Uveal melanoma is the most common primary intraocular malignant tumor among adults [1, 2]. The incidence of uveal melanoma is 4.3 cases per million population with a slightly higher rate among men (4.9 cases per million than women (3.7 cases per million) [3]. Uveal melanoma is more common in older age. The progressively increasing age-specific incidence peaks at age 70 years (24.5 cases per million among men, 17.8 cases per million among women) [3].

As many as 50% of patients have systemic metastasis after the initial diagnosis and treatment of primary uveal melanoma. Clinically evident metastatic disease at presentation is uncommon, however, indicating that early subclinical metastasis occurs in most cases [4, 5]. The liver is the predominant organ of involvement in 70–90% of patients with metastatic uveal melanoma and tends to be the first manifestation of metastatic disease [6–9]. Approximately 90% of patients with metastatic uveal melanoma die with hepatic metastasis [7]. In general, without treatment, survival after the development of hepatic metastasis of uveal melanoma is poor, the median survival period being 2–7 months [5, 6].

Treatment options for patients with metastatic uveal melanoma are limited. Only 9% of patients with hepatic metastasis are eligible for resection, predominantly because of the multiplicity of hepatic tumors when the patient comes to medical attention [9]. Furthermore, the standard systemic chemotherapies used for metastatic cutaneous melanoma have not been efficacious in the management of metastatic uveal melanoma [10]. However, survival has been lengthened with transarterial catheter-directed liver therapies, such as hepatic arterial chemoinfusion, transarterial chemoembolization (TACE), immunoembolization, and isolated hepatic perfusion [10–21]. Despite the development of various transarterial catheter-directed treatments, most patients with hepatic metastasis of uveal melanoma eventually progress through these first-line approaches. Therefore, subsequent salvage therapy is mandatory to lengthen the survival of patients with hepatic metastasis of uveal melanoma.

We evaluated our single-institution experience using radioembolization as salvage therapy in the care of patients with hepatic metastasis of uveal melanoma. Tumor response and related toxicity after radioembolization were assessed. The results were compared with those of other transarterial catheter-directed therapies to increase understanding of the efficacy of radioembolization in the management of hepatic metastasis of uveal melanoma.

From January 2007 through April 2009, 32 patients (14 men, 18 women; median age, 61 years; range, 29–89 years) with hepatic metastasis of uveal melanoma underwent radioembolization. Patient eligibility included tumor progression after treatment with immunoembolization or TACE, no extensive extrahepatic metastasis limiting life expectancy less than 3 months, Eastern Cooperative Oncology Group 0–2 performance status, and adequate liver function (bilirubin concentration < 1.8 mg/dL, albumin concentration > 3.0 g/dL, no ascites) and renal function (creatinine concentration ≤ 2.0 mg/dL). Exclusion criteria were uncorrectable arterial blood flow to the gastrointestinal tract diagnosed at arteriography or ^99mTc–macroaggregated albumin scintigraphy, lung shunting greater than 20%, portal vein thrombosis, biliary obstruction, and previous radiation therapy in which the liver was included in the radiation field. Before treatment, all patients underwent PET, contrast-enhanced abdominal MRI, and CT of the chest, abdomen, and pelvis. Pretreatment tumor burden was divided into the three following categories: less than 25% (n = 25), 25–50% (n = 5), and greater than 50% (n = 2). Tumor burden percentages were calculated with MRI and the following formula: volume of tumor / (volume of tumor + volume of normal liver). Pretreatment tumor burden was determined at radioactive microsphere dosimetry.

Thirty-two patients were treated with radioembolization as salvage therapy after failure of immunoembolization (median, 4 procedures; range 1–14, procedures) or TACE (median, 3 procedures; range, 1–14 procedures). The total number of procedures (immunoembolization and/or TACE) per patient ranged from 1 to 23 (median, 9). Thirty-one patients underwent radioembolization treatment without procedure-related complications. Common hepatic artery dissection occurred in one patient during preparatory arteriography. The patient received one fractionated whole-liver radioembolization treatment, but subsequent occlusion of the common hepatic artery precluded further treatment.

