Local control of melanoma brain metastases treated with stereotactic radiosurgery

Abstract

Purpose

To examine the effectiveness of stereotactic radiosurgery (SRS) for melanoma brain metastases, as the optimal management is unknown.

Materials and Methods

Patients with melanoma brain metastases treated between 1999 and 2013 with SRS as initial management were reviewed. Local control (LC), intracranial progression free survival, and overall survival were evaluated using the Kaplan Meier analysis and logistic regression.

Results

185 patients were identified with 435 treated brain metastases. 76% of metastases were controlled, with a median freedom from local failure of 23.4 months. Higher SRS dose (p=0.001) and smaller tumor volume (p=0.0007) were associated with improved LC on univariate analysis, but on multivariate analysis only smaller tumor volume remained significant (p=0.047). At analysis, 7.6% of patients were alive and the median time to death after SRS was 7.8 months.

Conclusions

SRS is an effective primary treatment for melanoma brain metastases. There was no benefit combining SRS and surgery or whole brain radiotherapy.

1. INTRODUCTION

Although melanoma accounts for less than 5% of skin cancers in the United States, it accounts for the majority of deaths from skin cancer [1]. Melanoma has a high propensity to metastasize and is the third most common cancer causing brain metastases [2-4]. The incidence of central nervous system disease in patients with metastatic melanoma is reported as 46% to 75% [5]. Overall survival (OS) once brain metastases are identified is typically less than one year, depending on patient age, performance status, number of metastases, and extent of systemic disease [6, 7]. Historically, patients with metastatic melanoma to the brain have had a survival of a few months [5]. However, more recent data shows the potential for prolonged survival [8-10].

Currently, brain metastases are treated with surgery, stereotactic radiosurgery (SRS), whole brain radiotherapy (WBRT) or a combination of these modalities [11-13]. In retrospective analyses, SRS alone for brain metastases has been reported to provide local control (LC) of up to 84% at one year and 77% at two years [14]. Retrospective [15] and randomized data [16, 17] show that there is no survival benefit when adding WBRT to SRS treatment, although patients who receive SRS alone are more likely to develop distant intracranial relapse requiring salvage therapy [16, 17]. Surgical resection followed by adjuvant SRS to the operative bed is also highly effective, with reported LC of 88% and 79% at six and twelve months respectively [18]. Importantly, these studies included patients with brain metastases from any cancer type, therefore optimal management for patients with melanoma brain metastases is not known.

Melanoma has classically been termed “radioresistant,” although retrospective studies of SRS for metastatic melanoma to the brain have demonstrated relatively durable outcomes [19-28]. However, many of the published retrospective series on SRS for melanoma brain metastases include small numbers of patients and treated tumors [19-22, 24, 27, 28]. One of the largest series out of Pittsburgh shows LC for melanoma metastases after SRS of 73%, although SRS was not always the first intervention for intracranial disease and many patients had prior WBRT or surgery [25]. Given the limited retrospective data and lack of prospective data on SRS used as first line therapy for patients with metastatic melanoma to the brain, we sought to review our institutional experience over the past 14 years.

2. MATERIALS AND METHODS

Our institutional, IRB-approved database was queried for all patients with a diagnosis of melanoma treated with SRS for brain metastases between 1999 and 2013, providing at least one year follow up. The primary intervention was limited to the first course of intracranial radiotherapy, including SRS with or without WBRT. Subsequent SRS treatment courses were excluded from the analysis. Patients with prior intracranial radiotherapy or short follow up were excluded.

Patients were treated using the Novalis Linear Accelerator with a BrainLab mask or a headframe for immobilization. The target volumes were treated in a single fraction, without a margin expansion, using a dynamic conformal arc technique with 6 MV photons. The process of simulation and treatment was standard for all patients; including the use of a BrainLab mask or a headframe for immobilization, a computed tomography (CT) scan for treatment planning, and a spoiled gradient echo thin slice magnetic resonance imaging (MRI) scan that was fused to the CT. Treatment volumes were delineated and measured in three dimensions, and planning was performed using the Brainscan or iPlan system (BrainLab; Munich, Germany). Our institutional practice is to select the prescription isodose line that covers at least 95% of the target volume, with 99% of the target volume covered by at least 95% of the prescription dose [29]. Dose prescription was determined by volume and size of the tumor, following the guidelines outlined by the RTOG 9005 study [30]. Small, medium and large lesions were defined by volume; 0-4.1888 cc, 4.1889-14.1372 cc, and 14.1373-33.51 cc, respectively. Small, medium, and large lesions were treated in a single fraction to 24 Gy, 18 Gy, and 15 Gy, respectively. The dose was occasionally reduced for those in close proximity to each other or normal structures. WBRT was delivered using opposed laterals with 6 MV photons. Patients who received WBRT had a median prescription dose of 3750 cGy, with a range from 900-3750 cGy. One patient did not complete the prescribed dose of 3000 cGy and stopped treatment at 900 cGy due to clinical decline.

