High-Dose Immunoembolization: Survival Benefit in Patients with Hepatic Metastases from Uveal Melanoma

Abstract

Purpose: To retrospectively evaluate prognostic factors for survival in patients with uveal melanoma who received chemoembolization (CE) with 1,3-bis (2-chloroethyl)-1-nitrosourea or immunoembolization (IE) with granulocyte-macrophage colony-stimulating factor (GM-CSF) for hepatic metastases.

Materials and Methods: Fifty-three consecutive patients with uveal melanoma were treated by using CE or IE in clinical trials approved by the Institutional Review Board. Prognostic factors associated with overall survival (OS) and progression-free survival (PFS) in the liver and extrahepatic (systemic) organs were retrospectively evaluated. Covariates of age, sex, preexisting extrahepatic metastases, liver enzyme levels, tumor volume, radiologic response in hepatic metastases, and treatment type were analyzed.

Results: Compared with CE, high-dose (≥1500 μg of GM-CSF) IE resulted in significantly better OS (20.4 vs 9.8 months, P = .005) and systemic PFS (12.4 vs 4.8 months, P = .001) at univariate analysis. Overall, women outlived men (14.4 vs 9.8 months, P = .01). Patients who achieved regression of hepatic metastases after embolization lived much longer than did those who did not achieve regression (27.2 vs 9.9 months, P < .001). At multivariate analysis, prolonged OS was confirmed for women, patients who underwent high-dose IE, younger patients (age < 60 years), and patients with regression of hepatic metastases. Independent predictors of longer systemic PFS included high-dose IE, younger age, and regression of hepatic metastases. No covariate predicted liver PFS except for hepatic response.

Conclusion: Treatment with high-dose IE prolonged survival of patients with uveal melanoma who received embolization of hepatic metastases and possibly delayed progression of extrahepatic metastases.

© RSNA, 2009

Introduction

Uveal melanoma is the most common noncutaneous melanoma (1) and has an estimated 5-year survival rate of between 50% and 70% (2). Roughly half of all patients will develop metastases (3), which disproportionately target the liver through hematogenous spread (4). After diagnosis of liver metastasis, median survival is reported to be between 2 and 7 months (3,5). Controlling hepatic metastases is therefore essential to extending patient survival. Systemic therapies that are used in patients with cutaneous melanoma have exceptionally poor activity against metastatic uveal melanoma (6), and as few as 9% of patients with metastatic uveal melanoma are candidates for surgical resection (7,8). Intrahepatic arterial chemotherapy and isolated hepatic perfusion have been shown to have some success in the local treatment of liver metastases (9), but no standard treatment protocol exists for patients with metastatic uveal melanoma.

Chemoembolization (CE) has been shown to be effective for the treatment of hepatocellular carcinoma (10,11), as well as that of metastatic hepatic tumors from other primary origins (12–15). In patients with metastatic uveal melanoma, the response rates for CE (overall up to 66% with cisplatin) are higher than those for traditional systemic intravenous chemotherapies (generally less than 20%) (16–18).

Our group conducted a phase II clinical trial of 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) after we were unable to confirm the previously published data on CE with cisplatin (19,20). We chose BCNU on the basis of its established efficacy for metastatic melanoma and the fact that its hepatic extraction rate is greater than six times higher than that of cisplatin (21). Since BCNU, unlike cisplatin, is oil soluble, it can also be directly reconstituted at high concentrations in a small amount of oily contrast agent. Despite the observed improvement in the control of hepatic metastases with CE with BCNU, patients who achieve stabilization of hepatic metastases subsequently develop systemic extrahepatic metastases within a median of 4.4 months (range, 1.1–17.1 months) (20). To delay or suppress the growth of extrahepatic metastases, investigators have used granulocyte macrophage colony–stimulating factor (GM-CSF) for immunoembolization (IE) (22). IE with GM-CSF is designed to destroy hepatic metastases through the ischemic effects of embolization, to stimulate antigen presenting cells (eg, dendritic and Kupffer cells) with the GM-CSF to facilitate antigen uptake, and to possibly enhance systemic immunity against tumor cells.

Prior studies (6,23–30) have examined factors that predict outcome in patients with metastatic uveal melanoma. However, prognostic factors in patients who receive hepatic arterial embolization have yet to be analyzed. In this study, we retrospectively investigated factors that were predicative of survival in patients with metastatic uveal melanoma who received embolization treatment with either BCNU or GM-CSF.

