Introduction

Malignant melanoma is an aggressive skin malignancy with estimated age-standardized incidence rates of 3 per 100,000, which is rising worldwide [1, 2]. Melanoma exhibits profound brain-specific tropism and is the third most common origin of brain metastases after lung and breast cancers. It is estimated that up to 60% of all patients with metastatic melanoma will develop brain metastasis during the course of their disease, including 25% with solitary brain metastasis, which may contribute significantly to disease-related morbidity and mortality [3, 4]. Historically, patients with melanoma brain metastasis (MBM) had a uniformly dismal prognosis with a median survival of less than 3–6 months [4]. However, development of novel systemic therapies, such as BRAF-targeted therapy and immunotherapy, and advancement in local therapy, such as stereotactic radiosurgery (SRS) and surgical techniques, have all collectively improved the outcome with MBM to a median survival of 1–2 years [5, 6]. Herein, we discuss the understanding and current management of MBM.

Biology and risk factors

Brain metastasis is an intricate, multistep process, originating and escaping from primary cancer that extravasate through the basement membrane into systemic circulation, survive in circulation, adhere to local brain vasculature, invade blood–brain barrier, extravasate through microenvironment, and thrive by angiogenesis to establish into a brain metastasis [7]. Various receptor and protein molecules are overexpressed in melanoma cells that can lead to high brain tropism and are strongly implicative in MBM formation; these include chemokine receptor type 4 (CCR4), tetraspanins, integrins, melanotransferrin, and S100A4 protein [8, 9]. Protein kinase B (AKT)/phosphoinositide 3-kinase (PI3K) and mitogen-activated protein kinase (MAPK) are the two parallel pathways that regulate cell survival and proliferation, and are strongly implicated in MBM as well [10]. Activation of the PI3K/AKT pathways is also known to mediate both de novo and acquired resistance to BRAF and MEK inhibitors [11].

Various studies identified risk factors for the development of brain metastasis, which include male sex; age > 60 years; melanomas arising on mucosal surfaces or the skin of the trunk, head, neck, or scalp; deeply invasive or ulcerated primary lesions; acral, lentiginous, or nodular histology; involvement of > 3 regional nodes at diagnosis or relapse; and visceral metastasis [12,13,14]. Molecular markers associated with MBM development include BRAF and NRAS mutation, expression of CCR4 on melanoma cells, and activation of the PI3K/AKT pathway [8, 15].

Predictors of survival in melanoma brain metastasis

To cluster patients with brain metastasis into similar prognostic groups, Gaspar et al. [16] published a seminal report and introduced Radiation Therapy Oncology Group’s (RTOG) recursive partitioning analysis (RPA) using pooled data of 1200 patients from three consecutive RTOG brain metastasis randomized trials. Based on three factors (Karnofsky Performance Score (KPS), status of extracranial disease, and patient age), patients were divided into 3 prognostic groups. The best survival (median, 7.1 months) was observed in class 1 patients < 65 years of age with a KPS ≥ 70 and a controlled primary tumor with the brain as the only site of metastases. The worst survival (median, 2.3 months) was seen in class 3 patients with a KPS < 70. All other patients (class 2) had relatively minor differences in observed survival, with a median of 4.2 months.

One limitation of RTOG RPA model is that it is not disease specific, although this was validated in various diagnoses including melanoma patients which had median survivals of 151, 71, and 21 days for RPA classes 1, 2, and 3, respectively (p < 0.001) [17]. Other limitations of RPA include omitting the number of brain metastases and estimating systemic disease control which was fraught with inconsistency due to variation in type and timing of imaging tests. Sperduto et al. [18] introduced graded prognostic assessment (GPA) using 4 factors—age, KPS, extracranial metastases (none and present), and number of metastases (1, 2–3, > 3)—to more accurately determine the prognosis of patients with brain metastases. GPA was subsequently refined to diagnosis-specific GPA (DS-GPA); for melanoma, prognostic factors were KPS and number of brain metastases, with a low KPS and > 3 brain metastases associated with poor outcome (median survival of 3.4 months) and a higher KPS with 1–3 metastases associated with better prognosis (median survival of 13.2 months) [19].

