New Therapeutic Perspectives in the Treatment of Uveal Melanoma: A Systematic Review

Uveal melanoma (UM) is a rare disease, but the most common primary intraocular cancer, mostly localized in the choroid. Currently, the first-line treatment options for UM are radiation therapy, resection, and enucleation. However, although these treatments could potentially be curative, half of all patients will develop metastatic disease, whose prognosis is still poor. Indeed, effective therapeutic options for patients with advanced or metastatic disease are still lacking. Recently, the development of new treatment modalities with a lower incidence of adverse events, a better disease control rate, and new therapeutic approaches, have merged as new potential and promising therapeutic strategies. Additionally, several clinical trials are ongoing to find new therapeutic options, mainly for those with metastatic disease. Many interventions are still in the preliminary phases of clinical development, being investigated in phase I trial or phase I/II. The success of these trials could be crucial for changing the prognosis of patients with advanced/metastatic UM. In this systematic review, we analyzed all emerging and available literature on the new perspectives in the treatment of UM and patient outcomes; furthermore, their current limitations and more common adverse events are summarized.


Introduction
Uveal melanoma (UM) is a rare disease (5% of all melanomas), but the most common primary intraocular cancer in adults (mean age of 60) [1,2]. It is more frequent in Caucasians, with an incidence of 0.69 per 100,000 person-year for males and 0. 54  For unpublished data, no effort was made to contact the corresponding authors.

Results
The results of the search strategy are summarized in Figure 1. From 75 articles extracted from the initial research, 73 abstracts were identified for screening and 64 of these met the inclusion/exclusion criteria for full-text review. Moreover, 11 articles were excluded, including 4 studies without efficacy data, 3 protocols, 1 without full-text, 1 nonoriginal article (Figure 1).  [18]. The 64 studies included in this systematic review were divided in the following groups: treatment of local complications (n = 3; Table 1), local treatment of primary tumor or metastases (n = 13; Table 2), and systemic therapy (n = 48; Table 3). Only one study was a randomized, double blind, placebo-controlled phase III trial; the other studies were phase I/pilot studies (n = 14), phase II studies (n = 34), other designs/design not available (n = 16). Tables 4-6 summarize the main efficacy and safety results of these studies. No data synthesis was possible for the heterogeneity of available data and the design of the available studies (i.e., case reports or case series). Thus, the current systematic review reports a qualitative analysis, detailed issue-by-issue below in narrative fashion. All patients were treated with iodine 125 plaque radiotherapy (8000 cGy to the tumor apex); adjuvant TTT applied to the tumor in most cases. Patients in the triamcinolone group received 3 periocular injections of triamcinolone (at the completion of plaque application, at 4-and 8-month follow-up) 18 months BCVA = best-corrected visual acuity; IVT = intravitreal; PRN = pro re nata; RCT = randomized clinical, trial; SD OCT = spectral-domain optical coherence tomography; TRP = targeted panretinal photocoagulation; TTP = time to progression; TTT = transpupillary thermotherapy. The median follow-up time was 6.5 months (range 4-9 months)           The triamcinolone group had a significantly lower risk of developing macular edema (hazard estimate, 0.35; 95% confidence interval, 0.11-0.58; p = 0.004). Other factors predictive of development of macular edema were largest tumor base (p = 0.001) and tumor thickness (p = 0.018). By multivariate analysis, triamcinolone treatment was the most significant factor associated with a lower risk of macular edema (hazard estimate, 0.45; 95% confidence interval, 0.19-0.70; p = 0.001). At 18-month follow-up, moderate vision loss occurred significantly less frequently in the triamcinolone group than in the control group (31% vs. 48%; chi-square, 4.25; 1 df; p = 0.039).
