FormalPara Key Points

Overview of the genetic background of uveal melanoma.

Summary of the current treatment landscape for metastatic disease.

Insights into potential innovations in therapeutic approaches and ongoing developments.

1 Introduction and Genetic Background of Uveal Melanoma

Uveal melanoma (UM) is the most common malignant tumor of the eye in adults, but it is an orphan tumor condition with a mean age-adjusted incidence of 5.2 per million in the Caucasian population [1]. For large tumors, enucleation is still recommended, but in recent years, significant advances have been made in the local treatment of the primary tumor. Preserving the eye with some of its remaining visual function can be achieved for smaller tumors through methods such as brachytherapy/plaque radiotherapy, stereotactic radiotherapy, proton beam therapy, or endoresection [2]. However, these improvements in the local treatment of the primary tumor have not affected the occurrence of distant metastases or overall survival (OS). Depending on the genetic background of the primary tumor, at least 50% of patients develop metastases, primarily in the liver [3, 4]. Once metastases develop, the survival rate decreases rapidly from 52% at 1 year to 25% at 2 years, and 13% at 3 years [5, 6].

UM is a distinct tumor entity and should be strictly distinguished from cutaneous melanoma (CM). According to The Cancer Genome Atlas (TCGA), CM can be categorized into four distinct genomic subtypes based on specific genetic alterations of the following oncogenes: BRAF, RAS (N/H/K), NF1, and wild-type (wt) for the three genes (referred to as “triple” wt) [7]. Interestingly, except for the triple wt subtype, more than 90% of CM showed a clear signature of exposure to ultraviolet (UV) radiation. This suggests that UV radiation plays a significant role in the development of these subtypes [7].

In contrast to CM, UM tumors show a distinct mutational spectrum, a very low mutational burden, with a rate of 0.46 mutations per megabase, an average of 32 coding mutations per tumor, and no UV radiation mutational signature [8, 9]. Robertson et al. classified UM into four distinct groups based on TCGA data: disomy 3 accompanied by EIF1AX mutations with a favorable prognosis, disomy 3 and SF3B1 mutation with intermediate prognosis, and monosomy 3 and BAP1 mutation with poor prognosis, which can be further divided into two subsets, each exhibiting distinct genomic aberrations and transcriptional features [10].

Monosomy 3 is associated with poor disease-free survival (DFS) and OS, while tumors with disomy 3 tend to develop limited metastasis and have extended DFS [11, 12]. The most frequent somatic mutations observed in UM are in GNAQ and GNA11, leading to the continuous activation of the mitogen-activated protein kinase (MAPK) pathway, which contributes to UM development [13]. However, these mutations do not seem to impact OS significantly [14]. Furthermore, hemizygous mutations in the BAP1 gene have been identified in monosomy 3 tumors, resulting in either loss or dysfunction of BAP1 expression [10, 15, 16]. A further study also showed a strong association between BAP1-mutated tumors and a peak in clinically detected metastases occurring 2 years after enucleation [17]. Additionally, somatic mutations in SF3B1 and EIF1AX are identified in tumors with disomy 3 and are associated with a more favorable disease course [10, 18, 19]. SF3B1 mutations are linked to a prolonged DFS and a delayed peak of metastases occurring after 7 years [15, 17]. EIF1AX mutations, as well as wt BAP1, SF3B1, and EIF1AX genes, are associated with favorable survival [10, 15]. Consequently, the specific genetic background of the tumor (genotype) plays a critical role in determining the overall prognosis, irrespective of treatment options. To date, the most straightforward and prognostically relevant differentiation in UM remains the presence or absence of monosomy 3.

There is evidence that a high mutational burden can be predictive of a positive response to immune checkpoint blockade (ICB) [20]. In metastatic UM, a hypermutator phenotype related to MBD4 has shown exceptional sensitivity to anti-programmed cell death protein 1 (anti-PD-1) therapy, which is present in up to 2% of UM patients only [21]. Furthermore, it has been discussed that tumors with SF3B1 mutations demonstrate increased sensitivity to ICB due to the generation of neoepitopes derived from splicing [19, 22]. These findings indicate a potential significance of genetic and mutational characteristics in guiding treatment decisions and predicting responses in UM.

