Approach to Liver Metastases in Palliative Oncology

Vaibhav Sahai

Mary F. Mulcahy

BURDEN OF DISEASE

The liver is the most common site of distant metastases in patients with gastrointestinal cancers (1). Patients with liver metastases may be asymptomatic or present with a range of symptoms depending on the volume of tumor, distribution of metastases, or type of primary carcinoma. Symptoms related to tumor burden may range from vague abdominal discomfort, fever, night sweats, pruritus, anorexia, and early satiety to severe localized pain and signs of liver failure such as ascites, jaundice, encephalopathy, and coagulopathy. Patients with extensive neuroendocrine metastases may experience the classic findings of flushing, wheezing, and diarrhea, as well as symptoms related to carcinoid heart disease. The presence of liver metastases confers a poor prognosis and for most patients represents incurable disease.

Outcomes of medical and/or surgical treatments have been expressed primarily in terms of response rates and disease-free survival, progression-free survival, and overall survival (OS). However, more important but less acknowledged additional measures of outcome are quality-adjusted life years and healthrelated quality of life (HRQoL) (2,3). These measures have primarily been used by health economists and rarely applied in oncologic clinical trials, even though the value-based HRQoL approach is conceptually similar to the index approach advocated by clinimetrics (4,5,6). The inclusion of quality of life measures is necessary to discern the overall impact of therapy in clinical trials and for effective communication with patients alongside response and survival outcomes (6).

Colorectal cancer is the third leading cause of cancer in the United States and the third highest in cancer-related deaths every year. It is estimated that approximately 143,460 new cases of colorectal cancer will be diagnosed in the United States in 2012 (7) and that 40% of these patients will develop liver metastasis (8). For many patients, involvement of the liver is the primary determinant of long-term survival. Surgery provides an opportunity for long-term survival (9,10), with an estimated OS of 30% at 5 years (11) and 22% to 23% at 10 years (12) for those that are amenable to surgical resection. Surgical resection is possible for only 15% to 25% of patients with metastatic colon cancer, limited to patients with liver-only metastatic disease that is resectable while leaving adequate liver mass (13). Alternative methods of controlling liver-only metastases are being pursued in an attempt to improve overall outcomes.

Neuroendocrine tumor (NET) presents with disseminated metastatic disease in a large proportion of patients with hepatic metastases present in 60% to 80% of patients (14,15,16). Patients with metastatic NETs have a 5-year survival rate of 22% (17), and less than 10% of patients have surgically resectable disease (18). The liver metastases usually lead to symptoms through mass effect or release of biologically active polypeptides or amines, which may cause a wide range and multitude of symptoms. The management of NETs with liver metastases is mainly focused on palliation of symptoms and prevention of future complications from the disease (19) as the disease course is usually protracted.

Therapeutic modalities are available for the management of liver metastasis when curative surgical resection is not an option. We can broadly divide them into five categories: 1) medical therapy to control symptoms and cancer, 2) palliative surgical debulking, 3) nonsurgical liver-directed therapy, 4) combination therapies, and 5) nonsurgical management of biliary obstruction. Nonsurgical liver-directed therapy includes local tumor ablative techniques (microwave ablation, radiofrequency ablation [RFA], laser photocoagulation, cryoablation) and transarterial therapy (bland transarterial embolization [TAE], transarterial chemoembolization [TACE], transarterial radioembolization, and hepatic artery infusion [HAI]).

EVALUATION

A detailed history and physical examination is the cornerstone of the evaluation of any oncologic patient. Characteristics to consider when evaluating a patient for liver-directed therapy include prior therapy received, subsequent therapy that may be available, underlying liver disease not related to the malignancy, and the goals of therapy. The goal of liverdirected therapy may be to downsize a tumor to allow for surgical resection, control the disease to improve survival, or treat symptoms related to the tumor burden. Laboratory studies evaluate liver function with regard to synthetic function and signs of portal hypertension and biliary obstruction. Depending on the distribution of disease to be treated, patients must have an adequate albumin, prothrombin time, platelet count, bilirubin, and alkaline phosphatase. Imaging modalities need to evaluate the number, size, and distribution of the metastatic lesions; the patency of the biliary tract; hepatic and portal venous systems; and the presence or absence of cirrhosis, ascites, and extrahepatic disease. Highquality imaging is paramount to decide what can technically be done and what might benefit the patient. This may consist of magnetic resonance imaging or triphasic liver computed
tomography, with or without an ultrasound to evaluate evidence of vascular invasion. In the absence of extrahepatic disease, cirrhosis, and portal hypertension, ablative techniques may be limited by location of the tumor and the number and size of the metastases. Tumor adjacent to the gallbladder, main bile ducts, or vena cava may not be amenable to RFA. Intra-arterial embolic therapy is limited to patients without portal vein invasion, as discussed below.

