Management of Melanoma Brain Metastases
Jack M. Qian and Veronica L. Chiang
Among all the different cancer primaries, brain metastases occur most commonly in patients with metastatic melanoma. Neurologic symptoms often range from headache to focal neurologic deficits to seizure. Not only have melanoma brain metastases (MBM) reportedly been found in 40% of patients during their disease course, but autopsy reports suggest an even higher incidence of up to 75% (1). Today, the prevalence of MBM is increasing, likely both because of improved imaging techniques that can detect brain metastases earlier, while still asymptomatic, as well as improved systemic therapies that have increased patient survival (2). However, of all visceral metastases from melanoma, brain metastases historically have carried the worst prognosis, with high rates of morbidity and median survival of 1 month if brain metastases are not treated (3).
TRADITIONAL MANAGEMENT OF BRAIN METASTASES
Whole Brain Radiation Therapy
Radiation in the form of whole brain radiation therapy (WBRT) has been the mainstay of treatment for brain metastases since the 1970s (4). In this era, imaging techniques were less sensitive, and most patients were found to have brain metastases only after the development of neurologic symptoms. Radiation, in conjunction with corticosteroids, proved to be quite effective at ameliorating these symptoms in several retrospective series, with symptom improvement reported in about 60% of cases (5,6). Patients treated with WBRT also appeared to live longer compared with patients managed by supportive care only, with an increased median survival of 3 to 6 months (7). Given the lack of alternative treatment options, adoption of WBRT grew quickly.
Despite obvious differences between brain metastases of different histologic types, much of the literature continues to address the management of brain metastases collectively as a single entity. Only more recent studies have now begun to consider specific subtypes individually for management. Compared to most other cancer types, melanoma has a higher propensity to produce multiple brain lesions (8), and brain metastases from melanoma often hemorrhage (9). Because of the larger number of metastatic lesions and higher rate of symptomatology, WBRT remains the treatment of choice for these patients at many institutions.
Unfortunately, melanoma, along with renal cell carcinoma and sarcoma, has traditionally also been considered to be relatively resistant to radiation. In vitro studies of melanoma cell lines have shown a large shoulder on the radiation survival curve, suggesting a high capacity to repair DNA damage (10,11). As expected, studies looking specifically at MBM patients showed median survival ranging from only 2.2 to 3.4 months with WBRT, with no clear evidence that WBRT actually increased survival compared with treatment with corticosteroids alone (12,13). Moreover, in one study, MBM was found to have a 0% objective response rate (ORR) after WBRT (i.e., in no cases did MBM shrink or disappear completely), whereas other histologies had ORR ranging from 46% to 81% following radiation (14). Such findings have contributed to the widespread belief that melanoma is inherently radioresistant to conventionally fractionated radiation. Attempts to increase the efficacy of WBRT in MBM by combining it with other central nervous system (CNS)-active therapies, such as temozolomide and thalidomide, have also been disappointing (15).
In addition, the most significant side effect of WBRT is long-term neurocognitive decline in the areas of memory, executive functioning, and processing speed. The risk of developing significant neurocognitive compromise after WBRT increases over time, often not manifesting until several months or years have passed (16). Given the historically short survival of patients with brain metastases, these sequelae were not well studied as most patients did not live long enough to develop them. However, with recent improvements in systemic therapies resulting in longer survival of patients, the neurocognitive effects of WBRT have gained increasing attention. In an increasing number of centers today, WBRT is being deferred initially in favor of other treatment modalities (to be discussed later), whenever possible.
The role of surgery in the management of brain metastases was controversial until two key studies were published in the 1990s. In the first study, 48 patients with a single brain metastasis were randomized to resection followed by WBRT, or biopsy followed by WBRT (17). The patients who underwent surgical resection had increased survival, decreased local recurrence, and an overall improved quality of life. A second study by the same group examined whether WBRT could be omitted. They found no difference in survival with the omission of WBRT, but WBRT decreased tumor recurrence anywhere in the brain and decreased the likelihood that patients died for neurologic reasons (18). These studies established surgical resection followed by consolidative WBRT as the standard of care for patients with good performance status and a single brain metastasis. It should be noted once again, though, that few studies examined the role of surgical management for MBM specifically, and practice guidelines continue to be extrapolated from studies evaluating brain metastases in general. The key indications for surgery in MBM therefore continue to include (4)
1. Surgical resection of large, symptomatic, single, or surgically accessible lesions in order to decompress associated mass effect and, therefore, provide relief of symptoms.
2. To provide tissue when diagnosis of MBM is not clear.
Nevertheless, the role of surgery in the management of multiple brain metastases remains poorly established, and several additional factors have kept surgery an unattractive option in many cases (4). These include the inherent morbidity and mortality associated with surgery, the greater delay to systemic treatment when compared with radiotherapy, and the increasing availability of equivalent alternative therapy (e.g., stereotactic radiosurgery [SRS]). Moreover, with improved imaging techniques and increased screening of patients for cranial metastases, more patients are being identified with small, asymptomatic multiple metastases.
