Impact of Previous Local Treatment for Brain Metastases on Response to Molecular Targeted Therapy in BRAF-Mutant Melanoma Brain Metastasis: A Systematic Review and Meta-Analysis

Background Melanoma brain metastases (BMs) are associated with poor prognosis and are the main cause of mortality in melanoma patients. BRAF inhibitors have shown intracranial activity in both treatment-naïve and previously treated BM patients. We aimed to investigate if there was any difference in response of BRAF inhibitors in these two cohorts. Materials and Methods Electronic database search included PubMed, Medline, and Cochrane library until March 2021 for studies with desired comparative outcomes. Outcomes of interest that were obtained for meta-analysis included intracranial response rate as the primary outcome and survival and safety outcomes as the secondary outcomes. Review Manager version 5.4 was used for data analysis. Results Three studies comprising 410 BRAF-mutated melanoma patients with BMs were included according to eligibility criteria. The comparative cohort included patients with treatment-naïve BMs (TN cohort; n = 255) and those who had progressive disease after receiving local brain treatment for BMs (PT cohort; n = 155). Meta-analysis revealed that BRAF inhibitors (vemurafenib and dabrafenib) and BRAF/MEK inhibitor combination (dabrafenib and trametinib) induced significantly higher intracranial disease control (OR 0.58 [95% CI: 0.34, 0.97], p = 0.04) and a trend toward improved progression-free survival (PFS) (HR 1.22 [95% CI: 0.98, 1.52], p = 0.08) in the PT cohort as compared to the TN cohort. Overall survival was not significantly different between the cohorts (HR 1.16 [95% CI: 0.89, 1.51], p = 0.28). Subgroup analysis revealed that PFS was significantly improved (HR 1.67 [95% CI: 1.06, 2.62], p = 0.03), and a trend toward improved OS (HR 1.62 [95% CI: 0.95, 2.75], p = 0.08) was achieved in patients receiving BRAF/MEK inhibitor combination and patients with BRAFv600K mutation receiving dabrafenib alone. No increase in overall adverse events (AEs), grade 3/4 AEs, and severe adverse events (SAEs) was observed between the cohorts. Conclusions BRAF inhibitors (plus MEK inhibitor) may achieve better intracranial disease stability in BRAF-mutant melanoma patients who have received previous local treatment for BMs. Systematic Review Registration https://www.crd.york.ac.uk/prospero/), identifier CRD42020185984.


INTRODUCTION
Melanoma is the fifth leading cause of cancer according to 2021 estimates of the American Cancer Society. An estimated 106,110 new cases of melanoma were diagnosed and 7,850 deaths from melanoma occurred in the United States in 2021 (1,2). Melanoma represents the third most common cancer type (10%) that metastasizes to the brain after lung (50%) and breast cancer (20%) (3,4). Approximately 20% of melanoma patients present with brain metastases (BMs) at diagnosis and approximately 10% to 44% develop BMs during the disease course (1)(2)(3)(4). The risk of developing BMs is increased up to 80% in patients with metastatic melanoma (5). Patients with melanoma brain metastases (MBMs) have a poor prognosis with a median survival of 3.8 to 7.69 months (5)(6)(7).
Several trials have demonstrated that BRAF inhibitors can induce intracranial responses in treatment-naïve and previously treated BMs derived from BRAF-mutant melanoma alike (32,34,35). Here, we attempt to systematically review and analyze the efficacy and safety outcomes as reported in these trials.

METHODS AND MATERIALS
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed in this review (38). A protocol of this study is registered on PROSPERO (https://www. crd.york.ac.uk/prospero/): CRD42020185984.

Eligibility Criteria
The study population comprised BRAF-mutant melanoma patients with untreated BMs (control group) or progressive BMs after previous local treatments (experimental group) receiving BRAF inhibitors and/or MEK inhibitors. Outcomes of interest included efficacy and safety results such as intracranial response rate, progression-free survival (PFS), overall survival (OS), and treatment-related adverse events (AEs). Only clinical trials (randomized/non-randomized) with prospective design and comparative outcomes were considered for inclusion. Language was restricted to English.

Information Sources
Explored databases included PubMed, Medline, and Cochrane library that were searched until April 10, 2021. Several search terms relevant to the eligibility criteria were employed accordingly. References of the relevant articles were further elaborated for potential studies.

Study Selection
Retrieved studies were organized and screened for duplication and eligibility. Two reviewers (SA and MK) independently carried out the evaluation and selection of eligible studies. Disagreements were resolved after consulting with a third reviewer (GL).

Data Extraction
A modified form of "The Cochrane Collaboration Data Collection form-RCTs and non-RCTs" was used for data extraction. Attributes of the studies included design, first author, publication year, number of participants, treatment regimens, and main efficacy and safety outcomes for the overall study group. Baseline characteristics of the patients included age, sex, performance status (Eastern Cooperative Oncology Group), BRAF mutation type, number of BMs, level of lactate dehydrogenase (LDH), and previous therapy. Furthermore, outcomes of interest (intracranial response rate, survival, and safety) for treatment differences were also extracted.

