Upfront molecular targeted therapy for the treatment of BRAF-mutant pediatric high-grade glioma

Abstract

Background

The prognosis for patients with pediatric high-grade glioma (pHGG) is poor despite aggressive multimodal therapy. Objective responses to targeted therapy with BRAF inhibitors have been reported in some patients with recurrent BRAF-mutant pHGG but are rarely sustained.

Methods

We performed a retrospective, multi-institutional review of patients with BRAF-mutant pHGG treated with off-label BRAF +/– MEK inhibitors as part of their initial therapy.

Results

Nineteen patients were identified, with a median age of 11.7 years (range, 2.3–21.4). Histologic diagnoses included HGG (n = 6), glioblastoma (n = 3), anaplastic ganglioglioma (n = 4), diffuse midline glioma (n = 3), high-grade neuroepithelial tumor (n = 1), anaplastic astrocytoma (n = 1), and anaplastic astroblastoma (n = 1). Recurrent concomitant oncogenic alterations included CDKN2A/B loss, H3 K27M, as well as mutations in ATRX, EGFR, and TERT. Eight patients received BRAF inhibitor monotherapy. Eleven patients received combination therapy with BRAF and MEK inhibitors. Most patients tolerated long-term treatment well with no grade 4–5 toxicities. Objective and durable imaging responses were seen in the majority of patients with measurable disease. At a median follow-up of 2.3 years (range, 0.3–6.5), three-year progression-free and overall survival for the cohort were 65% and 82%, respectively, and superior to a historical control cohort of BRAF-mutant pHGG patients treated with conventional therapies.

Conclusions

Upfront targeted therapy for patients with BRAF-mutant pHGG is feasible and effective, with superior clinical outcomes compared to historical data. This promising treatment paradigm is currently being evaluated prospectively in the Children’s Oncology Group ACNS1723 clinical trial.

Key Points.

• Outcome for patients with pHGG, including BRAF-mutant HGG, is poor.

• Efficacy of BRAF targeted therapy in the recurrent/refractory setting is limited.

• Upfront BRAF targeted therapy is associated with superior outcomes.

Importance of the Study.

The historical outcome for pHGG patients with standard multimodal therapy, including surgery, radiation therapy, and chemotherapy is poor, including a subset of patients with BRAF-mutant tumors. While BRAF targeted therapy for these patients in the recurrent setting may result in objective responses, the durability is generally short, and overall efficacy is limited. The upfront implementation of BRAF and MEK inhibitor therapy for pHGG patients is being investigated in an ongoing Children’s Oncology Group clinical trial (NCT03919071). We performed a multi-institutional retrospective analysis of BRAF-mutant pHGG patients treated with off-label BRAF +/– MEK targeted therapy and observed superior PFS and OS compared to a matched historical control cohort. These findings suggest that molecular targeted therapy should be strongly considered as adjuvant treatment in the upfront setting for all pediatric/young adult patients with BRAF-mutant HGG, including off-label targeted therapy for those who do not have access to, or are ineligible to participate in a suitable clinical trial.

Pediatric high-grade glioma (pHGG) constitutes a heterogenous group of WHO grade III and IV central nervous system (CNS) tumors of glial origin (excluding ependymoma). As a group, pHGG are the most common malignant pediatric CNS tumors and represent a major cause of cancer-related death among children.¹ Despite aggressive multimodal therapy, the outcome remains poor, with approximately 20% of children alive three years from diagnosis.^2,3 Recent advances in cancer genomics have uncovered molecularly and clinically distinct subgroups of pHGG, allowing for improved molecular classification and risk stratification. Importantly, these efforts have also led to the identification of potential therapeutic targets, including alterations in the mitogen-activated protein kinase (MAPK) pathway.^4–6

Approximately 5 to 10% of pHGG are driven by somatic MAPK pathway alterations, most commonly by somatic point mutations in the BRAF oncogene.^6–8BRAF mutations (such as the most common BRAF^V600E mutation) are recognized as important oncogenic drivers in various malignancies, including cutaneous melanoma, colorectal cancer, papillary thyroid carcinoma, nonsmall cell lung cancer, and pediatric low-grade glioma (pLGG).^9–11 These mutations lead to the constitutive activation of downstream MEK/ERK signaling, promoting cell growth, proliferation, and survival. Multiple BRAF inhibitors with high affinity to the mutant protein have been developed and tested in clinical trials, leading to the regulatory approval of dabrafenib and vemurafenib by the US Food and Drug Administration (FDA) for several indications.^12–15 However, despite promising initial responses, many patients eventually develop treatment resistance to monotherapy BRAF inhibition, leading to disease progression or recurrence.^16,17 Combination of BRAF and MEK inhibitors has consequently been investigated for the treatment of adult patients with BRAF-mutant melanoma, with some evidence pointing to the superior efficacy of combination therapy.^18–20 In human BRAF-mutant glioma, several mechanisms of resistance to BRAF inhibition have been proposed, including BRAF dimerization, loss of NF1, activation of receptor tyrosine kinases, CRAF upregulation, and loss of negative regulation of the PI3K/mTOR pathway.²¹