MATERIALS AND METHODS

Patients

Consecutive patients treated with embolization of the hepatic artery for metastatic uveal melanoma at Thomas Jefferson University between April 1995 and February 2004 were retrospectively evaluated (Table 1). Patients were treated with either CE with BCNU (phase II study [20], 1995–2000) or IE with GM-CSF (phase I/IIa study [22], 2000–2004) under two different protocols approved by the Clinical Cancer Research Review Committee and the Institutional Review Board of Thomas Jefferson University. All patients provided written informed consent.

Table 1.

Patient Demographics and Treatment Responses

Note.—Unless otherwise specified, data are numbers of patients, with percentages in parentheses.

^*Data are medians, with ranges in parentheses.

^†Systemic chemotherapy (n = 2) and arterial chemoinfusion (n = 1).

^‡Systemic chemotherapy (n = 1); interferon (n = 1); surgery and cancer vaccine (n = 2); and cancer vaccine, interleukin 2, and chemotherapy (n = 1).

^§Surgery (n = 1).

BCNU was obtained from the in-patient pharmacy of Thomas Jefferson University Hospital, and the cost was covered by the health insurance of each individual patient. Yeast-derived GM-CSF (sargramostim, Leukine; Bayer HealthCare Pharmaceuticals, Wayne, NJ) was provided by the manufacturer free of charge. Bayer HealthCare Pharmaceuticals and the Commonwealth of Pennsylvania (grant no. ME-01–329) provided the funding to support the phase I clinical trial of IE. The data collection and analysis were performed independently from Bayer HealthCare Pharmaceuticals.

For inclusion in the IE study, patients were required to have an Eastern Cooperative Oncology Group performance status score of 0 or 1; a patent portal vein; and adequate liver (serum bilirubin ≤ 2 mg/dL; albumin ≥ 3 g/dL), renal (creatinine ≤ 2 mg/dL), and bone marrow (granulocyte count > 1000/mm³; platelet count > 100 000/mm³) function. All patients had unresectable metastatic uveal melanoma involving less than 50% of the liver parenchyma. The percentage of liver involvement was qualitatively measured on contrast agent–enhanced computed tomographic (CT) or magnetic resonance (MR) images or both prior to the first treatment. Patients who had non–life threatening extrahepatic metastasis were allowed to participate. Patients who had received previous treatments were also enrolled, as long as the progression of hepatic metastases was radiologically confirmed. Some of the patients from the study of CE who fulfilled the above inclusion criteria and had less than 50% hepatic involvement were included in our retrospective study.

Procedures and Evaluation of Response

Embolization procedures were performed according to previously published methods (20,22). Briefly, for each lobe with a tumor, a hepatic artery that supplied it was cannulated, and either 100 mg of BCNU dissolved in ethiodized oil or various doses of GM-CSF (dose escalation from 25 to 2000 μg) emulsified in ethiodized oil was infused. The dose of BCNU for CE was chosen to be 100 mg on the basis of the standard intravenous dose of BCNU that could safely be given every 3 weeks. After the infusion, the cannulated hepatic artery was embolized with absorbable gelatin sponge particles until near stasis of antegrade blood flow was achieved.

Patients with disease confined to one lobe received treatment of that lobe, while patients with disease involving both lobes received treatment to one lobe at each session, with which lobe was treated alternating. When applicable, a more targeted (subselective) tumor embolization was performed. Treatment was performed every 3 weeks for CE and every 4 weeks for IE until progressive disease or unacceptable adverse events, including procedure-related serious complications (eg, dissection of the hepatic artery despite stabilization of disease), occurred.

Hepatic response was evaluated by using spiral CT of the chest, abdomen, and pelvis with and without contrast enhancement with a 5-mm contiguous reconstruction algorithm. MR imaging of the liver with and without contrast enhancement was also obtained after every other treatment. Hepatic tumor response was classified on the basis of the Response Evaluation Criteria in Solid Tumors, or RECIST, (31): Up to six target hepatic lesions were defined and measured prior to the initiation of treatment. The measurable target lesions had to be at least 10 mm in maximum diameter. The Response Evaluation Criteria in Solid Tumors criteria were also applied to assessment of extrahepatic metastases. The appearance of a new measurable (≥10-mm) extrahepatic metastasis or a 20% or greater increase in the sum of the longest diameters of existing measurable extrahepatic metastases was considered to be progression in extrahepatic sites. The response assessment was performed by independent board-certified radiologists. Best radiologic response in the liver was used for efficacy analysis.