However, the low mutational burden makes UM capable of evading the immune system and leads to a rapid progression to death. Unlike metastatic CM, effective therapies are very limited. As a result, median survival dramatically decreases when metastases occur and adjuvant therapies to prevent tumor spread do not exist.

2 Systemic Therapy for Metastatic Uveal Melanoma

2.1 Tebentafusp and Immune Checkpoint Blockade

The only approved therapy for metastatic UM from the Food and Drug Administration (FDA) and the European Medicines Agency (EMA) is currently tebentafusp (tebe) [23,24,25]. Tebe belongs to a novel class of anti-cancer drugs known as immune-mobilizing monoclonal T cell receptors against cancer (ImmTACs) designed for the treatment of HLA-A*02:01-positive patients with metastatic or unresectable UM. It is a bispecific molecule that combines an affinity-enhanced soluble T cell receptor targeting the peptide sequence “YLEPGPVTA” of the glycoprotein gp100 presented by major histocompatibility complex (MHC) class I molecules, along with an anti-CD3-specific single-chain variable fragment [23]. Consequently, tebe recruits T cells to gp100-expressing UM cells, leading to a polyfunctional T cell-mediated immune response, including tumor-specific cell killing and cytokine release [23, 26].

In the pivotal trial, tebe demonstrated, after a 3-year follow-up, a median OS of 21.6 months compared to 16.9 months in the control group, which received an investigator's choice of pembrolizumab, ipilimumab, or dacarbazine (hazard ratio [HR] 0.68, 95% confidence interval [CI] 0.54–0.87) [27]. The survival rates for patients treated with tebe at 1, 2, and 3 years were 72%, 45%, and 27%, respectively, while in the control group, the survival rates were 60%, 30%, and 18%, respectively. A phase II study involving 127 patients with treatment-refractory metastatic UM, treated with tebe, showed a 1-year survival rate of 62% (95% CI 53–70), with a median OS of 16.8 months (95% CI 12.9–21.3), indicating potential benefit in later lines of therapy [28]. Further, the data indicated a correlation between early reduction of circulating tumor DNA (ctDNA) and OS [28]. However, the observed response showed significant heterogeneity, both among patients and among metastases within individual patients, with no identified or obvious reasons for this heterogeneity. Additionally, this therapy is accessible to only approximately 45–50% of patients due to the HLA restriction.

Tebe was compared to monotherapy of pembrolizumab, ipilimumab, or dacarbazine, but not to double checkpoint blockade (DCB) with anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA-4) antibody and anti-PD-1 antibody [27]. Hence, Petzold et al. conducted a comprehensive meta-analysis of available systemic treatments regarding OS and progression-free survival (PFS) with a focus on the comparison of tebe and DCB [29]. This study provided robust evidence that tebe is the most beneficial therapy option for metastatic UM in terms of OS. Population-adjusted models, such as matching-adjusted indirect comparison (MAIC) (HR 0.465, 95% CI 0.276–0.781) and simulated treatment comparison (STC) (HR 0.420, 95% CI 0.272–0.647), as well as the unadjusted model (HR 0.641, 95% CI 0.449–0.915), revealed significantly better survival for tebe-treated patients compared to those receiving DCB. Comparing the pooled survival curves in numbers, tebe clearly outperformed other treatments: the median OS was 22.4 months, while DCB revealed a median OS of 15.7 months. The other treatment groups performed less favorably, with median OS ranging from 7.7 months (anti-CTLA-4 monotherapy) to 10.9 months (anti-PD-[l]-1 monotherapy). Additionally, an analysis adjusted for propensity scores comparing DCB with tebe also revealed a survival advantage for the latter (HR 0.52, 95% CI 0.35–0.78) [30]. However, it is important to highlight that this examination was limited to just one prospective trial of DCB, which showed a comparably poor OS.