The early success of BRAF inhibitors in adults, including that seen in BRAF^V600-mutant CNS metastatic disease and primary CNS tumors, provided critically important rationale for the initiation of pediatric-specific CNS trials. In the early-phase clinical trials of dabrafenib and vemurafenib in pediatrics, children and adolescents with BRAF^V600-mutant low-grade gliomas (LGG) treated with these drugs had excellent durable responses, with tolerable side effects.^22,23 Preliminary data from phase I/II studies with monotherapy or combination of dabrafenib and the MEK inhibitor trametinib have also indicated that these regimens are safe and efficacious in the pediatric population.^24–26 Recently, MEK inhibitor therapy with selumetinib has also been shown to be effective in the treatment of pediatric patients with recurrent/progressive LGG in a prospective phase II clinical trial, with more durable responses observed in BRAF-fusion-positive versus BRAF mutation-positive patients.²⁷ As a result, off-label use of BRAF and MEK inhibitors is increasingly being utilized in the treatment of children with recurrent/refractory BRAF-altered LGG.²⁸

In stark contrast to children with LGG, the efficacy of molecular targeted therapy for patients with BRAF-mutant pHGG patients treated in the recurrent setting appears to be limited, with a reported median progression-free survival (PFS) of approximately three months.²⁹ These findings mirror retrospective data from adult patients with HGG, showing a median PFS of approximately five months, with the vast majority of patients progressing within 12 months.^30,31 In a recently published prospective phase 2 “basket” trial, response rates of 33% to the combination of dabrafenib and trametinib were reported in adults with recurrent BRAF^V600E mutated HGG; however, median time-to-progression remained dismal (< 4 months).³² To explore the efficacy of targeted therapy in the adjuvant upfront setting, the ongoing Children’s Oncology Group (COG) nonrandomized phase II study (ACNS1723) utilizes combination therapy dabrafenib and trametinib following radiotherapy for BRAF^V600-mutant pHGG (ClinicalTrials.gov identifier NCT03919071). While the results from this trial might eventually help establish a new standard of care, they are not expected for several years. At the same time, given the commercial availability of BRAF and MEK inhibitors, their off-label use for pediatric patients with BRAF-mutant tumors is increasing.

Given the paucity of published data concerning the upfront use of targeted therapy for BRAF^V600-mutant pHGG, we conducted a multi-institutional retrospective study of pediatric patients with molecularly confirmed BRAF^V600-mutant pHGG who received molecular targeted therapy in the upfront setting. We evaluated the clinicopathological features, genomic landscape, imaging responses, treatment-related toxicity, and clinical outcomes. In addition, we compared survival outcomes of patients in our cohort with similar patients treated with standard therapy (ie, radiation therapy and chemotherapy) from the published literature, including prospective clinical trials and large retrospective studies.

Methods

Patient Selection

This is a multi-institutional retrospective study that included patients seen at Dana-Farber/Boston Children’s Cancer and Blood Disorders Center, Memorial Sloan Kettering Cancer Center, Children’s Healthcare of Atlanta, Cancer and Blood Disease Institute at Children’s Hospital Los Angeles, Children’s of Alabama, Washington University School of Medicine and other collaborating institutions. Pediatric patients (age 1–21 years) with BRAF^V600-mutant HGG were identified using institutional databases. Patients were included in this study if they were diagnosed with a HGG (WHO grade III–IV), had molecular confirmation of a BRAF^V600- mutation, and were treated with off-label upfront targeted therapy following surgery with or without adjuvant radiation therapy. Patients ≤ 1 year of age at diagnosis were excluded from this study, based on our current understanding of the distinct biology of infant HGG in this age group, including those with BRAF^V600E mutations, as described by Clarke et al.³³

The institutional review board of each respective collaborating institution approved this retrospective review with a waiver of individual consent (DFCI 19-548).

