Purpose:

This study aimed to identify baseline clinical features associated with the outcomes of patients enrolled in the COMBI-MB phase II study of dabrafenib and trametinib treatment in patients with V600 BRAF-mutant metastatic melanoma with melanoma brain metastases (MBM). Exploratory biomarker analysis was also conducted as part of the synergistic COMBI-BRV trial (BRV116521), to identify molecular and immunologic changes associated with dabrafenib in MBMs and extracranial metastases (ECM).

Patients and Methods:

Post hoc analysis was performed for baseline features of patients (n = 125) enrolled in COMBI-MB. Analyses were performed to identify baseline clinical features associated with intracranial response rate (ICRR), progression-free survival (PFS), and overall survival (OS).

Exploratory biomarker analysis was performed on biospecimen collected in the COMBI-BRV trial in which patients with BRAF-mutant, resectable MBM were treated with dabrafenib for 10 to 14 days prior to craniotomy. Accessible ECM were resected or biopsied at the time of craniotomy. Biospecimens underwent molecular and immunologic profiling for comparative analyses.

Results:

In COMBI-MB baseline treatment with corticosteroids was independently associated with lower ICRR [39% vs. 63%; OR, 0.323; 95 % confidence interval (CI), 0.105–0.996; P = 0.049] and shorter PFS (HR, 1.93; 95% CI, 1.06–3.51; P = 0.031). Additional significant associations identified in the multivariate analysis were improved PFS in patients with a BRAFV600E genotype (HR, 0.565; 95% CI, 0.321–0.996; P = 0.048) and improved OS in patients with Eastern Cooperative Oncology Group 0 (HR, 0.44; 95% CI, 0.25–0.78; P = 0.005).

Conclusions:

Corticosteroid treatment was associated with reduced ICRR and PFS in COMBI-MB, similar to results with immunotherapy for MBMs. Baseline corticosteroid treatment is a key factor to consider in MBM patient management and clinical trial design/interpretation.

This article is featured in Highlights of This Issue, p. 503

Translational Relevance

This study, along with emerging data from immunotherapy-treated cohorts, identifies the clear need to develop more effective strategies for patients with melanoma brain metastasis (MBM) who require steroids to control symptoms from these tumors. Such patients need to continue to be included in clinical trials, but likely should be considered and evaluated separately from patients with MBM who do not require steroids. This consistent observation of worse outcomes also supports the strategy to minimize and/or avoid steroid treatment in patients with MBM when possible (i.e., patients with asymptomatic cerebral edema), regardless of systemic therapy to be given.

Melanoma brain metastases (MBM) are present in up to 20% of patients at the time of diagnosis of stage IV disease and in up to 60% of patients by the time of melanoma-specific death (1–3). Historically, the median overall survival (OS) for patients with MBM was 4 to 5 months before the development of contemporary immune and targeted therapies (2). Markedly improved outcomes have been reported in recent years, particularly with combined anti—programmed cell death protein 1 (PD-1) and anti–CTLA-4 immune checkpoint blockade (ICB) therapies in patients with asymptomatic MBM (4, 5). However, worse outcomes have been reported with both single-agent and combination ICB therapy regimens in patients who require corticosteroid treatment to control their symptoms from MBM (4–7). Thus, improving outcomes in patients with MBM remains a major unmet need.

Approximately 50% of cutaneous melanomas harbor a hotspot mutation affecting the V600 codon in the BRAF serine-threonine kinase (8, 9). The presence of a BRAFV600 mutation results in marked hyperactivation of the MAPK signaling pathway (10). This molecular event was therapeutically exploited with the development of mutant-specific BRAF inhibitors, including vemurafenib (FDA approval 2011), dabrafenib (FDA approval, 2013), and encorafenib. Subsequent clinical trials demonstrated that combined treatment with BRAF and MEK inhibitors results in superior outcomes and tolerability, and ultimately led to the regulatory approval of dabrafenib and trametinib (2014), vemurafenib and cobimetinib (2015), and encorafenib and binimetinib (2018) for patients with BRAFV600 mutant, metastatic melanoma (11–15). Notably, none of the registration trials of the aforementioned agents allowed for participation of patients with untreated or progressing MBM.

The safety and activity of single-agent dabrafenib was confirmed in early phase trials, finding the treatment was active and could cross the blood brain barrier (16). With phase II trials confirming the activity of single-agent dabrafenib or vemurafenib treatment in patients with treated and untreated MBMs (17, 18). The COMBI-MB was the first prospective clinical trial reported to evaluate the safety and efficacy of an approved combination BRAF and MEK inhibitor targeted therapy regimen (dabrafenib and trametinib) in patients with MBM (15). The trial included a total of 125 patients with a BRAFV600 mutation and untreated or progressing MBM, who were enrolled into 4 cohorts based on BRAFV600 mutation status (BRAFV600E versus BRAFV600D/K/R), neurologic symptoms, and prior local (brain) therapy. In Cohort A (BRAFV600E mutation, asymptomatic brain metastases, no prior local therapy), the largest (n = 76) cohort in the study, treatment with dabrafenib and trametinib achieved an intracranial response rate (ICRR, objective intracranial complete or partial response as a proportion of all evaluable patients) of 58% and an intracranial disease control rate (objective intracranial complete, partial or stable disease as a proportion of all evaluable patients) of 78%, and no new or unexpected toxicities were observed (15). While these results were promising, the median intracranial duration of response was only 6.5 months. In contrast, pooled analysis of the COMBI-D and COMBI-V studies of dabrafenib plus trametinib in metastatic melanoma patients (n = 563), which excluded patients with active MBMs, identified a median progression-free survival (PFS) of 11.1 months (11, 12, 19). Notably the pooled analysis of COMBI-D/-V identified several factors associated with longer PFS and OS, including older age, female gender, BRAFV600E genotype, Eastern Cooperative Oncology Group performance status (ECOG PS) of 0, normal serum lactate dehydrogenase (LDH) level and less than 3 organ sites of metastases (19).

It is currently unknown what clinical or biological factors are associated with clinical outcomes with dabrafenib and trametinib in patients with MBM. The identification of such factors may help to guide the use of dabrafenib and trametinib in these patients, inform the design and interpretation of future clinical trials, and perhaps provide insights into the observed difference in response duration between intracranial and extracranial metastases (ECM). Thus, we report here clinical features associated with ICRR, PFS, and OS in patients with BRAF-mutant and MBM who were treated with dabrafenib and trametinib in COMBI-MB. This study also presents the exploratory biomarker analysis of COMBI-BRV trial (NCT01978236) of dabrafenib treatment in patient-matched MBM and ECMs as part of the previously unreported phase II clinical trial of preoperative treatment with dabrafenib in patients undergoing surgical resection of BRAF-mutant MBM. Synergistic exploratory analyses of the COMBI-BRV cohort were conducted to identify molecular and immunologic differences in the response of intracranial and extracranial melanoma metastases in patients treated with dabrafenib, to determine if site-specific responses contribute to these inferior outcomes seen in patients with MBM.

Study populations

Post hoc analysis was performed on the COMBI-MB open-label, multicohort, phase II trial which evaluated the activity and safety of dabrafenib plus trametinib in patients with MBM (Fig. 1A; ref. 15). Briefly, cohort A included patients with BRAFV600E mutant, asymptomatic MBM, without previous local brain-directed therapy, and an ECOG PS of 0 or 1; cohort B included patients with BRAFV600E mutant, asymptomatic MBM, with previous local therapy, and an ECOG PS of 0 or 1; cohort C included patients with BRAFV600D/K/R mutant, asymptomatic MBM, with or without previous local therapy, and an ECOG PS of 0 or 1; and cohort D included patients with BRAFV600D/E/K/R mutant, symptomatic MBM, with or without previous local therapy, and an ECOG PS of 0, 1, or 2 (15). Analysis was conducted across all cohorts.

Figure 1.

Consort diagram of the COMBI-MB and COMBI-BRV Trial. A, COMBI-MB trial profile; Cohort A = BRAFV600E-mutant, asymptomatic MBMs, without previous local brain-directed therapy, ECOG PS of 0 or 1. Cohort B = BRAFV600E-mutant, asymptomatic MBMs, with previous local therapy, ECOG PS of 0 or 1. Cohort C = BRAFV600D/K/R-mutant, asymptomatic MBMs, with or without previous local therapy, ECOG PS of 0 or 1. Cohort D = BRAFV600D/E/K/R-mutant, symptomatic MBMs, with or without previous local therapy ECOG PS of 0, 1, or 2. B, COMBI-BRV trial profile. Of the 18 biopsies, whole-exome sequencing (WES) analyses were performed on 12 samples after quality control analyses, with samples excluded for lack of tumor content (n = 5) or failure in DNA extraction in small biopsies (n = 1). While transcriptomic analyses were performed on 11 samples after quality control analyses (3 samples excluded for lack of viable tumor; 4 samples excluded due to insufficient tumor content based on BRAF mutation frequency; Supplementary Table S4). Multiplex IHC was performed on 11 biopsies, with samples excluded when no FFPE biopsies was collected (n = 4) or lack of tumor cells (> 100 melanoma cells, n = 3). PT, patient ID; PRE, baseline biopsy; √, included in analysis.