Quality Assessment
Quality assessment was carried out with the checklist developed by the National Institute for Health and Care Excellence (NICE) (39). The checklist comprised eight questions that addressed various aspects of clinical case series.

Measurement of Treatment Effect and Data Synthesis
The number of events for objective response rate (ORR) and disease control rate (DCR) at the intracranial and extracranial level were recorded from the studies to obtain odds ratio. Likewise, AEs were also recorded to obtain odds ratio. Hazard ratios for the survival outcomes (PFS and OS) were extracted from the Kaplan-Meier curves using the Digital Equalizer and methods for incorporating summary time-to-event data into the meta-analysis according to Tierney et al. (40,41). The acquired odds ratios or hazard ratios were pooled using the software "RevMan 5.4 software" (42,43). The Mantel-Haenszel method or inverse variance statistical method was applied for pooling obtained odds ratios and hazard ratios using the fixed effects analysis model, respectively (44). Heterogeneity was assessed using the chi-squared test and I 2 value. Heterogeneity was graded as low, moderate, and high according to I 2 values of 25%, 50%, and 75%, respectively (45). A random effects analysis model was used in case of moderate heterogeneity (≥50%). Significance level was set at p < 0.05.

RESULTS
A total of 1,271 articles were identified via an initial electronic database search. After exclusion for duplicity and eligibility, 35 articles were examined for detailed evaluation. Another 32 articles were excluded due to different reasons and the remaining 3 articles were finally selected for inclusion in this systematic review and meta-analysis (32,34,35). The search and selection process is detailed in Figure 1. All the three included studies were multicenter, open label, non-randomized, phase 2 clinical trials (32,34,35). Altogether, data of 410 BRAF-mutant melanoma brain metastatic patients treated with BRAF inhibitors (vemurafenib and dabrafenib) and/or MEK inhibitor (trametinib) were provided for analysis. Studies included mainly comprised two cohorts: patients with no local treatment administered to the brain (n = 255) and patients who had disease progression after having received local treatment to the brain (n = 155) in the form of surgery and/or radiation therapy (32,34,35). BMs were asymptomatic in two of the three trials (34,35). General characteristics are illustrated in Table 1.
Participants were predominantly male (n = 260) rather than female (n = 150). The majority of the patients (n = 374) were mainly BRAFV600E (Val600Glu); however, the study by Long et al. also had an additional comparative cohort (n = 33) with BRAFV600K (Val600Lys). The study by Long et al. provided data separately for Val600Glu and Val600Lys patients (35). Patients were evenly distributed according to age (p = 0.85), gender (p = 0.64), performance status (ECOG: 0 = 223, p = 0.92;  Moreover, the PT cohort was also more exposed to previous therapy than the TN cohort (OR 0.36 [95% CI: 0.23, 0.55], p < 0.00001). Baseline characteristics of the patients are outlined in Table 2. Quality assessment according to the eight items of the checklist are described in Supplementary Table 1. All clinical trials obtained a score of 7 out of 8 points.

Intracranial Response Rate
Intracranial response rate was reported in all three clinical trials involving 410 patients (32,34,35). All four parameters of intracranial response, namely, complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD), were reported across studies, and these were analyzed separately. Meta-analysis of overall intracranial response rate, defined as the combined number of patients achieving complete and partial response, revealed no significant difference between the cohorts (OR 1.22 [95% CI: 0.78, 1.93], p = 0.38) ( Figure 2). Nonetheless, overall disease control, defined as the combined number of patients achieving CR, PR, and SD, was significantly higher in patients with previously treated BMs (OR 0.58 [95% CI: 0.34, 0.97], p = 0.04) ( Figure 2).
Separate analysis of each intracranial response parameters showed that complete response rate was higher in treatmentnaïve patients, although not significant (OR 1.93 [95% CI: 0.41, 9.00], p = 0.40). No significant difference between the cohorts was seen with respect to PR as well (OR 1.14 [95% CI: 0.72, 1.80],   (Figure 2). Moreover, duration of intracranial response was reported in all three studies (32,34,35) (Supplementary Table 2). Vemurafenib-induced median duration of intracranial response was 4.7 months (range; 2.7-24.2) in the TN cohort and 6.6 months (range; 1.9-22.0) in the PT cohort (32). Median duration of intracranial response with dabrafenib was also slightly higher in the PT cohort compared to that in the TN cohort (BRAFV600E group: 28