In 8 patients (7%), raised IOP developed after the 1st or 2nd triamcinolone injection. Elevated IOP occurred in 15% of the triamcinolone group and in 7% of the control group (chi-square, 1.93; 1 df; p = 0.165). All cases were controlled with topical treatment. Rates of cataract progression were similar in both groups. No case of globe perforation associated with periocular injection. AEs = adverse events; BCVA = best-corrected visual acuity; IO= intraocular; IOP = intraocular pressure. Overall radiological response: 68% of patients (12% CR, 56% PR, 18% SD, and 15% PD); time to local progression: 7 months; 68% of patients developed extrahepatic metastases after a median of 13 months, and the median OS was 24 months. Significant survival advantage of 14 months (p = 0.029) when comparing these patients with a control group.
No postoperative mortality was observed. There were 3 major complications: 1 patient developed a systemic inflammatory response syndrome with respiratory insufficiency as well as renal and cardiovascular failure; 1 patient also developed respiratory insufficiency with pneumonia and pleural effusion; 1 patient perforated duodenal ulcer at the fourth postoperative day. The AUC of oxaliplatin at the MTD of 100 mg oxaliplatin ranged from 11.9 mg/L h to 16.5 mg/L h. All 4 patients treated at the MTD showed progressive disease 3 months after IHP. Only 8 patients were available for response evaluation of which 3 patients showed a PR, with a duration of response of 6.5-11.1 months. Median OS was 18.7 months.
Dose limiting sinusoidal obstruction syndrome (SOS) occurred at 150 mg oxaliplatin. No patient showed CR, 8 patients (57%) showed PR, 4 patients (29%) had SD and 2 patients (14%) had PD. Time to progression ranged between 5 and 35 months (median 8.5 months). Median survival of all patients was 11.5 months (3-69 months) following first TACE and 18.5 months (5-75 months) following diagnosis of liver metastases. At the time of data analysis, 10 patients had died and 4 patients were alive. The survival rate was 86% at 6 months, 50% at 12 months, 28% at 18 months, and 14% at 24 months following first TACE.
Symptoms of post-embolization syndrome were seen in all patients. In 2 patients, the application of additional morphine was necessary to overcome right upper quadrant pain. In 1 patient, acute renal insufficiency occurred after the second TACE. All patients had an objective response; 3 patients had a major response with evidence of metastases reduction of 90%, 3 had a reduction of 80% and 4 presented a reduction between 70% and 60%; 8 patients are alive at the time of this analysis.
The most important AE was abdominal pain during the procedure. 8 Voelter V et al., 2008 [42] Median survival for patients treated with fotemustine was 9 years (95% confidence interval (CI) 2.2-12.7) compared with 7.4 years in the control group (95% CI 5.4-12.7). The corresponding 5-year survival rates were 75 and 56%, respectively (p = 0.539). The estimated hazard ratio for death at 5 years was estimated at 0.98, with a 95% CI of 0.38-2.61, p = 0.981 5 patients (23%) received only the induction cycle (3-4 infusions), with early treatment cessation owing to hepatic toxicity (n = 4) and/or catheter-related complications (n = 2). The main side effect of the adjuvant treatment was drug-induced hepatitis; 3 patients experienced grade 3 gastric toxicity; 2 patients presented with grade 3 neutropenia and 1 with grade 3 thrombocytopenia. 9 van Iersel LB et al., 2008 [43] Of the 12 UM patients perfused, 4 (33%) had a PR, 6 (50%) patients had SD, and 2 (17%) patients were immediately progressive. Median DFS was 6.6 months with a median OS of 10.0 months. 50% of other primary tumors showed at least PR.
Major complications occurred in 3 patients in the week following the perfusion procedure: VOD in 2 patients and lung embolism in 1 patient. 11 Egerer G et al., 2001 [45] 2 patients achieved a PR, 3 had SD, and in 2 patients, the tumor progressed in the liver. Extrahepatic progression to lung was diagnosed in 1 patient. The median survival was 18 months (range, 3-43 months) with a median TTP of 16 months (range, 0 to 43 months) in 6 patients. The median survival from diagnosis of the primary tumor was 54 months (range, 31-122 months); 2 patients survived for more than 2 years, and 2 patients are alive at present with residual liver metastasis, at 3 and at 43 months.