Figure 1.

Consort diagram of the COMBI-MB and COMBI-BRV Trial. A, COMBI-MB trial profile; Cohort A = BRAFV600E-mutant, asymptomatic MBMs, without previous local brain-directed therapy, ECOG PS of 0 or 1. Cohort B = BRAFV600E-mutant, asymptomatic MBMs, with previous local therapy, ECOG PS of 0 or 1. Cohort C = BRAFV600D/K/R-mutant, asymptomatic MBMs, with or without previous local therapy, ECOG PS of 0 or 1. Cohort D = BRAFV600D/E/K/R-mutant, symptomatic MBMs, with or without previous local therapy ECOG PS of 0, 1, or 2. B, COMBI-BRV trial profile. Of the 18 biopsies, whole-exome sequencing (WES) analyses were performed on 12 samples after quality control analyses, with samples excluded for lack of tumor content (n = 5) or failure in DNA extraction in small biopsies (n = 1). While transcriptomic analyses were performed on 11 samples after quality control analyses (3 samples excluded for lack of viable tumor; 4 samples excluded due to insufficient tumor content based on BRAF mutation frequency; Supplementary Table S4). Multiplex IHC was performed on 11 biopsies, with samples excluded when no FFPE biopsies was collected (n = 4) or lack of tumor cells (> 100 melanoma cells, n = 3). PT, patient ID; PRE, baseline biopsy; √, included in analysis.

Close modal

In addition, analyses were conducted on biospecimens collected in an international, open-label study of stage IV BRAFV600E/K metastatic melanoma with untreated, resectable MBM (1–4 cm) and ECM (NCT01978236, Fig. 1B). The first cohort (Cohort A) of 15 patients were planned to receive single agent dabrafenib orally 150 mg twice a day for 7 to 14 days prior to surgery; the second cohort (Cohort B) of 15 patients was planned to receive the combination of dabrafenib 150 mg twice a day and trametinib 2 mg once daily for 7 to 14 days prior to surgery, followed by resection of MBM and safely accessible ECM. Patients with active disease after surgery received dabrafenib and trametinib until disease progression. Study sites included The University of Texas MD Anderson Cancer Center, Melanoma Institute Australia, and University of Pittsburgh Cancer Institute.

The study protocols were approved by the institutional review board or human research ethics committee at each participating institution. Furthermore, the study was conducted in accordance with both the Declaration of Helsinki and the International Conference of Harmonization Good Clinical Practice. Written informed consent was obtained for all patients.

Exome sequencing

DNA and RNA were isolated from formalin-fixed, paraffin-embedded (FFPE) tissue sections using the AllPrep DNA/RNA FFPE Kit (Qiagen) according to the manufacturer's instructions and quality controls conducted as previously described (20). Library preparation was performed using the Nextera Flex Enrichment (Illumina) using IDT XGen Exome Research panel probes and sequenced on a NovaSeq 6000 S1 2×100bp. Sequenced reads were qualitychecked using FastQC v0.11.8 (RRID:SCR_014583). DNA reads were aligned against the hg38 version of the human genome using bwa-mem v0.7.17 BWA (RRID:SCR_010910). Aligned reads were further processed for marking of duplicates with Picard v2.18.23 (RRID:SCR_006525) and base recalibration with GATK v4.1.5.0 (GATK, RRID:SCR_001876). For all samples with available exome sequencing, point mutations and small indels were called using HaplotypeCaller as provided in GATK (RRID:SCR_001876). The initial variant calling was used to confirm that sequenced tumors belonged to the same patient by measuring the Jaccard index of putative heterozygote germline SNPs between samples. SNVs and small InDels with a minimum of 10 reads of coverage and supported by at least two reads were kept for further processing. Variants were annotated using ANNOVAR (RRID:SCR_012821) version dated April 16, 2018 and those with an Exome Aggregation Consortium version 0.3 (RRID:SCR_004068). ExAC_ALL values below 1% were kept. Mutational hotspots were identified using the Cancer Hotspots database (21).

RNA sequencing

Libraries were prepared using the TruSeq RNA Exome kit and sequenced on a NovaSeq S2 2×100bp. Reads were aligned following the SCANB pipeline (22) and counts were generated using the prepDE.py script provided by the Stringtie team (23). Gene counts were normalized using edgeR (TMM; RRID:SCR_012802). Gene set scores were generated using single-sample gene set enrichment analysis (SingScore; ref. 24), based on the Hallmark gene sets from the Molecular Signatures Database v7.4 and the Pratilas MEK dependent signaling gene set (25).

Multiplex IHC

Multiplex IHC was performed on the lesion matching FFPE melanoma biopsies used for RNA and DNA sequencing. Tissue sections were cut and stained for multiple markers using and following the instructions outlined in the Opal 7 plex IHC kit (Akoya Biosciences: NEL811001KT, RRID:AB_2890927) for three separate panels as previously described (8, 26). Firstly, an oncogenic signaling panel included phospho-Akt (Cell Signaling Technology, catalog no. 3787, RRID:AB_331170) phospho-Erk1/2 (Cell Signaling Technology, catalog no. 4370, RRID:AB_2315112), phospho-p90RSK (Cell Signaling Technology, catalog no. 11989, RRID:AB_2687613), phospho-S6 Ribosomal Protein (Cell Signaling Technology, catalog no. 4858, RRID:AB_916156), and SOX10 (Biocare, catalog no. ACI3099C, RRID:AB_2861289). Secondly, a T-cell specific panel included antibodies specific to PD-1 (Cell Marque, catalog no. 315M-95, RRID:AB_1160824), FoxP3 (Abcam, catalog no. ab20034, RRID:AB_445284), CD8 (Cell Marque, catalog no. 108R-15, RRID:AB_2892088), CD3 (Cell Marque, catalog no. CM103R95, RRID:AB_1158162), SOX10 (Biocare, catalog no. ACI3099C, RRID:AB_2861289). Finally, B cell and macrophage panel included CD20 (Dako, catalog no. M075501–2, RRID:AB_2282030), CD68 (Cell Marque, catalog no. CM168M95, RRID:AB_1158188), programmed death-ligand 1 (PD-L1; Cell Signaling Technology, catalog no. 13684, RRID:AB_2687655), and SOX10 (Biocare, catalog no. ACI3099C, RRID:AB_2861289). Multispectral images were acquired using a Vectra 3 multispectral microscope (Akoya Biosciences). Individual markers were spectrally unmixed and expression in each cell was quantified using the Inform software (Akoya Biosciences). The quantitative data was exported and analyzed in Spotfire (Tibco). Samples with less than 100 melanoma (SOX10 positive) cells were excluded from further analysis.

Statistical analysis

Exploratory biomarker data was generated blinded to all clinical data. The associations between categorical variables and treatment type were tested using either the two-tailed Fisher exact test or the χ2 test as appropriate. Clinical outcomes analyzed were intracranial response determined as per RECIST V.1.1. Duration of intracranial, extracranial, and overall response, defined as the time from first documented complete or partial response until the time of disease progression; PFS, defined as the interval between the first dose of study treatment and the earliest date of disease progression or death from any cause; OS, defined as the time from first dose until death due to any cause. Summaries for response, PFS and OS were generated using Kaplan–Meier estimates along with two-sided 95% confidence intervals (CI) and log-rank tests. Univariate and multivariate analysis was conducted on clinical features to test for associations with outcomes using Cox models with Brookmeyer and Crowley method to calculate confidence intervals. Associations with ICR were identified using logistic regression modelling. Exploratory biomarker data was assessed with a linear mixed-effect model with a random intercept to account for intra-patient correlation. To control the Type I error rate given the small sample size (27), the models were fitted with the restricted maximum likelihood method using the lme4 package in R v4.1.0 (28) and P values derived using the Satterthwaite approximation as provided in the sjPlot R package (29). No correction for multiple testing was pursued due to the exploratory nature of this analysis and small sample size. A P value less than 0.05 was considered statistically significant.

Data availability

The data generated in this study is available in the European Genome-Phenome Archive (EGAC00001002614) database.

COMBI-MB population

The baseline characteristics of all 125 patients enrolled in COMBI-MB were pooled together (Table 1; Fig. 1A) to identify patient and disease factors associated with clinical outcomes in patients with MBMs treated with dabrafenib and trametinib. The representativeness of the study participants is summarized in supplementary table S1. The clinical outcomes for the patients have been reported previously (15). Standard clinical features assessed for associations with clinical outcomes included baseline age, gender, ECOG PS, BRAF mutation type (V600E or V600D/K/R), prior systemic anticancer treatment, and serum LDH. The number of brain metastases, size of brain metastases (largest intracranial lesion; sum of long diameters of target lesions), prior CNS-directed treatment (any; radiation; surgery), presence of uncontrolled symptoms from brain metastases, and treatment with corticosteroids at baseline were also assessed. At baseline, 33 patients were recorded as receiving steroids—16 on dexamethasone (median dose 4 mg/day), 12 on prednisone (median dose 40 mg/day), and for 5 patient details for the specific corticosteroid treatment were not available. All patients were treated with the standard dose of dabrafenib (150 mg twice daily) and trametinib (2 mg daily).

Table 1.