Data Collection

Patients’ demographics, histopathological diagnoses, molecular findings, imaging data, clinical data including disease staging, tumor-directed therapy, drug dosing, toxicity, duration of follow-up, and clinical outcomes were extracted from respective electronic medical records. Histopathologic diagnoses were based on reports from each respective institution, in accordance with the WHO 2016 classification of tumors of the CNS.³⁴ De-identified genomic data were collected, unified, and summarized from various platforms used. All clinical data were de-identified at the collaborating sites prior to secure, electronic data transfer between sites via Research Electronic Data Capture (REDCap).³⁵

Imaging, Pathology, and Toxicity Analysis

Sequential MRI studies were available for all but one patient with measurable disease after surgery and prior to initiation of targeted therapy. Baseline imaging was obtained prior to initiation of targeted therapy, either postsurgery or postradiotherapy according to the institutional choice of upfront treatment. MR images were centrally reviewed by a pediatric neuro-radiologist (S.P.), with best response and time to best response evaluated based on the RAPNO criteria for pHGG.³⁶

Complete histopathology and molecular reports were available for all patients and were centrally reviewed by a pediatric neuropathologist (S.A). Grading for toxicities was based on the NCI Common Terminology Criteria for Adverse Events (CTCAE) v5.0.

Statistical Analysis

PFS and overall survival (OS) were estimated using the Kaplan-Meier method. PFS was defined as the time from initial diagnosis to the time of disease progression or death, while OS was defined as the time from initial diagnosis to the time of death. Patients who did not experience disease progression and/or death were censored at the time of last follow-up.

Survival outcome data were compared with BRAF^V600E-mutant pHGG patients from previously published, molecularly and clinically annotated cohorts from the (i) prospective HERBY trial that randomized newly diagnosed, localized pHGG patients (excluding pleomorphic xanthoastrocytoma and anaplastic ganglioglioma) to receive standard of care chemoradiotherapy versus standard of care chemoradiotherapy with additional bevacizumab; (ii) retrospective analysis of uniformly treated pHGG patients by Korshunov et al.; and (iii) integrated pHGG meta-analysis by Mackay et al.^3,6,7,37 Pertinent patient-level data were extracted from the supplementary materials available from these publications and survival differences between cohorts were assessed using the log-rank test. Of note, patients from the Korshunov publication that were also included in Mackay’s meta-analysis were identified and removed from the comparative analysis. Additionally, patients that were ≤ 1 year of age were excluded from the historical control cohort to match our study inclusion criteria, as detailed above.

Statistical analyses were performed using GraphPad Prism v9 (GraphPad Software, San Diego, CA) and R 3.6.0. OncoPrint was generated using R.

Results

Patient Characteristics

Nineteen newly diagnosed patients with a median age at diagnosis of 11.7 years (range: 2.3–21.4) were identified and included in this study. Four patients had been followed with surveillance imaging for a prolonged period of time (range 16 months to 16 years) for a presumed (n = 2) or histologically confirmed (n = 2) low-grade tumor before being diagnosed with a HGG (“secondary HGG”, Table 1). Sixteen patients (16/19; 84.2%) had localized disease while three patients (3/19; 15.8%) had multifocal/metastatic disease at the time of HGG diagnosis. Most patients had tumor originating in the cerebral hemispheres (11/19; 57.9%), followed by midline location in six (6/19, 31.6%) and multifocal disease in two (2/19, 10.5%). Full neuroaxis imaging at diagnosis was available for all but two patients. All patients underwent a neurosurgical procedure for initial diagnosis of HGG; near or gross total resection was performed in eight, subtotal resection in three, and biopsy only in eight patients. Sixteen patients (16/19; 84.2%) received radiation therapy following surgery, of which fifteen received focal irradiation while one received craniospinal irradiation for metastatic disease. All patients received targeted therapy with a BRAF inhibitor, with (n = 11) or without (n = 8) the addition of a MEK inhibitor. Patient characteristics are summarized in Table 1.

Table 1.

