of the target volume, and local failure rates are high. Time is also an issue since, depending upon the device used, each burn requires 12 to 40 minutes.

RF ablation is extremely painful, and requires high levels of analgesia and sedation. Other than pain, side effects are minimal. Complications are similar to chemical ablation, but occur somewhat more frequently due to the larger needle size and longer dwell time in the liver. Careful placement of the grounding pads is important to avoid thermal injury to the skin.

Results of thermal ablation

Immediate results of RF ablation based on follow-up imaging indicate absence of any enhancing tumor in 60% to 90% of tumors, with lesion size being the major determinant of success (figure 10). Local recurrence rates have been reported to be as low as 2% to as high as 40% within 1 year.31 As with any local approach, new lesions are the predominant mode of failure, occurring in up to 65% of patients within 1 year. Comparison studies of RF ablation and alcohol ablation for hepatocellular carcinoma demonstrate complete ablation with fewer treatment sessions with RF ablation, and a lower local recurrence rate.32 These advantages are believed to offset the increased treatment time per session, the slight increase in complications, and the increased cost for the device, making RF ablation now the preferred treatment modality.

Combined regional and

local therapy

Chemoembolization, while treating a large volume, is limited in its ability to induce complete tumor necrosis. Conversely, RF ablation causes thorough tissue necrosis within a small volume, primarily limited by blood flow. Hence, combining the two techniques is an appealing approach to intermediate size lesions. Occlusion of the hepatic artery and/or portal vein during RF ablation results in substantially larger burn volumes. This is routinely done during intraoperative ablation using the Pringle maneuver to temporarily reduce hepatic blood flow. Chemoembolization, by devascularizing the tumor, allows effective RF ablation of tumors up to 8 cm in diameter (unpublished data, May 2000). Blood flow to the surrounding liver is still preserved via the portal vein, so a "surgical" margin is not achieved. Experience with this combined modality therapy is still preliminary, so long-term follow up for recurrence is not available, but the immediate response is promising. AR

References

1. Konno T: Targeting cancer chemotherapeutic agents by use of Lipiodol contrast medium. Cancer 66:1897-1903, 1990.

2. Egawa H, Maki A, Mori K: Effects of intraarterial chemotherapy with a new lipophilic anticancer agent, estradiol-chlorambucil (KM2210), dissolved in Lipiodol on experimental liver tumor in rats. J Surg Oncol 44:109-114, 1990.

3. Nakamura H, Hashimoto T, Oi H, et al: Transcatheter oily chemoembolization of hepatocellular carcinoma. Radiology 170:783-786, 1989.

4. Sasaki Y, Imaoka S, Kasugai H, et al: A new approach to chemoembolization therapy for hepatoma using ethiodized oil, cisplatin, and gelatin sponge. Cancer 60:1194-1203, 1987.

5. Kruskal JB, Hlaty L, Hahnfeldt P, et al: In-vivo and in-vitro analysis of the effectiveness of doxorubicin combined with temporary arterial occlusion in liver tumors. J Vasc Interv Radiol 4:741-748, 1993.

6. Daniels JR, Sternlicht M, Daniels AM: Collagen chemoembolization: Pharmacokinetics and tissue tolerance of cisplatin in liver and kidney. Cancer Res 48:2446, 1988, [abstract].

7. Pentecost MJ, Daniels JR, Teitelbaum, GP, et al: Hepatic chemoembolization: Safety with portal vein thrombosis. J Vasc Intervent Radiol 4:347-351, 1993.

8. Charnsangavej C: Chemoembolization of liver tumors. Sem Invest Radiol 10:150-160, 1993.

9. Onishi K, Tsuchiya S, Nakayama T, et al: Arterial chemoembolization of hepatocellular carcinoma with mitomycin C microcapsules. Radiology 152:51-55, 1984.

10. Nakao N, Miura K, Takahashi H, et al: Hepatocellular carcinoma: Combined hepatic arterial and portal venous embolization. Radiology 161:303-307, 1986.

11. Van Beers B, Roche A, Cauquil P, et al: Transcatheter arterial chemotherapy using doxorubicin, iodized oil and Gelfoam embolization in hepatocellular carcinoma. Acta Radiologica 30:415-418, 1989.

12. Venook AP, Stagg RJ, Lewis BJ, et al: Chemoembolization for hepatocellular carcinoma. J Clin Oncol 8:1108-1114, 1990.