Routine patient follow up included a history and physical examination by the treating physician, as well as a diagnostic MRI every 3 months, sooner if symptoms developed. Follow up imaging was retrospectively examined to determine the volume of tumor or operative bed, as measured in three dimensions and coded as either stable/decreased or failed/progressed based on growth of more than 25% in cross-sectional measurement. This information was correlated with the radiology report and the clinical follow up notes. If a discrepancy was encountered between the measurement and the radiology report or the physician interpretation, the default was to follow the treating physician’s interpretation. LC was determined as the time between SRS and last follow-up or date of failure. OS after SRS was determined using the last follow up or date of death from any cause. Patients were censored at last follow up if the endpoint was not reached. The use of WBRT was captured in the upfront setting as part of initial management, either before or after SRS treatment, with a median of 2.6 months between WBRT and SRS for those patients that received both. Patients who received WBRT for salvage were identified. The time to intracranial progression was determined as the time between initial SRS and intraparenchymal failure outside of the treated area. Subsequent metastases were treated at the discretion of the physician with either surgery, WBRT or SRS. Subsequent SRS treatments were not included in this analysis.

Statistical analysis was performed using MedCalc and Stats Direct. The Kaplan Meier method was used to determine the time to endpoint (LC, OS, and freedom from distant failure in the brain) and logistic regression analyses were performed to examine predictors of LC. A p value of <0.05 was considered significant.

3. RESULTS

3.1. Patient Population

A total of 185 patients were identified that met the inclusion criteria; patient demographics are displayed in Table 1. Eighty-three percent of patients received SRS alone (22 were surgical beds); the remaining 17% were treated with WBRT and SRS (3 were surgical beds) upfront as a planned part of therapy for their brain metastases. Of those patients who did not receive WBRT upfront, 28% received subsequent WBRT for salvage therapy. Sixty-six percent of patients that had SRS upfront had at least one subsequent intracranial therapy, including surgery, SRS or WBRT within a median of 4 months after initial SRS treatment.

Table 1.

Patient Demographics.

	Percent (number)
GPA Score
1	4% (8)
2	23% (43)
3	31% (57)
4	42% (77)
Sex
Male	70% (130)
Female	30% (55)
KPS at Consultation
100	18% (34)
90	53% (98)
80	17% (32)
70	9% (16)
60	1% (1)
N/A	2% (4)
WBRT
None	83% (154)
*salvage WBRT	*28% (51/154)
WBRT with initial treatment	17% (31)
Age
Mean	55 years
< 65 years old	68% (125)
> 65 years old	32% (60)
Number of metastases at intracranial diagnosis
Single	54% (100)
Two	17% (31)
Three or more	29% (54)
Median survival following first SRS treatment	7.8 months
Median survival after intracranial diagnosis	8.5 months
Median time to second intracranial treatment (surgery, SRS, WBRT) after SRS alone	4.1 months
Median time to WBRT after SRS alone	17.4 months

[GPA – graded prognostic assessment, KPS – Karnofsky Performance Status, N/A – not applicable/no information, WBRT – whole brain radiotherapy, SRS – stereotactic radiosurgery]

Forty two percent of patients had a graded prognostic assessment (GPA) score of 4, 31% a GPA score of 3, 23% a GPA score of 2, and 4% a GPA score of 1. Fifty-four percent of patients had a single metastasis at the time of diagnosis of intracranial metastatic disease, 17% had 2 metastases, and 29% had 3 or more metastases. The majority of treatments were delivered to intact metastases (92%), and 8% of treatments were operative beds after surgical resection. The median number of sites treated per session was 3 (range 1-12).

In total, 435 treated metastatic sites were analyzed; clinical characteristics are shown in Table 2. The median tumor volume was 0.5 cc (range 0.01-36.2 cc), and the median prescription dose was 2000 cGy (range 1200-2927 cGy). The initial therapy for 75% of metastases was SRS alone and for 25% was with SRS in combination with WBRT.

Table 2.

Clinical Characteristics of Treated Tumors.