It became increasingly clear in the 1990s that patients with brain metastases represented a heterogeneous population, not only because of different primary tumors but also because within each tumor type there seemed to be differences in tumor behavior. Gaspar et al. (19) were the first to perform a recursive partitioning analysis (RPA) on a cohort of 1,200 patients treated with WBRT in order to generate subclassifications of patients with different prognoses and therefore determine in whom it was appropriate to treat with WBRT. Class 1 patients had the best prognosis (median survival 7.1 months) and included those younger than 65, with a Karnofsky performance status (KPS) of at least 70, a controlled primary tumor, and no other extracranial disease. Class 3 patients, on the other hand, had the worst prognosis (median survival 2.3 months) and were those with a KPS of less than 70. All other patients were considered Class 2, with a median survival of 4.2 months. Notably, the vast majority of patients in this analysis had lung or breast cancer, and while the primary site was included as a variable in their analysis, it was not part of the final prognostic model. Subsequent analyses have validated the RPA in MBM patients (20); however, it has become increasingly evident that a large number of patients fall in the Class 2 category, which likely remains quite heterogeneous and requires further stratification (4).
Since then, several other classification systems have been created including the Score Index for Radiosurgery (SIR), the Basic Score for Brain Metastases (BSBM), the Rotterdam system, the Golden Grading System, and the 2 Rades classification (21). Most of these scales are based on clinical characteristics such as performance status, number of brain metastases, and presence and extent of extracranial disease, but again group all types of brain metastases together. The only prognostic scoring system that examines brain metastases by tumor type is the Diagnosis-Specific Graded Prognostic Assessment (DS-GPA) (22). Using this index, not only has it been shown that different clinical factors need to be used in prognostication for different tumor types but also that genetic mutational subgrouping also affects prognosis (23,24). Unfortunately, the only clinical factors that have been reported to impact prognostication for MBM patients are KPS and number of brain metastases at the time of diagnosis (22). Based on a score from 0 to 4.0, median survival of the entire melanoma cohort (patients treated between 1985 and 2007) was 6.7 months, but ranged from 3.4 months for a poor DS-GPA score (0–1.0) to 13.2 months for the best DS-GPA score (4.0) (25). What remains unclear is the impact of this scoring on decision making about treatment of MBM patients, as anecdotally it has been possible to see significant changes in KPS in response to newer systemic therapies such as BRAF inhibitors.
MANAGEMENT IN THE ERA OF STEREOTACTIC RADIOSURGERY
SRS was originally created as a method for treating patients with high-dose radiation in a single session with neurosurgical precision both in targeting to a specific volume and with delivery with sharp dose-fall off outside the target (26). Although radiosurgery can be delivered using many different radiation machines today, the Gamma Knife machine remains the gold standard for the delivery of radiosurgery because of its minimum number of moving parts, and the continued use of headframe immobilization allowing for submillimeter accuracy of targeting. Although the technique was initially developed in the 1950s and 1960s for the treatment of functional disorders, it was not until the late 1980s when SRS was first used for treating brain metastases (27).
Following initial promising retrospective studies, prospective studies were able to demonstrate that the addition of SRS to WBRT-treated lesions resulted in
1. Significantly increased duration of local lesional control independent of previously assumed tumor type radioresistance (28).
2. More rapid and durable lesional response and therefore better neurologic outcome with less use of steroids.
3. Increased overall survival in patients with single metastases (29,30).
In addition, in patients who survived long enough to develop new brain metastases after upfront WBRT, SRS was found to be a safe and effective salvage treatment (31). Given the success with treating new brain metastases with SRS alone after WBRT failure, it was then questioned whether SRS could be used alone without WBRT for the initial management of a limited number of brain metastases. Prospective studies have since shown that while the addition of WBRT following SRS does decrease distant CNS recurrences, it does not confer an overall survival benefit (32,33).
Controversy over the use of SRS as initial management for patients with more than four brain metastases therefore remains (4). Two studies, however, have moved the field toward the use of SRS alone without WBRT. The first was a randomized study by Chang et al. (34) that showed not only significantly worsened neurocognitive function at 4 months after WBRT than SRS alone (52% decline after WBRT, 24% decline after SRS), but also significantly decreased survival in those being treated with WBRT (29% at 4 months after WBRT, 13% after SRS). In addition, the largest retrospective study published to date was recently released by a Japanese group showing that in SRS-treated patients, survival is independent of the number of brain metastases, as long as all brain metastases are treatable with SRS (35).