MANAGEMENT

Medical

Systemic Therapy

Systemic chemotherapy is usually the mainstay of therapy for patients with colorectal or other primary cancers and liver metastasis who are not candidates for surgical resection. The chemotherapy indications, response rates, effect on OS, and adverse effects should be appropriately discussed with the patient prior to initiation of therapy. A detailed review of chemotherapy for metastatic cancer is beyond the scope of this chapter; however, some landmark studies comparing chemotherapy to best supportive care are discussed below.

A phase III study randomized patients with metastatic colorectal cancer to irinotecan plus supportive care versus supportive care alone. Improvements in global quality of life score (47.57 vs. 38.47, respectively; P = 0.009), survival without weight loss, survival without performance status deterioration, and pain-free survival were observed in the group treated with irinotecan. An improvement in OS (9.2 vs. 6.5 months, respectively; P = 0.0001) was also noted (20). Another phase III study randomized patients with metastatic colorectal cancer after progression on 5-fluorouracil to irinotecan versus continuous infusional 5-fluorouracil. An improvement in progression-free survival was observed in the group treated with irinotecan (4.2 vs. 2.9 months, respectively; P = 0.03). No improvement in quality of life score (53.90 vs. 53.0, respectively; P = 0.69) was noted but trends in symptom-free survival, pain-free survival, time to performance status deterioration, and loss of body weight were noted, along with a significant improvement in OS (10.8 vs. 8.5 months, respectively; P = 0.035) (21). More recently, an open-label, randomized trial of panitumumab plus best supportive care versus best supportive care showed improved median progression-free survival (8.0 vs. 7.3 weeks, respectively; hazard ratio [HR] 0.54, 95% CI = 0.44-0.66; P < 0.0001) and response outcome (10% vs. 0%, respectively; P< 0.0001) for those treated with panitumumab (22).

Systemic chemotherapy for metastatic NET has a limited role in view of the modest response and significant toxicity. Recently, a phase III randomized placebo-controlled study of sunitinib showed improved median progression-free survival (11.4 vs. 5.5 months, respectively; HR 0.42, 95% CI = 0.26-0.66; P < 0.001) and response rate (9.3% vs. 0%, respectively; P = 0.007) in patients with advanced pancreatic NETs treated with sunitinib (23). Another phase III randomized placebocontrolled study of everolimus showed improved median progression-free survival (11.0 vs. 4.6 months, respectively; HR 0.35, 95% CI = 0.27-0.45; P < 0.001) and response rate (5% vs. 2%, respectively) in patients with advanced pancreatic NETs treated with everolimus (24). Some of the emerging therapies include radionuclides, such as I-131-mIBG and Lu-111-Octreotide.

A large proportion of gastrohepatic NETs secrete biologically active peptides or amines, which may lead to a wide range of symptoms depending on the primary NET. Somatostatin analogues have been the mainstay of treatment for these symptoms. It can be delivered in a depot formulation once a month, with the dose tailored to control the patient’s symptoms. Until recently, the ability of these somatostatin analogs to control the growth of the well-differentiated metastatic midgut NETs was a matter of debate. A placebo-controlled, double blind, phase IIIB study in patients with metastatic midgut NETs showed that octreotide LAR significantly prolonged time to tumor progression compared with placebo (14.3 vs. 6.0 months, respectively; HR 0.34, 95% CI = 0.20-0.59; P = 0.000072) for both metabolically active and inactive NETs (25).