Baseline characteristics of COMBI-MB trial.

Cohort A (n = 76)Cohort B (n = 16)Cohort C (n = 16)Cohort D (n = 17)Total (n = 125)
Age 
 Median (range) 52·0 (23–84) 54·5 (36–74) 63·0 (44–84) 46·0 (23–68)  
 <65 60 (79%) 12 (75%) 9 (56%) 16 (94%) 97 (78%) 
 ≥65 16 (21%) 4 (25%) 7 (44%) 1 (6%) 28 (22%) 
Sex 
 Male 40 (53%) 10 (63%) 11 (69%) 11 (65%) 72 (58%) 
 Female 36 (47%) 6 (38%) 5 (31%) 6 (35%) 53 (42%) 
ECOG 
 0 50 (66%) 11 (69%) 12 (75%) 9 (53%) 82 (66%) 
 1 25 (33%) 5 (31%) 4 (25%) 6 (35%) 40 (32%) 
 2 1 (1%)a 2 (12%) 3 (2%) 
BRAF genotype 
 V600E 73 (96%) 16 (100%) 15 (88%) 104 (83%) 
 V600K 3 (4%)b 14 (88%) 1 (6%) 18 (14%) 
 V600R 2 (13%) 1 (6%) 3 (2.4%) 
 V600D 
Target brain metastases 
 1 41 (54%) 7 (44%) 7 (44%) 7 (41%) 62 (49.6%) 
 2 20 (26%) 7 (44%) 6 (38%) 7 (41%) 40 (32%) 
 3 7 (9%) 2 (13%) 2 (13%) 1 (6%) 12 (9.6%) 
 4 4 (5%) 1 (6%) 5 (4%) 
 5 4 (5%) 1 (6%) 1 (6%) 6 (4.8%) 
SLD of target intracranial lesions (mm) 
 20 (6–117) 14 (5–40) 20 (5–61) 33 (10–84) 
ECMs 
 No 8 (11%) 4 (25%) 5 (29%) 17 (14%) 
 Yes 68 (89%) 12 (75%) 16 (100%) 12 (71%) 108 (86%) 
LDH concentration 
 Normal (≤ULN) 48 (63%) 13 (81%) 10 (63%) 12 (71%) 83 (66%) 
 Elevated (>ULN) 28 (37%) 3 (19%) 6 (38%) 5 (29%) 42 (34%) 
Receiving steroid therapy 
 Yes 13 (17%) 3 (19%) 3 (19%) 14 (82%) 33 (26%) 
 No 63 (83%) 13 (81%) 13 (81%) 3 (18%) 92 (74%) 
Previous systemic anticancer treatment 
 No 59 (78%) 11 (69%) 13 (81%) 10 (59%) 93 (74%) 
 Yes 17 (22%) 5 (31%) 3 (19%) 7 (41%) 32 (26%) 
Cohort A (n = 76)Cohort B (n = 16)Cohort C (n = 16)Cohort D (n = 17)Total (n = 125)
Age 
 Median (range) 52·0 (23–84) 54·5 (36–74) 63·0 (44–84) 46·0 (23–68)  
 <65 60 (79%) 12 (75%) 9 (56%) 16 (94%) 97 (78%) 
 ≥65 16 (21%) 4 (25%) 7 (44%) 1 (6%) 28 (22%) 
Sex 
 Male 40 (53%) 10 (63%) 11 (69%) 11 (65%) 72 (58%) 
 Female 36 (47%) 6 (38%) 5 (31%) 6 (35%) 53 (42%) 
ECOG 
 0 50 (66%) 11 (69%) 12 (75%) 9 (53%) 82 (66%) 
 1 25 (33%) 5 (31%) 4 (25%) 6 (35%) 40 (32%) 
 2 1 (1%)a 2 (12%) 3 (2%) 
BRAF genotype 
 V600E 73 (96%) 16 (100%) 15 (88%) 104 (83%) 
 V600K 3 (4%)b 14 (88%) 1 (6%) 18 (14%) 
 V600R 2 (13%) 1 (6%) 3 (2.4%) 
 V600D 
Target brain metastases 
 1 41 (54%) 7 (44%) 7 (44%) 7 (41%) 62 (49.6%) 
 2 20 (26%) 7 (44%) 6 (38%) 7 (41%) 40 (32%) 
 3 7 (9%) 2 (13%) 2 (13%) 1 (6%) 12 (9.6%) 
 4 4 (5%) 1 (6%) 5 (4%) 
 5 4 (5%) 1 (6%) 1 (6%) 6 (4.8%) 
SLD of target intracranial lesions (mm) 
 20 (6–117) 14 (5–40) 20 (5–61) 33 (10–84) 
ECMs 
 No 8 (11%) 4 (25%) 5 (29%) 17 (14%) 
 Yes 68 (89%) 12 (75%) 16 (100%) 12 (71%) 108 (86%) 
LDH concentration 
 Normal (≤ULN) 48 (63%) 13 (81%) 10 (63%) 12 (71%) 83 (66%) 
 Elevated (>ULN) 28 (37%) 3 (19%) 6 (38%) 5 (29%) 42 (34%) 
Receiving steroid therapy 
 Yes 13 (17%) 3 (19%) 3 (19%) 14 (82%) 33 (26%) 
 No 63 (83%) 13 (81%) 13 (81%) 3 (18%) 92 (74%) 
Previous systemic anticancer treatment 
 No 59 (78%) 11 (69%) 13 (81%) 10 (59%) 93 (74%) 
 Yes 17 (22%) 5 (31%) 3 (19%) 7 (41%) 32 (26%) 

Note: Data are median (range) or n (%). Cohort A = BRAF V600E -mutant, asymptomatic MBM, without previous local brain-directed therapy, ECOG PS of 0 or 1. Cohort B = BRAF V600E-mutant, asymptomatic MBM, with previous local therapy, ECOG PS of 0 or 1. Cohort C = BRAF V600D/K/R -mutant, asymptomatic MBM, with or without previous local therapy, ECOG PS of 0 or 1. Cohort D = BRAF V600D/E/K/R -mutant, symptomatic MBM, with or without previous local therapy ECOG PS of 0, 1, or 2.

Abbreviations: SLD, sum of lesion diameters; ULN, upper limit of normal.

aPatient had ECOG PS 1 at time of screening and enrolment.

bPatients were enrolled based on BRAF V600E status but were found to be BRAF V600K on central confirmation.

Factors associated with clinical outcomes in COMBI-MB

Univariate and multivariate analysis identified a significant association of ICRR with baseline corticosteroid treatment. The associations between corticosteroid treatment and LDH levels and outcome are summarized in Fig. 2. Treatment with dabrafenib and trametinib achieved an ICRR of 39% in patients treated with corticosteroids (n = 33) and 63% in patients not on corticosteroids (n = 92; Fig. 2A and B respectively). This difference in ICRR was significant on both univariate (OR, 0.381; 95% CI, 0.168–0.862; P = 0.02) and multivariate analyses with all clinical variables (OR, 0.323; 95% CI, 0.105–0.996; P = 0.0491). No other feature was significantly associated with ICRR (Table 2). Features tested included age at treatment, gender, ECOG PS, BRAF mutation status (V600E vs. other), number of brain metastases, intracranial lesion size, presence of ECMs, LDH, prior treatment or uncontrolled symptoms from the intracranial lesions. Intracranial response duration (ICRD) was shorter for patients with uncontrolled symptoms from brain metastasis (n = 24, median 4.4 months) compared with those without symptoms on univariate analysis (n = 101, median 5.6 months; HR, 2.365; 95% CI, 1.184–4.726; P = 0.0148), but no clinical features were significantly associated with ICRD on multivariate analysis (Supplementary Table S2).

Figure 2.

Outcome analysis of the COMBI-MB trial. Waterfall plot of best intracranial response in (A) MBM patient on steroids at baseline, and (B) patients with MBM not on steroids at baseline. Kaplan–Meier curves demonstrate (C) PFS stratified by baseline steroid treatment status, (D) OS stratified by baseline steroid treatment status, (E) PFS stratified by baseline LDH, and (F) OS stratified by baseline LDH.

Figure 2.

Outcome analysis of the COMBI-MB trial. Waterfall plot of best intracranial response in (A) MBM patient on steroids at baseline, and (B) patients with MBM not on steroids at baseline. Kaplan–Meier curves demonstrate (C) PFS stratified by baseline steroid treatment status, (D) OS stratified by baseline steroid treatment status, (E) PFS stratified by baseline LDH, and (F) OS stratified by baseline LDH.

Close modal
Table 2.

Univariate and multivariate analyses of baseline factors associated with overall ICRR.