The toxicity of the locoregional chemotherapy was minimal in the 7 patients 12 Hussain RN et al., 2020 [46] No patients achieved CR or PR at any visit. All required enucleation. The study was terminated early, as alternative treatments were clearly superior for local tumor control.
/ The longest duration of tumor control = 6.5 years. 76% of melanomas were not growing at the end of the first year, 62% at the end of the second year, 38% with no signs of growth at the end of the fifth year. Three patients died during the 5-year period: 1 from melanoma metastases after 2 years, the second from a concurrent malignancy, and the third from heart failure after 1.7 years.  Most common TEAE were dysgeusia (49%), thrombocytopenia (48%), fatigue (26%), and nausea (25%). Most common grade 3/4 TEAEs were thrombocytopenia (35%) and anemia (6%). Dose-limiting toxicities included thrombocytopenia, gastrointestinal bleed, hypertension, fatigue, decreased appetite, and aspartate aminotransferase elevation. At dose levels 1a-3a, no relevant toxicity was observed. Acute toxicity was mainly a flu-like syndrome with fever, chills, arthromyalgia, headache, nausea, and vomiting, and diarrhea. The first patient at the level 5a experienced transaminitis grade 3 and grade 3 thrombocytopenia. 8 Carvajal RD et al., 2018 [55] The primary endpoint of PFS was not met. In the selumetinib + dacarbazine group, there were 82 events (85%) compared with 24 (75%) for placebo + dacarbazine; the HR for PFS was 0.78 (95% CI: 0.48 to 1.27; p value = 0.32), with no significant benefit of selumetinib. Median PFS was 2.8 months in the selumetinib + dacarbazine and 1.8 months in the placebo + dacarbazine group.
Incidence of AEs of special interest was more frequent with selumetinib plus dacarbazine; however, these were generally grade 1/2.
All patients experienced at least 1 AE of any grade, and 88% reported drug-related AEs. Grade ≥ 3 AEs were reported in 53%. Overall, 67% experienced rash. 14% experienced ophthalmic AEs (visual impairment, photopsia, blurred vision, photophobia, periorbital, and retinal edema); 27% experienced at least 1 SAE. 23% had SD for at least 16 weeks and were considered to have clinical benefits. Overall, SD was the best objective outcome. Median duration of SD in these patients was 8 weeks (range: 8-16 weeks). 6/13 patients had PD on first assessment. Median PFS from first date of treatment was 16 weeks (range: 7-23 weeks); median OS from first day of study treatment was 11 months (range: 4.5-28.5 months).
Most patients (n = 29, 96.7%) reported at least 1 AE related to treatment, and the total incidence of grade 3 or 4 AE was 66.7%. The most common AE was neutropenia (67.7%), with half of the patients experiencing grade 3 or 4. Non-hematologic toxicity was less common and less severe. The median number of treatment cycles was 1 (range 1-7 cycles). 7 patients (32%) had SD with a median duration of 3 months (range: 3-7 months). The median OS was 9.8 months.
Reversible drug-related severe (grade 3) AEs in 13/21 patients; anorexia and fatigue were mostly of mild or moderate severity. 22 Danielli R et al., 2012 [69] No objective responses were observed; however, 2 patients had SD, a third patient had SD after initial PD. Median OS was 36 weeks (range 2-172 weeks).
The most common TEAEs were rash/dermatitis acneiform (80 out of 97; 82%) and diarrhea (n = 44; 45%), most of which were grade 2 or lower. No objective responses were seen in any of the three patient cohorts. The best overall response was SD (31%); 71% had progressive disease at the 1 response assessment. In the phase II patient cohort, 23.5% patients had SD. The longest duration of SD in the cohort receiving the highest dose was 169 days. SDs >6 months occurred in patients with UM. The median duration of SD in the UM patients was 141 days (range: 57-337 days). The median TTP for the 29 evaluable patients was 57 days, whereas 113 days for the patients with UM.