Demographics, Clinical, and Molecular Characteristics

ID	Age at Dx (years)	Secondary HGG	Histologic diagnosis	Tumor location	Multifocal/ metastatic	Co-alterationsc	Extent of surgery	RT	Best response	Time to progression (months)	Current status (time from HGG diagnosis, months)
1	3.9	No	HGNET	Midline/multifocal	Yes	H3K27M	Biopsy	CSI	PR	2	DED (8)
2	10.7	No	HGG, NOS	Hemispheric	No		NTR	Focal	CR	N/A	NED (43)
3	8.4	No	DMG	Midline	No	H3K27M	Biopsy	Focal	PR	N/A	AWD (29)
4	5.8	Yesa	Anaplastic Ganglioglioma	Hemispheric	No	CDKN2A/B loss	GTR	Focal	N/A	N/A	NED (26)
5	20.3	Yesb	HGG, NOS	Multifocal	No	CDKN2A/B loss	STR	Focal	SD	N/A	AWD (28)
6	2.9	No	Anaplastic Ganglioglioma	Hemispheric	Yes		Biopsy	No	PR	N/A	AWD (10)
7	21.4	No	DMG	Midline	No	H3K27M ATRX mutation	Biopsy	Focal	SD	N/A	AWD (13.5)
8	8.8	No	Glioblastoma	Hemispheric	No	CDKN2A/B loss TERT mutation	NTR	Focal	CR	34	AWD (39)
9	12.5	No	Anaplastic Astroblastoma	Hemispheric	No	CDKN2A/B loss	GTR	Focal	N/A	24.3	DED (25)
10	11.7	No	Glioblastoma	Hemispheric	No	CDKN2A/B loss	STR	Focal	PR	42.4	DED (60.5)
11	2.3	No	HGG, NOS	Midline	No		Biopsy	Focal	PR	N/A	AWD (78)
12	12.9	Yesa	HGG, NOS	Hemispheric	Yes	CDKN2A/B loss	Biopsy	No	PR	N/A	AWD (54)
13	14.5	No	Anaplastic Ganglioglioma	Midline	No	EGFR mutation	Biopsy	Focal	PD	0.5	DED (11)
14	4.8	No	Anaplastic Ganglioglioma	Midline	No		Biopsy	No	PRd	N/A	AWD (48.7)
15	16.6	No	Anaplastic Astrocytoma	Hemispheric	No	CDKN2A/B loss TERT mutation	GTR	Focal	N/A	N/A	NED (25)
16	15.0	No	HGG, NOS	Hemispheric	No		GTR	Focal	N/A	39	AWD (60)
17	14.4	Yesb	Epithelioid Glioblastoma	Hemispheric	No	CDKN2A/B loss	GTR	Focal	SD	14	AWD (40)
18	15.0	No	HGG, NOS	Hemispheric	No	CDKN2A/B loss	GTR	Focal	N/A	N/A	NED (19)
19	4.1	No	DMG	Midline	No	H3K27M	STR	Focal	SD	N/A	AWD (4)

^aPresumed prior diagnosis of LGG based on imaging surveillance.

^bHistologically confirmed prior diagnosis of LGG.

^cCoalterations of interest evaluated were H3K27M, ATRX, TERT promoter and EGFR point mutation, and CDKN2A/B homozygous loss.

^dPer report—imaging not available for central review.

AWD, Alive with Disease; CR, Complete Response; DED, Dead of Disease; DMG, Diffuse Midline Glioma, H3K27 mutant; Dx, Diagnosis; HGG, NOS High-Grade Glioma, not otherwise specified; HGNET, High-Grade Glioneuronal Tumor; NED, No Evidence of Disease; PD, Progressive Disease; PR, partial response; SD, Stable Disease; Tx, Treatment.

Histologic Diagnosis and Molecular Characteristics

Initial histologic diagnoses included HGG (n = 6), glioblastoma (n = 3), anaplastic ganglioglioma (n = 4), diffuse midline glioma (n = 3), high-grade neuroepithelial tumor (n = 1), anaplastic astrocytoma (n = 1), and anaplastic astroblastoma (n = 1). All patients had molecular confirmation of a BRAF^V600E-mutant tumor. Additional molecular analyses were performed according to local institutional practice.^38–43 A detailed list of the molecular testing performed for each patient is summarized in Supplementary Table 1, and Supplementary Table 2 provides the details of somatic changes and copy number alterations detected.

Central review of all patients’ histopathology and molecular reports confirmed a high-grade tumor diagnosis. There were no significant discordances between institutional diagnoses and the central review.

CDKN2A/B status was available for seventeen patients, with nine tumors showing homozygous loss. Molecular analysis for H3 mutations was available for eleven patients, with four patients harboring the H3 K27M mutation. As expected, all four H3 K27M-mutant tumors were located in the midline. Other pertinent concomitant molecular aberrations included TERT promotor mutations in two of twelve evaluable patients, EGFR mutation in one of fourteen evaluable patients, and ATRX alteration in one of thirteen evaluable patients. DNA copy number analysis was available for fifteen patients and revealed broad alterations in nine patients. The most common alterations found were chromosome 7 polysomy, chromosome 10 broad changes, and chromosome 22 broad changes, each found in five patients (33.3%). Histology and pertinent molecular features are summarized in Figure 1.

