Brain metastases from malignant melanoma have a hemmorhagic rate of 25–50% [73, 83–96]; even small lesions can lead to significant hemorrhage. Intratumoral hemorrhage presents with similar subacute progressive symptoms as non-hemorrhagic brain metastases [86]. They may also present with an abrupt onset of a headache with focal neurological deficits progressing into obtundation, while other will have worsening of their previous neurological deficits [86, 89, 97, 98]. Patients may be completely asymptomatic with BM found on screening MRI [84]. Hypertension and thrombocytopenia do not lead to the intratumoral hemorrhage [95].

Macroscopic bleeding is present in up to 80% of patients with cerebral melanoma metastases [86, 90]. Patients with multiple metastases tend to have bleeding into most of their lesions rather than an isolated hemorrhage [86]. In addition, up to 25% of patients who were previously thought not to have hemorrhage were found to have histological evidence of prior hemorrhage [90]. Hemorrhage may be intratumoral, adjacent to the tumor, or may present as a larger intracerebral hematoma surrounding scattered tumor fragments [89]. Intratumoral hemorrhage may be secondary to loss of vessel integrity associated with inter- and intratumoral necrosis and neovascularization [95]. The lesion size does not seem to affect the risk of developing a hemorrhage; lesions ranging from microscopic to 8 cm in diameter can bleed [89].

Intratumoral hemorrhage usually has non-contiguous enhancing lesions, with hemorrhage occurring outside of the basal ganglia region [86, 99], and enhancement adjacent to the blood clot [86, 97]. Neoplastic hemorrhage usually does not resolve on serial images and tends to have a heterogenous intensity pattern on spin-echo sequences [100].

One report showed that up to 50% of patients with cerebral melanoma may develop subdural hematomas [101]. Most subdural hematomas are asymptomatic, with only one-fourth of patients presenting with symptoms. Usually, patients will present with acute confusion or lethargy, and only 10% have focal neurological deficits [86]. It is thought that the etiology is secondary to neovascularization of tumor cells that spread to the leptomeninges [98, 102]. Thrombocytopenia is likely not related to the development of subdural hematomas [86].

Like other systemic metastases to the CNS, melanoma metastases carry a poor prognosis. CNS metastases are present in up to 75% of patients who die from the disease and are the cause of over half of all melanoma deaths [14, 42, 50, 103–105]. Patients with initial melanoma metastasis in the CNS may have a better prognosis than patients with initial metastasis elsewhere in the body [14, 46, 50]. Metastases to the lungs or liver have been found to worsen the prognosis [14, 50]. There is also a relationship with the prognosis and the location of the primary lesion, with head and neck lesions shortening the survival time [50]. In patients with >4-mm-thick melanomas, ulcerated primaries had a significantly worse five-year survival rate, even when controlling for thickness, mitotic rate, histotype, vascular involvement, and lymph node status [49]. Early hemorrhage also worsens the patient’s prognosis [106]. Unsurprisingly, patients who respond to their initial management tend to have a better prognosis than those who fail their initial therapy [46].

The number and location of cerebral metastases discovered at diagnosis is another factor that worsens prognosis, with cerebellar and leptomeningeal metastases having a worse survival [46, 96]. The extent of extracranial metastases is associated with survival. The median survival in one study of patients with controlled extracranial disease activity was 12.7 months, which dropped to 3.9 months if the disease was active [96]. Hence, improving the prognosis for patients with CNS metastases necessitates also controlling systemic disease [96, 107].

The patient’s clinical status is another prognostic factor. Patients who present with asymptomatic brain metastases or have a systemic response after their first treatment have an overall better prognosis [108]. Patients with Karnofsky Performance Score (KPS) ≥90 do significantly better than patients with a score ≤80 [96, 107]. In addition, patients with a higher Recursive Partitioning Analysis (RPA) class had a worsened survival. In one study comparing patients’ RPA classes, those who were class 1 had a median survival of 12.7 months, class 2 had 4.9 months, and class 3 had 2.3 months [96].

When one study evaluated 17 patients surviving > three years from initial diagnosis of cerebral metastases, some shared characteristics included the lack of a primary lesion in the head and neck region, only a single CNS lesion discovered at diagnosis, no visceral metastases, and surgical resection performed as the initial treatment [50].

Patients with untreated brain metastases have a median survival between 3 and 8 weeks [52, 70, 109, 110]. Once a patient develops CNS metastases, it is important that palliative care be a part of therapy discussions [57]. Although current management can significantly prolong survival, in particular for solitary brain metastases, the overall prognosis remains poor [14]. This is largely due to melanoma being characteristically radioresistant and chemoresistant. Moreover, in most cases, multiple metastases are found at presentation [45].