No DLTs were seen in the first cohort (40 IU/m 2 ). In cohort 2 (80 IU/m 2 ), 1 grade 3 episode of arthralgia. No DLTs were observed in cohort 3 and enrollment onto the phase II part of the protocol was continued. MTD was not reached. Overall, the treatment was well tolerated; 6 grade 3 toxicities were observed in total. 27 Mahipal A et al., 2012 [74] 1 patient achieved a PR and 12 had a SD, with the duration of SD ranging from 2.1 to 29.2 months (median = 5.5 months). The patient who achieved a PR remained on treatment for 13 months. The median OS and PFS were 8.2 and 4.2 months, respectively; 3 patients had SD for more than 12 months with sunitinib after failing previous treatments.
The most common AEs were fatigue (90%), diarrhea (60%), hemorrhage (55%), anorexia (45%), hand-foot syndrome (25%), hypothyroidism (25%), and rash (25%); 11 patients required dose reduction due to grade 3 AEs. 28 Lane A total of 9 patients (75%) received imatinib for 8 weeks; 25% discontinued because of disease progression. No patient achieved an objective response; the best clinical response was a SD in 1 patient, which lasted for 52 weeks. Thus, the median PFS was not calculated. The median OS of all patients was 6.8 months. For the 8 patients who received imatinib as 1 line therapy, the median OS was 7.8 months. The 4 patients treated in a 2-line setting had a median OS of 4.9 months.
Abdominal pain and vomiting was the most common toxicity, resulting in a dose reduction in 2 cases (17%); 1 patient had facial edema. There was no significant hematologic toxicity. Most of the side effects were grade 1-2 neurological, gastrointestinal, or constitutional. Nausea was the most common side effect. It occurred in about 75% of the patients and was usually mild. Neurologic AEs were constipation, hypoesthesia, anxiety, paresthesia, and peripheral neuropathy. The hematologic side effects were mild (mostly grade 1/2 neutropenia). None developed grade 3/4 thrombocytopenia. 31 Penel N et al., 2008 [95] No objective response and only 1 SD with duration of 5 months were noted. No patient was found to be free of disease progression 6 months after the initiation of treatment. The overall survival was 10.8 months.
5 and 1 out of 13 enrolled patients experienced grade 3 and grade 4 toxicities, respectively. The most common severe AEs were abdominal pain. 32 Adjei AA et al., 2008 [78] 19 patients (33%) had SD at the end of cycle 2, and 9 patients (16%) had SD for ≥ 5 months; 1 patient with medullary thyroid cancer experienced SD for 19 cycles, whereas 1 patient with both UM and renal cell carcinoma had SD for 22 cycles.
Rash was the most frequent toxicity and DLT, occurring in 74% of all patients, and precluded dose escalation greater than 300 mg bid. Of the 43 episodes of skin rash, 34 were of maximum grade 1-2, and 9 were grade 3-4. Mild to moderate diarrhea was the principal gastrointestinal toxicity (56% of patients). Mild-moderate reversible ALT and AST elevation occurred in 14%; 14% experienced SAEs, including hypoxia, pneumonitis, bradycardia, renal insufficiency, and exfoliative dermatitis. 33 Schmittel A et al., 2006 [79] 7 confirmed SDs and 1 PR were observed in 24 patients treated with the GeT regimen, whereas no PR and only 3 SDs were observed in the T arm (p = 0.08). Median PFS was 3 months (95% CI 1.1-4.9) and 2 months (95% CI 1.7-2.3) in the GeT and T arm (p = 0.008, log-rank); 6 and 12 months PFS was 34.8% and 17.9% and 16.7% and 0% always favoring the GeT arm.