Extracranial Response Rate
Extracranial response rate was reported in only two clinical trials involving 211 patients (32,35). These studies involved vemurafenib and the combination of dabrafenib and trametinib. Meta-analysis of overall extracranial response rate revealed no significant difference between the cohorts ( Figure 3.
Duration of extracranial response was reported in two studies wherein vemurafenib and dabrafenib plus trametinib were administered to patients (32, 35) (Supplementary Table 2

Overall Response Rate
All three studies reported overall response rate involving 410 patients (32,34,35). Overall response rate was defined as the proportion of patients with a best response of CR or PR when both extracranial and intracranial disease were assessed in them. Metaanalysis of overall response rate showed no significant difference between the cohorts (OR 1.05 [95% CI: 0.65, 1.65], p = 0.85) ( Figure 4). Likewise, no difference was noted in the overall disease control (OR 0.67 [95% CI: 0.34, 1.30], p = 0.23) reported in two studies involving 264 patients (34,35).
Duration of overall response (intracranial and extracranial response) was reported in only one study with patients receiving FIGURE 2 | Forest plot of meta-analysis of intracranial response rate comparison between treatment-naïve (TN) and previously treated brain metastases (PT) cohorts in BRAF-mutant melanoma patients with brain metastases receiving BRAF/MEK inhibitors.

Progression-Free Survival
Meta-analysis of PFS involving 410 participants resulted in slightly better PFS for MBM patients with previously treated BMs ( Figure 5)

Overall Survival
Meta-analysis of OS involving 410 participants showed no difference between the cohorts ( Figure 6). A hazard ratio of 1. 16

Adverse Events
All three included studies reported safety of these drugs in detail (32,34,35). AEs were assessed according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0) grading system. Meta-analysis of any grade toxicity revealed no significant difference between the cohorts (OR 1.14 [95% CI: 0.57, 2.28], p = 0.70) (Figure 7). Likewise, there was no difference in grade 3/4 toxicity between the cohorts (OR 0.99 [95% CI: 0.62, 1.59], p = 0.98).

Fatal SAEs
Meta-analysis of fatality caused by SAEs revealed no difference (OR 1.01 [95% CI: 0.14, 7.13], p = 0.99) between the cohorts based on data from the two studies (n = 238) (32,35) (Figure 7). No treatment-related death was reported in the study by Long et al. (34). AEs causing death in patients receiving vemurafenib were pneumonia (2 deaths; one in each cohort) and glioma (1 death in cohort B) (32). Intracranial tumor hemorrhage was the cause of the only SAE-related death in the cohort A receiving dabrafenib plus trametinib (35).

Discontinuation/Dose Interruptions
Discontinuation and dose interruptions of study treatment on account of SAEs were recorded in all the three studies (32,34,35). Based on the data from two studies (n = 238), discontinuation resulted in no significant difference between cohorts (OR 0.66 [95% CI: 0.23, 1.89], p = 0.44) (32, 35) (Figure 7). The study by Long et al. also mentioned discontinuation by 4 patients on account of treatment-related AEs (34).

Publication Bias
Publication bias was assessed using a funnel plot for all the main outcomes, including intracranial/extracranial response rate, survival outcomes, and AEs. All results were within the 95% CI, indicating no evidence of publication bias ( Supplementary  Figures 1-6).

