Impact of Novel Treatments in Patients with Melanoma Brain Metastasis: Real-World Data

Simple Summary Cutaneous melanoma is highly prone to metastasize to the brain, with a historically poor overall survival of only 4–5 months. Over the past decade, novel drugs such as targeted therapies and immune checkpoint inhibitors (ICIs) have revolutionized the treatment of advanced melanoma. However, most prospective studies have excluded patients with melanoma brain metastasis (MBM) or included only patients with stable (steroid-free) or asymptomatic MBM. To assess the impact of novel treatments in patients with MBM in a real-world setting, we studied a cohort of consecutive patients over a period of sixteen years (2005 to 2021) in a large, tertiary referral center for melanoma (Erasmus MC, Rotterdam, the Netherlands). We found a significant improvement in overall survival in the modern (post-2015) era, associated with stereotactic radiotherapy and especially ICIs. These findings support that ICIs, if clinically feasible, should be considered first in clinical practice after a diagnosis of MBM. Abstract Background: Melanoma brain metastasis (MBM) is associated with poor outcome, but targeted therapies (TTs) and immune checkpoint inhibitors (ICIs) have revolutionized treatment over the past decade. We assessed the impact of these treatments in a real-world setting. Methods: A single-center cohort study was performed at a large, tertiary referral center for melanoma (Erasmus MC, Rotterdam, the Netherlands). Overall survival (OS) was assessed before and after 2015, after which TTs and ICIs were increasingly prescribed. Results: There were 430 patients with MBM included; 152 pre-2015 and 278 post-2015. Median OS improved from 4.4 to 6.9 months (HR 0.67, p < 0.001) after 2015. TTs and ICIs prior to MBM diagnosis were associated with poorer median OS as compared to no prior systemic treatment (TTs: 2.0 vs. 10.9 and ICIs: 4.2 vs. 7.9 months, p < 0.001). ICIs directly after MBM diagnosis were associated with improved median OS as compared to no direct ICIs (21.5 vs. 4.2 months, p < 0.001). Stereotactic radiotherapy (SRT; HR 0.49, p = 0.013) and ICIs (HR 0.32, p < 0.001) were independently associated with improved OS. Conclusion: After 2015, OS significantly improved for patients with MBM, especially with SRT and ICIs. Demonstrating a large survival benefit, ICIs should be considered first after MBM diagnosis, if clinically feasible.


Introduction
Cutaneous melanoma has the highest risk of brain metastasis (MBM) of all solid tumors, spreading to the brain in over 50% of patients with metastatic disease [1][2][3]. Overall survival (OS) used to be <5 months, but several novel treatments have become available in the last decade, which have changed the perspectives of patients with MBM [2]. Whole brain radiotherapy (WBRT) and chemotherapy have limited efficacy in MBM, and WBRT is associated with cognitive impairment in the few long-term survivors [4][5][6]. These days, stereotactic radiotherapy (SRT) and especially targeted therapies (TTs; BRAF-MEK inhibitors) and immune checkpoint inhibitors (ICIs; anti-PD[L]1 and anti-CTLA4) have revolutionized treatment for patients with MBM [7][8][9][10][11][12][13][14]. Combined dabrafenibtrametinib showed an intracranial response rate (IRR) of 58%, but with limited duration of response (6.5 months) [7]. More durable responses are reported for ICIs, especially for combined nivolumab-ipilimumab. The ABC trial reported an IRR of 51% with a 5-year OS of 51% [8][9][10], and the CHECKMATE-204 trial had an IRR of 57% with a 1-year OS of 81.5% for this combination [11,15]. However, only a few phase II and III trials included limited numbers of patients with previously treated and/or symptomatic MBM, and these reported considerably lower IRRs (between 6% and 17% for ICIs) [8,11,15,16].
To assess the eligibility of individual patients for these novel treatments, the Melanoma-molGPA is often used in clinical practice as a prognostic scoring tool. This index is based on age, Karnofsky performance status (KPS), presence of extracranial metastasis (ECM), number of MBMs and, more recently, BRAF-mutation status [17][18][19]. However, since the index was derived from a retrospective cohort (n = 823) of patients between 2006 and 2015, its validity for modern cohorts is uncertain [17,20].
In the Netherlands, TTs and ICIs have been implemented in clinical practice for patients with MBM since their reimbursement almost a decade ago [21]. To assess the real-world impact of these treatments on MBM, we studied OS in a cohort of consecutive patients diagnosed with MBM over a period of sixteen years (2005 to 2021). Additionally, we examined the validity of the Melanoma-molGPA index for patients diagnosed with MBM in the modern treatment era.