Grades 3 and 4 leukopenia only occurred in the GeT arm (17%; p = 0.001). 8% experienced a febrile neutropenia. Frequencies of anemia, nausea, vomiting, and infections were not significantly different in both treatment arms. 34 Richtig E et al., 2006 [80] In 3 patients, therapy had to be withdrawn because of the appearance of metastases. Neither a univariate approach nor a multivariate approach could show a protective effect of interferon treatment on survival.
For 46% the initial dose had to be reduced due to leukopenia, thrombocytopenia, cardiac symptoms, elevated of liver function, or vertigo. In 5 patients, therapy had to be withdrawn because of serious side effects. 14 patients are evaluable for response. 4 patients completed all 6 cycles of chemotherapy. Of these, all 4 achieved SD after 3 cycles but 2 patients had progressed at re-assessment after cycle 6. The other 10 patients all had PD. Median PFS from the first cycle of chemotherapy was 12 weeks (2-26 weeks) and median OS was 30 weeks (2-64 weeks).
The treosulfan/dacarbazine combination was generally tolerated well. The major toxicities were hematological. Grade 1 or 2 thrombocytopenia was seen in 8 patients and grade 3 thrombocytopenia was seen in 3 patients. Grade 4 thrombocytopenia was also seen in 1 patient and Grade 4 neutropenia occurred in 1 patient. Non-hematological toxicity was generally mild with 2 patients experiencing grade 3 nausea and vomiting while 1 patient experienced grade 3 lethargy. 36 Schmittel A et al., 2005a [82] In cohort 1 with a treosulfan dose of < or = 3000 mg/m2, no objective response was observed. Of the patients treated with > or = 3500 mg/m2 in cohort 2, 1 had PR (5%), 10 showed SD and 8 PD. An increased survival was observed in the second cohort with higher treosulfan doses, with median survival times of 6.0 versus 9.0 months (p = 0.03) in cohort 1 and 2, respectively, and a 1-year survival of 7.1% versus 47.3%, respectively.
Grade 3 and 4 leukopenia was observed in 9/19 patients. Grade 3 and 4 thrombocytopenia and leukopenia occurred in 8 and 9 patients, respectively. Grade 3 nausea, vomiting, and mucositis occurred in 1 patient each. All 9 evaluable patients had progressive disease. 1 patient with progression experienced clinically significant symptom relief and therefore received 6 cycles. 2 patients died after the first cycle because of progressive disease.
Grade 3 and 4 toxicity consisted of anemia, thrombocytopenia, and leukocytopenia in 2, 1, and 2 patients, respectively. 3 patients showed grade 2 nausea, 1 patient grade 2 diarrhea and 1 patient grade 2 to 3 drug fever. For cases treated at dose levels 1/2, no objective responses were observed, whereas 2 patients (UM, renal cancer) on dose level 3 and 1 patient on dose level 4 (ovarian cancer) had a PR. Furthermore, we observed a stabilization of disease for more than 3 months in 15 patients with UM. A significant trend for improved OS with higher treosulfan doses was recorded.
Chemotherapy was generally well tolerated. Acute toxicity consisted of mild nausea in 9 patients. The predominant delayed toxicity was myelotoxicity. Grade 3 or 4 thrombocytopenia was observed in 3 and 1 patient, respectively. On dose levels 3 and 4, 1 and 2 patients, respectively, developed thrombocytopenia requiring a dose reduction. Grade 3 leukopenia was observed in 1 patient on dose level 3. Non-hematological side effects > grade 2 were alopecia and neutropenic fever. 41 Terheyden P et al., 2014 [87] No patient achieved an objective response, 25% of patients (95% confidence interval, 8.6-49.1%) had stabilization of disease. The median time to progression for the patients achieving a SD was 187 days (range 182-316 days), with a prolonged median OS of 17 months compared with 7 months for the patients with PD. The median OS was worst in patients receiving treosulfan/gemcitabine as first-line therapy, i.e., 206.5 days (range 25-491 days).
The combination therapy of treosulfan/gemcitabine was well tolerated with no common toxicity criteria grade 2-4 non-hematological AEs.