Recently, Sperduto et al. updated the DS-GPA specific for MBMs by including molecular markers (Melanoma-molGPA) in a larger and current cohort (n = 823) diagnosed from 2006 to 2015. Melanoma-molGPA scores of 4.0 and 0.0 were associated with the best and worst prognoses, like DS-GPA indices. There were 5 significant prognostic factors for survival (age, KPS, extracranial metastases, number of brain metastases, and BRAF status), whereas only KPS and number of brain metastases were included in the original Melanoma-GPA. Using Melanoma-molGPA, median survival times for patients for scores 0–1, 1.5–2, 2.5–3, and 3.5–4 were 4.9, 8.3, 15.8, and 34.1 months, respectively [20••]. The development of such prognostic tools has facilitated clinical decision-making by helping physicians differentiate patients based on expected survival and also useful for stratification of clinical trials.

Approach to management

Management of melanoma brain metastasis involves multimodality approach and increasing personalized treatment planning, with the integration of systemic therapy combined with traditionally local therapy such as SRS and/or surgery. CNS is previously considered a sanctuary site that rendered systemic therapy ineffective due to poor BBB penetration. However, recent data have shown improved intracranial response with newer classes of systemic therapies such as BRAF/MAP kinase inhibitors (BRAFi/MEKi) and immunotherapy (anti-programmed cell death protein 1 or anti-PD1, anti-cytotoxic T lymphocyte–associated protein 4 or anti-CTLA-4). The choice of local therapy (surgery, SRS) depends on patient performance, location, size, metastasis-induced neurological symptoms, and number of brain metastasis. The advantages of surgery include rapid relief of pressure effect on surrounding normal brain structures and evacuation of intra- or peri-tumoral hemorrhage, as well as procurement of tissue for diagnosis and molecular studies to select appropriate systemic therapy. It is a frequent practice in patients with multiple MBM to combine surgery for larger and symptomatic lesion(s), and SRS for the remainder lesions. Following resection, post-operative radiotherapy to resection cavity is recommended to improve local control by eliminating micrometastasis or residual metastasis. In the next sections, we describe each treatment modality and recent evidence that led to its present-day use.

Surgery

Surgical resection is a therapeutic option if metastasis is solitary or limited in number or symptomatic due to mass effect on the surrounding brain, and above all if located in a surgically accessible region [21]. Moreover, surgery can provide a tissue diagnosis to identify new genetic drivers to potentially guide appropriate systemic therapy, given a growing evidence that MBMs possess unique molecular characteristics compared with primary disease or metastases at other sites [22•]. Surgical resection of brain metastasis has been shown to improve overall survival in various studies [23, 24]. In a population study with > 4200 patients, metastasectomy improved median (12 vs. 5 months) and 5-year overall survival (16% vs. 7%, p < 0.001). Although the role of surgical resection is well established for solitary brain metastasis, in multiple brain metastases, the pendulum sways away from surgery towards radiotherapy as mainstay local treatment or used in combination with surgery. The number of lesions that can be safely removed depends primarily on safety and clinical justification. The selective patient population who underwent metastasectomy of multiple MBMs may have survival benefit, although there is insufficient prospective data to support this approach [23,24,25]. Generally, for patients with up to 3–4 brain metastases in accessible locations, good performance status (KPS > 70), and controlled systemic disease, surgical resection may be a viable option [25, 26]. Depending on extent of excision and size and number of metastases, post-operative radiation therapy to resection cavity or whole brain radiation therapy is always recommended [27, 28].