Overall survival (OS) was measured from the first embolization treatment to death. Progression-free survival (PFS) was defined as the time from first embolization to documentation of disease progression or death. Liver PFS (time from first embolization to progression of liver tumor or death, whichever came first) and systemic PFS (time from first embolization to progression of extrahepatic metastases or death, whichever came first) were separately analyzed.

Variables for Statistical Analysis

Information regarding sex, age, preexisting extrahepatic metastases, lactate dehydrogenase, tumor volume in the liver (<20% vs 20%–50% involvement), aspartate aminotransferase, alkaline phosphatase, treatment type (CE or IE), and hepatic response to treatment was collected. Categorical data (ie, sex, extrahepatic metastases, tumor volume, treatment type, hepatic response) were summarized with frequencies and percentages. Continuous variables (ie, lactate dehydrogenase, aspartate aminotransferase, alkaline phosphatase) were dichotomized into two groups: values above the normal limit and values equal to or below the normal limit. Age was dichotomized into two groups (<60 years and ≥60 years) on the basis of the median age of all patients (58 years) rounded to the nearest decade. Hepatic response was categorized as regression (including complete and partial responses) or nonregression (including stable and progressive disease). IE doses were categorized as high (≥ 1500 μg) or low (< 1500 μg).

Statistical Analysis

The intent-to-treat analyses of OS, liver PFS, and systemic PFS were performed by using the Kaplan-Meier method. Univariate associations between survival, categorical risk factors, and treatment type were evaluated by using the log-rank test, and a P value of less than .05 was considered to indicate a significant difference. Multivariate analyses were performed by using the Cox proportional hazards model. For the fitted Cox models, the proportional hazards assumptions were verified (32). Statistical analyses were performed by using software (SAS, version 9.1; SAS Institute, Cary, NC).

RESULTS

Clinical Characteristics

A total of 53 consecutive patients with uveal melanoma with less than 50% metastatic involvement of the liver were evaluated in this study. Eighteen patients underwent CE between April 1995 and June 1999 in a phase II clinical trial, and 34 patients underwent IE between February 2000 and February 2004 in a subsequent phase I/IIa clinical trial. Additionally, one patient who was scheduled to undergo IE in 2000 was instead treated with CE because of issues with insurance coverage. Therefore, a total of 19 patients were treated with CE (Table 1). There was no overlap in the patient accrual periods, except for this case.

At the time of our analysis, only one patient was still alive (40.8 months after treatment). This patient had received 2000 μg of GM-CSF and had progression of hepatic and extrahepatic metastases. One patient who underwent CE was withdrawn from the study prior to radiologic evaluation owing to deteriorating medical condition. Two patients in the low-dose IE group did not receive the two IE treatments required for assessment of response owing to development of brain metastases and protocol violation (preexisting substantial liver dysfunction). One patient in the high-dose IE group was removed from the study and placed in hospice care before radiologic evaluation owing to substantial liver pain. Overall, 18 of the 19 patients in the CE group and 31 of the 34 patients in the IE group were evaluated for radiologic hepatic response. All patients were included in intent-to-treat survival analyses.

Among the 19 patients who underwent CE, the majority were men (14 patients, 74%), while among the 34 patients who underwent IE, the majority were women (22 patients, 65%) (P = .01 by using the two-tailed Fisher exact probability test). There was no significant difference in gender between the low- and high-dose IE groups. Age, preexisting extrahepatic metastases, percentage liver involvement, and baseline enzyme levels before the first treatment were not significantly different between these groups. The overall response rate for the hepatic metastases on the basis of the Response Evaluation Criteria in Solid Tumors criteria (complete and partial response) was 17% in the CE group (n = 18) and 32% in the IE group (n = 31), with no significant difference between high- and low-dose IE groups.

Univariate Analysis of Survival

As shown in Table 2, OS was significantly longer in patients who underwent high-dose IE (median, 20.4 months; 95% CI: 11.4, 28.0) than in those who underwent CE (median, 9.8 months; 95% CI: 4.7, 14.1) (P = .005). Patients who underwent low-dose IE showed no significant difference in OS from those who underwent CE. Additionally, sex significantly affected OS. Women had a median OS of 14.4 months (95% CI: 12.7, 22.8), whereas men had a median OS of 9.8 months (95% CI: 5.1, 12.4) (P = .01). Of note, patients who achieved regression (complete or partial response) of hepatic metastases after embolization lived longer (median, 27.2 months; 95% CI; 22.8, 35.9) than did those who had stable or progressive disease (median, 9.9 months; 95% CI: 7.5, 13.7) (P < .001).