Patient Selection
This retrospective cohort study was performed at Erasmus MC Cancer Institute (Rotterdam, The Netherlands), a large, tertiary referral center for patients with melanoma. In this center, the treatment plan for patients with MBM is routinely discussed in a local multidisciplinary board which includes experienced medical oncologists, neuro-oncologists, radiotherapists, and neurosurgeons. We identified all consecutive patients with a diagnosis of cutaneous or mucosal melanoma and MBM referred to the Erasmus MC between 1 January 2005 and 1 July 2021 ( Figure A1). The study was approved by the local Institutional Review Board (MEC-2020-0681). The following baseline characteristics were collected: age, sex, KPS, lactate dehydrogenase (LDH) level, BRAF-status, presence of symptoms of MBM (e.g., headache, nausea, epilepsy, focal deficits), number of MBMs, and status of ECM. Diagnosis of MBM was the date of first brain imaging (magnetic resonance imaging or computed tomography (CT)) that confirmed parenchymal MBM. Diagnosis of ECM was the date of first imaging (CT or 2-deoxy-2-[ 18 F]fluoro-D-glucose-positron emission tomography-CT) of at least thorax and abdomen that confirmed metastasis (M1-disease). Melanoma brain metastasis was considered synchronous when diagnosed within one month of diagnosis of ECM and metachronous when diagnosed at least one month after diagnosis of ECM. The Melanoma-molGPA was calculated for each individual patient.

Statistical Analysis
Data cut-off was 4 March 2022. We created a total cohort and two consecutive time cohorts split 1 January 2015, as novel systemic treatments were increasingly prescribed after 2015 in the Netherlands (Table A1, Figure A3). The primary outcome was OS, defined as the time between diagnosis of MBM and death of any cause.
Continuous variables were described with medians (interquartile range (IQR)) and categorical variables with frequencies. The Mann-Whitney U test was used for continuous, and the Chi-squared test for categorical variables. For categorical variables with groups < 10 patients, the Fisher exact test was used. Missing data were omitted from analysis.
Kaplan-Meier (KM) and Cox Proportional Hazards (CPH) methods were used for timeto-event analysis and assessed with the Logrank and Likelihood ratio tests, respectively. For multivariate CPH modeling, backward elimination (i.e., including variables with p < 0.2 in univariate analysis) was applied, and interaction testing was performed. No correction for multiple testing was performed in these exploratory analyses. A two-sided p-value of 0.05 was taken as statistically significant. All analyses were performed using R version 4.0.2 (the R-Project, Auckland, New Zealand).

Baseline Characteristics
We included 430 patients in total; 152 (35.3%) patients were categorized in the pre-2015 cohort and 278 (64.7%) in the post-2015 cohort ( Figure A1). The majority of patients had a follow-up of at least 12 months (420 of 430 patients, 97.6%). Baseline characteristics at diagnosis of MBM are shown in Table 1

Overall Survival
In the total cohort, median OS was 5.  Table A2).
For patients receiving SRT and surgical resection directly after diagnosis of MBM, OS improved after 2015 (Table A3). Systemic treatments were not analyzed over time since their frequencies significantly changed over time ( Table 2). To assess the impact of treatments in the modern era, we analyzed treatment subgroups in the post-2015 cohort (n = 278) only.
In these univariate analyses, a specific treatment before the diagnosis of MBM or directly after was compared to not having that specific treatment at that time. In addition, patients with symptomatic MBM were analyzed separately.

Systemic Treatments Post-2015
In patients with BRAF V600E+/K-mutated melanoma (n = 157), TTs prior to the diagnosis of MBM were associated with a shorter median OS as compared to no prior TTs (2.0 (IQR 0.8-7.1) vs. 10.9 (IQR 5.2-27.0) months, HR2.67, p < 0.001; Figure 4a, Table A4). In patients with prior TTs, 58.3% of patients had also been previously treated with ICIs (Table A5a). Of all patients with TTs directly after diagnosis of MBM (n = 89), 64 (71.9%) had symptomatic MBM ( Figure A5a). No significantly different OS was found between patients with and without TTs directly after MBM diagnosis   Table A2).   For patients receiving SRT and surgical resection directly after diagnosis of MBM, O improved after 2015 (Table A3). Systemic treatments were not analyzed over time sinc their frequencies significantly changed over time (Table 2). To assess the impact of trea ments in the modern era, we analyzed treatment subgroups in the post-2015 cohort (n 278) only.
In these univariate analyses, a specific treatment before the diagnosis of MBM or d rectly after was compared to not having that specific treatment at that time. In addition patients with symptomatic MBM were analyzed separately.