Comparing the results based on unprocessed data from two published prospective single-arm trials investigating DCB with those from the pivotal trial of tebe (Table 1), the median OS was 12.7 and 19.1 months, respectively, for DCB and 21.6 months for tebe [27, 31,32,33]. The 1-year survival rates of 51.9% and 56%, respectively, for DCB contrast most strikingly with the rate of 73% for tebe. On the other hand, the objective response rate (ORR) for tebe was 11% compared to 11.5% and 18.2% for DCB, respectively. Given these data regarding tebe therapy, it is not feasible to make predictions about OS based on the radiologically assessed ORR.

Table 1 Overview of prospective studies on immune checkpoint blockade and tebentafusp
Full size table

There is a considerable difference in OS between the two prospective trials involving DCB. Piulats et al. exclusively enrolled treatment-naïve patients and reported a notably lower OS (12.7 vs. 19.1 months), PFS (3 vs. 5.5 months), and ORR (11.5% vs. 18.2%) in comparison to Pelster et al., who included patients with any number of prior treatments [31, 32]. Furthermore, Piulats et al. enrolled a higher proportion of patients with Eastern Cooperative Oncology Group (ECOG) performance status 0 (84.6% vs. 71%) and a lower percentage of patients with elevated lactate dehydrogenase (LDH) levels (32% vs. 43%). Paradoxically, this resulted in unfavorable outcome parameters (OS, PFS, ORR) within a population that is considered to have a more favorable prognosis based on these prognostic parameters [34, 35]. Regarding the metastatic patterns observed in both studies, the survival benefit might be attributed to the fact that Pelster et al. included a greater number of patients with extrahepatic metastases (69% vs. 58%) and a lower number with hepatic metastases solely (31% vs. 42%) [31, 32].

A panel of retrospective analyses from our group provides evidence that hepatic metastasis only represents a negative prognostic factor for OS, while extrahepatic disease serves as a prognostic factor for improved OS [35, 36]. We also demonstrated that exclusive hepatic metastasis was associated with a lower response to ICB therapy. In a separate exploration, we could show that the best response to ICB, particularly a partial response (PR) based on Response Evaluation Criteria in Solid Tumors (RECIST), serves as a strong prognostic factor (p < 0.001) for extended OS [35]. This observation was irrespective of whether the patients underwent PD-1 inhibitor monotherapy or DCB. Our findings, along with those of others, highlight that DCB yields an improved ORR compared to ICB monotherapy [37, 38]. Additionally, we could demonstrate in another study that re-induction of ICB after primary resistance or development of toxicities may yield a clinical benefit for a small subgroup of patients with advanced UM when no other treatment options are left [38]. In a further retrospective multicenter study, patients treated with tebe following ICB showed a trend for a prolonged OS compared to patients treated with second-line ICB after tebe therapy (28 vs. 24 months, p = 0.257) [42].

In addition to efficacy, the toxicity of immunotherapeutic approaches is an important issue that could limit their use. During treatment with tebe, grade 3 or 4 treatment-related adverse events (AEs) occurred in 47% of patients [27]. The most prevalent AE is cytokine release syndrome (89%), and the most probable cytokine-mediated AEs are pyrexia (76%), chills (47%), nausea (43%), and hypotension (38%) [33, 39]. Further frequent AEs include skin-related AEs (summarized as rash, 83%), pruritus (70%), and fatigue (41%) [33]. However, most AEs typically occur within the initial 4 weeks of treatment during the step-up dosing regimen, which commences with a 20-mcg dose, followed by an increase to 30 mcg, and culminates in the third week with the recommended phase 2 dose of 68 mcg [40]. AEs show a decreased frequency and severity with subsequent doses. In 2% of cases, treatment must be discontinued due to AEs.

DCB carries a higher risk of severe immune-mediated AEs due to its broad activation of the immune system, resulting in grade 3 or 4 treatment-related AEs in up to 57.7% in UM [31, 32]. These AEs have the potential to affect nearly any organ system within the body, including the gastrointestinal, respiratory, cardiovascular, musculoskeletal, nervous, hepatic, renal, cutaneous, and hematological systems. While most of these AEs can be completely resolved with a prompt clinical diagnosis and early intervention, it is worth noting that immune-related endocrine toxicities are in about 50% of cases irreversible [41]. Overall, the therapeutic benefit of DCB in metastatic UM is predominantly unfavorable, especially in view of the relatively common, partially irreversible severe AEs.