Radioembolization treatments were administered as either unfractionated whole-liver (eight patients), fractionated whole-liver (two patients), or consecutive right and left lobe (19 patients) treatments separated by 3–5 weeks. Three patients underwent single-lobe treatments, two of whom could not undergo the second radioembolization treatment because of deterioration of hepatic function related to progression of disease. One patient underwent only one radioembolization treatment of tumor isolated to the right lobe. Lobar treatment was generally preferred. Whole-liver treatment was performed in accordance with the arterial anatomy at the discretion of the treating interventional radiologist. Three of the patients who received unfractionated whole-liver treatments underwent left lobectomy for tumor resection before initiation of transarterial catheter-directed therapy. The median dose delivered was 1.08 GBq (29.28 mCi) per patient (range, 0.63–1.86 GBq [16.77–50.10 mCi]).

There were no procedure- or treatment-related deaths. Before radioembolization, 13 patients had preexisting liver dysfunction from previous transarterial catheter-directed therapy or intrahepatic tumor burden (Table 1). Hepatic toxicity grades after the procedure are shown in Table 1. Grade 3 and 4 liver toxicity (n = 4) 1 month after radioembolization was caused by tumor progression confirmed with either CT or MRI and was not related to the treatment itself. In the four patients with grade 3 or 4 toxicity after treatment, grades 1 (n = 1), 2 (n = 2), and 3 (n = 1) hepatic dysfunction was present before radioembolization. Only one patient with pretreatment grade 1 hepatic dysfunction had a two-level increase in hepatic dysfunction (grade 3), which was due to marked progression of liver metastasis after radioembolization. Systemic nonhepatic toxicity after radioembolization was mainly grade 1 or 2 mild tiredness (n = 9) and gastrointestinal symptoms such as indigestion (n = 2) and mild abdominal discomfort (n = 5), all of which were self-limiting.

The clinical follow-up lasted 1.0–29.0 months (median, 10.0 months). At the end of the follow-up period (September 2009), 10 patients were alive 4.7–27.0 months (median, 9.4 months) after treatment. Twenty-two patients died 1.0–29.0 months (median, 5.8 months) after radioembolization owing to progression of liver disease (n = 13), extrahepatic disease (n = 4), or both (n = 5). Two patients underwent treatment to a single lobe then died 1.0 and 3.3 months after radioembolization before completing therapy to all hepatic metastatic lesions. The causes of death were progression of hepatic metastasis and extrahepatic disease (pulmonary metastasis). The causes of death of the two patients were determined with imaging (CT and MRI) and laboratory and clinical evaluation. The patient who had common hepatic artery dissection during preparatory arteriography underwent one fractionated whole-liver treatment and died 9.8 months after radioembolization owing to progression of liver disease. These three patients were included in the intent-to-treat analysis of overall survival. Therefore, overall survival of the entire patient sample (n = 32) ranged from 1.0 to 29.0 months (median, 10.0 months) (Fig. 1). Median overall survival was found to be significantly longer for patients with a pretreatment tumor burden less than 25% (10.5 months; range, 3.1–29.0 months) than for those with a pretreatment tumor burden of 25% or greater (3.9 months; range, 1.0–12.1 months) (p = 0.0003) (Fig. 2A).

With respect to best radiologic response of hepatic metastasis, one patient had a complete response, one patient had a partial response, 18 patients had stable disease, and 12 patients had disease progression. The median overall survival period among patients with complete response, partial response, or stable disease (14.7 months) was significantly longer than the median overall survival among patients with disease progression (4.9 months) (p = 0.0006) (Fig. 2B). The median progression-free survival of hepatic metastasis for the entire patient sample (n = 32) was 4.7 months (range, 1.0–26.5 months) (Fig. 1). Patients who had a pretreatment tumor burden less than 25% had significantly longer median progression-free survival of hepatic metastasis than did patients with a 25% or greater pretreatment tumor burden (6.4 versus 3.0 months) (p = 0.03) (Fig. 3A). The median period of progression-free survival of hepatic metastasis among patients who had a complete response, partial response, or stable disease also was significantly longer than that among patients who had disease progression after radioembolization (7.9 versus 3.1 months) (p < 0.0001) (Fig. 3B).

