A n O v e r v i e w o f L i v e r D i r e c t e d L o c o r e g i o n a l T h e r a p i e s

An overview of all liver-directed locoregional therapies, including surgical resection for melanoma liver metastases (MLMs), is provided. MLM patients are divided by their primary melanoma location; cutaneous, uvea (eye), and mucosal melanoma. If patients with isolated cutaneous MLMs are considered for surgical resection, treatment with systemic therapy should be part of the treatment course. For uveal MLMs, complete surgical or ablative treatment of all MLMs suggests superior results compared with other liver-directed or systemic therapies, based on current evidence, no recommendations for any liver-directed regional therapy in the treatment of mucosal MLMs can be made.

metastasizes, it most often (77%) is to the locoregional skin and approximately 14% to the liver (Fig. 1B). 3 Liver metastases from cutaneous melanoma almost always are seen in combination with metastases at other sites (13%) and only rarely are confined to the liver (1% [Fig.1C]). 3The median time to metastasis of cutaneous  3 and uveal 5 melanoma (C) Degree of liver involvement in metastatic cutaneous 3 and uveal 5 melanoma.(D) Time to metastasis in cutaneous 4 and uveal 6 melanoma.The diamond shape represents the median and the line represents the corresponding interquartile range.a Including subcutaneous tissue and regional lymph nodes.
melanoma is 2 years (Fig. 1D). 4 Compared with cutaneous melanoma, uveal melanoma is rare (5.1/1,000,000/y [Fig.1A]).Uveal melanoma often metastasizes to the liver, which is involved in 89% of patients with metastatic uveal melanoma and represents the only site of metastasis in 46% (Fig. 1B, C). 5 The median time to metastasis of uveal melanoma is 3.4 years (Fig. 1D). 6If uveal melanoma is rare, mucosal melanoma is extremely rare (2.3/1,000,000/y). 2 Mucosal melanoma has a metastatic pattern more comparable to cutaneous melanoma, with the lung the most frequent metastatic site (40%). 7Metastases in the liver are seen in approximately 36% of metastatic mucosal melanoma patients, with 26% of patients presenting with liver-only metastases. 7The median time to metastasis of mucosal melanoma is 3.5 years. 8This article provides a systematic review (Fig. 2) of all liver-directed locoregional therapies, including surgical resection for melanoma liver metastases (MLMs), and briefly discusses current efficacy of systemic therapy in order to aid evidence-based clinical decision making in the surgical oncologists' and clinicians' practice.

PARTIAL LIVER RESECTION Treatment Specifics
The study details, demographics, and treatment specifics of 37 studies investigating surgical treatment of MLMs are reported in Supplementary Data [Table 1].  Bas on 12 studies that reported the complete diagnostic workup, 12,14,15,21,23,25,26,28,32,35,36,41 a total of 5859 patients were diagnosed with MLMs.Of 5859 patients, 343 patients (6%) diagnosed with MLMs eventually were treated with curative intent resection (Rx).An important reason for patients not undergoing Rx was extrahepatic disease and more extensive liver involvement discovered during surgery compared with preoperative assessments, especially miliary disease in patients with MLMs from uveal origin.Large heterogeneity existed between studies concerning preoperative and postoperative treatment strategies; systemic therapy of any Treatment of Melanoma Liver Metastases kind (SYS), hepatic arterial infusion chemotherapy (HAI), and/or transarterial chemoembolization (TACE) were administered both preoperatively and postoperatively across studies.Concurrent ablation was not often utilized (overall 9%, range 0%-31%), whereas major hepatectomy frequently was performed (overall 44%, range 0%-85%).