DISCUSSION
Our results comprising 410 BRAF-mutated melanoma brain metastatic patients indicate that BRAF inhibitors (vemurafenib and dabrafenib) alone or in combination with MEK inhibitor (trametinib) are active and safe in BRAF-mutated melanoma with treatment-naïve or previously treated BMs. Even though the treatment-naïve cohort consisted of significantly more single BMs and fewer 2-4 BMs, intracranial DCR was significantly higher (p = 0.04) in previously treated brain metastatic patients as compared to treatment-naïve BMs. The main difference was observed in intracranial disease stability (high in previously treated BMs; p = 0.0009) and progression (high in treatmentnaïve BMs; p = 0.07). Consequently, a slight trend towards better PFS (p = 0.08) was observed with no significant improvement in OS. Subgroup analysis revealed that a significant improvement in PFS (p = 0.03) and an almost significant OS (p = 0.08) could be achieved in previously treated BMs when analysis was restricted to patients receiving a combination of dabrafenib and trametinib and patients with BRAFV600K (Val600Lys) mutation.
Vemurafenib, dabrafenib, and dabrafenib plus trametinib have shown intracranial responses in previous studies in which participants were either untreated or heavily treated or were of the mixed type (23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37). For example, vemurafenib has previously shown intracranial responses in heavily pretreated MBMs. In a pilot study of vemurafenib, a partial response and stable disease in four patients was achieved in a total of five heavily pretreated MBM patients (29). Likewise, in a phase 2 trial of 24 pretreated MBM patients, both intracranial (n = 19; PR: 3, SD: 13) and extracranial responses (n = 21; PR: 13, SD: 6) were observed with 3.9 (95% CI: 3.0, 5.5) months of PFS and a median OS of 5.3 (95% CI: 3.9, 6.6) months (30). A separate trial of vemurafenib that included mixed BRAF-mutated MBM patients (treatment-naïve or previously treated BMs) also showed vemurafenib activity (n = 66; CR:1, PR:11, SD:34, and PD:10) with a PFS of 4.8 months (95% CI: 3.7, 5.7) and OS of 7.9 (95% CI: 5.9, 9.3) (31). A similar objective response and survival outcome (PFS and OS) for vemurafenib were demonstrated in both treatment-naïve and previously treated BMs in the study by McArthur et al., which was included in our meta-analysis (32). Previously, dabrafenib was only investigated in a small cohort of 10 BRAF-mutated melanoma patients with untreated BMs, which revealed excellent intracranial (n = 10; CR: 4, PR: 9) and extracranial (n = 10; PR: 9) activity, and yielded 4.2 months (95% CI: 3.3, 5.3) of PFS (33). BREAK MB trial reveals that like vemurafenib, dabrafenib is also active and safe in both cohorts of BMs. COMBI-MB trial revealed efficacy and safety of dabrafenib for the first time in a combination of trametinib in both cohorts. The combination has been previously proven to be more effective than vemurafenib and dabrafenib alone in metastatic melanoma without BMs (21,46). Although the odds of intracranial response DCR were similar in the PT cohort, the combination of dabrafenib and trametinib was associated with slightly improved PFS and OS (35). In fact, extracranial ORR and DCR were also slightly favoring the treatment-naïve patients in their study (35). Nonetheless, no patient with progressive disease (0/16) and a small cohort may have caused improved PFS and OS (35). Therefore, this outcome needs to be validated in studies with larger cohorts. On the other hand, improved PFS and OS in BRAFV600K was in coherence with its intracranial efficacy in a cohort with previously treated BMs (34). Although overall response showed no difference between cohorts, extracranial activity alone was not reported in these studies. Nonetheless, duration of response was also longer in the PT cohort across all studies.
The activity and efficacy observed with BRAF/MEK inhibitors in both cohorts appeared to be superior and in slight contrast to the intracranial activity and survival benefit achieved with temozolomide in MBM patients (47). An objective response of 7% (8 patients) and stable disease of 29% (34 patients) was demonstrated for patients who had not received previous treatment for brain lesions with a median survival of 3.5 months. Response in previously treated BMs was slightly lower with temozolomide, revealing a partial response and 18% (6 patients) stable disease with a median survival of 2.2 months (47). Patients with previous radiotherapy have benefited in preclinical and clinical studies for immune checkpoint blockade (48,49). In a secondary analysis of the KEYNOTE-001 phase 1 trial, the benefit for NSCLC patients with previous radiotherapy receiving pembrolizumab was evident in terms of PFS (HR 0.56 [95% CI: 0.34, 0.91], p = 0.019) and OS (HR 0.58 [95% CI: 0.36, 0.94], p = 0.026) (49). In a metaanalysis of BRAF-mutant MBM patients, those administered BRAF inhibitors concurrently or after SRS induction showed superior benefit than patients receiving prior SRS (HR 0.39 [95% CI: 0.24, 0.65], p = 0.0003) (50). Similarly, in a retrospective study, MBM patients receiving BRAF inhibitor after SRS were noted to have better survival compared to those receiving BRAF inhibitors concurrently (24 months vs. 10.1 months, p = 0.007) (51). Hence, it appears that patients with previous radiotherapy may be at an advantage for deriving better outcomes compared to treatmentnaïve cancer patients. However, the underlying mechanism for this finding remains to be elucidated.
No increase in overall AEs, grade 3/4 AEs, and SAEs were reported between the cohorts that used a single agent or a combination of agents. Only increased dose reductions or interruptions were required with dabrafenib and/or trametinib. Our results were limited by several factors. The number of studies and total population of our study was small. In the study by Davies et al., the PT cohort had a small number of patients (n = 16) (35). All the trials lacked randomization and blinding, which may have rendered these studies prone to selection, performance, and detection biases (32,34,35). Moreover, agents used were different and patients had both types of BRAF mutation.

CONCLUSIONS
Vemurafenib, dabrafenib, and dabrafenib plus trametinib demonstrate intracranial activity and safety in BRAF-mutated melanoma (BRAFV600E) patients with treatment-naïve or previously treated BMs. Intracranial disease control was significantly higher in patients with previously treated BMs. Moreover, patients with previously treated BMs had improved PFS and/or OS with dabrafenib plus trametinib compared to treatment-naïve patients. In BRAFV600K (Val600Lys) mutated melanoma, dabrafenib may improve PFS and OS in patients with previously treated BMs. Randomized controlled trials (RCTs) with larger cohorts would, however, be required to validate these outcomes accordingly.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.