Symptom Management

Pruritus is a complex process that involves stimulation of free nerve endings found superficially in skin. Many chemicals are pruritogenic, including bilirubin, histamine, opioids, serotonin, and cytokines. In patients with hyperbilirubinemia, treatment involves topical and systemic antihistamines, cholestyramine, corticosteroids, local anesthetics, calcineurin inhibitors, or methods to substitute another sensation for itch, which may include a combination of cooling, heating, scratching, or application of a moisturizing lotion. Cholestyramine is a non-absorbable, anion exchange resin that can bind bile acids in the intestinal lumen, thus depleting serum bile salt pool. Cholestyramine has also been shown to be useful in other conditions, such as polycythemia rubra and uremia and, therefore, likely blocks absorption of other compounds (26). However, cholestyramine is not universally effective and associated gastrointestinal side effects can cause intolerance (27). Interestingly, rifampin at 10 mg/kg oral daily has been shown to lower intra-hepatocyte bile salt concentration by competing for uptake, with subjective improvement in pruritus (26). Other agents used for moderate-severe pruritus are opioid antagonists, such as naltrexone. Its use is based on the hypothesis that there is a higher level of endogenous opioids in patients with cholestasis, and use of an opioid antagonist will reduce the central neurotransmission of pruritic signals. Up to 50 mg oral dose once daily resulted in a significant improvement in itching and sleep in patients with pruritus resistant to cholestyramine. Nausea may be limited by using an initial dose of naltrexone 25 mg once daily, followed by subsequent titration. Caution must be exercised when used in conjunction with opioids for pain management, as the effects may be contradictory.

Patients may experience liver-related abdominal discomfort due to ascites or more severe abdominal pain as a result of stretching of the liver capsule due to liver metastases. Opioids are the mainstay of treatment, even in patients with moderate liver impairment. Concern about side effects
of opioids, such as sedation, constipation, confusion, or even hesitation from caregivers, may limit titration for effective palliation. Corticosteroids provide effective adjuvant pain relief by decreasing inflammation and edema. They may also improve appetite and ameliorate constitutional symptoms of fatigue, fever, and night sweats. Extensive liver metastasis can replace most of the liver parenchyma and lead to signs and symptoms of liver failure, such as coagulopathy, encephalopathy, and hypoalbuminemia, and require appropriate medical management for this terminal stage of disease.

Palliative Surgical Resection

Surgical resection may employ hemihepatectomy, segmentectomy, or wedge resection of the metastasis. Surgical resection may provide an opportunity for long-term survival in patients with liver-only metastases from colorectal carcinoma (9,10). However, as mentioned before, only 15% to 25% patients are eligible (13). Contraindications for liver resection, mostly based on a large retrospective multi-institutional review (28), include patients with more than three lesions, bilobar distribution of metastases, portal lymph node or extrahepatic metastases, or inability to achieve 1 cm surgical margins. However, with major technical advances in surgical procedures, the associated morbidity and mortality has improved (29). As a result, surgical and medical oncologists are pushing the envelope and exploring the role of cytoreduction in non-curative patient populations (30). Palliative liver resection is occasionally offered to patients to debulk biochemically active NETs or bypass biliary obstruction. However, debulking or cytoreduction for non-biochemically active cancers, such as colorectal cancers, may not translate into improvement in OS. Some of the newer advances in surgery include sequential hepatic resection (31) and portal vein embolization (32) to allow for hypertrophy of healthy liver parenchyma and permit an aggressive surgery. A combination of liver-directed therapy and systemic chemotherapy (33) has also been used as a “neoadjuvant” approach prior to surgical resection.

Nonsurgical Liver-Directed Therapy

Liver-directed therapy, compared with surgical approaches, is less limited by patient comorbidities and lesion characteristics and, therefore, presents a palliative management option for those not eligible for surgical resection or systemic chemotherapy. Over the past few decades, transcatheter intraarterial and ablative therapies have been utilized in patients with liver metastases to prolong survival and/or improve quality of life. Liver-directed therapy may provide symptom relief, especially in patients with functional neuroendocrine carcinoma with liver metastasis. Many of the constraints for surgical resection (inadequate liver functional reserve, extrahepatic disease, lesion characteristics, multiple bilobar lesions, or patient comorbidities) are less constraining for nonsurgical liver-directed therapy. However, selection of patients is of major importance to limit toxicity, side effects, and premature death as well as cost to the patient and the health-care system.

Nonsurgical liver-directed therapy includes local tumor ablation (microwave or RFA, laser photocoagulation, cryoablation) and transarterial therapy (bland TAE, TACE, transarterial radioembolization, and HAI) (see Table 20.1).

Since the integration of liver-directed therapy into routine clinical practice, major improvements in catheter, device, and imaging technology have translated into improved outcomes for patients with metastatic lesions to the liver. Unfortunately, there are no standardized treatment protocols for liver-directed therapy, especially when delivered in combination with systemic chemotherapy.