2.2 Liver-Directed Therapies

Given the frequent occurrence of hepatic metastases, the liver's immunosuppressive microenvironment, and the lack of effective systemic treatments, liver-directed therapies (LDTs) are frequently utilized in advanced UM [42].

Well circumscribed and solitary lesions of the liver can be treated interventionally with radiofrequency or microwave ablation or surgically resected [43, 44]. Another option for larger solitary lesions (> 3–4 cm) is embolization procedures such as the transarterial chemoembolization (TACE). Nevertheless, the lack of comparative clinical trials and standardized treatment protocols pose challenges [45]. In cases of multiple liver metastases, as well as depending on the distribution pattern and location-specific availability of therapeutic options, selective internal radiotherapy (SIRT), isolated hepatic perfusion (IHP), and percutaneous hepatic perfusion (PHP, also chemosaturation) are typically employed [42].

In a phase II trial, SIRT exhibited a median OS of 18.5 months and a PFS of 8.1 months. Additionally, a retrospective analysis involving 71 patients revealed a median OS of 12.3 months and a PFS of 5.9 months [46]. A systematic review including 11 studies (9/11 retrospective) and 268 patients reported an OS of 12.3 months and mostly low-grade (Common Terminology Criteria for Adverse Events [CTCAE] 1–2) AEs in 90.9% [47]. However, two SIRT-related deaths occurred, due to liver failure [47].

PHP with melphalan is a regional therapy procedure in which the liver is isolated from the systemic circulation to enable hepatic perfusion via a double-balloon catheter. Subsequently, venous blood is aspirated and purified of melphalan by an extracorporeal filtration system, thereby mitigating systemic toxicity [48,49,50,51]. Hughes et al. demonstrated a significantly improved hepatic response and PFS compared to best alternative care (BAC) in a randomized phase III study including 93 patients [52]. However, in this study, with about 40% of patients suffering from extrahepatic metastases, no survival benefit could be shown (PHP 10.6 vs. BAC 10 months). Furthermore, 9% (4/44) treatment-related deaths occurred in the PHP group, with most of them being related to bone marrow suppression [52]. To reduce hematological toxicity, an improved filter system (second generation [GEN2]) was established. In a prospective phase II study of 35 patients with liver metastases only, an OS of 19.1 months was reported, and there were no treatment-related deaths [53, 54]. Hematological grade 3/4 AEs were observed in the majority of patients, but the authors reported an acceptable safety and toxicity profile [53,54,55]. In another retrospective analysis, by Tong et al., of 101 patients with metastatic UM undergoing PHP treatment, a significant survival benefit was evident in patients with a complete response (CR)/PR or stable disease (SD) compared to patients with progressive disease (PD, 27 vs. 21 vs. 8 months, respectively) [56]. The authors discuss that the best candidates for PHP treatment are patients with a good non-cancer-related health status, no cardiovascular disease, early-stage primary UM with hepatic metastases only, and low LDH levels [56]. In another phase III trial (FOCUS) comparing PHP with BAC (investigator’s choice of TACE, pembrolizumab, ipilimumab, or dacarbazine) including 144 patients (102 PHP vs. 42 BAC), PHP demonstrated a significant survival benefit (20.53 vs. 14.06 months, respectively) [57]. Systemic therapies such as tebe and DCB were not included. However, patients are still being followed for survival, and evaluations of subsequent systemic therapies have not been reported yet [57]. In a German, two-center, retrospective analysis, PHP demonstrated a median hepatic PFS of 12.4 months and a median OS of 18.4 months [58].

IHP includes surgical isolation of the liver and the connection of arterial and venous catheters to a heart-lung machine. Following this, the liver undergoes perfusion with melphalan for 60 min [59]. In a randomized, open-label, multicenter, phase III trial (the SCANDIUM trial), patients with previously untreated hepatic metastases showed a significant improvement of PFS with a duration of 7.4 months compared to 3.3 months in the control group receiving chemotherapy (49%), ICB (39%), or other LDT (9%) (p > 0.001) [60]. Severe AEs were reported in 19.5% (8/41) and treatment-related death in one case. More recently, follow-up data were published reporting an OS of 21.7 months in the IHC group and 17.6 months in the control, with an HR of 0.64 (95% CI 0.37–1.10) [61].