Switching from Targeted Therapy to Immune Checkpoint Inhibition
Here, 18 of 430 (4.2%) patients initiated with TTs directly after diagnosis of MBM and switched to ICIs when their performance status improved and/or imaging showed tumor response at a median time of 6.1 months (IQR 3.9-7.0) after the start of TTs (Table A7). Three of eighteen (16.7%) patients remained stable since the switch to ICIs and were alive at data cut-off, with a median survival time of 25.5 months (IQR 21.7-27.2) since ICI initiation. Three of eighteen (16.7%) patients had progressive disease after ICI initiation but received no further treatment due to poor clinical condition, with a median OS of 11.9 months (IQR 7.6-16.3) since ICI initiation. Twelve of eighteen (66.7%) patients had progressive disease on ICIs and switched back to TTs, with a median OS of 11.6 months (IQR 6.5-25.5) since ICI initiation.

Discussion
Since 2015, novel systemic therapies (TTs and ICIs) and SRT have replaced chemotherapy and WBRT for patients with MBM, resulting in a significantly improved OS. Compared to clinical trials, real-world patients with MBM usually have a worse clinical condition, which is illustrated by the high number of patients with poor KPS and symptomatic MBM in this cohort.
Surgical resection and SRT remained important treatment options after 2015, although their frequency directly after diagnosis of MBM did not significantly increase over time. Especially SRT was associated with a beneficial impact on survival (median OS 30.5

Switching from Targeted Therapy to Immune Checkpoint Inhibition
Here, 18 of 430 (4.2%) patients initiated with TTs directly after diagnosis of MBM and switched to ICIs when their performance status improved and/or imaging showed tumor response at a median time of 6.1 months (IQR 3.9-7.0) after the start of TTs (Table A7). Three of eighteen (16.7%) patients remained stable since the switch to ICIs and were alive at data cut-off, with a median survival time of 25.5 months (IQR 21.7-27.2) since ICI initiation. Three of eighteen (16.7%) patients had progressive disease after ICI initiation but received no further treatment due to poor clinical condition, with a median OS of 11.9 months (IQR 7.6-16.3) since ICI initiation. Twelve of eighteen (66.7%) patients had progressive disease on ICIs and switched back to TTs, with a median OS of 11.6 months (IQR 6.5-25.5) since ICI initiation.

Discussion
Since 2015, novel systemic therapies (TTs and ICIs) and SRT have replaced chemotherapy and WBRT for patients with MBM, resulting in a significantly improved OS. Compared to clinical trials, real-world patients with MBM usually have a worse clinical condition, which is illustrated by the high number of patients with poor KPS and symptomatic MBM in this cohort.
Surgical resection and SRT remained important treatment options after 2015, although their frequency directly after diagnosis of MBM did not significantly increase over time. Especially SRT was associated with a beneficial impact on survival (median OS 30.5 months) and remained independently associated with improved OS. Although the efficacy of SRT has been demonstrated in patients with up to 10 MBMs [22,23], SRT was almost exclusively reserved for patients with <4 MBMs in the current cohort.
The increased use of TTs (10.5% to 33.8%) and ICIs (2.6% to 35.3%) post-2015, directly after diagnosis of MBM, reflects the clinical approval of different novel drugs by the European Medicines Agency (EMA) and their reimbursement in the Netherlands since 2015 [24]. Bander et al. showed an even higher use of ICIs (77%) in their American cohort, likely resulting from the earlier approval of these drugs by the Food and Drug Administration (FDA) and from differences in therapeutic approaches between centers [25,26].
Patients who received TTs or ICIs to treat ECM, before the diagnosis of MBM, had poor survival (median OS of 2.0 and 4.2 months, respectively) as they had secondary resistance with intracranial disease progression on/after these treatments. Other real-world studies confirm the poor OS of patients with systemic treatments prior to the diagnosis of MBM [26,27]. Most importantly, a significantly favorable OS was seen with ICIs directly after the diagnosis of MBM (median OS 21.5 months), even in symptomatic patients (18.5 months). Although median follow-up is limited, the tail of the OS curve (Figure 4d) showed a considerable group of patients with long-term survival, as illustrated by >17% of patients with ICIs who lived >5 years after diagnosis of MBM. This confirms in a realworld setting the efficacy of ICIs in MBM, as reported by the benchmark phase II and III trials [10,15]. Therefore, ICIs should be considered first after diagnosis of MBM, if clinically feasible.
The Melanoma-molGPA, based on an MBM cohort between 2006 and 2015, remained a valid prognostic tool in patients diagnosed with MBM between 2015 and 2021 [17,18]. Therefore, clinicians may continue to use this tool to predict prognosis and assess the eligibility of individual patients for specific treatments.
Since TTs can induce rapid tumor responses, these drugs were frequently administered (>70%) to patients with symptomatic MBM and a poor performance status. However, TTs lack durable responses and it was recently demonstrated that first-line nivolumabipilimumab followed by BRAF/MEK-inhibition on progression had the most favorable OS in metastatic melanoma [28]. Nevertheless, in patients with a BRAF V600E+/K-mutation and an initial poor performance status, induction with TTs provides an opportunity for effective ICI treatment after performance has improved [29]. Although our study was not designed to compare treatment strategies, we assessed the outcome of 18 patients who were first treated with induction TTs after the diagnosis of MBM, followed by ICIs. Only three (16.7%) patients benefitted from the switch to ICIs, with stable disease and a median OS of 25.5 months since ICIs. Although (pre-)clinical studies have shown that an immune-resistant phenotype might arise after progression on BRAF/MEK-inhibition, it is unknown whether the switch to ICIs before progression on TTs could be effective, and it is worthwhile to further investigate this prospectively [8,30].
This study was designed to assess the potential survival gain in the brain for patients with MBM in a real-world setting after the introduction of novel systemic treatments. The retrospective design is a limitation, as is the lack of information on steroid use. The inclusion of a single center might provide additional bias. However, this tertiary center receives referrals from a large region in the Netherlands, and each consecutive patient is treated and followed in this center until end-of-life care or death. Therefore, this data set provides detailed information over a large period of time, fully capturing the changing treatment landscape.