Outcomes
,15,25,27,28,43 A total of 947 patients treated with Rx are described, 302 (32%) with MLMs from cutaneous, 489 (52%) with MLMs from uveal, 3 (<1%) with MLMs from mucosal, and 16 (2%) with MLMs from unknown primary origin.In the remaining 137 (14%) patients treated with Rx, MLM origin was not (clearly) reported. The 7 studies reportng on debulking describe a total of 221 patients, 6 (3%) patients with MLMs from cutaneous and 187 (86%) patients with MLMs from uveal origin; MLM origin was not (clearly) reported for 28 (13%) patients.This discrepancy in MLM origin between patients treated with Rx and debulking can be attributed to debulking (with adjuvant HAI/SYS) being standard of care treatment of disseminated liver disease in 2 dedicated uveal melanoma centers, whose series together account for all 187 patients with MLMs from uveal origin treated with debulking.Complete microscopic resection of all MLMs (R0) was achieved in 83% (range 40%-100%) of patients treated with Rx.Perioperative mortality and morbidity after surgical treatment of MLMs (both Rx and debulking) were approximately 2% (range 0%-8%) and 14% (range 0%-36%), respectively.In the patients rendered free of disease after Rx, recurrent disease was observed in a majority of patients (overall 75%, range 40%-100%).Median progression-free survival (PFS) (range of medians) for Rx was 12 (range 5-20) months.Median overall survival (OS) was 26 (range 10-100) months after Rx versus 11 (range 5-18) months after debulking.All but 1 of the studies that reported a median OS of greater than 36 months after Rx were small, highly selected series of less than 20 patients.14,18,21,26,28,40,44 The study by Groeschl and colleagues, 30 which reported a median OS of 39 months in 31 patients, is a pooled multicenter retrospective analysis of 4 major liver centers, making it highly unlikely that 20 or more patients were treated in a single center.Three of the 4 studies comparing Rx to SYS/best supportive care (BSC) reported improved survival after Rx, 32,35,36 with the study that did not find a survival benefit comparing only 5 patients treated with Rx to 12 treated with SYS/BSC.13 Similarly, all studies that compared Rx to debulking with or without adjuvant HAI or SYS reported an improved survival for Rx.11,14,15,25,27,28,43 The only study that directly compared Rx (with HAI), debulking with HAI and SYS/BSC found OS for Rx (with HAI) to be significantly improved compared with the other treatments, with no difference observed between debulking with HAI and SYS/BSC.15 Four studies directly compared patients with MLMs from cutaneous and uveal origin, with 3 reporting no survival differences between cutaneous and uveal origin 17,33,38 and 1 suggesting improved survival for patients with MLMs from uveal origin.20 Lastly, 30 of the 37 studies reported long-term survivors (longer than 5 years) after Rx, suggesting possibility of cure.9,11,12,[14][15][16][17][18][19][20][23][24][25][26][28][29][30][31]33,[35][36][37][38][39][40][41][42][43][44][45] Individual patient survival data could be extracted from 14 studies for 99 patients treated with Rx for MLMs and are shown in Fig. 3.

ABLATION AND STEREOTACTIC RADIATION THERAPY
7][48][49][50] There were 5 retrospective cohort studies and 1 prospective phase Ib/II trial. 50One study investigated SRTx 49 ; the other 5 reported on ablation, all of which treated patients using radiofrequency ablation (RFA) 38,[46][47][48]50 ; 2studies also treated patients using cryoablation, 38,48 and 1 study also used microwave ablation. 38In all the retrospective series, complete treatment of all known (liver) lesions was performed using ablation, SRTx, or Rx (2 patients), 38,[46][47][48][49] whereas in the prospective phase Ib/II trial by Rozeman and colleagues 50 only 1 liver lesion was treated by RFA and patients subsequently received 4 courses of systemic ipilimumab at varying dosages (0.3 mg/kg, 3 mg/kg, or 10 mg/kg).In this trial it was hypothesized that through an abscopal effect of treating a single MLM with RFA the efficacy of systemic ipilimumab therapy could increase. A total f 134 patients treated with ablation or SRTx are described, 55 (41%) of whom had MLMs from cutaneous and 79 (59%) uveal origin.No treatment-related mortality was reported.Treatment-related morbidity was approximately 20% (range 0%-42%), with the highest morbidity rate of 42% reported in the prospective phase Ib/II trial of RFA plus systemic ipilimumab.50 Median PFS and OS were 7 (range 3-11) months and 19 (range 11-46) months, respectively.Reported survival was higher in the retrospective studies compared with the prospective phase Ib/II trial of RFA plus systemic ipilimumab.The trial correspondingly concluded that "Combining RFA with ipilimumab 3 mg/kg was well tolerated, but showed very limited clinical activity in uveal melanoma."50 Akyuz and colleagues 46 directly compared ablation (and laparoscopic Rx in 2 patients) to systemic chemotherapy and found OS to be significantly improved in the patients treated with ablation.Doussot and colleagues 38 directly compared ablation with Rx (see results in Supplementary Data [Table 2] and demonstrated comparable survival outcomes.