Whole brain radiotherapy

Radiation therapy was traditionally delivered as whole brain radiotherapy (WBRT), especially for multiple MBMs. The rationale of WBRT is that it comprehensively treats both macroscopic and microscopic metastases in the brain. Nevertheless, prognosis is dismal, with the median overall survival following WBRT approximately 2–5 months, and 1-year survival less than 10–12% [6, 17, 29]. WBRT for MBM has played a central role for decades until recent years when stereotactic radiosurgery (SRS) was introduced. The role of WBRT is shrinking mainly due to its detrimental effect on neurocognition and quality of life and lack of survival benefit. Nevertheless, it continues to play an important role for palliation in patients who have numerous symptomatic brain metastases, extensive metastasis not amendable for radiosurgery, and symptomatic leptomeningeal carcinomatosis, and for those whose performance status is poor [30]. The most common used WBRT fractionation regimen is 30 Gy in 10 fractions or 20 Gy in 5 fractions delivered daily to the whole brain. Various interventions have been attempted to mitigate the WBRT-induced neurocognitive decline. The concomitant use of memantine hydrochloride, a non-competitive N-methyl-d-aspartate (NMDA) receptor antagonist, has shown to be neuroprotective by binding to and inhibiting ion channels of NMDA receptors located in cortical and hippocampal neurons, thus preserving cognitive functions [31]. This was further investigated in a large (n = 508) placebo-controlled trial (RTOG 0614) demonstrating that patients who received memantine during and following WBRT had lesser neurocognitive failure (53.8% vs. 64.9%) and reduced rate of decline in memory, executive function, and processing speed as compared with those who did not receive memantine [31]. Preclinical studies have demonstrated that modest doses of radiation can cause a significant decline in neurogenesis in the hippocampus in sub-granular zone which is associated with the suppression of new memory formation and impaired recall [32]. Studies have shown a dose-response relationship between radiation dose received by the hippocampus and risk of postradiotherapy decline in recent memory [33]. This incited growing interest of hippocampal-avoidance WBRT in reducing neurocognitive decline [34]. RTOG 0933, a single-arm multi-institutional phase II study, showed hippocampal-avoidance WBRT is associated with a better preservation of memory and QOL as compared with historical series [35]. NRG CC001, a recently reported phase III trial of WBRT (30 Gy in 10 fractions) plus memantine with or without hippocampal avoidance, showed better preservation of neurocognitive function and patient-reported symptoms, with similar intracranial control and survival. Authors recommended hippocampal avoidance WBRT should be considered a standard of care for patients with good performance status who receive WBRT for BM with no metastases in the hippocampal region [36, 37•]. WBRT may not provide long-term intracranial control, as many patients live longer with immunotherapy and/or targeted therapy and develop recurrence, and therefore, continued surveillance is recommended.

Stereotactic radiosurgery

With technological advancement in precision radiation oncology and imaging modalities such as high-resolution MRI in the last 2–3 decades, it became feasible to treat metastatic lesion(s) with a large ablative radiation dose, aka SRS, effectively and safely while sparing surrounding normal brain tissues. Melanoma being a radioresistant cancer, ablative dose with SRS has a strong radiobiological rationale for improved cell killing, and clinical studies since the 1990s has shown SRS as a very effective local therapy for MBMs [38]. SRS is currently the standard local therapy over WBRT for ≤ 4 brain metastases, and ongoing prospective trials are evaluating the role of SRS for 5–15 brain metastases [30, 39, 40]. In selective patients with 1–4 metastases measuring less than 3–4 cm, SRS yields an excellent local control with response rates of as high as 90%, median survival of 5–11 months, and 1-year survival of 25% [41, 42]. Although no large randomized study comparing surgery versus SRS, there are a few small randomized studies and retrospective series which showed SRS to be equally efficacious as surgery [43, 44]. A phase III randomized study of 33 patients, which was stopped prematurely due to poor accrual, compared Gamma Knife SRS alone versus surgery plus WBRT in patients with a single, ≤ 3-cm brain metastasis and showed similar local tumor control [43]. Generally, SRS can be used as an alternative to surgery in patients with lesion size(s) of < 3 cm each, deep tumor locations, minimal pressure symptoms or midline shift, or hydrocephalus, and in those who are unable to undergo surgery [41]. While existing evidence for the use of SRS for ≤ 4 brain metastases is sufficient, there are growing interest and studies for ≥ 10 lesions, including an ongoing phase III trial evaluating SRS for 5–15 brain metastases versus hippocampal-avoidance WBRT [30, 39, 40]. While effective with excellent local control of MBM up to 85–95%, randomized trials did not demonstrate improved overall survival following SRS treatment [45, 46]. Although patients treated with SRS alone are more likely to develop distant intracranial relapse, randomized and retrospective data did not show survival benefit when adding WBRT to SRS treatment [45,46,47]. The landmark Alliance trial (NCT00377156) randomized 213 patients to SRS alone vs. SRS + WBRT, which showed lesser cognitive deterioration (63.5% vs 28%) and better quality of life favoring SRS alone. Although the time to intracranial failure was significantly shorter for SRS as compared with SRS + WBRT, there was no difference in overall survival (p = 0.9) [48••]. Therefore, SRS alone with imaging surveillance without WBRT has become a standard therapy for limited number of brain metastases ≤ 3 cm.