UnivariateMultivariate
CategoryPatients (N)N responding (%)OR (95% CI)POR (95% CI)P
Age 
 <54 59 34 (58) 1.066 (0.524–2.166) 0.8599 1.143 (0.512–2.554) 0.7443 
 ≥54 66 37 (56)     
>Gender 
 Female 53 28 (53) 0.755 (0.369–1.546) 0.4425 0.735 (0.334–1.618) 0.4447 
 Male 72 43 (60)     
>ECOG 
 0 82 50 (61) 1.637 (0.777–3.447) 0.1946 1.034 (0.399–2.681) 0.9456 
 ≥1 43 21 (49)     
>BRAF mutation status 
 V600E 104 61 (59) 1.56 (0.609–3.999) 0.3541 1.686 (0.574–4.957) 0.3421 
 Other 21 10 (48)     
>Brain metastases 
 1 62 37 (60) 1.138 (0.432–2.997) 0.568 1.287 (0.355–4.668) 0.5891 
 2 40 21 (53) 0.85 (0.303–2.386) 0.5717 0.988 (0.286–3.410) 0.7598 
 ≥3 23 13 (57)     
>SLD of target intracranial lesion 
 <median 62 35 (56) 0.972 (0.479–1.973) 0.9378 0.632 (0.184–2.167) 0.4655 
 ≥median 63 36 (57)     
>Largest intracranial lesion 
 <median 60 35 (58) 1.128 (0.555–2.291) 0.7396 1.177 (0.392–3.530) 0.7717 
 ≥median 65 36 (55)     
>ECMs 
 Yes 97 55 (57) 0.982 (0.420–2.297) 0.9669 1.212 (0.435–3.379) 0.7135 
 No 28 16 (57)     
>Elevated serum LDH 
 Yes 42 21 (50) 0.660 (0.312–1.394) 0.2761 0.71 (0.302–1.676) 0.436 
 No 83 50 (60)     
>Steroid use at baseline 
Yes 33 13 (39) 0.381 (0.168–0.862) 0.0206 0.323 (0.105–0.996) 0.0491 
 No 92 58 (63)     
>Previously treatment anticancer 
 Yes 32 21 (66) 1.642 (0.712–3.786) 0.2449 1.831 (0.710–4.719) 0.2105 
 No 93 50 (54)     
>Previous treatment to brain 
 No 97 57 (59)     
 Yes 28 14 (50) 0.702 (0.302–1.632) 0.4108 1.186 (0.385–3.654) 0.7669 
>XRT to brain 
 Yes 34 15 (44) 0.493 (0.222–1.096) 0.0827 0.477 (0.156–1.464) 0.196 
 No 91 56 (62)     
>Previous Surgery to brain 
 Yes 121 69 (57) 1.327 (0.181–9.734) 0.7809 1.487 (0.172–12.850) 0.7182 
 No 2 (50)     
>Presence of uncontrolled symptoms from brain 
 Yes 24 13 (54) 0.876 (0.358–2.143) 0.7721 1.603 (0.479–5.367) 0.4438 
 No 101 58 (57)     
UnivariateMultivariate
CategoryPatients (N)N responding (%)OR (95% CI)POR (95% CI)P
Age 
 <54 59 34 (58) 1.066 (0.524–2.166) 0.8599 1.143 (0.512–2.554) 0.7443 
 ≥54 66 37 (56)     
>Gender 
 Female 53 28 (53) 0.755 (0.369–1.546) 0.4425 0.735 (0.334–1.618) 0.4447 
 Male 72 43 (60)     
>ECOG 
 0 82 50 (61) 1.637 (0.777–3.447) 0.1946 1.034 (0.399–2.681) 0.9456 
 ≥1 43 21 (49)     
>BRAF mutation status 
 V600E 104 61 (59) 1.56 (0.609–3.999) 0.3541 1.686 (0.574–4.957) 0.3421 
 Other 21 10 (48)     
>Brain metastases 
 1 62 37 (60) 1.138 (0.432–2.997) 0.568 1.287 (0.355–4.668) 0.5891 
 2 40 21 (53) 0.85 (0.303–2.386) 0.5717 0.988 (0.286–3.410) 0.7598 
 ≥3 23 13 (57)     
>SLD of target intracranial lesion 
 <median 62 35 (56) 0.972 (0.479–1.973) 0.9378 0.632 (0.184–2.167) 0.4655 
 ≥median 63 36 (57)     
>Largest intracranial lesion 
 <median 60 35 (58) 1.128 (0.555–2.291) 0.7396 1.177 (0.392–3.530) 0.7717 
 ≥median 65 36 (55)     
>ECMs 
 Yes 97 55 (57) 0.982 (0.420–2.297) 0.9669 1.212 (0.435–3.379) 0.7135 
 No 28 16 (57)     
>Elevated serum LDH 
 Yes 42 21 (50) 0.660 (0.312–1.394) 0.2761 0.71 (0.302–1.676) 0.436 
 No 83 50 (60)     
>Steroid use at baseline 
Yes 33 13 (39) 0.381 (0.168–0.862) 0.0206 0.323 (0.105–0.996) 0.0491 
 No 92 58 (63)     
>Previously treatment anticancer 
 Yes 32 21 (66) 1.642 (0.712–3.786) 0.2449 1.831 (0.710–4.719) 0.2105 
 No 93 50 (54)     
>Previous treatment to brain 
 No 97 57 (59)     
 Yes 28 14 (50) 0.702 (0.302–1.632) 0.4108 1.186 (0.385–3.654) 0.7669 
>XRT to brain 
 Yes 34 15 (44) 0.493 (0.222–1.096) 0.0827 0.477 (0.156–1.464) 0.196 
 No 91 56 (62)     
>Previous Surgery to brain 
 Yes 121 69 (57) 1.327 (0.181–9.734) 0.7809 1.487 (0.172–12.850) 0.7182 
 No 2 (50)     
>Presence of uncontrolled symptoms from brain 
 Yes 24 13 (54) 0.876 (0.358–2.143) 0.7721 1.603 (0.479–5.367) 0.4438 
 No 101 58 (57)     

Consistent with the association with ICRR, patients on corticosteroids had a shorter PFS (median 4.3 vs. 6.2 months) on both univariate (HR, 1.788; 95% CI, 1.152–2.774; P = 0.0095) and multivariate (HR, 1.931; 95% CI, 1.061–3.513, P = 0.0312) analyses (Table 3; Fig. 2C). The median PFS was significantly longer for patients with an ECOG PS of 0 (6.5 months) compared with those with ECOG ≥ 1 on univariate analysis only (3.8 months; HR, 0.617; 95% CI, 0.411–0.927; P = 0.02; Table 3). While the median PFS was not significantly different for patients with elevated or normal LDH (HR, 1.16; 95% CI, 0.768–1.749; P = 0.4821; Fig. 2E). Multivariate analysis identified longer PFS in patients with a BRAFV600E (5.9 months) mutation vs. BRAFV600D/K/R mutant disease (4.2 months; HR, 0.565; 95% CI, 0.321–0.996; P = 0.0483).

Table 3.

Univariate and multivariate analyses of baseline factors associated with PFS.