Transarterial Therapy

Secondary liver tumors derive their blood supply from the hepatic artery (34) while the normal liver parenchyma obtains at least 50% of the oxygen supply from the portal system (35). This makes the hepatic artery a promising conduit for intra-arterial techniques. Despite the addition of these increasingly popular nonsurgical therapies to the armamentarium, there are limited data describing outcomes, quality of life measures, and OS compared with surgical metastasectomy.

TACE or Hepatic Artery Chemoembolization. Chemoembolization involves intra-arterial delivery of chemotherapy followed by embolization of the vascular supply to the tumor, resulting in selective ischemia and enhanced chemotherapeutic effect on the lesion. Chemotherapy given via the hepatic artery achieves a 10 times greater intra-tumoral concentration than when delivered via the portal vein (36). Diagnostic angiography of the celiac and mesenteric arteries is performed by the interventional radiologist prior to TACE to evaluate the hepatic arterial anatomy and extrahepatic perfusion. Extrahepatic delivery of chemotherapy can lead to a multitude of adverse effects that can be controlled, or limited, with coil embolization of aberrant vessels or distal catheter placement (37). In most institutions, TACE requires a 1- to 3-day hospital stay for each treatment. The chemotherapy to be infused is suspended in an emulsion with lipiodol (an iodized oil), which is selectively retained by the tumor (39). The lipiodol acts as both a vehicle for the cytotoxic drugs and an agent for vessel occlusion to reduce systemic toxicity. After infusion of this viscous chemotherapeutic mixture, embolization of the arterial blood supply to the tumor is completed using embolic agents, including but not limited to gelatin sponge particles, polyvinyl alcohol particles, or hydrophilic, polyacrylamide microporous beads, known as microspheres (38,39). TACE is typically delivered as a series of treatments, with the number determined by the tumor burden and localization, as well as patient tolerance.

TACE is not usually recommended for patients with portal vein thrombosis and must be used with caution in patients with higher degrees of portal hypertension. It is contraindicated in patients with significant aberrant perfusion that

could lead to extrahepatic distribution, bleeding diathesis, greater than 75% hepatic parenchymal involvement, severe liver dysfunction, pregnancy, severe cardiac abnormalities or contraindication to the angiographic or selective visceral catheterization (37).

TABLE 20.1 Liver-directed therapies

Type of Procedure	Description of Procedure	Contraindications	Adverse Effects
Radioembolization	Intra-arterial catheter-directed administration of polymer, resin, or glass microspheres, incorporating radioisotopes into the hepatic artery directly targeting the tumor, which leads to local radiotherapeutic effect	Allergy to contrast, uncorrectable bleeding diathesis, vascular abnormalities, portal vein thrombosis without hepatopetal flow, renal insufficiency, severe liver dysfunction, pulmonary insufficiency, or pregnancy	Acute hepatitis, pancreatitis, gastritis, or ulceration, radiation pneumonitis, acute cholecystitis, vague abdominal pain, nausea/vomiting
Transarterial chemoembolization (TACE)	Intra-arterial catheter-directed administration of chemotherapeutic agent/s into the hepatic artery directly targeting the tumor, with embolization of the vascular supply to the tumor resulting in selective ischemia and therefore enhanced chemotherapeutic effect on the metastasis	Allergy to contrast, uncorrectable bleeding diathesis, vascular abnormalities, portal vein thrombosis without hepatopetal flow, renal insufficiency, severe liver dysfunction, pulmonary insufficiency, or pregnancy	Post-embolization syndrome (nausea, vomiting, fever, abdominal pain with transaminitis, liver abscess, acute liver failure, acute cholecystitis, biliary duct injury, renal dysfunction, gastrointestinal bleed, cardiac toxicity
Transarterial embolization	Intra-arterial catheter-directed administration of an embolic agent such as lipiodol, polyvinyl alcohol, angiostat, or gel foam, which results in devascularization and consequent ischemic injury to the lesion	Similar to TACE	Similar to TACE
Radiofrequency ablation	Percutaneous, laparoscopic, or intraoperative insertion of a conductive probe or electrode into the tumor through imaging guidance following which high-frequency alternating current is transmitted to the immediate tissue, which leads to a calorific effect and coagulative necrosis of the tumor and its surrounding microvasculature	Tumor volume > 50% of liver or significant impairment of hepatic function, lesions near hilum, vessels, or capsule	Biliary leakage, stricture, hemorrhage, thrombosis, abscess, pleural effusion, damage to vascular system, colon perforation, post-ablation syndrome (fever, chills, nausea, vomiting, malaise, abdominal pain)
Cryoablation	Percutaneous, laparoscopic, or intra-operative insertion of a conductive probe or electrode into the tumor following which rapid freezing process leads to local tissue destruction over multiple freeze-thaw cycles	Tumor volume > 50% of liver or significant impairment of hepatic function, extrahepatic disease, or lesions greater than 10 cm in diameter	Biliary leakage, stricture, hemorrhage, “cryoshock phenomenon,” abscess, or damage to the vascular system
Hepatic arterial infusion	Transcutaneously placed hepatic arterial catheter-directed administration of chemotherapeutic agent/s directly targeting the tumor	Unable to undergo surgical placement of catheter or with hepatic arterial anatomy suitable for pump placement. Portal vein thrombosis, more than 70% liver replacement by tumor, or significant impairment of hepatic function	Hepatic misperfusion, thrombosis of the hepatic artery, and catheter dislodgement or chemotherapy-related complications including biliary sclerosis specifically with fluorodeoxyuridine therapy. Transaminitis and gastrointestinal toxicity secondary to extrahepatic perfusion