Table 2.

Univariate Survival Analysis

^*Data are median, with 95% confidence interval (CI) in parentheses.

^†Obtained by using the log-rank test.

^‡Three-way comparison: CE versus high-dose IE versus low-dose IE.

^§Versus CE alone.

In terms of PFS, the median systemic PFS in the high-dose IE group was 12.4 months (95% CI: 11.2, 18.4), which was significantly longer than that in the CE group (median, 4.8 months; 95% CI: 2.7, 6.4) (P = .001) (Table 2). The analysis also showed a significant difference in median systemic PFS between the low-dose (median, 5.6 months) and high-dose (median, 12.4 months) IE groups (P = .007). Patients aged 60 years or older showed shorter systemic PFS (median, 5.6 months) than did those younger than 60 years (median, 7.5 months) (P = .02).

In contrast, there were no significant differences between the CE, high-dose IE, and low-dose IE groups with respect to liver PFS. OS, systemic PFS, and liver PFS were not significantly different between groups when stratified solely by extrahepatic metastases, lactate dehydrogenase, liver involvement, aspartate aminotransferase, or alkaline phosphatase (Table 2).

Multivariate Analysis of Survival

OS.—The fitted Cox proportional hazards model for OS had appropriate goodness of fit (Wald, score, and likelihood-ratio tests of the global hypothesis of no covariate effect, all P < .0001), fair predictive power (R² = 0.593), and no indication of violation of proportional hazards assumptions globally or in any covariate (as indicated by nonsignificant tests of the proportional hazards) (32). Results from this model are reported in Table 3.

Table 3.

Multivariate Survival Analysis

^*For all treatment groups.

^†For men and women.

In the multivariate Cox model for OS, the difference between patients who underwent high-dose IE and those who underwent CE was significant (hazard ratio = 0.23; 95% CI: 0.09, 0.57; P = .001) and favored of high-dose IE (Table 3, Fig 1a). Regression (complete and partial response on the basis of the Response Evaluation Criteria in Solid Tumors criteria) of hepatic metastases was a critically important factor for longer OS (hazard ratio = 0.11; 95% CI: 0.04, 0.27; P < .001) (Table 3, Fig 1b). OS in men was shorter (hazard ratio = 3.02; 95% CI: 1.46, 6.22; P = .003) than that in women. The effect of older (≥60 years) age was also significant (hazard ratio = 2.04; 95% CI: 1.02, 4.10; P = .045).

Figure 1a:

Survival curves, adjusted for effects of other covariates in the multivariate Cox models, show OS (a) by treatment type and (b) by radiologic response in the liver after treatment (nonregression [SD, PD]vs regression [CR, PR]). CR = complete response, HDIE = high-dose IE, LDIE = low-dose IE, PD = progressive disease, PR = partial response, SD = stable disease.

Figure 1b:

Survival curves, adjusted for effects of other covariates in the multivariate Cox models, show OS (a) by treatment type and (b) by radiologic response in the liver after treatment (nonregression [SD, PD]vs regression [CR, PR]). CR = complete response, HDIE = high-dose IE, LDIE = low-dose IE, PD = progressive disease, PR = partial response, SD = stable disease.

Liver PFS.—Overall, the Cox model for liver PFS (Table 3) was adequate (Wald, score, and likelihood-ratio tests of the global hypothesis of no covariate effect, all P ≤ .007) and somewhat predictive of liver PFS (R² = 0.31). The global test and individual covariate tests of the proportional hazards assumptions were not significant. Only hepatic treatment response (complete or partial response vs stable or progressive disease) was a significant predictor of liver PFS (hazard ratio = 0.25; 95% CI: 0.12, 0.54; P < .001).

Systemic PFS.—For systemic PFS, the results of the Cox model of the effects of age, sex, treatment type, and treatment response are reported in Table 3. The model includes a nonsignificant interaction between treatment and sex (P = .085 for high-dose IE by sex; P = .703 for low-dose IE by sex), which was required to satisfy proportional hazards assumptions. The resulting Cox model with treatment-by-gender interaction had appropriate goodness of fit (Wald, score, and likelihood-ratio tests of the global hypothesis of no covariate effect, all P < .0001) and fair predictive power (R² = 0.487).