Fig. 1.

Oncoplot displaying molecular characteristics and clinical features of 19 patients with BRAF^V600E-mutant pHGG. Genomic alterations were detected using next generation sequencing and/or micro-array analysis (“Array”), and the relative incidence of alterations is shown. Abbreviations: CSI, craniospinal irradiation; DMG, diffuse midline glioma; EOR, extent of resection; GG, ganglioglioma; HGG, high-grade glioma; NGS, next generation sequencing.

Choice of Upfront Treatment and Reported Adverse Events

Eight (8/19; 42.1%) patients received adjuvant treatment with BRAF inhibitor monotherapy (five with vemurafenib, three with dabrafenib), while eleven patients (11/19; 57.9%) were treated with a combination of dabrafenib and trametinib (MEK inhibitor).

Nine patients experienced drug-related toxicity of any grade (1–5). The most common toxicities reported were fatigue, skin toxicity, and diarrhea. There were no grade 4 or 5 toxicities reported. Grade 3 toxicities were reported in four patients, which included neutropenia, weight loss, fatigue, and photosensitivity. Additional grade 1 and 2 toxicities reported included fever, rash, alopecia, diarrhea, and hypothyroidism. Toxicity led to treatment cessation in one patient and dose reduction in two patients. Treatment details, including individual choice of drugs, time from surgery or RT to targeted therapy initiation and toxicities are as summarized in Table 2.

Table 2.

Treatment Details

ID	Targeted therapya	Timing of therapy start (months)b	Toxicity (CTCAE grade)	Dose reduction for toxicity (reason)	Treatment cessation (reason)	Time on therapy (months)
1	Vemurafenib	1	No	No	No	2
2	Dabrafenib Trametinib	1	Neutropenia (3) Fatigue (2) Rash (1)	Yes (Neutropenia)	Yes (elective stop)	29
3	Dabrafenib Trametinib	1.5	Fever (1)	No	No	25
4	Dabrafenib Trametinib	1	Fatigue (1) Rash (1)	No	No	25
5	Dabrafenib Trametinib	1	Weight loss (3) Hypothyroidism (2)	Yes (Weight loss)	No	28
6	Dabrafenib	< 1	No	No	No	9
7	Dabrafenib Trametinib	1	No	No	No	11
8	Dabrafenib Trametinib	1.5	No	No	Yes (elective stop)	20.7
9	Dabrafenib Trametinib	1	Fatigue (3)	No	Yes (toxicity)	21
10	Dabrafenib Trametinib	1	N/A	N/A	No	42
11	Vemurafenib	4.5	Photosensitivity (3) Diarrhea (2) Hair loss (1) Rash (1)	No	No	66
12	Vemurafenib	< 1	Infection (2) Diarrhea (2) Photosensitivity (1) Fatigue (1)	No	No	54
13	Vemurafenib	1	Rash (1)	No	No	1
14	Dabrafenib Trametinib	< 1	Delayed growth (N/A)	No	No	49
15	Dabrafenib Trametinib	1	No	No	No	25
16	Dabrafenib	1	No	No	Yes (elective stop)	34
17	Vemurafenib	1	No	No	No	14
18	Dabrafenib	1.3	No	No	No	19
19	Dabrafenib Trametinib	1.3	No	No	No	4

^aDosing based on accepted pediatric recommended phase 2 dose (RP2D) or FDA approved adult dose.

^bTime periods provided are from completion of radiotherapy, or from surgery if no RT was utilized upfront.

Centrally Reviewed Imaging Response

Fourteen patients had evaluable disease at the initiation of targeted therapy. Among these, thirteen had deidentified images available for central review. Best response included complete response (CR) in two, partial response (PR) in six, and stable disease (SD) in four patients. One patient had progressive disease (PD) as best response. Deidentified images were not available for review in one patient, that had reported PR. For patients with evaluable responses (CR/PR), the median time to best response was 2.5 months (range: 1.5–11). Durable response (defined as treatment response lasting for ≥6 months) to treatment was observed in eight of the nine patients with PR/CR as best response. Imaging responses are summarized in Table 1.