In addition to tumor destruction, TACE may also cause liver decompensation (40). Approximately 80% of patients develop a post-embolization syndrome characterized by transient abdominal pain, fever, nausea, and vomiting. This is usually self-limited and typically resolves in 7 to 10 days. Serious complications may also occur; a 30-day mortality of 4.3% has been reported, primarily due to hepatic failure or infection (41). After hospital discharge, patients may require 2 to 3 weeks of convalescence prior to the next treatment.

The effect of TACE on survival is difficult to assess due to the variability of techniques, chemotherapy and embolic agents utilized, and retreatment schedules. A prospective study of 463 patients with hepatic metastases from metastatic colorectal cancer by Vogl et al. (39) showed that the median survival was 14 months from date of TACE compared with 7 to 8 months for untreated patients (42). In another prospective non-randomized study by Sanz-Altamira et al. (43), 40 patients underwent TACE and had a median OS of 10 months (see Table 20.2).

TAE or Hepatic Arterial Embolization. The hepatic arterial supply to the tumor is embolized via materials such as lipiodol, polyvinyl alcohol, angiostat, or gel foam, which results in devascularization and consequent ischemic injury to the lesion. Patient selection and adverse effects are similar to those observed with TACE. Similar to TACE, diagnostic angiography of celiac and mesenteric arteries is performed by the interventional radiologist to evaluate the hepatic arterial anatomy.

Randomized controlled trials and other retrospective studies comparing TAE with TACE have demonstrated no advantage of one technique over the other in patients with metastasis from colorectal cancer (58,59) as well as NET (41,60,61) (see Table 20.3).

Transarterial Radioembolization. Radioembolization involves the intra-arterial delivery of either glass or resin microspheres containing radioisotopes into the tumor to produce a local radiotherapeutic effect. There are two different radioisotopes containing commercial microspheres: TheraSphere (MDS Nordion, Canada) which consists of non-biodegradable glass microspheres and SIR-Spheres (Sirtex, USA) which consists of biocompatible polymer microspheres. Yttrium-90 (Y-90), a radioisotope, is an integral constituent of these spheres. Yttrium is a pure betaemitter and has a physical half-life of 64.2 hours (2.68 days) and decays to stable zirconium-90. The average energy of the beta emissions from Y-90 is 0.9367 MeV. The average tissue range of the radiation is 2.5 mm, with a maximum range less than 1 cm. The microspheres are unable to traverse the tumor microvasculature and exert a local radiotherapeutic effect with relatively limited concurrent injury to the surrounding normal tissue.

Diagnostic angiography is performed by the interventional radiologist to evaluate the hepatic arterial anatomy. Technetium-99m macroaggregated albumin (Tc-99 MAA) hepatic arterial perfusion scintigraphy is completed prior to the procedure to detect shunting of blood to the lungs or gastrointestinal tract. Radioembolization is contraindicated if there is excessive shunting that cannot be corrected by angiographic techniques or if the shunting of blood to lungs results in delivery of greater than 16.6 mCi of radiation to the lungs. To avoid reflux of microspheres into the gastric vasculature, the gastroduodenal artery may be occluded by coil embolization techniques or the catheter advanced beyond the gastroduodenal artery at the time of infusion of the radiomicrospheres.