In the multivariate Cox model, treatment with high-dose IE resulted in significant improvement in systemic PFS (P = .001) in men and women considered together, while the overall sex difference was also significant (P = .047). Improvement for high-dose IE versus CE was more substantial in men (hazard ratio = 0.13; 95% CI: 0.04, 0.41; P < .001) than in women (hazard ratio = 0.53; 95% CI: 0.17, 1.68; P = .28), although the nonsignificant finding in women is generally owing to the limited sample size (only four women in the CE group). Systemic PFS by treatment type and sex are shown in Figure 2. Systemic PFS for patients who underwent low-dose IE was similar to that in patients who underwent CE. Longer systemic PFS was associated with regression of hepatic metastases (complete or partial response) (hazard ratio = 0.41; 95% CI: 0.19, 0.88; P = .022). Additionally, shorter systemic PFS was associated with older age (>60 years) (hazard ratio = 3.30; 95% CI: 1.45, 7.51; P = .005).

Figure 2a:

Survival curves, adjusted for the effects of other covariates in the multivariate Cox models, show systemic PFS by treatment type for (a) men and (b) women. HDIE = high-dose IE, LDIE = low-dose IE.

Figure 2b:

Survival curves, adjusted for the effects of other covariates in the multivariate Cox models, show systemic PFS by treatment type for (a) men and (b) women. HDIE = high-dose IE, LDIE = low-dose IE.

DISCUSSION

In this study, we identified several prognostic factors for survival in patients with primary uveal melanoma metastatic to the liver who were treated with one of two forms of embolization of the hepatic artery. As shown in univariate and multivariate analyses, patients who underwent high-dose IE (GM-CSF ≥ 1500 μg) had significantly longer OS and systemic PFS than did patients who underwent CE. The critical role of high-dose GM-CSF in IE is further supported by the fact that there was no significant difference in OS or systemic PFS between patients who underwent CE and those who underwent low-dose IE (GM-CSF < 1500 μg). These results are extremely encouraging because the IE technique was developed to control liver metastases and, at the same time, to delay the development or progression of extrahepatic metastases.

Despite the presence of numerous antigen-presenting cells, (eg, sinusoidal endothelial cells, liver macrophages [Kupffer cells], and dendritic cells), the liver is often a site of tumor metastasis. Because the liver is exposed to food-derived antigens and probiotics from the gastrointestinal tract through the portal vein, it tends to have induced peripheral tolerance rather than immunity to avoid unnecessary activation of the immune system (33). It has also been reported (34) that antigens from apoptotic tumor cells presented by liver sinusoidal endothelial cells lead to tumor tolerance. Furthermore, tumor cells could produce inhibitory cytokines (eg, interleukin 10, transforming growth factor beta) to actively suppress antitumor immune response (35,36).

To overcome the tolerogenic microenvironment in the liver, some type of stimulatory change should be introduced to the tumor microenvironment. This might be achieved by the destruction of tumor with radiofrequency ablation, cryoablation, or embolization. Addition of immunostimulatory cytokines (eg, GM-CSF) or blockage of inhibitory cytokines would further facilitate the remodeling of the immunologic environment in the liver and might result in the development of local and systemic immune responses against tumor cells. In immunoembolization, hepatic tumors and surrounding normal hepatic tissues were exposed to GM-CSF. GM-CSF is a glycoprotein secreted principally by activated T cells that stimulates antigen-presenting cells, such as macrophages and dendritic cells (37). We hypothesize that ischemic injury caused by embolization of the hepatic artery allows hepatic antigen-presenting cells to take up tumor antigens and present them on the surface to stimulate tumor-specific T cells (38). The infusion of GM-CSF at the time of embolization may facilitate this process by inducing an inflammatory response in the tumor and surrounding tissue, which stimulates the production of T helper type 1 cytokines (39). This might result in the enhancement of an antitumor immune response locally. Although further preclinical and clinical investigations are needed, it is possible that the delay in systemic extrahepatic progression could be related to the development of a systemic immune response against tumors following treatment with high-dose IE. Despite our hypothesis, it is possible that GM-CSF released into the systemic circulation after intrahepatic arterial infusion directly stimulated the systemic immune system. Therefore, the importance of intrahepatic arterial infusion of GM-CSF prior to embolization may need to be investigated in a future clinical trial in which patients are randomized to undergo either IE with GM-CSF or plain embolization with systemic administration of GM-CSF.