For larger size lesion (> 3 cm) or close to vital structures (brain stem, optic chiasm), single-fraction SRS are associated with a higher risk of radiation injury [49]. In such cases, fractionated (3–5 fractions) stereotactic radiotherapy (FSRT) has been used to exploit differential radiation repair capacity of tumor and normal tissues, thus permitting treatment with a higher effective total dose while minimizing toxicity. FSRT has shown good local control of > 85% in published studies, and ongoing prospective trials such as a dose-escalation phase I study for large unresectable lesions (NCT02054689) are underway [50, 51]. However, patients with large brain metastases are usually symptomatic due to pressure effect and may require upfront surgical decompression. Surgical resection followed by adjuvant SRS to the surgical cavity is highly effective, with 1-year local control of 70–90% [52,53,54]. While surgical cavities of large metastases may involute to a size amenable to treatment with single-fraction SRS (NCCTG N107C/CEC·3 study only included < 5 cm cavity) [54], FSRT is often considered in cases where the cavity remains large. A randomized trial (NCT04114981) comparing post-op SRS and FSRT is being planned by Alliance for Clinical Trials in Oncology [55] (Fig. 1).

Fig. 1
figure 1

Patient with 8 melanoma brain metastases was treated with stereotactic radiosurgery (SRS). a Baseline MRI showing metastatic lesions. b SRS planning with bird’s-eye view and planning isodose lines. c Follow-up MRI after 5 months showing near complete resolution of all lesions. However, this patient developed new cerebellar lesion 8 months after initial SRS which was also treated with SRS

Full size image

Systemic therapy

The impermeability of the BBB confers limited responses of MBMs to chemotherapeutic agents, and therefore, in the past decades, local therapies (surgery, WBRT, SRS) were the pillars of MBM management. However, with a better understanding of melanoma molecular biology, discoveries of driver mutations, and advances in immunotherapies and targeted therapies, there are evolving clinical evidence demonstrating considerable intracranial activity with these agents. Since 2011, the US Food and Drug Administration (FDA) has approved indication for 7 new drugs as systemic therapy of metastatic melanoma. For MBMs, these newer systemic therapies have been shown to have intracranial activities and in conjugation with local therapy may provide better intracranial control.

Cytotoxic chemotherapy

Larger chemotherapy molecules, including dacarbazine, a DNA-alkylating agent, which has long been historical standard cytotoxic chemotherapy for melanoma, are not able to cross the BBB to incite appropriate response for the management of MBMs. Not surprisingly, most studies exploring the role of cytotoxic chemotherapy in MBMs, such as dacarbazine, temozolomide, paclitaxel, carboplatin, did not show survival advantage or intracranial response in metastatic melanoma [56,57,58]. Temozolomide, a derivative of dacarbazine which readily penetrates the blood–brain barrier to achieve cerebrospinal fluid concentration up to 30% of plasma, only demonstrated clinical responses in roughly 10% as a single agent or combined with WBRT or other agents [56].