Survival Outcomes

Median follow-up period for the entire cohort was 2.3 years (range: 0.3–6.5). The estimated three- and five-year PFS was 65% (95% CI 43–98) and 44% (95% CI 22–87), respectively, while the three- and five-year OS were both 82% [95% CI 65–100, (Figure 2A)]. Eleven patients (11/19, 57.9%) remain on upfront targeted therapy with a median time on therapy of 25 months (range: 4–73). Three patients suffered disease progression while on therapy, and one had imaging changes concerning for recurrence vs. radiation necrosis, leading to treatment change. Of the remaining four patients taken off therapy for various reasons (planned treatment completion, toxicity), three suffered subsequent disease recurrence, while one patient remains disease-free 14 months following cessation of therapy (Figure 3).

Fig. 2.

(A) Progression-free and overall survival of the entire study cohort. (B) Progression-free survival according to CDKN2A/B status (n = 17). Comparison of (C) Progression-free survival and (D) overall survival between BRAF-mutant and BRAF-wildtype patients in the pHGG historical control population. Comparison of (E) progression-free survival and (F) overall survival between the study cohort versus BRAF-mutant historical control cohort (BRAF^V600E).

Fig. 3.

Swimmer’s plot showing duration of treatment, timing and reason for treatment cessation, and timing of disease recurrence.

Assessment of known clinical risk factors in pHGG did not reveal a statistically significant difference in survival outcomes when comparing stage of disease (metastatic vs. localized disease), extent of resection (GTR/NTR vs. STR/biopsy), and BRAF monotherapy vs. BRAF/MEK combination therapy, though the numbers in each group were small. Among 17 patients with available CDKN2A/B status, no statistically significant difference in PFS was noted for patients harboring homozygous loss of CDKN2A/B versus CDKN2A/B intact (Figure 2B).

Among 13 patients with evidence of residual disease after initial surgical resection, the objective response rate (defined as the proportion of patients with CR or PR to treatment) was 69%. Nearly all six patients who underwent initial gross total resection maintained prolonged disease remission on targeted therapy, with one patient having concerns for progression vs. radiation necrosis leading to treatment change after 14 months, while the other five remained on upfront therapy without evidence for disease (range: 18–34 months). Strikingly, of the seven recurrences/progressions in our cohort, three occurred following discontinuation of targeted therapy (Figure 3).

Comparison to Historical Controls

BRAF-mutant pHGG is a rare disease, and published outcome data for patients treated with standard therapy is therefore limited. We identified 35 patients with BRAF-mutant pHGG from relevant published patient cohorts of pHGG for which molecular data were available. The HERBY trial was a phase II, open-label, randomized trial for newly diagnosed, localized pHGG, which evaluated the addition of bevacizumab to standard of care chemoradiotherapy.³ Korshunov and others assessed the prognostic significance of genomic and epigenetic alterations among 202 uniformly treated pHGG patients.⁷ Lastly, Mackay et al. described an integrated analysis of more than 1,000 pHGG patients, but specific details on treatment were not included.⁶ PFS and OS were available for the HERBY trial and the Korshunov publication, while only OS was reported in the Mackay publication. Eighteen patients from the Korshunov publication that were also included in Mackay’s meta-analysis were removed from our analysis to avoid duplicate data.

We first evaluated the outcomes within the historical cohort, comparing the survival outcomes based on BRAF mutation status. In total, 288 patients were identified for which BRAF mutation status by sequencing and survival outcomes including PFS were known [BRAF-mutant (n = 28); BRAF wildtype (n = 260)]. For this historical cohort, the 3-year PFS for known BRAF-mutant versus BRAF-wildtype patients were 17% (95% CI 5–36) and 18% (95% CI 14–24), respectively (Figure 2C, P = .15); whereas the 3-year OS for known BRAF-mutant versus BRAF-wildtype patients were 57% (95% CI 40–75) and 30% (95% CI 23–36), respectively (Figure 2D, P = .003).

For the comparative analysis, we combined patients from all three cohorts. In total, the HERBY trial enrolled 121 patients, seven of which had a confirmed BRAF^V600E mutation. the Korshunov paper described a total of 202 pHGG patients, 19 of whom had BRAF^V600E-mutant tumors. The Mackay paper added 9 unique BRAF^V600E-mutant pHGG patients (excluding 18 from the Korshunov et al. paper). We combined these patients with BRAF^V600E mutations (n = 35) for comparative survival analysis. The 18-month PFS rate was 83% (95% CI 67–100) in our cohort, compared to 42% (95% CI 26–66) in the BRAF-mutant historical cohort (Figure 2E, P = .001). Three-year OS was 82% (95% CI 65–100) in our cohort, compared to 44% (95% CI 29–67) in the BRAF-mutant control cohort (Figure 2F, P = .03).