SELECTIVE INTERNAL RADIATION THERAPY
0][101][102][103][104][105][106][107][108][109][110] There were 12 retrospective cohort studies 69,[99][100][101][102][103][104][105][106][107]109,110 and 1 prospective phase II trial. 108 Allstudies performed SIRT with yttrium 90 (Y90)-labeled microspheres. Twostudies combined SIRT Y90 with systemic immunotherapy.107,109 In total, 287 patients treated with SIRT Y90 for MLMs are described, which were of cutaneous origin in 23 patients (8%), of uveal origin in 259 patients (90%), of mucosal origin in 3 patients (1%), and the origin was unknown for 2 (<1%).Treatment-related mortality and morbidity were 2% (range 0%-13%) and 13% (range 0%-32%), respectively.The study that reported the highest mortality rate of 13% was a small retrospective cohort study with only 8 patients, of whom 1 died.103 The response rate to SIRT Y90 was 26% (range 0%-77%).The 2 studies that reported exceptionally high response rates of 62% 101 and 77% 99 were both small retrospective series with 13 or fewer patients. Contrasingly, the largest retrospective series of 71 patients reported a response rate of 8%.104 Median PFS and OS were 5 (range 1-15) months and 11 (range 4-26) months, respectively.The longest median OS of 26 months was observed in a small retrospective series of 12 patients treated with SIRT Y90 and systemic immunotherapy.109 That same study also analyzed 12 patients who received only SIRT Y90 and reported an improved survival in the patients treated with SIRT Y90 plus systemic immunotherapy.109 Two retrospective studies directly compared SIRT Y90 to SYS 106 or BSC.105 Although both reported improved survival for SIRT Y90, the patients receiving SYS or BSC were treated in the same centers during the same period and, therefore, by default were not considered for or refused SIRT Y90.105,106 Furthermore the patients treated with SIRT Y90 or BSC all were patients with progressive disease after prior SYS.105 The prospective phase II trial compared SIRT Y90 in treatment-naı ¨ve patients (n 5 23) to SIRT Y90 in patients with progressive disease after immunoembolization.108 A difference in neither response rate nor survival was observed between the 2 groups.The trial, therefore, concluded that SIRT Y90 is safe and effective as either first-line or second-line treatment of MLMs of uveal origin.108

OTHER LOCOREGIONAL THERAPIES
2][113][114] There are no series about liver transplantation for unresectable MLMs, although an open nonrandomized clinical trial investigating liver transplantation for uveal MLMs has been registered (NCT01311466, clinicaltrials.gov)and reportedly recruited 2 patients from 2011 to 2017.