13. Nakao N, Kamino K, Miura K, et al: Recurrent hepatocellular carcinoma after partial hepatectomy: value of treatment with transcatheter arterial chemoembolization. AJR 156:1177-1179,1991.

14. Stokes KR, Stuart K, Jenkins RL, Clouse ME: Chemoembolization with Adriamycin/ Ethiodol in hepatocellular carcinoma. J Vasc Interv Radiol 3:17, 1992, [abstract].

15. Groupe D'Etude de Traitment du Carcinome Hepatocellulaire: A comparison of lipiodol chemoembolization and conservative treatment for unrtesectabvle hepatocellular carcinoma. N Engl J Med 332:1256-1261, 1995.

16. Bronwicki JP, Vetter D, Dumas F, et al: Transcatheter oily chemoembolization for hepatocellular carcinoma. A 4-year study of 127 French patients. Cancer 74:16-24, 1994.

17. Lang EK, Brown CL Jr: Colorectal metastases to liver: Selective chemoembolization. Radiology 189:417-422, 1993.

18. Sanz-Altamira PM, Spence LD, Huberman MS, et al: Selective chemoembolization in the management of hepatic metastases in refractory colorectal carcinoma. Dis Colon Rectum 40:770-775, 1997.

19. Tellez C, Benson AB III, Lyster MY, et al: Phase II trial of chemoembolization for the treatment of metastatic colorectal carcinoma to the liver and review of the literature. Cancer 82:1250-1259, 1998.

20. Tuite CM, Soulen MC, Baum RA, et al: Hepatic metastases from colorectal cancer treated with CAM/Ethiodol/PVA chemoembolization: Evaluation of survival, biologic and morphologic response. J Vasc Interv Radiol 10(No. 2 Part 2):260, 1999, [abstract].

21. Mavligit GM, Charnsangavej C, Carrasco CH, et al: Regression of ocular melanoma metastatic to the liver after hepatic arterial chemoembolization with cisplatin and polyvinyl sponge. JAMA 260:974-976, 1988.

22. Vogelzang RL, Nemcek AA Jr., Lyster M, et al: Hepatic artery chemoembolization for treatment of hepatic malignancy: Results with microfibrillar collagen and triple drug therapy. J Vasc Intervent Radiol 4:61, 1993, [abstract].

23. Hajarizadeh H, Ivancev K, Mueller CR, et al: Effective palliative treatment of metastatic carcinoid tumors with intraarterial chemotherapy/ chemoembolization combined with octreotide acetate. Am J Surg 163:479-483, 1992.

24. Therrase E, Breittmayer F, Roche A, et al: Transcatheter chemoembolization of progressive carcinoid liver metastasis. Radiology 189:541-547, 1993.

25. Stokes KR, Stuart K, Clouse ME: Hepatic arterial chemoembolization for metastatic endocrine tumors. J Vasc Interv Radiol 4:341-345, 1993.

26. Mavligit GM, Zukwiski AA, Ellis LM, et al: Gastrointestinal leiomyosarcoma metastatic to the liver. Durable tumor regression by hepatic chemoembolization infusion with cisplatin and vinblastine. Cancer 75:2083-2088, 1995.

27. Rajan DK, Soulen MC, Baum RA, et al: Sarcomas metastatic to the liver: Response and

survival following CAM/Ethiodol/ PVA chemoembolization. J Vasc Interv Radiol 11(No. 2 Part 2):187, 2000.

28. Dodd III GD, Soulen MC, Kane RA, et al: Minimally invasive treatment of malignant hepatic tumors: At the threshold of a major breakthrough. Radiographics 20:9-27, 2000.

29. Livraghi T, Giorgio A, Marin G, et al: Hepatocellular carcinoma and cirrhosis in 746 patients: Long-term results of percutaneous ethanol injection. Radiology 197:101-108, 1995.

30. Lencioni R, Bartolozzi C, Caramella D, et al: Treatment of small hepatocellular carcinoma with percutaneous ethanol injection: Analysis of prognostic factors in 105 Western patients. Cancer 76:1737-1746, 1995.

31. Curley SA, Izzo F, Delrio P, et al: Radiofrequency ablation of unresectable primary and metastatic hepatic malignancies: Results in 123 patients. Ann Surg 230:1-8, 1999.

32. Lencioni RA, Cioni D, Donati F, et al: Percutaneous treatment of small hepatocellular carcinoma in cirrhosis: Radiofrequency thermal ablation vs ethanol injection - a prospective randomized trial (final report). Radiology 213(P):123, 1999.