UnivariateMultivariate
Category(N)eventsMedian MonthsHR 95% CIPHR 95% CIP
Age 
 <54 59 47 5.6(5.3–7.3) 1.096 (0.740–1.623) 0.6479 1.093(0.705–1.696) 0.6908 
 ≥54 66 54 5.7(5.4–7.3)     
Gender 
 Female 53 43 5.6(4.2–7.4) 1.056 (0.711–1.568) 0.7884 0.869(0.557–1.356) 0.5368 
 Male 72 58 5.8(5.5–7.3)     
ECOG 
0 82 63 6.5(5.6–7.5) 0.617 (0.411–0.927) 0.02 0.692(0.401–1.197) 0.1881 
 ≥1 43 38 3.8(3.5–5.9)     
BRAF mutation status 
 V600E 104 83 5.9(5.5–7.3) 0.638 (0.381–1.066) 0.0859 0.565(0.321–0.996) 0.0483 
 Other 21 18 4.2(3.5–9.1)     
Brain metastases 
 1 62 48 7.2(5.5–9.1) 0.592 (0.348–1.006) 0.0528 0.673(0.322–1.403) 0.2903 
 2 40 33 5.5(4.2–6.8) 0.808 (0.463–1.411) 0.4537 1.014(0.503–2.045) 0.9685 
 ≥3 23 20 5.5(3.6–7.4)     
SLD of target intracranial lesion 
 median 62 49 5.9(5.3–7.3) 0.996 (0.674–1.472) 0.9841 1.418(0.667–3.014) 0.3637 
 ≥median 63 52 5.6(4.7–7.3)     
Largest intracranial lesion 
 median 60 46 5.9(5.5–7.3) 0.951 (0.642–1.408) 0.801 1.018(0.523–1.983) 0.9572 
 65 55 5.6(4.3–7.3)     
ECMs 
 Yes 97 82 5.6(4.7–6.7) 1.45 (0.879–2.391) 0.1452 1.249(0.722–2.162) 0.4261 
 No 28 19 7.3(5.6–14.6)     
Elevated serum LDH 
 Yes 42 35 5.6(3.7–7.5) 1.159 (0.768- 1.749) 0.4821 0.811(0.514–1.280) 0.368 
 No 83 66 5.7(5.5–7.2)     
Steroid use at baseline 
Yes 33 29 4.3(3.5–6.4) 1.788 (1.152–2.774) 0.0095 1.931(1.061–3.513) 0.0312 
 No 92 72 6.2(5.6–7.3)     
Previously treatment anticancer 
 Yes 32 27 7.4(5.6–12.0) 0.72 (0.768- 1.749) 0.1485 0.716(0.435–1.179) 0.1895 
 No 93 74 5.5(4.7–6.2)     
Previous treatment to brain 
 Yes 28 21 7.2(5.5–13.4) 0.778 (0.481- 1.260) 0.3077 0.796(0.441–1.435) 0.4473 
 No 97 80 5.6(5.3–6.7)     
XRT to brain 
 Yes 34 26 5.3(4.3–12.2) 0.781 (0.497- 1.227) 0.2838 0.766(0.422–1.390) 0.3804 
 No 91 75 5.9(5.5–7.2)     
Previous Surgery to brain 
 Yes 121 99 5.6(5.4–7.2) 1.795 (0.441- 7.302) 0.4141 1.817(0.420–7.859) 0.4243 
 No 9.1(6.2–9.1)     
Presence of uncontrolled symptoms from brain 
 Yes 24 21 5.3(3.7–7.5) 1.504 (0.921- 2.456) 0.103 1.171(0.611–2.241) 0.6346 
 No 101 80 5.8(5.5–7.2)     
UnivariateMultivariate
Category(N)eventsMedian MonthsHR 95% CIPHR 95% CIP
Age 
 <54 59 47 5.6(5.3–7.3) 1.096 (0.740–1.623) 0.6479 1.093(0.705–1.696) 0.6908 
 ≥54 66 54 5.7(5.4–7.3)     
Gender 
 Female 53 43 5.6(4.2–7.4) 1.056 (0.711–1.568) 0.7884 0.869(0.557–1.356) 0.5368 
 Male 72 58 5.8(5.5–7.3)     
ECOG 
0 82 63 6.5(5.6–7.5) 0.617 (0.411–0.927) 0.02 0.692(0.401–1.197) 0.1881 
 ≥1 43 38 3.8(3.5–5.9)     
BRAF mutation status 
 V600E 104 83 5.9(5.5–7.3) 0.638 (0.381–1.066) 0.0859 0.565(0.321–0.996) 0.0483 
 Other 21 18 4.2(3.5–9.1)     
Brain metastases 
 1 62 48 7.2(5.5–9.1) 0.592 (0.348–1.006) 0.0528 0.673(0.322–1.403) 0.2903 
 2 40 33 5.5(4.2–6.8) 0.808 (0.463–1.411) 0.4537 1.014(0.503–2.045) 0.9685 
 ≥3 23 20 5.5(3.6–7.4)     
SLD of target intracranial lesion 
 median 62 49 5.9(5.3–7.3) 0.996 (0.674–1.472) 0.9841 1.418(0.667–3.014) 0.3637 
 ≥median 63 52 5.6(4.7–7.3)     
Largest intracranial lesion 
 median 60 46 5.9(5.5–7.3) 0.951 (0.642–1.408) 0.801 1.018(0.523–1.983) 0.9572 
 65 55 5.6(4.3–7.3)     
ECMs 
 Yes 97 82 5.6(4.7–6.7) 1.45 (0.879–2.391) 0.1452 1.249(0.722–2.162) 0.4261 
 No 28 19 7.3(5.6–14.6)     
Elevated serum LDH 
 Yes 42 35 5.6(3.7–7.5) 1.159 (0.768- 1.749) 0.4821 0.811(0.514–1.280) 0.368 
 No 83 66 5.7(5.5–7.2)     
Steroid use at baseline 
Yes 33 29 4.3(3.5–6.4) 1.788 (1.152–2.774) 0.0095 1.931(1.061–3.513) 0.0312 
 No 92 72 6.2(5.6–7.3)     
Previously treatment anticancer 
 Yes 32 27 7.4(5.6–12.0) 0.72 (0.768- 1.749) 0.1485 0.716(0.435–1.179) 0.1895 
 No 93 74 5.5(4.7–6.2)     
Previous treatment to brain 
 Yes 28 21 7.2(5.5–13.4) 0.778 (0.481- 1.260) 0.3077 0.796(0.441–1.435) 0.4473 
 No 97 80 5.6(5.3–6.7)     
XRT to brain 
 Yes 34 26 5.3(4.3–12.2) 0.781 (0.497- 1.227) 0.2838 0.766(0.422–1.390) 0.3804 
 No 91 75 5.9(5.5–7.2)     
Previous Surgery to brain 
 Yes 121 99 5.6(5.4–7.2) 1.795 (0.441- 7.302) 0.4141 1.817(0.420–7.859) 0.4243 
 No 9.1(6.2–9.1)     
Presence of uncontrolled symptoms from brain 
 Yes 24 21 5.3(3.7–7.5) 1.504 (0.921- 2.456) 0.103 1.171(0.611–2.241) 0.6346 
 No 101 80 5.8(5.5–7.2)     

Several factors were associated with OS on univariate analysis (Supplementary Table S3). OS was shorter in patients treated with corticosteroids (median 9.3 versus 13.5 months; HR, 1.642; 95% CI, 1.037–2.598; P = 0.0343; Fig. 2D) and those with elevated LDH (median 10.6 versus 12.7 months, HR 1.595, 95% CI, 1.045–2.434, P = 0.0305; Fig. 2F). OS was longer in patients with ECOG PS of 0 compared with > = 1 (median 18.9 versus 8.2 months, HR 0.435, 95% CI, 0.285–0.664, P = 0.0001). Only ECOG PS remained significant on multivariate analysis (HR, 0.441; 95% CI, 0.249–0.779; P = 0.0048).

Molecular and immune analysis of melanoma brain and ECMs during dabrafenib treatment on the COMBI-BRV trial

COMBI-BRV was a clinical trial designed to evaluate and compare the molecular and immune effects of dabrafenib ± trametinib in MBMs versus ECMs. The study enrolled patients with BRAFV600E/K mutant metastatic melanoma, with a planned craniotomy for 1 or more previously untreated MBM(s). Safely accessible ECM were biopsied prior to the start of treatment (“PRE”). Patients were then treated for 7 to 14 days with dabrafenib (Cohort A) or dabrafenib + trametinib (Cohort B) prior to craniotomy (last dose given the night before surgery). Samples collected on the day of craniotomy were designated as early during treatment (“EDT”). Six patients were enrolled in the study, all of whom entered cohort A and received dabrafenib monotherapy (Supplementary Table 3). Due to slow enrollment, the trial was halted after 6 patients completed treatment, and as such no patients received dabrafenib plus trametinib prior to craniotomy (Fig. 1B). All patients underwent planned craniotomy; safely accessible ECM were also biopsied or excised. Two patients on the COMBI-BRV trial (PT4 and PT62) received corticosteroids prior to craniotomy. Biospecimens were available for PRE ECM, EDT ECM and EDT MBM for 6 patients (PT3 did not have a PRE ECM biopsy). Response data, biospecimen availability, and variables used for analysis are summarized in Supplementary Table S4 and Supplementary Table S5.

Oncogenic signaling was examined within the PRE and EDT melanoma biopsies via whole transcriptome sequencing (Fig. 3A). Of the total 18 biopsies, transcriptomic analysis were performed on 11 samples after quality control analyses (3 samples excluded for lack of viable tumor; 4 samples excluded due to insufficient tumor content based on BRAF mutation frequency) (Supplementary Table S5). Subsequent single sample pathway analysis (Fig. 3B; Supplementary Table S6) showed a trend of decreased MAPK (MEK dependent) signaling from PRE ECM to EDT ECM biopsy sites (Mean difference = −0.14, P = 0.081; Supplementary Fig. S1A and B) and a significant decrease from PRE ECM to EDT MBM (Mean difference = −0.11, P = 0.03; Supplementary Fig. S1A and B). MAPK signaling did not differ between EDT ECM and EDT MBM biopsy sites (Mean difference = 0.01, P = 0.678). A significant increase in MTOR signaling pathway score from PRE ECM to EDT ECM biopsy sites (Mean difference = 0.04, P = 0.024); and a significant decrease of cell cycle score from PRE ECM to EDT ECM (Mean difference = −0.08, P = 0.015) and a trend versus EDT MBM biopsy sites (Mean difference = −0.07, P = 0.061), were observed. Whilst oxidative phosphorylation and PI3K/MTOR signaling were higher in EDT MBM than EDT ECM, no significant differences were observed between any biopsy site/timepoint in this small cohort (Supplementary Table S6, Supplementary Fig. S1A).

Figure 3.

Analysis of biospecimen from the COMBI-BRV trial. A, Unsupervised clustering of gene expression signatures. B, Changes in MEK dependent gene expression and phosphorylated S6 protein expression. C, Changes in HALLMARK interferon gamma gene expression and intratumoral cytotoxic T-cell (CD8+) densities. D, Waterfall plot depicting changes in intratumoral T-cell (CD3+) densities between treatment timepoints and colored via steroid treatment status.

Figure 3.

Analysis of biospecimen from the COMBI-BRV trial. A, Unsupervised clustering of gene expression signatures. B, Changes in MEK dependent gene expression and phosphorylated S6 protein expression. C, Changes in HALLMARK interferon gamma gene expression and intratumoral cytotoxic T-cell (CD8+) densities. D, Waterfall plot depicting changes in intratumoral T-cell (CD3+) densities between treatment timepoints and colored via steroid treatment status.