	Percent (number)
Intact Metastasis	92% (400)
Surgical Bed	8% (35)
Metastases treated with:
SRS alone upfront	75% (327)
SRS + WBRT upfront	25% (108)
Mean Prescription Dose	1979.02 cGy (Range 1200-2927 cGy)
Mean Treatment Volume	2.41 cc (Range 0.01-36.24 cc)
Treated Metastases
Controlled	76% (330)
Failed	24% (105)

[WBRT – whole brain radiotherapy, SRS – stereotactic radiosurgery]

[GPA – graded prognostic assessment, KPS – Karnofsky Performance Status, N/A – not applicable/no information, WBRT – whole brain radiotherapy, SRS – stereotactic radiosurgery]

4. Local Control

After initial treatment with SRS, 76% of metastases were controlled locally and 24% failed. The median time to local failure (TTLF) for an individual metastatic site was 23.4 months (Figure 1A). LC was not affected by age, sex, GPA, prior surgical resection, the number of areas treated at diagnosis of intracranial disease, or the addition of WBRT to the initial SRS treatment (Figure 1B). The median TTLF for small, medium, and large treatment volumes was 33.5, 14.3, and 7.8 months respectively (p=0.06) (Figure 1C). The median TTLF for those treated with doses of 19-24 Gy, 16-18 Gy and 15 Gy or less was 33.5, 18.1, and 8.4 months respectively (p=0.03) (Figure 1D). On univariate analysis (Table 3), LC was significantly associated with a small volume compared to a medium volume (p=0.0007) and higher dose compared with lower dose (p=0.001). On multivariate analysis (Table 3), volume remained a significant predictor of LC when comparing small volume to medium volume (p=0.047), although dose did not remain significant (p=0.2).

Figure 1.

Local Control (LC)

A) LC for all treated tumors with a median time to local failure (TTLF) of 23.4 months.

B) LC for tumors treated with whole brain radiation (WBRT) and stereotactic radiosurgery (SRS) upfront or SRS alone upfront. Median TTLF for WBRT and SRS upfront was 14.3 months, and 23.4 months for SRS alone, p=0.2.

C) LC stratified by tumor volume. Median TTLF for small lesions (blue) measuring 0-4.1888 cc was 33.5 months, medium lesions (red) measuring 4.1889-14.1372 cc was 14.3 months, and large lesions (orange) measuring 14.1373-33.51 cc was 7.8 months, p=0.06.

D) LC by SRS dose. Median TTLF for tumors treated to 19-24 Gy (blue) was 33.5 months, for those treated to 16-18 Gy was 18.1 months and for those treated to 15 Gy or less was 8.4 months, p=0.03.

Table 3.

Predictors of Local Control.

Variable	Bivariate Analysis			Multivariate Analysis
	OR	95% CI	p value	OR**	95% CI	p value
Age >65 years Age <65 years	1 0.9793	0.6048, 1.5857	0.9323	1.1306	0.6604, 1.9356	0.6544
Single Metastasis Multiple Metastases	1 0.7409	0.4634, 1.1847	0.2104	2.1480	0.7311, 6.3111	0.1644
Intact Metastasis Surgical Bed	1 1.2842	0.5954, 2.7699	0.5236	0.5251	0.2047, 1.3475	0.1803
SRS Dose 19-24 Gy SRS Dose 16-18 Gy SRS Dose <15 Gy	1 1.5529 3.2408	0.8917, 2.7044 1.5805, 6.6452	0.1199 0.0013	1.0987 1.9588	0.4894, 2.4667 0.7001, 5.4806	0.8196 0.2003
Small Tumor Volume Medium Tumor Volume Large Tumor Volume	1 2.6525 2.5845	1.5132, 4.6496 0.8913, 7.4939	0.0007 0.0804	2.2479 1.9482	1.0123, 4.9918 0.4947, 7.6715	0.0466 0.3402
WBRT + SRS upfront SRS alone upfront	1 0.8800	0.5335, 1.4516	0.6166	0.8613	0.4406, 1.6837	0.6625

^*Results reflect an analysis modeling failure.

^**multivariate odds ratios derive from a model that includes all of the variables shown, with additional adjustment for GPA, sex and number of metastases.

[OR odds radio, CI confidence interval, GPA graded prognostic assessment, SRS stereotactic radiosurgery, WBRT whole brain radiotherapy, small tumor volume = 0-4.189 cc, medium tumor volume = 4.19-14.137 cc, large tumor volume = 14.138-33.51cc]

5. Overall Survival

At the end analysis, 14 patients were alive with a median follow up of 35.9 months (Range 17 143 months). The median OS after initial SRS was 7.8 months (Figure 2).