Utilizing the available molecular data, we compared the outcomes between our cohort and the combined historical control cohort based on two additional molecular alterations/biomarkers of interest (CDKN2A/B and histone 3 status). For patients with known CDKN2A/B loss, our cohort had significantly better PFS (18-month PFS of 89% versus 14%; P = .001) and superior OS (3-year OS of 88% versus 21%; P = .04), when compared to the historical cohort (Figure 4A,B). For patients with known intact CDKN2A/B, our cohort had significantly better PFS (18-month PFS of 88% versus 52%; P = .025) and a longer OS although not statistically significant (3-year OS of 86% versus 51%; P = .19), when compared to the historical cohort (Figure 4C,D). For patients known to be histone 3 wildtype, our cohort had superior PFS and OS (Figure 4E,F), with 18-month PFS of 86% versus 43% (P = .02) and 3-year OS of 100% versus 50% (P = .04). For the four patients with concurrent K27M mutation in our cohort, median PFS and OS were not reached. This result appears favorable compared to historical controls (n = 3), however, the numbers were too small to reach a formal conclusion (Supplementary Figure 1A,B).

Fig. 4.

Comparison of progression-free survival between the study cohort and the BRAF-mutant historical control for patients with known (A) CDKN2A/B loss, (C) intact CDKN2A/B, and (E) histone 3 wildtype. Comparison of overall survival between the study cohort and the BRAF-mutant historical control cohort for patients with known (B) CDKN2A/B loss, (D) intact CDKN2A/B, and (F) histone 3 wildtype.

Discussion

In our cohort of 19 patients with BRAF-mutant pHGG treated upfront with molecular targeted therapy including a BRAF inhibitor +/– MEK inhibitor, we observed durable responses in the vast majority of patients. This novel treatment approach was associated with excellent 18-month PFS and three-year OS rates of >80%, compared to approximately 40% in a combined historical cohort treated with conventional therapy. Strikingly, of the seven recurrences in our cohort, three occurred following treatment discontinuation, suggesting the need for continued targeted therapy to sustain responses. Importantly, molecular targeted therapy was generally well-tolerated, with anticipated and manageable toxicities that led to permanent dose reductions in two patients only and treatment cessation in one.

Consistent with previous publications on BRAF-mutant pHGG, the most common co-existing alteration found in our patient population was homozygous deletion of CDKN2A/B, which was present in almost half of our cohort.⁶ Importantly, loss of CDKN2A/B has emerged as an unfavorable prognostic biomarker across a wide spectrum of CNS tumors, including pediatric LGG, pediatric HGG, ependymoma, and IDH-mutant glioma when treated with conventional therapy.^44–48 In our study cohort treated with molecular targeted therapy, CDKN2A/B loss was not associated with inferior survival outcome, but we acknowledge the small sample size may contribute to this finding. Of note, within the CDKN2A/B loss subgroup, four patients had been followed with surveillance imaging for a prolonged period of time for a presumed (n = 2) or histologically confirmed (n = 2) low-grade tumor before being diagnosed with a HGG. This observation is in line with previous publications highlighting a risk of malignant transformation in this patient population.^45,49

Tumors from four patients in our cohort were found to have co-existing BRAF^V600E and H3 K27M mutations, which is considered a rare occurrence in pHGG.⁶ Two of these patients had PR as best response, while the other two continue to have stable disease as best response. One of the patients with an initial PR suffered rapid disease progression thereafter. Our observation suggests that BRAF targeted therapy administered upfront may also be effective for patients with BRAF^V600E co-mutated diffuse midline glioma harboring the H3 K27M mutation, which is generally considered to carry a dismal prognosis.

To compare survival outcomes from our study cohort with historical controls, we utilized three previously published pHGG patient cohorts.^3,6,7,37 Importantly, within the historical cohort, patients with BRAF-mutant tumors had similarly poor PFS when compared to those with BRAF-wildtype tumors. While a modestly better OS was noted for patients with BRAF-mutant tumors within the historical cohort, this observation needs to be interpreted carefully, as data regarding the use of salvage regimens is lacking. Compared to the historical cohort, PFS and OS from our study cohort were significantly superior, supporting a survival advantage when utilizing targeted therapy in the upfront setting. Notably, the HERBY trial excluded patients with multifocal disease and all seven patients with BRAF^V600E mutations underwent a gross or near-total resection of their tumor prior to starting radiation therapy, whereas the majority of patients in our study underwent subtotal resection or biopsy only.

Two patients in our study suffered rapid progression: patient 1, whose tumor also harbored an H3 K27M mutation, and patient 13, whose tumor was found to have an additional EGFR activating mutation (COSV51769339). These cases highlight potential intrinsic resistance mechanisms and could help direct future studies assessing tumor resistance to molecular targeted therapy in pHGG.