Close modal

Lesion-matched FFPE biopsies were available for all patients for multiplex IHC and image analysis (Supplementary Table S5 and Supplementary Fig. S1C). While the protein expression of pS6 and pp90RSK were reduced in melanoma cells from the PRE biopsies compared with both EDT biopsy sites (Supplementary Table S6), only the reduction in pS6 from pretreatment to EDT MBM reached significance (Mean difference = −0.39, P = 0.043; Figs. 3B; Supplementary Fig. S1B). Other qualitative changes in expression, such as reduced proliferation (Ki-67) from PRE ECM to both EDT sites, were observed but did not reach statistical significance (Supplementary Table S5).

A trend of increased interferon gamma signaling activity was observed following treatment from PRE ECM to EDT ECM (Mean difference = 0.06, P = 0.073) and to EDT MBM (Mean difference = 0.09, P = 0.051; Fig. 3C; Supplementary Fig. S1B; Supplementary Table S6). The total (CD3+) and cytotoxic (CD8+) T-cell densities measured by mIHC tended to increase from PRE ECM to both EDT sites, and macrophage densities (CD68+) tended to decrease, but no differences reached significance (Fig. 3C; Supplementary Fig. S1D). In addition, no significant differences were observed between EDT ECM and EDT MBM biospecimen. Interestingly, the two patients (PT4 and PT62) treated with corticosteroids in the preoperative period experienced a reduction in immune cell densities (T cells, cytotoxic T cells, and PD-L1 positivity) from their PRE to their EDT ECM biopsies (Fig. 3D). PT4 had the highest densities of cytotoxic T cells in the PRE biopsies, then experienced a reduction in T-cell and PD-L1 positive cell densities, down to the lowest levels of the cohort in both EDT ECM and EDT MBM biopsies (Supplementary Fig. S1E and F, Supplementary Table S5).

Exome sequencing was possible for 5 patients’ tumors [1 sample failed to extract adequate DNA; 4 melanoma biopsies lacked enough tumor content to detect the BRAF mutation and were excluded from analysis (Supplementary Table S5). BRAFV600E mutations were detected in all remaining tumors except PT62, whose tumors carried a BRAFV600K mutation (Supplementary Table S7). Screening for known genetic mechanisms of MAPK inhibitor resistance (30, 31) revealed a BRAF amplification in the PRE ECM biopsy of patient PT2, which was not present in the EDT biopsies, while the remaining mutations were concordant between all biopsies for this patient. PT4 contained MAP2K1P124S and CDKN2AQ50X mutations and loss of heterozygosity of chromosome 10 in both EDT lesions, with an additional PTENR130X mutation detected exclusively in the MBM (Supplementary Table S7). In addition, the EDT MBM for PT62 (the only biopsy profiled for this patient) contained an allelic imbalance of the long arm of chromosome 10 and a deleterious mutation of PTEN, suggesting inactivation of this gene (Supplementary Table S7).

There remains a critical need to improve our understanding of the determinants of benefit of BRAF plus MEK inhibitor targeted therapy for MBMs. This study reveals for the first time that baseline treatment with corticosteroids is associated with significantly reduced ICRR and shorter PFS in patients with BRAF-mutant MBMs who were treated with dabrafenib and trametinib. Further, our analysis of biospecimens in the COMBI-BRV trial is the first to explore molecular and immune differences between intracranial and ECMs in patients receiving a BRAF inhibitor (dabrafenib).

The need for corticosteroids has previously been shown to be associated with worse clinical outcomes in clinical trials of immune checkpoint inhibitors for patients with MBM. In the phase II trial that evaluated ipilimumab 10 mg/kg, the ICRR was 18% in patients with asymptomatic MBMs but only 5% in patients who required corticosteroids to control symptoms (7). In the ABC trial, the response rate with nivolumab in patients (n = 25) with asymptomatic, previously untreated brain metastases was 20% (4, 32). The ABC trial also evaluated nivolumab in a cohort (Cohort C) of patients with MBM who either had previous CNS-directed therapy, neurologic symptoms (n = 10), or leptomeningeal disease (LMD; n = 4). Only 1 patient in this cohort had an intracranial response, which was noted to be a patient with neurologic symptoms (ICRR 10%). While the response rates even among asymptomatic patients in each of these trials were low, much more impressive results have been seen with combination immunotherapy with ipilimumab and nivolumab. The ICRR for ipilimumab and nivolumab in patients with asymptomatic brain metastases was 54% in CheckMate-204 (Cohort A; n = 101) and 59% in ABC (Cohort A; n = 27 drug-treatment naïve) studies (4, 5, 32). However, the ICRR in patients with symptomatic brain metastases, including in patients (n = 12) requiring up to 4 mg/day of dexamethasone, in CheckMate-204 (Cohort B, n = 18) was 22% (5).

In this study, we show that the association of corticosteroids with poor clinical outcomes in melanoma patients with brain metastases also applies to targeted therapy. In the COMBI-MB trial, baseline treatment with corticosteroids was associated with a 2-month reduction in the median PFS that was independent of measures of tumor burden, including the number of MBM, ECM/MBM lesion size, or serum LDH (15). Although the specific mechanism behind corticosteroid-driven immunosuppression and impaired response to immunotherapies is elusive (33), it has been well established that host response also plays a critical role in augmenting the response to targeted therapies, albeit not specifically in patients with brain metastases (34, 35). For example, treatment of metastatic melanoma with BRAF inhibitors has been shown to increase T-cell infiltration and upregulation of melanoma antigen expression (MART-1, TYRP1/2, and GP100) in the melanoma biopsies from patients early during treatment (10–14 days; ref. 36). These studies have led to combination trials that aim to optimize the scheduling of combination molecular inhibitors with immunotherapies to take advantage of this immunogenic window (37).

Together the data suggests that patients with symptomatic MBMs who are treated with corticosteroids have poor outcomes, regardless of whether they receive immune or targeted. Whether use of corticosteroids is simply a surrogate for aggressive tumor biology or specifically antagonizes the effects of targeted therapy in MBMs remains to be determined. Notably, the available data strongly supports that symptomatic patients with MBM who require corticosteroids need to be included in clinical trials, as there is an unmet need to identify strategies that will improve their poor outcomes. However, this patient cohort requires specific consideration in cohort design and trial analysis for both targeted and immune therapies. Interestingly, recently reported initial results for the TRICOTEL study of combined treatment with vemurafenib, cobimetinib, and atezolizumab in patients with BRAF-mutant brain metastases showed comparable outcomes for patients with symptomatic (n = 24, ICRR 46%) and asymptomatic (n = 41, ICRR 39%) MBMs (38). These results, combined with the inferior outcomes reported here for targeted therapy alone, and previously for immunotherapy alone, support the rationale to continue to explore combinatorial approaches in the challenging setting.

The COMBI-BRV trial offered the unique opportunity to investigate oncogenic and immunologic signaling at EDT in ECM and MBM. As the study was ultimately limited in recruitment, this analysis should be considered exploratory in nature due to the small sample size. Despite this limitation, the interrogation of the available samples following short-term treatment with dabrafenib demonstrated comparable inhibition of MEK dependent signaling, as well as stimulation of interferon gamma signaling, in both ECMs and MBM sites early on treatment. While prior studies in larger unmatched cohorts detected upregulation of PI3K/AKT/mTOR signaling (39) and increased oxidative phosphorylation in MBM compared with ECM (40), we observed trends but no significant differences from PRE to either EDT sites in the COMBI-BRV cohort. However, as noted above these analyses were limited by small numbers and the heterogeneity observed among these samples. In addition, the MBM samples within the COMBI-BRV cohort represent early on treatment (7–14 days) biopsies, with the expected decrease in viable tumor content and influx of immune cells potentially affecting the bulk gene expression signatures. Prior studies have also reported a decrease in interferon gamma gene signatures in MBMs versus same-patient ECMs (41). While Fischer and colleagues observed an immunosuppressed TME, with lower T-cell densities and lower immunoscores, in MBM compared with ECM sites in untreated patients, the results here suggest that treatment with MAPKi may alter this suppressive TME in MBMs, at least at this early timepoint (40). Therefore, this data confirms the MEK dependent inhibition across both MBMs and ECMs, whilst highlighting increased immunogenicity of the MBM lesions early during BRAF inhibitor treatment.

Together these findings highlight the continued challenges to improving outcomes in patients with MBMs, particularly those with symptomatic disease. These results, combined with the inferior outcomes reported here for targeted therapy alone, and previously for immunotherapy alone, support the rationale for continued research and for further exploration of combinatorial approaches for patients who require corticosteroid treatment. These findings also raise the need for a restrained use of steroids to treat asymptomatic patients with peritumoral edema detected via medical imaging. Therefore, regardless of the treatment strategy, the use of corticosteroids to treat symptomatic MBM is associated with poor outcomes and highlights the urgent need for new, more effective strategies for these patients (42).