Some positive strides have been made in the past few decades in the treatment of CNS melanoma. One study evaluating melanoma patients with cerebral metastases from 1986 to 2004 found that patients diagnosed before 1996 had an average survival of 4.14 months, while those diagnosed after 1996 had an average survival of 5.92 months. The increase may be due to a lead time bias since the use of MRIs has increased during this period of time; however, the number of patients who underwent SRS has also doubled likely improving the overall prognosis [46]. Current treatment options include surgical resection, stereotactic radiosurgery (SRS), whole-brain radiotherapy (WBRT), systemic therapy, and supportive care. An additional confounder of such retrospective can be selection bias, whereby therapies are often chosen based on a patient’s burden of disease. Patients with less than three lesions and low tumor burden are usually treated with surgical resection or SRS. The patient with diffuse metastases and high tumor burden is usually treated with whole-brain radiotherapy, chemotherapy, or supportive care [46]. In general, WBRT has limited efficacy for the treatment of melanoma due to radioresistance. Radiosurgery is often more effective due to the ability to deliver a large fraction in a single dose, though this technique is often limited by the number and size of lesions.

SRS and surgical resection are both targeted methods for treating patients with few cerebral metastases to provide immediate relief. Both SRS and surgery seem to have similar tumor control rates with or without fractionated radiotherapy [117–123]. To date, there is no known randomized trial addressing the superiority of one over the other; however, there are some considerations which should be made before choosing a therapy. SRS is less expensive than surgery and may have an overall reduced morbidity [124]. Because SRS is associated with peritumoral edema or radiation necrosis oftentimes necessitating steroids, there is concern that the efficacy of immunotherapy will be hindered. Thus, in patients who are planning to receive immunotherapy, surgery may be preferred [125].

Systemic therapy for melanoma CNS metastasis has many challenges. Extracranial melanoma is already very resistant to chemotherapy, and agents used must cross the blood–brain barrier [61]. In addition, because of the poor prognosis of patients with cerebral metastases, patients with CNS metastases tend to be excluded from clinical trials for systemic therapy, making the data on efficacy limited [126]. Until recently, even systemic melanoma responded poorly to the available systemic treatments, in part due to the intrinsic resistance of the disease.

Fotemustine, dacarbazine (DTIC), and temozolomide (TMZ) are chemotherapy agents used to treat extracranial melanoma metastases. All are alkylating agents with penetration into the CNS. Unfortunately, phase II trials showed that they were not as effective as hoped; TMZ and fotemustine had a response rate of 6 and 5.9%, respectively [57, 127]. Of note, the fotemustine study compared it to dacarbazine (DTIC), which at the time was the primary treatment for malignant melanoma. DTIC had a CNS response rate of 0%, and the study showed a trend toward fotemustine prolonging the time to brain metastases (22.7 vs. 7.2 months; p = 0.059) [126].

A multicenter, open-label, phase II trial with TMZ enrolled 151 patients with histologically confirmed melanoma metastases to the brain without prior radiotherapy or radiosurgery. Previously untreated patients were started on TMZ 200 mg/m²/day for five days, and previously treated patients were started on TMZ 150 mg/m²/day for five days every 28 days. Of the 117 patients who had no prior treatment, one patient had a complete response, seven had a partial response, and thirty-four had stable disease of the brain. Of the 34 patients who had previous therapy, one had a partial response and six had stable disease of the brain. Kaplan–Meier estimates of these groups showed progression-free survival of 1.2 and 1.0 months, and median survival or 3.5 and 2.2 months, for the previously untreated and treated groups, respectively. The improved response of the melanoma to TMZ in patients without prior treatment was speculated to be due to the melanoma cells being naïve to the chemotherapy. Disease progression occurred in 74% of patients and was the most common reason for discontinuation of the treatment. Of note, 6% of the patients had a response to temozolomide, comparable to the rates in the other visceral sites, suggesting it has similar efficacy for both. This also suggests that the reason other chemotherapy may have been failing at treating melanoma metastases was the difficulty in penetrating the blood–brain barrier [57]. One phase II pilot study substituting TMZ for DTIC reduced the incidence of CNS relapse, but it did not affect overall survival [128].

Thalidomide is an agent with both antiangiogenic and immunomodulatory effects. A phase II trial of 35 patients who had a WHO performance scale of ≤2 examined the effect of a thalidomide at doses starting at 100–400 mg/day over a span of four weeks. No patient had any response to brain metastases, although three had a partial response to their extracranial lesions. Four of the patients had stable CNS disease for over four months, suggesting some CNS activities. Twenty patients were placed on the maximum dose of 400 mg, with seven requiring dose reductions due to side effects that were reversible [129].

The potential to target molecular aberrations and mutations in melanoma brain metastasis has been the focus of recent research. Current studies are evaluating BRAF, NRAS, and KIT for melanoma metastases and BRAF for cerebral metastases [61]. The BRAF mutation is a particularly promising target with 50% of melanoma patients harboring the mutation [56]. There are two known mutation types: 71.9% have a substation of valine with glutamic acid at position 600, BRAF^Val600E, while 22.5% have a substitution with lysine, BRAF^Val600K [56]. There is currently no evidence of a relationship between KIT inhibitors and cerebral metastases [61].