SYSTEMIC THERAPY Systemic Therapy in Metastatic Cutaneous Melanoma
Until recently, dacarbazine was the chemotherapeutic agent used most often.Large clinical trials investigating dacarbazine monotherapy reported response rates of approximately 7%, with median survival times of just 6 months. 115,1168][119] In 2010, ipilimumab, a CTLA-4 inhibitor, was proved to prolong survival in previously treated patients with metastatic cutaneous melanoma, with a response rate of 11% and a median survival of 10 months. 120Besides prolonging survival, treatment with ipilimumab also produced durable responses with long-term survivors; pooled analysis of 1861 patients treated with ipilimumab showed overall survival plateauing at 20% from 3 years on, all the way up to 10 years. 121In 2014, nivolumab and pembrolizumab, 2 anti-PD-1 antibodies, were approved for the treatment of metastatic melanoma.In a randomized phase III study of 834 advanced cutaneous melanoma patients, pembrolizumab proved superior to ipilimumab in terms of PFS and OS, with less toxicity. 122Combining CTLA-4 and PD-1 checkpoint inhibitors has proved even more effective.In a randomized phase III trial, 945 patients with unresectable or metastatic cutaneous melanoma were randomized 1:1:1 to either first-line nivolumab, nivolumab plus ipilimumab, or ipilimumab. 123,124Recent 5-year results demonstrated a striking 5-year PFS and OS for combination therapy of Ho ¨ppener et al 36% and 52%, respectively, compared with 29% and 44%, respectively, for nivolumab and 8% and 26%, respectively, for ipilimumab. 125Importantly, reported results are best in the BRAF-mutant patients, with 5-year OS rates of 60% for combination therapy, 46% for nivolumab, and 30% for ipilimumab. 125Preliminary results for the combination of pembrolizumab and ipilimumab report similar efficacy. 126Immune checkpoint inhibitors also have proved efficacy in the adjuvant setting, with adjuvant nivolumab demonstrating better performance than ipilimumab in resected stage III or stage IV cutaneous melanoma. 127Combination therapy, however, again proves superior with recent results reporting a significantly improved 2-year recurrence-free survival of 70% for adjuvant nivolumab with ipilimumab in resected stage IV cutaneous melanoma versus 42% for nivolumab monotherapy and 14% for placebo. 128In parallel, BRAF inhibitors (dabrafenib, encorafenib, and vemurafenib) and MEK inhibitors (trametinib, binimetinib, and cobimetinib) also have demonstrated considerable efficacy in BRAF-mutated patients with metastatic cutaneous melanoma, which are present in approximately 50%.1][132][133] Results may be inferior (cave selection bias) compared with combination immune checkpoint inhibitor therapy, with reported 5-year PFS and OS rates of combined BRAF and MEK inhibitors of 19% and 34%, respectively. 134

Systemic Therapy in Metastatic Uveal Melanoma
Similarly to metastatic cutaneous melanoma, dacarbazine long has been considered the standard and still is used as comparative treatment in contemporary clinical trials 135,136 Results of dacarbazine in the treatment of metastatic uveal melanoma are poor, with objective response seldom to never observed and a median OS of 9 months in clinical trial populations. 135,1366][137][138][139][140][141][142] The CTLA-4 checkpoint inhibitors ipilumumab and tremelimumab both have been studied in metastatic uveal melanoma but both did not considerably improve survival; median OS was 7 months for ipilimumab 143,144 and 13 months for tremelimumab. 145Concerning the combined treatment of nivolumab and ipilimumab, preliminary results report a median OS of 13 months, but final results from phase II studies have yet to be published. 146A recent meta-analysis has pooled the individual patient data of many of these phase II trials, including several trials investigating liver-directed therapies, in order to create benchmarks for PFS and OS for future trials. 147In total, individual patient data of 29 trials were included, resulting in a total of 912 patients. 147Median PFS and OS were 3 months and 10 months, respectively, with a 1-year OS rate of 43%.These outcomes represent the benchmark against which (future) therapies for metastatic uveal melanoma should be gauged. 147

Systemic Therapy in Metastatic Mucosal Melanoma
There is a high paucity of evidence concerning the systemic treatment of metastatic mucosal melanoma.Only retrospective cohort series and pooled analyses have been published.Reported response rates with (a combination of) dacarbazinebased chemotherapy range from 8% to 26%, with a median OS of 10 months to 12 months. 148,149In contrast to uveal melanoma, there is evidence to suggest efficacy of immune checkpoint inhibitors in metastatic mucosal melanoma, with a reported response rate of 23% and a median PFS of 3 months for nivolumab monotherapy and 37% and 6 months, respectively, for nivolumab and ipilimumab combination therapy. 150,151There is as yet no report on OS in these patients.