H. Tawbi reports grants and personal fees from Novartis during the conduct of the study; grants and personal fees from BMS, Merck, and Genentech; grants from Eisai and GSK; and personal fees from Karyopharm, Boxer Capital, Jazz Pharma, and Pfizer outside the submitted work. I.A. Vergara reports other support from Melanoma Institute Australia during the conduct of the study; in addition, I.A. Vergara has a patent for US11145412B2 issued. P. Saiag reports personal fees and nonfinancial support from Novartis outside the submitted work. C. Robert reports personal fees from Novartis during the conduct of the study; consulting or advisory roles for Roche, BMS, MSD, AstraZeneca, Sanofi, and Pierre Fabre; and personal fees from Ribonexus outside the submitted work. J.J. Grob reports personal fees and nonfinancial support from Novartis, BMS, MSD, and Pierre Fabre; and personal fees from Philogen and Sanofi outside the submitted work. L.H. Butterfield reports that the following activities are unrelated to the present work but are listed here for total transparency: Advisory activities (honoraria): Calidi Scientific and Medical Advisory Board, April 6, 2017 to present; KaliVir, Scientific Advisory Board, 2018 to 2021; Torque Therapeutics, Scientific Advisory Board, 2018 to 2020; Khloris, Scientific Advisory Board, 2019 to present; Pyxis, Scientific Advisory Board, 2019 to present; CytomX, Scientific Advisory Board, 2019 to present; DCprime, Scientific Advisory Board meeting, November 2020; RAPT, Scientific Advisory Board, 2020 to present; Takeda, Scientific Advisor, 2020 to present; EnaraBio scientific advisor, Feb. 2021; Federation Bio scientific advisor September to October 2022; Pfizer scientific advisor October, 2022. R.A. Scolyer reports grants from National Health and Medical Research Council of Australia (NHMRC) and from National Health and Medical Research Council of Australia during the conduct of the study; and personal fees from MetaOptima Technology Inc., F. Hoffmann-La Roche Ltd, Evaxion, Provectus Biopharmaceuticals Australia, Qbiotics, Novartis, Merck Sharp & Dohme, NeraCare, Amgen, Bristol-Myers Squibb, Myriad Genetics, and GlaxoSmithKline outside the submitted work. G.V. Long reports personal fees from Agenus, Amgen, Array Biopharma, AstraZeneca, Boehringer Ingelheim, BMS, Evaxion, Hexal AG, Highlight Therapeutics, Innovent Biologics, MSD, Novartis, OncoSec, PHMR, Pierre Fabre, Provectus, QBiotics, and Regeneron outside the submitted work. M.A. Davies reports grants from Melanoma Research Alliance and GlaxoSmithKline during the conduct of the study; and grants and personal fees from ABM Therapeutics, Array/Pfizer, Genentech/Roche, BMS, Novartis, Iovance, Eisai, and Apexigen; nonfinancial support from NanoString Technologies; and grants from AstraZeneca, GlaxoSmithKline, Lead Pharma, Merck, Myriad, Oncothyreon, and Sanofi Aventis outside the submitted work; and GlaxoSmithKline provided financial support for the clinical trials that are the source of the data used in this manuscript. No disclosures were reported by the other authors.

J.S. Wilmott: Resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. H. Tawbi: Conceptualization, resources, data curation, investigation, writing–original draft, writing–review and editing. J.A. Engh: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, writing–review and editing. N.M. Amankulor: Conceptualization, resources, data curation, investigation, writing–review and editing. B. Shivalingam: Resources, data curation, software, formal analysis, investigation, writing–review and editing. H. Banerjee: Conceptualization, resources, data curation, software, formal analysis, methodology, writing–review and editing. I.A. Vergara: Data curation, software, formal analysis, validation, methodology, writing–original draft, writing–review and editing. H. Lee: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. P.A. Johansson: Resources, data curation, software, formal analysis, methodology, writing–original draft, writing–review and editing. P.M. Ferguson: Data curation, formal analysis, writing–original draft, writing–review and editing. P. Saiag: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, writing–review and editing. C. Robert: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, writing–review and editing. J.J. Grob: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. L.H. Butterfield: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. R.A. Scolyer: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J.M. Kirkwood: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. G.V. Long: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M.A. Davies: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