A meta-analysis comparing IHP and PHP showed a median OS for IHP of 17.1 months and for PHP of 17.3 months [62]. The median PFS was 7.2 and 9.6 months, and the median hepatic PFS was 10 and 9.5 months, respectively. However, PHP holds an advantage in its repeatability [62].

In a single-center study comparing PHP with SIRT, the median PFS was 4.25 months for SIRT and 13.6 months for PHP (p = 0.090) [63]. Additionally, a survival advantage was observed, with a median OS of 10 months for SIRT and 17.2 months for PHP (p = 0.006) [63].

Combining ICB with SIRT has shown the potential to enhance therapy responses [42, 64, 65]. The ORR to any ICB was significantly higher in the cohort receiving ICB plus LDT compared to the cohort receiving ICB only (16.7% vs. 3.8%, p = 0.0073). Additionally, the median OS was extended in the cohort undergoing combination therapy (20.1 vs. 13.8 months; p = 0.0016) [42]. A further retrospective analysis compared DCB and SIRT to SIRT only and demonstrated a significant prolonged OS in the combination group with ICB (46.6 vs. 11.8 months, p = 0.039) [65]. Another retrospective study demonstrated a notable survival advantage of the combination of LDT and ICB compared to predominantly ICB alone (22.5 vs. 11.4 months, p = 0.036) [66]. In addition, an open-label, single-center, phase Ib/randomized phase II trial on the safety and efficacy of the combination of PHP with DCB is currently under investigation, and first data demonstrated a good safety profile [67, 68].

However, interpretation and comparison of such studies should be performed with caution, as these studies might exhibit inter-site variability in adherence to protocols and some are retrospective analyses.

2.3 Conclusion

In summary, we recommend tebe be administered as first-line therapy in patients with a positive HLA-A*02:01 subtype (Fig. 1). The pivotal trial has indicated that patients with high tumoral burden (elevated LDH) and progressive disease (elevated bilirubin and liver enzymes) may not benefit from this treatment [33, 69]. For patients with only liver metastasis, LDT like SIRT, PHP, or IHP may be considered to reduce the tumor burden [42, 53, 60, 63]. LDT can also be administered concomitantly with or rapidly sequenced to tebe, but studies investigating the potential benefits of this combination are not evident yet. However, “ideal” patients for tebe are those with a good general condition, presenting a low ECOG performance status (0 or 1) and a low tumor burden with normal levels of LDH. Commencing treatment at the earliest opportunity seems to be crucial, as these patients exhibit the greatest benefit in terms of OS [33]. Despite the clinical benefit of OS, the ORR for tebe remains very low and the ideal treatment duration is oftentimes unclear, with a lack of data and valid biomarkers guiding when to cease or change treatment. Thus, therapy may be continued beyond modest progression. In cases of rapid tumor spread and growth, discontinuation of tebe is recommended and alternative therapy options should be deliberated within an interdisciplinary tumor board.

Fig. 1
figure 1

The figure presents the author's proposed decision flowchart for managing metastatic uveal melanoma. DCB Double checkpoint blockade, ICB mono Immune checkpoint blockade monotherapy. 

Full size image

For patients with negative HLA-A*02:01 typing or tebe resistance, the evaluation of clinical trial options is essential. Additionally, ICB represents currently the only alternative available systemic therapy with potentially sustainable clinical benefits. Despite the absence of sequential studies (under investigation NCT05549297), it is reasonable, from a tumor immunological perspective, to administer ICB sequentially to stimulate the T cells activated by tebe. A solid body of evidence indicates that DCB is a more effective therapeutic option than ICB monotherapy, regardless of the higher risk of AEs. The decision to proceed with DCB should be based on the patient's fitness for potential toxicity, necessitating thorough discussion and shared decision-making. Furthermore, depending on local availability, the evaluation of a combination therapy with ICB and LDT is recommended.