Figure 2.

Overall Survival (OS) for all treated patients. Median OS was 7.8 months.

6. Intracranial Progression Free Survival

Overall, 75% of patients had intracranial failure outside the treated region with a median time to intracranial progression of 2.9 months. Age, sex, GPA score, or the use of WBRT upfront did not affect progression free survival (PFS). For those patients who were not treated with WBRT upfront, 33% received salvage WBRT, with a median freedom from WBRT of 17.4 months. Overall, of the 154 patients treated with SRS alone upfront, 101 received at least one additional intracranial therapy, either surgery, SRS or WBRT. The median time to second treatment was 4.1 months.

The cause of death was unclear for a majority of patients since many went on hospice for progressive disease or did not have diagnostic imaging studies at end of life. Of the 171 patients who expired in the follow-up period, 19 died of intracranial progression, 30 died of systemic progression, 9 died of both intracranial and systemic progression, and 100 had no specific documentation with regards to cause of death.

7. DISCUSSION

Melanoma is the third most common malignancy causing brain metastases [2-4]. Autopsy studies have shown that the incidence of brain metastases in patients with metastatic melanoma is upwards of 75% [5]. Once metastatic intracranial disease is diagnosed, survival is reported anywhere from 2 to 7 months [2, 28], and intervention with surgery or radiotherapy improves outcomes [2, 12, 15, 17-21, 24, 25, 27-29, 31-37]. Patients are offered surgery, radiotherapy, or the combination for a new diagnosis of brain metastases [13].

Recently, the treatment of brain metastases has evolved, although the optimal paradigm is still controversial. In the 1980s, a randomized study enrolled patients with a single brain metastasis, demonstrating that patients treated with surgery and WBRT had improved median survival and decreased local recurrence compared to WBRT alone [37]. Another randomized study from the 1990s, with similar enrollment criteria, demonstrated that WBRT following gross total resection reduced recurrence rates at the treated site and elsewhere in the brain [36]. While the adoption of WBRT in addition to surgery delayed death from neurologic causes compared to surgery alone, there was no survival benefit [36]. In addition, WBRT has been shown to result in detrimental neurocognitive side effects [38] and a reduction in quality of life [16]. Therefore, investigators have attempted to identify a cohort of patients in whom WBRT can be omitted, and SRS alone can be utilized in the definitive setting. A dose escalation study in the 1990s used SRS alone in patients who failed prior WBRT and defined the maximum tolerated doses for the treatment of brain disease, reporting a LC rate of 52% for all treated patients [30]. Retrospective analyses of patients treated with SRS alone have quoted excellent LC at one year of 91% [39], and there is a cohort of patients who will achieve long-term survival after SRS alone for their intracranial disease [34]. A large retrospective analysis of patients with 1-3 brain metastases and RPA class 1-2 showed improvement in OS for patients after SRS alone compared to WBRT [40]. In the adjuvant setting, retrospective series have shown that SRS alone to the resection bed provides excellent LC [18, 41, 42]. A randomized trial published by the European Organization for Research and Treatment of Cancer (EORTC) explored the use of surgery or SRS upfront with or without WBRT for patients with 1-3 brain metastases and the combined use of SRS and WBRT provided superior LC [17]. The next best outcomes, from the use of SRS alone or surgery and WBRT, were equivalent, although inferior to that of SRS and WBRT [17]. Despite this level one evidence, the National Comprehensive Cancer Network (NCCN) Guidelines’ category 1 recommendation, “supported by high level evidence and uniform consensus” [13], recommends surgery and WBRT for patients with 1-3 metastases, and SRS and WBRT for patients with a single metastasis [13]. The category 2A recommendation is for SRS alone, suggesting lower level evidence and uniform consensus [13]. Even among experts, there is still controversy as to the optimal treatment for intracranial metastatic disease.