This study was supported by a Team Science Grant from the Melanoma Research Alliance (#269996). GSK supported the two clinical trials and provided the data/samples we used in the study. R.A. Scolyer and G.V. Long are supported by National Health and Medical Research Council of Australia (NHMRC) Practitioner Fellowships, and their research is supported by an NHMRC Program grant (APP2006415 and APP2018514). Support from CLEARbridge Foundation, The Cameron Family and colleagues at Melanoma Institute Australia and Royal Prince Alfred Hospital are also gratefully acknowledged. G.V. Long is supported by the University of Sydney Medical Foundation. J.S. Wilmott is supported by an NHMRC investigator fellowship (APP1174325), Melanoma Research Alliance young investigator fellowship (#700455), Cancer Institute NSW (TPG2114), and the University of Sydney. M.A. Davies is supported by the NCI of the NIH under grant award number P50CA221703, Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the AIM at Melanoma Foundation, the American Cancer Society and the Melanoma Research Alliance, Cancer Fighters of Houston, the Anne and John Mendelsohn Chair for Cancer Research, and philanthropic contributions to the Melanoma Moon Shots Program of MD Anderson.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Patel
JK
,
Didolkar
MS
,
Pickren
JW
,
Moore
RH
.
Metastatic pattern of malignant melanoma: a study of 216 autopsy cases
.
Am J Surg
1978
;
135
:
807
10
.
2.
Davies
MA
,
Liu
P
,
McIntyre
S
,
Kim
KB
,
Papadopoulos
N
,
Hwu
WJ
, et al
.
Prognostic factors for survival in melanoma patients with brain metastases
.
Cancer
2011
;
117
:
1687
96
.
3.
Pires da Silva
I
,
Lo
S
,
Quek
C
,
Gonzalez
M
,
Carlino
MS
,
Long
GV
, et al
.
Site-specific response patterns, pseudoprogression, and acquired resistance in patients with melanoma treated with ipilimumab combined with anti–PD-1 therapy
.
Cancer
2020
;
126
:
86
97
.
4.
Long
GV
,
Atkinson
V
,
Lo
S
,
Sandhu
S
,
Guminski
AD
,
Brown
MP
, et al
.
Combination nivolumab and ipilimumab or nivolumab alone in melanoma brain metastases: a multicenter randomized phase II study
.
Lancet Oncol
2018
;
19
:
672
81
.
5.
Tawbi
HA
,
Forsyth
PA
,
Hodi
FS
,
Lao
CD
,
Moschos
SJ
,
Hamid
O
, et al
.
Safety and efficacy of the combination of nivolumab plus ipilimumab in patients with melanoma and asymptomatic or symptomatic brain metastases (CheckMate 204)
.
Neuro-oncol
2021
;
23
:
1961
73
.
6.
Tawbi
HA
,
Forsyth
PA
,
Algazi
A
,
Hamid
O
,
Hodi
FS
,
Moschos
SJ
, et al
.
Combined nivolumab and ipilimumab in melanoma metastatic to the brain
.
N Engl J Med
2018
;
379
:
722
30
.
7.
Margolin
K
,
Ernstoff
MS
,
Hamid
O
,
Lawrence
D
,
McDermott
D
,
Puzanov
I
, et al
.
Ipilimumab in patients with melanoma and brain metastases: an open-label, phase II trial
.
Lancet Oncol
2012
;
13
:
459
65
.
8.
Hayward
NK
,
Wilmott
JS
,
Waddell
N
,
Johansson
PA
,
Field
MA
,
Nones
K
, et al
.
Whole-genome landscapes of major melanoma subtypes
.
Nature
2017
;
545
:
175
80
.
9.
Cancer Genome Atlas N
.
Genomic classification of cutaneous melanoma
.
Cell
2015
;
161
:
1681
96
.
10.
Davies
H
,
Bignell
GR
,
Cox
C
,
Stephens
P
,
Edkins
S
,
Clegg
S
, et al
.
Mutations of the BRAF gene in human cancer
.
Nature
2002
;
417
:
949
54
.
11.
Robert
C
,
Grob
JJ
,
Stroyakovskiy
D
,
Karaszewska
B
,
Hauschild
A
,
Levchenko
E
, et al
.
Five-year outcomes with dabrafenib plus trametinib in metastatic melanoma
.
N Engl J Med
2019
;
381
:
626
36
.
12.
Grob
JJ
,
Amonkar
MM
,
Karaszewska
B
,
Schachter
J
,
Dummer
R
,
Mackiewicz
A
, et al
.
Comparison of dabrafenib and trametinib combination therapy with vemurafenib monotherapy on health-related quality of life in patients with unresectable or metastatic cutaneous BRAF Val600-mutation-positive melanoma (COMBI-v): results of a phase III, open-label, randomized trial
.
Lancet Oncol
2015
;
16
:
1389
98
.
13.
Ascierto
PA
,
McArthur
GA
,
Dréno
B
,
Atkinson
V
,
Liszkay
G
,
Di Giacomo
AM
, et al
.
Cobimetinib combined with vemurafenib in advanced BRAFV600-mutant melanoma (coBRIM): updated efficacy results from a randomized, double-blind, phase III trial
.
Lancet Oncol
2016
;
17
:
1248
60
.
14.
Dummer
R
,
Ascierto
PA
,
Gogas
HJ
,
Arance
A
,
Mandala
M
,
Liszkay
G
, et al
.
Encorafenib plus binimetinib versus vemurafenib or encorafenib in patients with BRAF-mutant melanoma (COLUMBUS): a multicenter, open-label, randomized phase III trial
.
Lancet Oncol
2018
;
19
:
603
15
.
15.
Davies
MA
,
Saiag
P
,
Robert
C
,
Grob
JJ
,
Flaherty
KT
,
Arance
A
, et al
.
Dabrafenib plus trametinib in patients with BRAF(V600)-mutant melanoma brain metastases (COMBI-MB): a multicenter, multicohort, open-label, phase II trial
.
Lancet Oncol
2017
;
18
:
863
73
.
16.
Falchook
GS
,
Long
GV
,
Kurzrock
R
,
Kim
KB
,
Arkenau
TH
,
Brown
MP
, et al
.
Dabrafenib in patients with melanoma, untreated brain metastases, and other solid tumors: a phase I dose-escalation trial
.
Lancet
2012
;
379
:
1893
901
.
17.
Long
GV
,
Trefzer
U
,
Davies
MA
,
Kefford
RF
,
Ascierto
PA
,
Chapman
PB
, et al
.
Dabrafenib in patients with Val600Glu or Val600Lys BRAF-mutant melanoma metastatic to the brain (BREAK-MB): a multicenter, open-label, phase II trial
.
Lancet Oncol
2012
;
13
:
1087
95
.
18.
McArthur
GA
,
Maio
M
,
Arance
A
,
Nathan
P
,
Blank
C
,
Avril
MF
, et al
.
Vemurafenib in metastatic melanoma patients with brain metastases: an open-label, single-arm, phase II, multicenter study
.
Ann Oncol
2017
;
28
:
634
41
.
19.
Long
GV
,
Grob
JJ
,
Nathan
P
,
Ribas
A
,
Robert
C
,
Schadendorf
D
, et al
.
Factors predictive of response, disease progression, and overall survival after dabrafenib and trametinib combination treatment: a pooled analysis of individual patient data from randomized trials
.
Lancet Oncol
2016
;
17
:
1743
54
.
20.
De Paoli-Iseppi
R
,
Johansson
PA
,
Menzies
AM
,
Dias
KR
,
Pupo
GM
,
Kakavand
H
, et al
.
Comparison of whole-exome sequencing of matched fresh and formalin fixed, paraffin embedded melanoma tumors: implications for clinical decision making
.
Pathology
2016
;
48
:
261
6
.
21.
Chang
MT
,
Bhattarai
TS
,
Schram
AM
,
Bielski
CM
,
Donoghue
MTA
,
Jonsson
P
, et al
.
Accelerating discovery of functional mutant alleles in cancer
.
Cancer Discov
2018
;
8
:
174
.
22.
Brueffer
C
,
Gladchuk
S
,
Winter
C
,
Vallon-Christersson
J
,
Hegardt
C
,
Häkkinen
J
, et al
.
The mutational landscape of the SCAN-B real-world primary breast cancer transcriptome
.
EMBO Mol Med
2020
;
12
:
e12118
.
23.
Pertea
M
,
Kim
D
,
Pertea
GM
,
Leek
JT
,
Salzberg
SL
.
Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie, and ballgown
.
Nat Protoc
2016
;
11
:
1650
67
.
24.
Foroutan
M
,
Bhuva
DD
,
Lyu
R
,
Horan
K
,
Cursons
J
,
Davis
MJ
.
Single sample scoring of molecular phenotypes
.
BMC Bioinf
2018
;
19
:
404
.
25.
Pratilas
CA
,
Taylor
BS
,
Ye
Q
,
Viale
A
,
Sander
C
,
Solit
DB
, et al
.
V600EBRAF is associated with disabled feedback inhibition of RAF–MEK signaling and elevated transcriptional output of the pathway
.
Proc Natl Acad Sci
2009
;
106
:
4519
24
.
26.
Gide
TN
,
Quek
C
,
Menzies
AM
,
Tasker
AT
,
Shang
P
,
Holst
J
, et al
.
Distinct immune cell populations define response to anti–PD-1 monotherapy and anti–PD-1/anti–CTLA-4 combined therapy
.
Cancer Cell
2019
;
35
:
238
55
.
27.
Luke
SG
.
Evaluating significance in linear mixed-effects models in R
.
Behav Res Methods
2017
;
49
:
1494
502
.
28.
Bates
D
,
Mächler
M
,
Bolker
B
,
Walker
S
.
Fitting linear mixed-effects models using lme4
.
J Stat Softw
2015
;
67
:
1
48
.
29.
Lüdecke
D
.
2021 sjPlot: Data visualization for statistics in Social Science
.
Available from:
<https://CRAN.R-project.org/package=sjPlot>.
30.
Rizos
H
,
Menzies
AM
,
Pupo
GM
,
Carlino
MS
,
Fung
C
,
Hyman
J
, et al
.
BRAF inhibitor resistance mechanisms in metastatic melanoma: spectrum and clinical impact
.
Clin Cancer Res
2014
;
20
:
1965
.
31.
Carlino
MS
,
Fung
C
,
Shahheydari
H
,
Todd
JR
,
Boyd
SC
,
Irvine
M
, et al
.
Preexisting MEK1P124 mutations diminish response to BRAF inhibitors in metastatic melanoma patients
.
Clin Cancer Res
2015
;
21
:
98
105
.
32.
Long
GV
,
Atkinson
V
,
Lo
S
,
Guminski
AD
,
Sandhu
SK
,
Brown
MP
, et al
.
Five-year overall survival from the anti–PD-1 brain collaboration (ABC Study): randomized phase II study of nivolumab (nivo) or nivo+ipilimumab (ipi) in patients (pts) with melanoma brain metastases (mets)
.
J Clin Oncol
2021
;
39
:
9508
.
33.
Giles
AJ
,
Hutchinson
M-KND
,
Sonnemann
HM
,
Jung
J
,
Fecci
PE
,
Ratnam
NM
, et al
.
Dexamethasone-induced immunosuppression: mechanisms and implications for immunotherapy
.
J Immunother Cancer
2018
;
6
:
51
.
34.
Knight
DA
,
Ngiow
SF
,
Li
M
,
Parmenter
T
,
Mok
S
,
Cass
A
, et al
.
Host immunity contributes to the anti-melanoma activity of BRAF inhibitors
.
J Clin Invest
2013
;
123
:
1371
81
.
35.
Wilmott
JS
,
Long
GV
,
Howle
JR
,
Haydu
LE
,
Sharma
RN
,
Thompson
JF
, et al
.
Selective BRAF inhibitors induce marked T-cell infiltration into human metastatic melanoma
.
Clin Cancer Res
2012
;
18
:
1386
94
.
36.
Frederick
DT
,
Piris
A
,
Cogdill
AP
,
Cooper
ZA
,
Lezcano
C
,
Ferrone
CR
, et al
.
BRAF inhibition is associated with enhanced melanoma antigen expression and a more favorable tumor microenvironment in patients with metastatic melanoma
.
Clin Cancer Res
2013
;
19
:
1225
31
.
37.
Long
GV
,
Carlino
MS
,
Au-Yeung
G
,
Spillane
AJ
,
Shannon
KF
,
Gyorki
DE
, et al
.
NeoTrio: randomized trial of neoadjuvant (NAT) pembrolizumab (Pembro) alone, in sequence (SEQ) with, or concurrent (CON) with dabrafenib plus trametinib (D+T) in resectable BRAF-mutant stage III melanoma to determine optimal combination of therapy
.
J Clin Oncol
2022
;
40
:
9503
-.
38.
Dummer
R
,
Queirolo
P
,
Abajo Guijarro
AM
,
Hu
Y
,
Wang
D
,
de Azevedo
SJ
, et al
.
Atezolizumab, vemurafenib, and cobimetinib in patients with melanoma with CNS metastases (TRICOTEL): a multicenter, open-label, single-arm, phase II study
.
Lancet Oncol
2022
;
23
:
1145
55
.
39.
Chen
G
,
Chakravarti
N
,
Aardalen
K
,
Lazar
AJ
,
Tetzlaff
MT
,
Wubbenhorst
B
, et al
.
Molecular profiling of patient-matched brain and extracranial melanoma metastases implicates the PI3K pathway as a therapeutic target
.
Clin Cancer Res
2014
;
20
:
5537
.
40.
Fischer
GM
,
Jalali
A
,
Kircher
DA
,
Lee
W-C
,
McQuade
JL
,
Haydu
LE
, et al
.
Molecular profiling reveals unique immune and metabolic features of melanoma brain metastases
.
Cancer Discov
2019
;
9
:
628
.
41.
Trembath
DG
,
Davis
ES
,
Rao
S
,
Bradler
E
,
Saada
AF
,
Midkiff
BR
, et al
.
Brain tumor microenvironment and angiogenesis in melanoma brain metastases
.
Front Oncol
2021
;
10
:
2899
.
42.
Banks
PD
,
Lasocki
A
,
Lau
PKH
,
Sandhu
S
,
McArthur
G
,
Shackleton
M
.
Bevacizumab as a steroid-sparing agent during immunotherapy for melanoma brain metastases: a case series
.
Health Sci Rep
2019
;
2
:
e115
.

Supplementary data