Targeted therapy

Genetic mapping of melanoma identified approximately 50% of melanomas harboring the driver mutation in the serine-threonine protein kinase B-RAF (BRAF) gene resulting in uncontrolled activation of the MAPK pathway [59]. BRAF inhibition blocks the MAPK signaling pathway that normally promotes cell proliferation. In melanoma, over 90% BRAF mutations are at codon 600, and among them, > 90% are BRAFV600E (substitution of glutamic acid for valine), followed by BRAFV600K (substituting lysine for valine), that represents 5–6% [60]. To date, three BRAF inhibitors—vemurafenib, dabrafenib, and encorafenib— (in combination with the MEK inhibitor binimetinib) have been FDA approved for metastatic melanoma. Mouse model showed lower CSF level of BRAF inhibitors which are substrates for the BBB drug efflux pumps leading to 2–3-fold lower concentration of targeted agent in CSF than plasma [61]. Vemurafenib has a much lower brain penetration than dabrafenib. However, BBB is frequently violated in MBM and intracranial response was observed in responding cancer [62].

The initial phase 1 and phase II studies established the safety profile of vemurafenib (PLX4032), the first potent BRAFV600 inhibitor, which showed overall response rate of 53% with median overall survival of 15.9 months [59, 63]. However, these early studies excluded patients with active brain metastases. Vemurafenib for brain metastasis has shown intracranial overall response rate of 18–50% in early studies [64, 65]. A phase I trial first established safety and efficacy of dabrafenib, another BRAFV600 inhibitor, for MBM in treatment-naive MBM demonstrating that nine of ten patients had reductions in size of brain lesions following dabrafenib, including four complete remissions [66]. This led to a larger open-label phase II BREAK-MB study evaluating efficacy of dabrafenib in 172 patients with BRAF mutated MBMs. Among BRAF V600E mutated patients, median survival of 33 weeks and overall intracranial response of 39.2% were noted in patients without prior local therapy; and median survival of 31.4 weeks and overall intracranial response of 30.8% were reported in patients with prior local treatment. Patients with a BRAF V600K mutation had substantially lower overall response rates; however, all patients had acceptable safety profile with dabrafinib [67]. Another study establishing efficacy of dabrafenib for intra- and extracranial metastasis showed high and statistically similar response rates for intracranial (78%) versus extracranial (90%) sites, with identical median site-specific progression-free survival [68].

The BRAF inhibitors show good single-agent activity with a promising initial response; however, resistance is almost a rule, secondary either to inadequate MAPK inhibition or reactivation of the MAPK pathway through preferential molecular mechanisms (MEK mutations, BRAF-splice mutants, NRAS-mutations, adaptive RTK signaling) [69, 70]. Therefore, it is the rationale for BRAF-MEK inhibitor combination therapy to target the PI3K-AKT pathway concomitantly with MAPK inhibition. Combing BRAF/MEK inhibitors was supported by recent trials. A randomized trial of 423 untreated unresected stage IIIC and IV melanoma patients treated with dabrafenib and trametinib versus dabrafenib only showed median overall survival of 25.1 vs 18.7 months favoring dabrafenib and trametinib group (HR 0.71, p = 0.01) and improved median progression-free survival (PFS) (11.0 months vs. 8.8 months (HR 0.67, p = 0.0004)) [71]. For patients with MBM, results from COMBI-MB (NCT02039947) trial showed that dabrafenib–trametinib combination therapy had higher intracranial response (58%) than previously reported (39%, BREAK-MB study) single-agent BRAF therapy in patients with asymptomatic BRAF V600E-mutated MBM without prior local therapy [72••]. For symptomatic MBM patients vs. patients with prior local therapy, intracranial response rates were similar (59% vs. 56%). Encouragingly, response rate to BRAF-MEK inhibitor combination therapy in the brain was similar to that observed at extracranial sites and no additional brain-specific side effects were reported [72••]. Despite the promising results of these studies, the intracranial responses were relatively shorter than extracranial sites and the brain shown to be the major site of treatment failure [71]. This suggests a role for the brain microenvironment in therapeutic escape and BRAF-MEK inhibitor resistance. Various and emerging novel combination “parallel-pathway inhibition” treatment strategies are being explored to curb this resistance mechanism [73].