The drugs were well tolerated. The most common side effects were leukocytopenia and thrombocytopenia. 3 patients were withdrawn because of toxicity (thrombocytopenia grade 4). Grade 4 neutropenia occurred in only 1 patient. 43 Bedikian AY et al., 2003 [89] No CR or PR were observed. SD was achieved in two patients. The median survival of the group was 6.7 months, with a range of 1-12.7 months. The median TTP was 1.84 months, with a range of 0.7-3.8 months.
Hematological toxicity was moderate; 3 patients developed grade 4 neutropenia, and 2 of these also developed grade 4 thrombocytopenia; 1 patient had grade 4 thrombocytopenia. Gastrointestinal side effects were the most common non-hematological toxic effects.
The most prominent side effect due to fotemustine was thrombocytopenia but never exceeded grade 3. A more prominent systemic toxicity of IV infusion was the more common occurrence of leukocytopenia. Thrombocytopenia was observed in 12 patients within the IV group in contrast to 4 patients in the HIA group (p = 0.028); 2 patients receiving intra-arterial fotemustine developed gastroenteric complications. 47 Ellerhorst JA et al., 2002 [93] No CR or PR were observed. SD was achieved in 4 individuals (15%) for durations of 3, 4, 6, and 8 months; 2 of these patients had UM and 2 had cutaneous primaries. Disease progressed in spite of treatment in 22 individuals (85%).
17.9% developed grade 4 neutropenia and 7.1% grade 4 thrombocytopenia. 43% experienced grade 3 or 4 diarrhea and 18% grade 3 or 4 vomiting. Dehydration secondary to gastrointestinal toxicity lead to 4 hospitalizations. Myalgia and fatigue were also common, but were usually described as mild to moderate in intensity. 48 Mertens WC et al., 1996 [94] 1 PR was achieved for an ORR of 6% (95% CI 0-29) in a patient with UM metastatic to the liver, after 4 months of therapy, and lasted a further 8 months; 7 patients had SD (range 2-8 months), and lasting 3 months; 9 patients were found to have PD 6 weeks after initiation of treatment.
10 patients were able to escalate to 75 mg 3 times daily, but, of those, 3 required dose reductions because of toxicity. Of the 7 other patients, 5 could not escalate to a higher dose; 2 other patients required decrease in dosage of indomethacin. ADRs = adverse drug reactions; AEs = adverse events; CP = carboplatinum/paclitaxel; CR = complete response; DCR = disease control rate; DoR = duration of response; DTL = dose-limiting toxicity; GeT = gemcitabine plus treosulfan; IrAEs = immune-related AEs; MFS = metastasis-free survival; MTD = maximum tolerated dose; ORR = overall response rate; OS = overall survival; QD = once daily; PFS = progression free survival; PD = progressive disease; PR = partial response; RR = response rate; SAEs = serious adverse events; SD = stable disease; T = treosulfan alone; TEAEs = treatment emergent adverse events; TTP = time to progression; UM = uveal melanoma.
Among the 13 studies about the local treatment of the primary tumor or metastases from UM, we found 4 clinical trials-3 phase II studies and 1 phase III study. One of these trials assessed the local tumor control of the intravitreal administration of ranibizumab [33], but all patients required enucleation, and the study was terminated early due to the lack of therapeutic advantages.
Olofsson R et al. [23] observed an overall radiological response of 68% in 34 patients with liver metastasis from UM treated with isolated hepatic perfusion (IHP) within a phase II trial. The time to local progression was 7 months and the median overall survival (OS) 24 months, with a significant survival advantage compared to the control group (National Patient Register; p = 0.029). All patients enrolled in the phase II study of Fiorentini G et al. [28] obtained an objective response of liver metastasis treated with hepatic transarterial chemoembolization (TACE) adopting irinotecan-loaded microspheres. Finally, the prospective, randomized, phase III trial of Leyvraz S et al. [24] compared the IV or intraarterial hepatic (I.a.H.) fotemustine administration in 171 patients with liver metastases from UM followed for a median of 5 years. I.a.H. did not improve OS (median 14.6 months) compared to IV administration (median 13.8 months; p = 0.59). However, there was a benefit on progression free survival (PFS) for HIA (median PFS of 4.5 versus 3.5 months, respectively; 1-year PFS rate 24% versus 8%), and on response rate (10.5% versus 2.4%).