Transarterial catheter-directed therapies for hepatic metastasis are most advantageous when systemic chemotherapy is not effective or beneficial, the liver is the only or predominant site of involvement, or there is urgent need to control hepatic progression for symptom control or palliation. These criteria are often applicable to patients with hepatic metastasis of uveal melanoma. Therefore, various transarterial catheter-directed therapies have been investigated for the management of these metastatic lesions, including hepatic arterial infusion, isolated hepatic perfusion, TACE, and immunoembolization. Results of clinical studies of transarterial catheter-directed liver treatments for hepatic metastasis of uveal melanoma are shown in Table 2. Although the number of patients in the individual studies is small, the reported overall survival among patients who received hepatic arterial infusion, TACE, immunoembolization, or isolated hepatic perfusion as first-line therapy for hepatic metastasis of uveal melanoma ranges from 6.0 to 21.0 months (median, ≈1 year) (Table 2).

Radioembolization with the pure β-emitting isotope ⁹⁰Y has been used for several years to manage both primary and metastatic hepatic tumors [25–28]. The therapeutic advantage of radioembolization is based on two important principles: first, hepatic tumors receive 80–100% of their blood supply exclusively from the hepatic artery and, second, tumor neovascularity is denser than the vasculature supplying the normal hepatic parenchyma [29, 30]. Therefore, ⁹⁰Y microspheres infused into the hepatic artery concentrate in much higher doses in hepatic tumors than in the normal surrounding liver [30]. Yttrium-90 microspheres, which have a half-life of 64.2 hours, are permanently embedded in hepatic tumors and have an average energy of 0.94 MeV [22, 30]. Most of the radiation dose is released over 14 days at approximately 50 cGy/min [31]. The mean tissue penetration of ⁹⁰Y is 2.5 mm with a maximum penetration of approximately 10 mm [22, 31]. Most of the radiation dose is administered directly to the tumor with relative sparing of the normal liver parenchyma. Because radioactive plaque is one of the standards of care in the management of primary uveal melanoma, radioembolization is a reasonable approach to the management of hepatic metastasis of uveal melanoma.

Kennedy et al. [31] reported their experience treating a small number of patients with hepatic metastasis of uveal melanoma using ⁹⁰Y radioactive microspheres. In that retrospective multicenter study, 11 patients underwent 12 radioembolization procedures with a median activity of 1.55 GBq delivered per treatment. Toxicity was minimal. Only one patient experienced a grade 3 event, a gastric ulcer that healed uneventfully within 6 weeks with supportive care. No radiation-induced liver disease was detected. CT and PET of hepatic metastatic lesions 3 months after treatment showed one complete response, six partial responses, one case of stable disease, and one case of disease progression in nine evaluable patients. Among the 10 patients with available clinical follow-up information, the 1-year survival rate was 80%. The relatively low response rate (one complete response, one partial response, and 18 cases of stable disease in 32 patients) in our study compared with that found by Kennedy et al. may be explained in part by the inclusion of previously treated patients with bulky progressive hepatic metastatic lesions. Pretreatment tumor burden has been found to be an important predictor of response after TACE. In our phase 2 trial of 1,3-bis (2-chloroethyl)-1-nitrosourea (BCNU) for TACE [15], the overall survival period among patients with less than 20% pretreatment tumor burden was found to be significantly longer (median, 19.0 months; range, 3.8–27.6 months) than that among patients with 20–50% (median, 5.6 months; range, 0.1–14.0 months) or greater than 50% pretreatment tumor burden (median, 2.1 months; range, 0.6–7.5 months). Huppert et al. [16] reported similar results of TACE using cisplatin and carboplatin in the care of 14 patients with hepatic metastasis of uveal melanoma. Median survival was 17 and 11 months among patients with pretreatment tumor burdens of less than 25% (n = 7) and 25% or greater (n = 7) (p = 0.02). These results correlate well with those of a study by Dunfee et al. [32], who evaluated prognostic indicators associated with survival after radioembolization [32]. In that study, 130 patients underwent radioembolization therapy for either primary (n = 7) or secondary (n = 123) malignant disease of the liver after lack of response to standard-of-care therapy. In that patient sample, large pretreatment tumor burden (> 50%) was found to be a statistically significant negative prognostic indicator of survival. On the basis of those observations, radioembolization may have limited benefit in the care of patients with a large pretreatment tumor burden, and, therefore, exclusion from future clinical trials should be considered. Furthermore, the median ⁹⁰Y dose delivered in our study was 1.08 GBq compared with 1.55 GBq in the report by Kennedy et al. [31]. The relatively low dose of ⁹⁰Y, which was deemed necessary to avoid adverse events in our heavily pretreated population, may have limited the effectiveness of our radioembolization treatments and may explain the lower overall response rate among our patients.