Immunotherapy

Melanoma is highly immunogenic cancer, with several reports of immune-mediated spontaneous regression [74]. High-dose interleukin-2 (IL-2) was the first immune therapy used in stage 4 melanoma patients. IL-2, a cytokine produced by activated T cells, causes proliferation of cytotoxic T cells, natural killer cells, and monocytes. High-dose IL-2 achieved overall response, including durable complete remission for a small percentage of metastatic melanoma patients (< 10%) [75]. Unfortunately, IL-2 treatments were associated with severe toxicity and its efficacy against brain metastases was disappointing with 6% response rates in previously untreated MBMs [76]. The second immune drug approved by the FDA was interferon-α (IFN-α) and was associated with overall response rates of 22% and complete response < 4%. However, responses of IFN- α were limited to patients with low-tumor burden and did not show efficacy against MBMs. The clinical benefit of IL-2 and INF-α has provided the “prototype concept” for further research to establish the role of immunotherapy in metastatic melanoma.

Adoptive cell therapy entails harvesting lymphocytes from the melanoma patients either from blood or tumor-infiltrating lymphocytes, followed by in vitro selection, expansion (with or without genetic manipulation) and their activation with subsequent re-infusion of these processed lymphocytes into patients after lymphodepletion to induce an immune antimelanoma responses. A study of 26 patients with treatment-naïve MBMs showed 7 patients (41%) achieved intracranial complete response following adoptive cell therapy with tumor-infiltrating lymphocytes [77]. The flipside of adoptive cell therapy is inhibitory cost, limited availability, and complexity involved in the development of these “custom-made therapy” with the duration of in vitro culture, skilled man-hours, and patient preparation for lymphodepletion with high-dose chemotherapy and/or total body irradiation. Not surprisingly, despite these encouraging results, adoptive cell therapy has largely been replaced by newer class of immunotherapy such as checkpoint inhibitors in the treatment of metastatic melanoma.

In 1987, James P. Allison identified cytotoxic T lymphocyte antigen 4 (CTLA-4), a key negative regulator preventing T cells from attacking tumor cells, and in the 1990s, Okazaki et al. discovered a molecule on T cells called programmed death 1 (PD-1) [78]. Subsequent preclinical and early phase clinical trials established that blockage of CTLA4 or PD-1 would potentiate anti-tumor T cell activity [79]. The approval of checkpoint inhibitors, including human monoclonal antibody ipilimumab that blocks CTLA-4 and nivolumab and pembrolizumab that blocks PD-1 thereby stimulating antitumor T cell response, changed the landscape of immunotherapy. The two large phase III randomized trials comparing ipilimumab against standard of care in previously treated (NCT00094653/MDX010-20) and treatment-naive (NCT00324155/CA184-024) metastatic melanoma patients showed significant improvement in overall survival with ipilimumab, and thereby established ipilimumab as the new standard of care [80, 81]. The landmark study MDX010-20 was a 3-arm randomized trial with 676 patients, comparing ipilimumab plus gp100 vaccine, ipilimumab alone, and gp100 alone, and included 77 (11%) patients with stable MBMs not requiring steroids. The median survival was significantly improved in both arms receiving ipilimumab [80]. Similar benefits were seen in patients with or without brain metastases, with hazard ratios for death in patients with brain metastases 0.70 (95% CI 0.41–1.20) for 46 patients receiving ipilimumab plus gp100 vaccine and 0.76 (0.38–1.54) for 15 patients given ipilimumab alone.

A multi-institutional phase 2 study (NCT00623766) assessed the role of ipilimumab in 72 patients with MBMs, of which 51 individuals (cohort A) were neurologically asymptomatic and steroid free, and 21 patients (cohort B) were symptomatic and on stable doses of steroids. The study showed similar proportion of patients achieving overall response rate for intracranial (24% and 10%) and extracranial (27% and 5%) disease sites in cohorts A and B, respectively [82]. Global and CNS immune-related responses were higher in asymptomatic patients who did not require steroids. Median survival was 7.0 months in asymptomatic (cohort A) versus 3.7 months in symptomatic (cohort B) patients [82]. Although immune therapies have poor penetration through the intact blood–brain barrier, studies showed that activated T cells can pass through BBB, and thus immune therapies that stimulate T cell responses may be effective against MBMs [83].