3 Approaches in Drug Development

The high prevalence of GNAQ/GNA11 mutations in metastatic UM designates the MAPK pathway as a suitable target for new therapeutic approaches. Inhibition of the protein kinase C (PKC) pathway downregulates the MAPK-activated pathway by GNAQ/GNA11 mutations [70, 71]. The oral PKC inhibitor darovasertib (LXS196) is the first of its class to inhibit both novel and conventional isoforms of protein kinase C (α, β, δ, ϵ, η, θ). In a recently published phase I study evaluating darovasertib in 68 patients with metastatic UM, a clinical response (CR/PR) was achieved in six of 66 patients (9.1%), and SD was observed in 45 of 66 patients (68.2%) [72]. Another drug combination pairs darovasertib with the c-MET inhibitor crizotinib and showed a confirmed PR in 30% and tumor shrinkage in 92% in both first-line and pretreated metastatic UM [73]. Notably, the activity was independent of the HLA status of patients and the study included patients with a comparable large tumor burden and elevated LDH. The safety profile was manageable, with predominantly grade 1–2 AEs (30%) and only 9% severe AEs.

Another interesting target is phosphoinositide 3-kinase delta (PI3Kδ), which plays a role in leukocyte activation and cancer proliferation [74]. In a phase I study, the highly selective oral allosteric modulator of PI3Kδ (roginolisib, IOA-244) demonstrated favorable tolerability and some efficacy [75]. The trial involved patients with solid tumors, including UM (9/16; 56%). These patients showed a PR of 5% and SD in 80%, indicating potential anti-tumor activity. Specifically, roginolisib induced phenotypic changes that promoted the infiltration of CD8+ and natural killer cells, while concurrently reducing the presence of suppressive immune cells [75]. As stated by the manufacturers, roginolisib can sensitize solid tumors to anti-PD-1 therapy through its immune-modulatory properties.

A different experimental approach is ImmTAC therapy with IMC-F106C, a bispecific molecule (same class as tebe) that combines an affinity-enhanced soluble T-cell receptor (TCR) targeting a peptide sequence of the preferentially expressed antigen in melanoma (PRAME) protein presented by MHC class I molecules, along with an anti-CD3-specific single-chain variable fragment [76, 77]. Preliminary data have shown promising results in terms of T cell activation and tolerability. Treatment-related AEs have been manageable, with none leading to treatment discontinuation or death. The efficacy of IMC-F106C extends to diverse tumor types, including UM, showcasing durable responses that have persisted for up to 9 months. Specifically, in UM, a PR was observed in 50% (3/6) of tebe-naïve patients and SD in the remaining 50% (3/6). Furthermore, in tebe-pretreated patients, SD was observed in 100% (5/5). Across all tumor types, 90% of evaluable patients have shown a reduction in ctDNA, with early reductions being associated with clinical benefits [77]. However, the encouraging early findings of IMC-F106C need to be confirmed in further studies involving larger patient cohorts and survival analyses.

Further clinical trials investigating tumor-infiltrating lymphocytes (TILs) (NCT00338377), RNA vaccines (NCT04455620), G protein-coupled estrogen receptor (GPER) agonists (LNS8801, NCT04130516), genetically modified herpes simplex virus (RP2, NCT04336241), vascular endothelial growth factor (VEGF) inhibitors (lenvatinib, NCT05282901), lymphocyte activation gene 3 (LAG-3) inhibitors (relatlimab, NCT04552223), and further PKC inhibitors (DYP688, NCT05415072) are still ongoing, and the potential of these approaches remains uncertain.

4 Conclusion

Tebe is the first agent to demonstrate a survival advantage in metastatic UM through a clinically controlled trial and also represents a novel anti-cancer drug class. Despite the evident survival benefit, the restriction to HLA-A*02:01-positive patients and the low response rate highlight the urgent need for further therapies. Nevertheless, there are promising approaches in drug development, such as the combination of PKC and c-MET inhibitors, which raise hope for additional approvals in metastatic UM.