Radioembolization is contraindicated in patients in whom hepatic artery catheterization is contraindicated, such as those patients with vascular abnormalities, uncorrectable bleeding diathesis, uncorrectable allergy to contrast dye, as well as portal vein thrombosis without hepatopetal flow, or patients with renal insufficiency that cannot undergo treatment using alternatives to iodinated contrast media (CO₂, gadolinium). The procedure is also contraindicated in patients with severe liver dysfunction, pulmonary insufficiency, or pregnancy (66). Clinical trials have excluded patients with elevated bilirubin, except when the tumor can be isolated from a vascular standpoint.

Radioembolization has been shown to cause abdominal pain, nausea, vomiting, ulceration, and bleeding from introduction of microspheres into the gastrointestinal microvasculature (67). Pulmonary vascular shunting can cause pulmonary edema and fibrosis or radiation pneumonitis, which may be irreversible (67). Radiation pneumonitis has been seen in patients with shunting that has resulted in doses greater than 30 Gy delivered to the lungs in a single treatment. A significant deposition of radiomicrospheres can occur in the lungs in patients with arteriovenous malformations that allow the particles to pass directly from the arterial circulation to the venous system without being trapped in the hepatic capillary bed. Radioembolization may also lead to transient fever and abdominal discomfort for a few hours immediately following the procedure. However, typically the side effects experienced with radioembolization are of lower intensity compared with TACE or TAE, with postembolization syndrome rarely reported (66). Also, radioembolization is an outpatient procedure compared with TAE or TACE, which necessitates 24 to 72 hours hospitalization for post-radioembolization syndrome. However, an evaluation of reimbursement, cost, and profit comparing TACE versus radioembolization showed that the cost for radioembolization was substantially higher (36).

Radioembolization is recognized as a treatment option by the National Comprehensive Cancer Network for metastatic NET. Although the data are limited, recent studies have shown radiologic response between 39% and 64% (68,69,70,71,72) (see Table 20.4).

TABLE 20.2 Outcome comparison with transarterial chemoembolization

Study	Indication	N	Chemotherapy	Radiologic Response % (CR+PR)	Median Overall Survival (months)
*Hepatic metastasis from colorectal carcinoma*
Albert et al. (44)	Salvage	121	Mitomycin C cisplatin, doxorubicin	2	9
Vogl et al. (39)	Salvage	243	Mitomycin C	13.6	14
	Salvage	153	Mitomycin C, gemcitabine	11.1	13.9
	Salvage	67	Mitomycin C, irinotecan	19.4	14
Voigt et al. (45)	Salvage	11	Mitomycin, oxaliplatin, IFN-α2b, dexamethasone, 5-FU, folinic acid	33	Not reported
Tellez et al. (46)	Salvage	30	Mitomycin, cisplatin, doxorubicin	63	8.6
Sanz-Altamira et al. (43)	Salvage	40	5-FU, mitomycin	22.8	10
Lang et al. (47)	Not reported	46	Doxorubicin	Not reported	23
*Hepatic metastasis from neuroendocrine carcinoma*
Dong et al. (48)	Salvage	123	Doxorubicin or streptozocin	62	39.6
de Baere (49)	Not reported	20	Doxorubicin eluting beads	80	Not reached
Marrache et al. (50)	Salvage	80	Doxorubicin or streptozocin	37	61
Fiorentini et al. (51)	Salvage	10	Lipiodol, mitomycin, cisplatin, epirubicin	70	22
Kress et al. (52)	Salvage	26	Doxorubicin	7	53.5
Gupta et al. (53)	Salvage	31	Cisplatin, vinblastine, floxuridine, doxorubicin, mitomycin, or their combination	44.4	Not reported
Roche et al. (54)	Salvage	14	Doxorubicin	43	Not reported
Desai et al. (55)	Not reported	34	Adriamycin, mitomycin	32.3	Mean OS 8 mo
Dominguez et al. (56)	Salvage	15	Streptozocin	53	Not reported
Kim et al. (57)	Salvage	30	Cisplatin, doxorubicin	37	15
CR, complete response; PR, partial response; IFN, interferon; 5-FU, 5-fluorouracil; OS, overall survival.