A single-arm phase 2 trial (NIBIT-M1) investigated combination of ipilimumab plus fotemustine in 86 patients with metastatic melanoma, including 20 patients with asymptomatic brain metastases. A 3-year follow-up analysis showed median OS of 12.7 and 12.9 months in patients with MBM and for entire cohort, respectively [84, 85]. Ongoing phase III NIBIT-M2 trial (NCT02460068) is evaluating contribution of fotemustine in a randomized 3-arm study comparing fotemustine alone, fotemustine plus ipilimumab, or ipilimumab plus nivolumab in patients with asymptomatic MBM [86].

Pembrolizumab, a PD-1 inhibitor, has shown 22–26% durable response rates in untreated MBMs (5–20 mm), with an acceptable safety profile, in a recent phase II trial. In this study, brain metastasis and systemic responses were concordant, with median progression-free and overall survival of 2 and 17 months, respectively, which was similar to patients without brain metastasis treated with anti-PD1 agents [87, 88•].

A phase II study (CheckMate 204, NCT02320058) investigated the combination of nivolumab and ipilimumab in 94 patients with no neurological symptoms and ≥ 1 measurable, non-irradiated MBM(s) measuring 0.5–3.0 cm. With median follow-up of 14.0 months, the intracranial response was 57%, including 26% complete response rates, which was concordant with extracranial activity. Treatment-related grade 3 or 4 adverse events were reported in 55% of the patients, including one death from immune-related myocarditis [89•]. A recent update of CheckMate 204 evaluated patients with no neurologic symptoms or steroid Rx (asymptomatic; cohort A) and those with neurologic symptoms, whether or not they were receiving steroid Rx (symptomatic; cohort B). With median follow-up of 20.6 months, the intracranial clinical benefit rate, i.e., proportion of patients with complete response + partial response + stable disease ≥ 6 months, in Cohort A was 58.4%. In cohort B, with a median follow-up of 5.2 months, intracranial clinical benefit rate was 22.2% [90]. A multicenter, open-label, randomized phase II Anti-PD1 Brain Collaboration trial (NCT02374242) investigated combination nivolumab plus ipilimumab (cohort A, n = 36) versus nivolumab alone (cohort B, n = 27) in patients with asymptomatic MBMs with no prior local radiotherapy and symptomatic patients (cohort C, n = 16), which showed intracranial response rate of 46% in cohort A, 20% in cohort B, and 6% in cohort C. Intracranial complete responses occurred in 17% of patients in combination cohort A. Treatment-related grade 3/4 adverse events occurred in 54% patients in combination cohort A, and 16% in cohort B, and 26%, and 5% of patients, respectively, discontinued due to an AE [91]. Current evidence from phase II studies as summarized above showed that combination immunotherapy with ipilimumab and nivolumab demonstrated durable intracranial response for patients with asymptomatic MBM, but poor response was evident in symptomatic MBM patients following combination immunotherapy. Further studies are warranted to confirm response and elucidate resistance mechanism with these treatment strategies for patients with asymptomatic MBM. Additional investigation will also be needed to improve therapy for patients with symptomatic MBM (Tables 1 and 2).

Table 1 Summary of pertinent clinical studies evaluating systemic and radiation therapies for melanoma brain metastasis
Full size table
Table 2 Ongoing clinical trials of melanoma brain metastasis
Full size table