Overall, we found a low response rate and limited advantage in terms of survival. The study by Lane AM et al. [62] did not demonstrate an advantage of adjuvant interferon treatment in terms of melanoma-related mortality compared to historical controls in 121 patients with choroidal or ciliary body melanoma during a long-term follow-up (approximately 9 years). However, Binkley E et al. [38] recently reported a survival benefit of adjuvant therapy based on sequential low-dose dacarbazine and interferon-alpha in 33 patients with high-risk UM (5-year median OS of 66% (45-80, median not observed) in treated patients and 37% (19-55, median 54 months) in control).
The trials that assessed the role of immunotherapy in patients with unresectable or metastatic UM showed a low rate of response, a median OS of 1 year with nivolumab and pembrolizumab [35,39], and 6 months with ipilimumab [51]. One patient treated with pembrolizumab reached a complete response (CR, ongoing at 25.5 months), but the drug was stopped after the first dose due to the onset of a severe form of diabetes (grade 4).
The only randomized, double blind, placebo-controlled phase III trial enrolled 129 patients with metastatic UM to receive either selumetinib or placebo plus dacarbazine [42]. The primary endpoint of PFS advantage was not met (median PFS was 2.8 months in the selumetinib + dacarbazine and 1.8 months in the placebo + dacarbazine group).
Finally, we found more than 140 interventional clinical trials (49 ongoing, 79 completed, 1 suspended, 4 withdrawn, and 9 with unknown status) on uveal melanoma listed on the clinicaltrials.gov database. Among the ongoing trials, 11 foresee the enrollment of patients with local disease and 38 patients with metastatic or unresectable uveal melanoma (Table 7a,b).

Discussion
To date, no drugs have been specifically approved for the treatment of non-metastatic uveal melanoma.
Pharmacological treatments result ineffective, likely due to the incapacity to reach enough concentration into the tumor area in the eye, as result of the characteristics of the posterior segment and the blood-retinal barrier [96]. It is possible that local delivery at the ocular site will obtain better results, in terms of efficacy and safety. Therefore, researchers are developing new drug delivery systems for uveal melanoma and other ophthalmological diseases, thanks in part to nanotechnology [97][98][99].
However, the possibility of using effective drug delivery still represents a big challenge, and further studies are needed to establish whether this new technology could help in the fight against uveal melanoma [100][101][102][103].
Despite advances in diagnosis and local treatment, the overall survival (OS) of patients with uveal melanoma remains poor because of the progression into metastatic disease. Indeed, up to 50% of cases develop metastasis, especially in the liver, at approximately 5 years after treatment of the primary tumor [104][105][106]. This time is shorter in patients with larger neoplasm, especially in those with a higher grade of malignancy [107,108].
Metastatic disease is particularly difficult to treat; available systemic therapy rarely produces durable responses or significant survival benefits. Actually, the reported median survival after detection of metastatic disease is less than 1 year [109].
Moreover, no adjuvant therapy, which may be more active in treating microscopic metastatic tumor, was shown to reduce the risk of disease spread or survival improvement, and would need further studies [104].
The most updated clinical practice guidelines [111,112] recommend the enrollment of patients with metastatic disease in clinical trials, if possible. Otherwise, systemic therapies used to treat cutaneous melanoma can be considered, although no regimens demonstrated improved overall survival in uveal disease.
Immunotherapy has dramatically improved outcomes for patients with advanced cutaneous melanoma, but this clinical benefit has not been observed in metastatic uveal melanoma, probably due to a low mutational burden and low PD-L1 expression [116][117][118].