Radiologic response in hepatic metastases was an independent prognostic factor for OS. Patients who achieved regression of hepatic metastasis (complete or partial response) after embolization lived much longer compared with those in whom this goal was not achieved, indicating that control of hepatic metastases remains a critical factor for prolonging OS. In this regard, there was no significant difference in liver PFS or radiologic response of hepatic metastases between the three treatment types, suggesting that, regardless of treatment type, regression of hepatic metastases after embolization predicted better OS. The above observations also explain the fact that the difference in treatment interval and follow-up frequency (IE every 4 weeks with evaluation every 8 weeks vs CE every 3 weeks with evaluation every 6 weeks) did not significantly influence the control of hepatic metastases. Because OS and PFS in the low-dose IE group were not better than those in the CE group, the differences in treatment and follow-up interval do not explain the survival benefit seen in the high-dose IE group.

Univariate and multivariate analyses showed that men had significantly shorter OS. The negative impact of male sex on survival has been noted in some (25,29,40,41), but not all (6,23,24,27,30), other studies that examined patients with metastatic uveal melanoma. Since there is an unexpected uneven distribution of sex in CE and IE studies and only four women treated with CE in our study, it is not clear whether the difference in survival improvement between men and women is real. Stratification of patients by gender might be needed in future clinical studies to answer this question.

Another significant prognostic factor was age at initiation of embolization treatment. This is especially true for systemic PFS (Table 3) and may be explained by age-related deterioration of the immune system. It has been reported that aging could lead to the increased susceptibility of the elderly to infection, autoimmune disease, and cancer and could contribute significantly to morbidity and mortality (42,43). Specific changes are the reduction of natural killer cell function, decreased T-lymphocyte response to proliferative stimuli, and hyperfunction of T helper type 2 cells (44). Alteration in the surveillance system in older patients may have allowed uveal melanoma cells to grow in extrahepatic organs after embolization of hepatic metastases.

In contrast to the findings of the majority of previous studies (6,23,24,27,40,45) that examined prognostic factors for survival of uveal melanoma metastatic to the liver, serum levels of lactate dehydrogenase, often thought to reflect the extent of tumor burden, did not emerge as a predictor of survival in our study. Our population of patients with hepatic metastasis encompassing less than 50% of the liver parenchyma might have had better disease control with embolization despite elevation of lactate dehydrogenase. This speculation is supported by the fact that tumor volume in the liver (<20% vs ≥20%) did not influence OS or liver PFS. Interestingly, a higher tumor volume (≥20%) or an abnormal lactate dehydrogenase level had a marginal impact on systemic PFS (P = .076 and .078, respectively), possibly indicating a negative impact of tumor volume on the systemic antitumor immune response.

Although this was a retrospective analysis of consecutive patients with uveal melanoma and hepatic metastases who were treated with different types and doses of medications in conjunction with standard embolization, a benefit to survival was shown in patients who received IE with GM-CSF, especially at higher doses (≥1500 μg). Because the prognosis in patients with hepatic metastases is uniformly ominous, the median OS of patients who underwent high-dose IE (20.4 months) is extremely encouraging. Since this study was a retrospective nonrandomized trial, unavoidable and nonquantifiable bias might have influenced the final conclusion. Therefore, the findings observed in this study are being further investigated in an ongoing double-blinded randomized phase II clinical study, in which patients will be randomized to receive embolization of the hepatic artery with or without administration of GM-CSF (2000 μg).

In conclusion, IE with high-dose (≥1500 μg) GM-CSF prolonged OS of patients with uveal melanoma with hepatic metastases and possibly delayed the progression of extrahepatic systemic metastases, as compared with CE with BCNU. Further investigation on efficacy and the mechanism of high-dose IE is warranted.

ADVANCES IN KNOWLEDGE

• Hepatic immunoembolization (IE) with granulocyte-macrophage colony-stimulating factor (GM-CSF), especially at higher doses, prolonged overall survival of patients with metastatic uveal melanoma, as compared with chemoembolization with 1,3-bis (2-chloroethyl)-1-nitrosourea.

• High-dose IE prolonged progression-free survival in systemic (extrahepatic) organs, as compared with chemoembolization.

IMPLICATIONS FOR PATIENT CARE

• IE with GM-CSF prolonged survival of patients with metastatic uveal melanoma, and this approach may be useful for other hepatic tumors.

• IE with GM-CSF at higher doses might prevent or delay progression of extrahepatic metastases.

Immunoembolization with high-dose (≥1500 μg) granulocyte-macrophage colony-stimulating factor prolonged overall survival of patients with uveal melanoma with hepatic metastases and possibly delayed the progression of extrahepatic systemic metastases, as compared with chemoembolization with 1,3-bis (2-chloroethyl)-1-nitrosourea.