CONSIDERATIONS
106][107][108][109][110][111][112][113][114] Most were retrospective cohort studies with the majority performed prior to immune checkpoint inhibitor systemic therapy.There currently are 3 prospective RCTs published, which almost exclusively included metastatic uveal melanoma patients. 82,95,114All 3 RCTs failed to definitively demonstrate a survival benefit of liver-directed regional therapies. 82,95,114Objectively the best results for any liver-directed therapy were seen when all MLMs were treated using local therapies, either with complete surgical resection or using ablative techniques.Herein, complete removal or ablation of all metastatic disease remains key, because multiple studies showed no discernible clinical benefit after surgical reduction of tumor load and subsequent adjuvant therapy. 11,14,15,25,27,28,43,50Although many of these patients have recurrence of disease after curative treatment, repeat local treatment has been described and, therefore, should be considered in eligible patients. 152In the 37 studies investigating surgical treatment of MLMs, 5 reported patients underwent repeat resection or ablation. 12,28,30,36,42The absence of randomized data comparing complete local therapy remains impeding.All specialized centers performing these complex liver-directed therapies most certainly also perform surgical treatment of MLMs.This makes it highly likely that all patients eligible for complete local therapy for MLMs are by default not included in any of the studies investigating other liverdirected therapies.This selection bias likely overstates results observed for complete local treatment.Malignant melanoma of cutaneous, uveal, and mucosal origin should be regarded as separate entities, each with distinct biology.

Cutaneous Melanoma Liver Metastases
Compared with uveal MLMs, there are few studies that specifically evaluated liverdirected therapies for cutaneous MLMs.Although there are several studies investigating the surgical management, ablation, and radiotherapy of cutaneous MLMs (see Supplementary Data [Tables 1-3]), only 3 studies of other liver-directed therapies either had a majority of patients with cutaneous MLM, 69 or specifically reported results for patients with primary cutaneous melanoma. 60,77Of these, 1 studied TACE, 60 another studied HAI, 77 and the third evaluated both TACE and PHP. 69][125][126][128][129][130][131][132][133][134]153,154 The reported 5-year PFS and OS rates for ipilimumab and nivolumab combination therapy of 32% and 52%, respectively, [123][124][125] in part achieved in patients with visceral cutaneous melanoma metastases, are however unparalleled compared with previous systemic therapies, liverdirected therapies, and even complete surgical resection (see Supplementary Data [Table 2], Fig. 3).Isolated (resectable) hepatic metastases in cutaneous melanoma are exceedingly rare (see Fig. 1). 3 They are so rare that in principle other metastatic sites always are part of the problem and so in principle systemic therapy always is considered as first-line treatment.Patients with cutaneous MLMs, therefore, always should be referred to a medical oncologist.The most effective systemic treatment of metastatic cutaneous melanoma is combination therapy with ipilimumab and nivolumab and is the treatment of choice in the absence of contraindications for immune checkpoint inhibitor systemic therapy.In some patients, solitary progression may become the target of a surgical salvage approach.If patients with isolated cutaneous

Ho ¨ppener et al
MLMs are considered for surgical resection, adjuvant systemic treatment with immune checkpoint inhibition should be part of the treatment course. 127,128Future studies are needed to investigate whether the combination of (aggressive) surgical treatment and immune checkpoint inhibition can improve survival further.

Uveal Melanoma Liver Metastases
Studies investigating liver-directed locoregional therapies for MLMs almost exclusively were performed in patients with primary uveal melanoma.This can be attributed to the unique hepatotropic metastatic behavior of uveal melanoma 5 (see Fig. 1), with the hepatic metastases a driver of survival and poor results obtained with systemic therapy, 147 which has prompted many liver-directed therapies to focus on this rare disease.Uveal MLMs are the only MLM type with data to support complete local/surgical intervention and might be considered for selective patients in which a complete resection/ablation can be achieved.Herein it is important to recognize that in approximately half of all surgically explored (uveal) MLM patients, complete local treatment proves impossible (see Supplementary Data [Table 1]), meaning large uncertainty lies in a percutaneous approach.Aggressive strategies with combined reduction of tumor load and subsequent adjuvant therapy have not been shown to improve survival. 11,14,15,25,27,28,43Novel immune checkpoint inhibitors have shown little to no clinical activity in metastatic uveal melanoma, although results of combined ipilimumab and nivolumab still are expected. 146Despite the consistently higher response rates observed for IHP or PHP, 95 no single liver-directed or systemic therapy has proved superior effectiveness. 147Treatment decisions in patients with metastatic uveal melanoma not eligible for complete local treatment, therefore, should be based more on a center's own experience and patient preference.