BRAF works by phosphorylating and activating MEK proteins, which then activate MAP kinases. The MAP kinase pathway is responsible for regulating proliferation and survival of tumor cells [130]. BRAF inhibitors are currently the standard of care for patients who have BRAF^V600E-positive metastatic melanoma [54]. Dabrafenib acts as an ATP-competitive inhibitor against BRAF kinase and seems to work similarly on intracranial and extracranial metastases [131]. To what extent penetrates the blood–brain barrier remains unclear. A phase II trial investigated the effects of dabrafenib on patients harboring BRAF^V600E or BRAF^V600K mutations and at least one cerebral metastases between 5 and 40 mm. The patients were divided into two cohorts: cohort A contained 89 patients who had not received any previous treatments and cohort B contained 83 patients who had cerebral metastases with progression after receiving local treatment. The patients were started on dabrafenib 150 mg twice a day until disease progression, intolerable toxicity, or death. Using a modified version of the Response Evaluation Criteria in Solid Tumors (RECIST) , for patients with BRAF^V600E, an intracranial response was found in 39.2% (29 of 74) in cohort A and 30.8% (20 of 65) in cohort B. Two of the patients in cohort A had a complete response to the therapy. The patients who were BRAF^V600K-positive had fewer overall responses, 1 of 15 in cohort A and 4 of 18 in cohort B. The global disease control rate was 80% in BRAF^V600E-positive and 50% in BRAF^V600K-positive patients. In these patients, serum LDH level made an impact on prognosis, with higher levels predicting slightly worsened response, and marginally shorter median progression-free survival and overall survival. The median survival was 33 weeks for cohort A and 31 weeks in cohort B [54].

Another inhibitor of BRAF kinase is vemurafenib. A phase III trial enrolled 371 patients with BRAF^V600 and included patients with poor performance status, to provide vemurafenib to patients in medical need and to evaluate the tumor response. The patients were given oral vemurafenib 960 mg twice a day, and patients were followed up at a median of 2.8 months. The overall response rate was 54%, including 13 of 31 patients who had an ECOG PS of 2 or 3 and who normally would have been excluded from other studies for their poor prognosis. 20% of the patients had seizures; however, these were likely to be pre-existing and not related to therapy [132]. Of note, vemurafenib was only tested in patients with a positive 4800 BRAF^V600 mutation, an allele-specific PCR test for BRAF^V600E. Therefore, unlike for dabrafenib which has been tested in BRAF^V600E and BRAF^V600K, its effectiveness against other mutation types is unknown [61]. There has been some evidence of patients with cerebral metastases on either dabrafenib or vemurafenib with a rare BRAF^V600R mutation responding to therapy [133].

Interleukin-2 (IL-2) has been used in patients with metastatic melanoma with some success; however, its use for cerebral metastases has been limited due to concerns of cerebral edema, intracerebral hemorrhage from thrombocytopenia, and neurotoxicity [61, 134].

Ipilimumab is a human IgG-1 monoclonal antibody that blocks cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) , an inhibitor of T-cell activity. A phase 3 study was done to determine whether ipilimumab with DTIC improved overall survival. This study found significant improvement in survival; however, patients with cerebral metastases were excluded from the trial [135]. To further investigate this, a phase 2 study evaluated the treatment of ipilimumab in patients with evidence of cerebral metastasis. This study gave patients four days of ipilimumab 10 mg/kg/day every three weeks; patients stable at 24 weeks received additional doses every 12 weeks. The study enrolled 72 patients, dividing them into 51 patients without neurological symptoms (cohort A) and 21 patients who were symptomatic and receiving steroids (cohort B). After 12 weeks, 18% of the patients in cohort A and 5% in cohort B had disease control. The median survival in cohort A was 7 months, with two-year survival rate of 25%, while cohort B had a median survival of 3 months and a two-year survival of 10% (see Table 29.1 for further results from this study [136]).

To investigate whether chemotherapy-induced release of tumor antigens could amplify the effect of ipilimumab, a phase II study enrolled 86 patients with metastatic melanoma, 20 of which had asymptomatic brain metastases, for treatment with the combination of ipilimumab with fotemustine. Patients were given 10 mg/kg of ipilimumab every three weeks a total of four times in addition to fotemustine IV 100 mg/m² weekly for three weeks followed by once every three weeks from week 9 to week 24. Patients who had a clinical response were eligible for maintenance therapy every 12 weeks with ipilimumab from week 24 and every 3 weeks with fotemustine. By the end of the study, 46.5% of the patient achieved disease control, including 50% of those with brain metastases. The improved survival was theorized to be secondary to chemotherapy-induced release of tumor antigens amplifying the effect of ipilimumab [126].

The effect of radiotherapy on the tumorigenicity of ipilimumab via the potential release of tumor antigens has also been investigated. In this study, the median survival of patients who received ipilimumab in addition to radiotherapy was 18.3 months compared to 5.3 months (with radiotherapy alone). In addition, patients who underwent radiotherapy first had an overall survival of 18.4 months, while patients who received ipilimumab first had a survival of 8.1 months. Furthermore, four of the ten patients who received ipilimumab before radiotherapy had a partial response to the radiotherapy, while only 2 out of 22 who did not receive ipilimumab had a partial response to the radiotherapy [137].