Given this new data and the knowledge that the high doses of radiation that are delivered by SRS can overcome the previously noted radioresistance of melanoma cells, radiosurgery has become the standard of care treatment for patients with MBM, with WBRT now being used predominantly for salvage cases, or cases in which the number of brain lesions present are too numerous to treat individually (36). In the following sections, results of studies of the use of SRS specifically for MBM will be reviewed.
Initial retrospective series examining the treatment of MBM with SRS were published starting in the 1990s. Comparisons across these studies is difficult, as some included patients who were treated with both SRS and WBRT (37–39), whereas other studies reported the results of patients who were managed with SRS alone, both for recurrent metastases after a prior course of WBRT as well as for the initial management of newly diagnosed CNS disease (40–42). Moreover, these studies did not report local control in a consistent way—some considered local control rates based on the last imaging follow-up for each patient, whereas others specified local control rates at a specified time point after SRS treatment (e.g., 6 or 12 months). Nevertheless, all these studies showed promising results overall, with local control rates of 57% to 97%, and median survival in the range of 7 to 9 months. Although this compared favorably to older studies of WBRT, selection biases may have played a significant role as well, since patients treated with SRS during this era typically only had at most a few brain metastases that were each smaller than 35 mm in diameter. Larger retrospective series were published in the early 2000s, which in general showed similar results regarding local control and overall survival for patients treated with SRS for MBM (43–47). Additional findings from these same studies included the idea that smaller metastases were more likely to show better response to SRS. Therefore, given the high incidence of brain metastases developing among melanoma patients, it has become standard practice in our institution to perform frequent surveillance brain MRIs to screen for new MBMs.
SRS and Ipilimumab
A significant breakthrough in the systemic treatment of metastatic melanoma came in 2011, with the approval of ipilimumab. Unlike traditional cytotoxic chemotherapies, ipilimumab acts instead on the immune system through immune checkpoint inhibition (see Chapter 15). Although the CNS has traditionally been thought of as an immune-privileged site, Wilson et al. (48) were able to show that activated T cells can indeed cross the blood–brain barrier. Hodi et al. (49) also demonstrated that antitumor immune infiltrates were present in the CNS following administration of ipilimumab in the case of a patient with melanoma metastatic to the brain and spine; and a subsequent phase II trial demonstrated that ipilimumab can have similar levels of initial clinical response in the CNS as it does in the rest of the body (50).
Independent of immunotherapy, radiation has also been shown to induce pro-inflammatory responses secondary to modulation of antigen presentation and immune signaling pathways, even when delivered to the CNS (51–53). Therefore, there has been much suggestion that radiation and immunotherapy may be able to act synergistically in the treatment of cancer (54). The first study to examine the impact of ipilimumab on outcomes of patients with MBM treated with SRS was by Knisely et al. (55). Out of a cohort of 77 patients with MBM treated by SRS, 27 patients also received ipilimumab. Patients who received ipilimumab and SRS had a significantly longer median survival of 21.3 months, compared with 4.9 months among those who did not receive ipilimumab. Of particular note, there was no difference in the DS-GPA scores between patients who did and did not receive ipilimumab, and ipilimumab independently improved survival both for patients with a high DS-GPA score (3–4) and a low DS-GPA score (0–2). Other retrospective series published since then have reported similar results. Silk et al. (56) studied a cohort of patients that received either SRS or WBRT for MBM, and found that ipilimumab increased median survival in the patients treated with SRS (19.9 months vs. 4.0 months) but not those treated with WBRT (3.1 months vs. 5.3 months). Shoukat et al. (57) found similarly increased median survival with the addition of ipilimumab (28.0 months vs. 7.0 months), and Tazi et al. (58) noted that among stage IV melanoma patients treated with ipilimumab, those with brain metastases treated with SRS had similar survival to patients without brain metastases. Two studies, however, found no difference in outcomes with the addition of ipilimumab to SRS (59,60), but all these studies did find that combining ipilimumab with SRS was safe, with no clear increases in treatment-related adverse events. What remains unclear from all of these studies is the mechanism leading to the improved survival. Although it is postulated that SRS might mediate an abscopal effect that can be enhanced by immunotherapy (61,62), it is equally likely that SRS-eligible patients have lower burdens of CNS disease, have good control of this disease using SRS, and therefore benefit most from the extended survival expected from treatment of their systemic disease with ipilimumab.
Another question that is not answered by the previously noted studies pertains to the relative timing of the use of SRS and ipilimumab. In most of the previous studies, patients were included if they received SRS and ipilimumab at any time point during their clinical course. It therefore remains unclear when the best time to administer radiation would be relative to immunotherapy, in order to maximize this potential synergism. Preclinical studies have suggested that concurrent treatment is most effective (63), but the few studies noted previously that did address timing only separated treatment as SRS either before or after ipilimumab (55,56). Kiess et al. (64) first studied this issue in more depth in a cohort of 46 MBM patients (64