Combination of SRS with targeted therapy/immunotherapy

SRS delivers high ablative radiation dose leading to cell apoptosis and might have a synergistic immunomodulatory effect when combined with immunotherapy. To date, there have been several retrospective studies that explored safety and efficacy of concurrent or sequential SRS with targeted therapy or immunotherapy showing excellent local control rates and safety profile. However, prospective data are lacking. In a study reviewing 77 MBM patients, combining ipilimumab with SRS was associated with significant improvement of median overall survival (OS) from 4·9 months to 21·3 months and 2-year overall survival from 19·7% to 47·2% compared with SRS alone [92]. Ahmed et al. [93] reviewed 314 MBM patients treated with SRS within 3 months of anti-PD-1, anti-CTLA-4, BRAF/MEK inhibitors, or conventional chemotherapy. Treatment with anti-PD-1, anti-CTLA-4, or BRAF/MEKi significantly improved OS when compared with conventional chemotherapy. Local MBM control rate at 12 months was 83%, with no significant differences among the groups. The risk of radionecrosis is reported to be higher in at least one retrospective study suggesting that approximately 10% of patients given ipilimumab in combination with focal radiotherapy or WBRT might be at risk, with a peak incidence around 12–15 months after radiation [94]. Further investigation is warranted.

Acharya et al. [95] investigated SRS within 3 months of immunotherapy or targeted therapy in 72 patients with 233 MBMs. One-year local control for SRS, SRS + immunotherapy, and SRS + targeted therapy were 66%, 85%, and 72%, respectively (p = .04). One-year distant intracranial control rates for SRS, SRS + immunotherapy, and SRS + targeted therapy were 11.5%, 60%, and 10%, respectively (p < .001). This study suggested SRS + immunotherapy was associated with a significant reduction in distant intracranial failure compared with SRS alone or SRS + targeted therapy.

There are little data regarding the timing of targeted therapy dosage relative to SRS, but oncologists commonly withhold targeted therapy for 3–5 days surrounding SRS treatment. Improved brain tumor control and OS were seen in concurrent or post-SRS BRAF/MEK inhibitors, suggesting perturbation of BBB by SRS leading to increased intracranial delivery of these drugs [93, 95, 96]. However, concurrent vemurafenib or dabrafenib and SRS have been reported to increase risk for radiation necrosis and grade ≥ 3 adverse events, especially skin toxicities, and this should be monitored [97,98,99].

Activation of an anti-tumor immune response after radiotherapy-induced cell death can lead to regression of tumors distant from the irradiated region (known as the abscopal effect). Several studies have reported abscopal effects when radiotherapy is combined with anti-CTLA-4 inhibitors [100]. However, abscopal effect is unpredictable, and its effect for MBMs is largely unknown.

The combination strategy of systemic therapies and SRS needs to be investigated prospectively to assess the optimal sequence of radiation and systemic therapy, radiation (and systemic therapy) dose and radiation fractionation, and risks including radionecrosis. Until then, based on observational non-randomized data, SRS concurrently or sequentially with immunotherapy is well tolerated; a washout period of 3–5 days is usually recommended when combined with targeted therapy [99].

Conclusions

For patients with melanoma brain metastases, treatment decision has become more individualized. A multidisciplinary approach involving medical oncologists, radiation oncologists, and neurosurgery to derive a personalized treatment plan based on prognostic groups, molecular classification, and number and size of brain metastasis can promise possibility of long-term intracranial control. For patients with a good prognosis, the focus is on effectively managing brain metastases while minimizing neurological toxicity and maintaining the quality of life. Surgery should be considered in symptomatic and large MBMs, while SRS considered in 1–4 lesions, and now also being explored for up to 15 brain metastases for improved local control. The role of WBRT is shrinking but yet to be re-defined, given its neurocognitive side effects, except for patients with diffuse brain involvement as a palliative measure. Immunotherapy has become the cornerstone of management for metastatic melanoma. For patients with asymptomatic MBMs, ipilimumab and nivolumab have shown excellent intracranial response based on results of recent trials including CheckMate 204. For patients with BRAF-mutant melanoma, BRAF/MEK-targeted agents can be considered. Combination of systemic therapy with SRS seems safe and effective based on retrospective data; potential benefits and risks should be evaluated prospectively. Given increasingly complex treatment options, patients with melanoma brain metastases should ideally be evaluated in a multidisciplinary setting for optimal treatment approach and encouraged to participate in clinical trials.