Conclusions
Overall survival has improved for patients with MBM after 2015 and is associated with the use of SRT and ICIs directly after diagnosis of MBM. Immune checkpoint inhibitors (ICIs) showed an important survival benefit and should be considered first after the diagnosis of MBM, if clinically feasible. The Melanoma-molGPA remains a valid prognostic tool for clinicians in the modern treatment era for patients with MBM. To further improve the prognosis of patients with MBM, future research should focus on optimizing treatment sequencing, such as switching from TTs to ICIs, especially in patients with an initial poor performance.   Table A1. Yearly incidence of patients with newly diagnosed melanoma brain metastasis (MBM) referred to the Erasmus MC, and yearly treatment prescriptions (at any time after MBM diagnosis).  Table A1. Yearly incidence of patients with newly diagnosed melanoma brain metastasis (MBM) referred to the Erasmus MC, and yearly treatment prescriptions (at any time after MBM diagnosis).

Immune
Detailed Description Table A6: Switching from TT to ICI Treatment Eighteen of four hundred and thirty (4.2%) patients of the total cohort, all with a diagnosis of melanoma brain metastasis (MBM) post-2015, initiated with targeted therapies (TTs) directly after diagnosis of MBM and later switched to immune checkpoint inhibitors (ICIs), after a generally improved clinical condition and/or improving disease status on imaging (Table A6). In 8 of these 18 (44.4%) patients, TTs were combined with surgical resection plus radiotherapy (SRT n = 2, WBRT n = 1), surgical resection alone (n = 2), and radiotherapy alone (SRT n = 2, WBRT n = 1). Median time between diagnosis of MBM and the start of TTs was 0.76 months. Patients switched to ICIs after a median time of 6.11 months (IQR 3.9-7.0) since the start of TTs. In all 18 patients, TTs were discontinued within one week prior to the start of ICIs.
In three (16.7%) patients, the switch from TTs to ICIs was successful since they had stable disease since ICI initiation until at least data cut-off. These three patients had a median survival time of 25.5 months (IQR 21.7-27.2) since ICI initiation.
Thirteen of eighteen (72.2%) patients had intracranial disease progression after ICI initiation, and four of eighteen (22.2%) patients had extracranial disease progression. Ultimately, 12 of 18 (66.7%) patients switched back from ICIs to TTs at a median time of 2.6 months (IQR 2.5-4.8) since ICI initiation. These 12 patients had a median OS of 11.6 months (IQR 6.5-25.5) since ICI initiation.
Three of eighteen (16.7%) patients progressed after ICI initiation, receiving no further treatment due to poor clinical condition: two patients died of progressive ECM (stable MBM) and one patient of combined progressive MBM and ECM. These three patients had a median OS of 11.9 months (IQR 7.6-16.3) since ICI initiation. months (IQR 2.5-4.8) since ICI initiation. These 12 patients had a median OS of 11.6 months (IQR 6.5-25.5) since ICI initiation.
Three of eighteen (16.7%) patients progressed after ICI initiation, receiving no further treatment due to poor clinical condition: two patients died of progressive ECM (stable MBM) and one patient of combined progressive MBM and ECM. These three patients had a median OS of 11.9 months (IQR 7.6-16.3) since ICI initiation.