Melanoma has been classically defined as “radioresistant,” demonstrated by in vitro studies of melanoma cell lines compared to Chinese hamster ovary cell lines [43]. In an attempt to improve outcomes in these patients, dose escalation has been studied with the hypothesis that a higher dose per fraction improves radiation response. As early as the 1980s, twelve patients with “radioresistant” tumors were treated with external beam radiation for brain metastases with doses ranging from 20 to 30 Gy, and no recurrences were observed [21]. At MD Anderson Cancer Center higher doses per fraction were used in the adjuvant treatment of patients with cutaneous melanoma of the head and neck, and LC rates were higher than had been reported using conventional fractionation [44]. In the setting of recurrent or metastatic melanoma to the skin, viscera, mucosa or lymphoid tissue, higher dose per fraction correlated with an improved response [45]. This data can be extrapolated to intracranial metastatic disease, suggesting that higher doses per fraction may improve outcomes for patients with central nervous system metastases from melanoma. SRS allows for the delivery of focal, high doses of radiotherapy, as opposed to WBRT, which typically is delivered in a fractionated course with relatively low doses per fraction. Recent retrospective studies of patients with metastatic melanoma to the brain treated with SRS have reported LC rates from 49 to 94% [9, 19, 20, 22, 24-27, 44], and our previously published institutional review of patients with metastatic melanoma and renal cell carcinoma treated with SRS alone for intracranial disease demonstrated LC of 63.6% [28].

In our current analysis, LC was 76% for metastatic melanoma patients after SRS upfront, and this outcome is consistent with previously published retrospective data [9, 19, 20, 22, 24-27, 44]. Our local failure rate at the initial site of treatment was 24% after SRS alone and 26% after SRS and WBRT. This is similar to the outcomes of the randomized EORTC study for all histologies, reporting a failure rate at the initial site of treatment of 31% after SRS alone and 19% after SRS with WBRT [17]. Interestingly, in our analysis, there was no significant LC benefit with the addition of WBRT to SRS in the upfront setting, with a median TTLF with SRS alone of 23.4 months compared 14.3 months with SRS and WBRT. This finding suggests that WBRT may not be as effective for melanoma as it is for other histologies, and the utility of WBRT in addition to SRS for melanoma brain metastases is the subject of an ongoing randomized study in Australia [46]. In our study, WBRT did not affect PFS in the upfront setting. This could be explained by the fact that this group of patients had multiple intracranial metastases at diagnosis, lower GPA scores, and the majority of our patients had died at the time of analysis. This supports the hypothesis that SRS alone is the optimal initial therapy for patients with melanoma brain metastases as opposed to the combination of SRS and WBRT. Treating initially with SRS alone reserves the use of WBRT for salvage treatment for progression or high volume disease, and in our analysis, patients had a median freedom from WBRT of 17.4 months

This analysis was limited due to the retrospective nature of the study, the relatively short follow up time for some patients, and the short median survival. Strengths of the study include the large number of patients and treated tumors. Another limitation is that we did not examine the impact of systemic therapy, which could independently affect outcomes. In recent years there have been many publications, both prospective and retrospective, examining the safety and efficacy of new systemic agents for metastatic melanoma on brain metastases [47-50]. While some of the data is conflicting [47, 48], there have been some reports of increased toxicity with these agents and SRS [49, 50] and the EORTC recently published guidelines for the use of BRAF inhibitors concurrently with radiotherapy [51]. We await longer follow up to determine whether these agents have activity in the brain independent of radiotherapy, and data on the effects of the combined treatments is becoming increasingly important.

As new systemic agents continue to emerge for the treatment of metastatic melanoma showing improved survival [23], LC of intracranial metastases and freedom from WBRT will become more important outcome measures. WBRT can cause a detriment to neurocognition and quality of life [16, 38], and sparing patients this side effect should be an important consideration when discussing treatment options for newly diagnosed brain metastases from melanoma, especially as survival increases. Newer systemic agents are hypothesized to have action in the central nervous system [23], and may work in conjunction with SRS. The combination of improved systemic agents reaching the brain and high dose per fraction SRS could improve intracranial disease control even further. Our institutional experience provides evidence that patients with multiple intracranial metastases from melanoma can have prolonged LC when treated with SRS upfront.

8. CONCLUSIONS

Although there are many treatment options for patients with metastatic melanoma and limited intracranial metastases including surgery, WBRT, SRS, and a combination of these modalities, the optimal management is not known. In our institutional experience, SRS alone provides excellent LC and the addition of WBRT does not provide a LC, OS or PFS advantage in these patients. The population in which SRS alone is appropriate may expand as systemic therapies improve and treatments emerge with the potential to stimulate a response in the central nervous system. The use of SRS for upfront management of limited intracranial metastases from melanoma should be discussed with all patients.

Glossary

Nomenclature

SRS

stereotactic radiosurgery

LC

local control

OS

overall survival

WBRT

whole brain radiation therapy

CT

computed tomography

MRI

magnetic resonance imaging

GPA

graded prognostic assessment

Gy

gray

cGy

centigray

TTLF

time to local failure

PFS

progression free survival

EORTC

European Organization for Research and Treatment of Cancer

NCCN

National Comprehensive Cancer Network