Mucosal Melanoma Liver Metastases
Only 13 patients with MLMs from mucosal origin are described in all 106 studies combined with no single study reporting specific outcomes for these patients.Based on current evidence, no recommendations for any liver-directed regional therapy in the treatment of MLMs from mucosal origin can be made.Although evidence is limited, immune checkpoint inhibitions should be the treatment of choice in metastatic mucosal melanoma, either as first-line treatment or possibly in the adjuvant setting. 150,151

THE MEDICAL ONCOLOGIST'S PERSPECTIVE
Because the introduction of immune checkpoint inhibition and targeted therapy has significantly improved the survival of patients with advanced cutaneous melanoma, surgical treatment of advanced cutaneous MLMs has become complementary, whereas surgical treatment still is the cornerstone of the treatment of uveal MLMs.In an era of novel systemic therapies, surgical treatment can complement systemic therapy for cutaneous MLMs.In particular, local treatment of cutaneous MLMs can be considered for oligoprogression, which is defined as progressive disease at a limited number of disease sites after an initial response to systemic therapy. 155Local treatment of oligoprogressive MLMs may eliminate drug resistant clones, thereby achieving stable disease or even complete response.After treatment with immune checkpoint inhibitors, pseudoprogression also can mimic oligoprogression.Pseudoprogression is a phenomenon in which metastases show an increase in size on imaging due to T-cell infiltration, which can be confirmed by a surgical biopsy. 156On the other hand, surgical treatment may support the efficacy of systemic therapy by Treatment of Melanoma Liver Metastases debulking large or symptomatic MLMs.Because low tumor burden, which is reflected by low serum lactate dehydrogenase, is associated with better outcome of patients with cutaneous melanoma during immune checkpoint inhibition, 157 surgical debulking may facilitate tumor response to these drugs. 158Although the combination of surgical treatment of cutaneous MLMs and immune checkpoint inhibition has not yet been investigated prospectively, surgeons increasingly are consulted to complement systemic therapy with surgical treatment.Because immune checkpoint inhibitors have shown improved OS of greater than 10 years in patients with metastatic cutaneous melanoma, 121,125 treatment is shifting slightly from palliative toward more curative intent.This paradigm shift in the treatment of cutaneous metastatic melanoma requires a multidisciplinary approach, involving experienced radiologists, dermatologists, medical oncologists, radiotherapists, and surgeons.

DISCLOSURE
Authors have nothing to disclose.

Fig. 1 .
Fig. 1. (A) Incidence of cutaneous melanoma and uveal melanoma in the United States.(B) Metastatic pattern of cutaneous 3 and uveal 5 melanoma (C) Degree of liver involvement in metastatic cutaneous 3 and uveal 5 melanoma.(D) Time to metastasis in cutaneous 4 and uveal 6 melanoma.The diamond shape represents the median and the line represents the corresponding interquartile range.a Including subcutaneous tissue and regional lymph nodes.

Fig. 2 .
Fig. 2. Overview and flowchart of the article screening and selection process.Prelim, preliminary.* Three studies reported on multiple treatments: one on surgery and HAI, one on surgery and ablation, and one on TACE, SIRT, and IHP.

Fig. 3 .
Fig. 3. Kaplan-Meier survival curves for OS and PFS after Rx of MLMs based on individual patient data of 99 patients extracted from 14 studies.PFS data were available for 76 patients.