As summarized in Table 2, most studies show significantly longer overall survival among treatment responders than among non-responders. For example, in the previously discussed TACE trial of BCNU in the treatment of 30 patients with hepatic metastasis of uveal melanoma [15], median overall survival was significantly longer for patients with complete response and partial response (21.9 months; range, 7.4–27.6 months) than for patients with disease progression (3.3 months; range, 1.6–5.6 months). In our phase 1 clinical trial [17, 18] of immunoembolization with granulocyte–macrophage colony-stimulating factor (GM-CSF) in the management of hepatic metastasis of uveal melanoma in 34 patients with less than 50% tumor burden, overall survival for patients who achieved complete or partial response (33.7 months) was significantly longer than that among patients with stable disease or disease progression (12.4 months) (p = 0.0043). Furthermore, in the study by Dunfee et al. [32], response to radioembolization was found to be a statistically significant positive prognostic indicator of survival and correlated well with the findings of our study.

Information on the benefit of salvage transarterial catheter-directed therapy for hepatic metastasis of uveal melanoma is limited. In 1995, Bedikian et al. [10] reported their results in the care of 20 patients with metastatic uveal melanoma treated with TACE with a cisplatin-based regimen after failure of previous first-line therapy. Five of 20 patients treated with salvage TACE had a partial response, and two patients had stable disease. However, the overall survival of this patient sample was only 5 months, inferior to the 10.0-month median overall survival of our patient sample who underwent radioembolization salvage therapy.

To better understand the efficacy of radioembolization as salvage therapy for hepatic metastasis of uveal melanoma, we compared the results of the current study with those of previous clinical trials performed at our institution: immunoembolization with GM-CSF and TACE with BCNU [15, 17, 18]. For patients with limited (< 50%) pretreatment tumor burden, the overall survival and progression-free survival of hepatic metastasis after first-line immunoembolization and TACE were similar to the results of radioembolization as salvage therapy (Table 3). Therefore, our institutional data indicate that radioembolization may be beneficial when previous immunoembolization or TACE has failed. Furthermore, because of its safety profile and potential efficacy, the addition of radioembolization to our treatment armamentarium affords a useful means of further lengthening the survival of patients with metastatic uveal melanoma.

It can be argued that our results might have been better if radioembolization had been used as first-line instead of salvage therapy. Distal hepatic artery attenuation and obliteration of segmental hepatic artery branches have been reported after transarterial catheter-directed therapy [33]. Because our patients were heavily pretreated with either immunoembolization or TACE (median, 9 procedures), adequate delivery of radioactive microspheres to metastatic lesions of uveal melanoma might have been hampered, limiting the effectiveness of the radioembolization treatments. In this regard, we have started a new phase 2 clinical trial to prospectively investigate the safety and efficacy of radioembolization with ⁹⁰Y resin microspheres as first-line therapy for hepatic metastasis of uveal melanoma. In that clinical trial, patient stratification will be based on history of transarterial catheter-directed therapy (first-line vs salvage). The safety and efficacy of treatments will be separately analyzed in these two patient cohorts. Furthermore, we will investigate whether the molecular and genetic characteristics of hepatic metastasis of uveal melanoma correlate with the outcome of radioembolization [34]. Biopsy specimens of hepatic metastatic lesions obtained before radioembolization may provide insight on the biologic activity of these tumors and thereby aid in prediction of treatment response. Because of the retrospective nature of this study, genetic analysis of hepatic metastatic lesions of uveal melanoma before radioembolization treatments was not performed, and that is a limitation.

To our knowledge, the current study included the largest number of patients with metastatic uveal melanoma treated with radioembolization at a single institution. The results of this study indicate that radioembolization is safe and effective salvage therapy for metastatic uveal melanoma in patients with a limited pretreatment tumor burden. The addition of radioembolization to our treatment armamentarium constitutes a useful means of lengthening survival among patients with metastatic uveal melanoma after failure of immunoembolization and TACE. Further investigation is needed to determine whether radioembolization should be used as an alternative first-line therapy for hepatic metastasis of uveal melanoma.

Address correspondence to C. F. Gonsalves ([email protected]).