The randomized phase III trial that led to the approval of ipilimumab did not include patients with uveal melanoma, and subsequent smaller studies found a low response rate (0-5%) and an OS of less than 10 months [64,[119][120][121][122].
Even the phase III CheckMate-067 trial, comparing the concomitant use of nivolumab plus ipilimumab versus the monotherapy alone, excluded uveal melanoma patients [123].
A large series of patients with metastatic uveal melanoma treated with PD-1 and PD-L1 antibodies (pembrolizumab N = 38; nivolumab N = 16; atezolizumab N = 2) showed a partial response rate of 3.6%, a median progression free survival (PFS) of 2.8 months, and an OS of 7.6 months [124].
Recently, the results of a single arm phase II trial demonstrated an overall response rate (ORR) of 18%, a median PFS of 5.5 months, and a median OS of 19.1 months in 33 patients with metastatic uveal melanoma treated with nivolumab plus ipilimumab [125]. Considering these results, the usefulness of immunotherapy in uveal melanoma requires additional investigation and many clinical trials are currently ongoing.
A novel bispecific molecule targeting T-cells (tebentafusp, IMCgp100) showed clinical benefit in patients with metastatic uveal melanoma in phase II, and recently, a phase III study [126].
The mechanism of action consists in the redirection of T cells to target the gp100 protein, highly expressed in melanocytes and melanoma cells. The phase III trial assessed OS as the primary endpoint in 378 naïve patients with metastatic uveal melanoma, randomized 2:1 to receive tebentafusp or the investigator's choice among dacarbazine, ipilimumab, or pembrolizumab.
OS was statistically significantly improved in patients randomized in the experimental group compared to the control group in the first pre-planned interim analysis (OS hazard ratio of 0.51, and estimated 1-year OS rate of 73% for the study drug versus 58% with the investigator's choice) [127,128]. These data confirm the positive survival benefit of the phase II clinical trial, and might likely support the use of this drug as a potential new treatment for cancer patients with this highly unmet need. Moreover, the drug was granted the fast-track and orphan drug designation by the FDA for uveal melanoma [129,130] and Promising Innovative Medicine designation under the UK Early Access to Medicines Scheme.
In regard to target therapies, BRAF and KIT inhibitors are not included among treatment options, as uveal melanomas usually lack BRAF and KIT mutations. Conversely, the typical mutations in GNAQ and GNA11 genes lead to constitutive activation of the MAPK and PI3K/Akt pathways and therapies that target downstream effectors, such as MEK, Akt, and protein kinase C (PKC) are under investigation, even with disappointing results so far [113].
For example, selumetinib, a potent and highly selective inhibitor of MEK, associated with dacarbazine, showed no significant improvement in terms of PFS compared to dacarbazine alone (2.8 versus 1.8 months, p = 0.32) in the phase III SUMIT trial [55]. Similarly, there was no significant difference in ORR (3.1 versus 0%, p = 0.36).
According to the underlying molecular mechanisms, target therapies could probably be improved by combinatory strategies [131].
To date, several clinical trials are ongoing to find new therapeutic options, mainly for those with metastatic disease [110]. Many interventions are still in the preliminary phases of clinical development, being investigated in phase I trial or phase I/II (Table 7).
Additionally, the possibility to exploit a possible ocular pharmaceutical RNA-based treatment against differentially expressed miRNAs in different ocular diseases [132,133], including UM, together with the success of these trials, could be crucial for changing the prognosis of patients with advanced/metastatic UM.

Conclusions
This systematic review shows the lack of well-designed randomized clinical trials so far and confirms the limited advantages, in terms of response and survival of treatment options for UM. Despite the progress in the development of new effective therapeutic strategies, to date, all treatments for UM are still unsatisfactory and patients have a poor long-term prognosis. The future success of ongoing trials could hopefully change the outcome of patients with advanced/metastatic UM.

Conflicts of Interest:
The authors declare no conflict of interest.