Despite the inherent limitations of our retrospective study, we showed that patients with BRAF^V600E-mutant pHGG treated with upfront targeted therapy following standard of care chemoradiotherapy had improved outcomes with significantly prolonged duration of disease control and survival time compared to their historical counterparts. Also encouraging was the small group of patients who showed durable responses to targeted therapy following surgery only, though we acknowledge this apparent benefit requires validation in larger, prospective trials. Lastly, upfront molecular targeted therapy appeared to be beneficial across the entire clinical and biological spectrum of BRAF-mutant pHGGs except pleomorphic xanthoastrocytoma, including those with known clinical and molecular risk factors associated with particularly poor outcome, for example, subtotal resection, multifocal disease, biallelic CDKN2A/B loss, and H3 K27M mutation.

Our study has several limitations, predominantly related to the retrospective study design and rarity of the disease. The retrospective nature of this study increases the likelihood of selection and information bias that could affect the observed results, and treatments received were not uniform. Another important limitation is the lack of long-term safety data when using novel targeted agents in the pediatric population, especially since our data suggest clinical benefit for prolonged use of these agents. The restricted sample size limits our ability to detect associations between specific clinical, histological, and molecular features within our patient population and patient outcomes, nor does it allow us to compare the efficacy and toxicity of specific targeted therapy regimens, for example, choice of a specific agent or difference between BRAF inhibitor monotherapy versus combination BRAK/MEK inhibitors. An additional weakness of our study is the lack of central pathology review, allowing for potential inter-observer variability between the different participating institutions. Lastly, the historical cohorts used for comparison include a large number of patients that underwent comprehensive molecular diagnostics, but we acknowledge that the incorporation of different patient cohorts, nonstandardized inclusion criteria, and lack of treatment uniformity between these papers introduce potential bias into our comparison.

Nevertheless, we believe that our study results provide a compelling rationale to include molecular targeted therapy as part of upfront treatment for all pediatric, adolescent, and young adult patients with BRAF-mutant HGG, considering the limited efficacy of molecular targeted therapy reported in the recurrent setting.^29–32 While the ongoing COG nonrandomized phase II trial ACNS1723 for patients with BRAFV600-mutant pHGG is applying the combination therapy with dabrafenib and trametinib following radiotherapy (ClinicalTrials.gov identifier NCT03919071), it is only open to patients treated at COG member institutions, and excludes patients with multifocal/metastatic or infratentorial disease. For patients who do not have access to participation in a suitable clinic trial or do not qualify, treating physicians are currently faced with the dilemma of recommending conventional versus off-label molecular targeted therapy. We believe that our data strongly favor the latter. While our study does not inform a particular choice of targeted agents, the findings suggest that once initiated, molecular targeted therapy should be continued until disease progression or unacceptable toxicity.

Conclusions

The outcome for patients with pHGG, including BRAF-mutant tumors, is poor when treated with standard therapy, and BRAF targeted therapy in the recurrent setting is of limited efficacy. We show that upfront molecular targeted therapy including a BRAF inhibitor +/– concomitant MEK inhibitor, following standard of care surgery and chemoradiotherapy, is well-tolerated and associated with superior PFS and OS compared to historical controls. Our findings suggest that incorporation of molecular targeted therapy should be strongly considered in all newly diagnosed pediatric, adolescent, and young adult patients with BRAF-mutant HGG, including those who may be considered too young for radiation therapy. Molecular testing for therapeutically targetable alterations, including BRAF mutations, should be performed at the time of diagnosis for all pediatric and young adult patients with HGG, and participation in suitable clinical trials should be encouraged.

Funding

This work was funded in part by the Marie-Josée and Henry R. Kravis Center for Molecular Oncology and the National Cancer Institute Cancer Center Core Grant P30 CA008748.

Conflict of interest statement. The authors have no conflicts of interest to declare that are relevant to the material or content of this article.

Authorship statement. Experimental Design: K.K.Y., T.R., S.N.C., M.A.K. Implementation: K.K.Y., T.R., M.J., S.F.S., K.P., M.K.R., J.K.B., G.M., H.N.T., S.D., K.K., G.F., J.I.K., M.A.H., A.K., K.W., D.A., T.M., M.A.K. Analysis: K.K.Y., T.R., A.M., S.A., S.P.P., J.W.T., S.M., S.N.C., M.A.K. Interpretation of the data: All authors. Writing of manuscript: All authors.