Rare Pediatric Invasive Gliofibroma Has BRAF^V600E Mutation and Transiently Responds to Targeted Therapy Before Progressive Clonal Evolution

CASE HISTORY

A previously healthy 20-month-old boy presented with 3 weeks of difficulty walking, emesis, headaches, and asymmetric smile after a viral illness. Physical examination demonstrated left-sided extremity weakness and right-sided facial weakness. Magnetic resonance imaging (MRI) showed a large enhancing infiltrative mass (4.7 × 5.8 × 3.9 cm) within the left midbrain/tectum extending into the left temporal lobe and superior cerebellar vermis (Fig 1) with severe obstructive hydrocephalus and no brain/spine leptomeningeal spread. Significant residual remained after partial resection (Fig 1). A ventriculo-peritoneal shunt was subsequently placed to relieve the hydrocephalus.

FIG 1.

Magnetic resonance imaging (MRI) demonstrates a large tumor that improved with targeted therapy. Axial and sagittal T1-weighted FLAIR (fluid attenuation inversion recovery) images of post–contrast MRI at the times indicated. Sections shown are matched at sagittal and axial views chosen to demonstrate maximal extent of residual tumor. (A and B) Preoperative imaging at diagnosis. MRI demonstrates a large enhancing mass centered within the left midbrain/tectum extending into the left temporal lobe and superior cerebellar vermis. (C and D) MRI 1 day after partial tumor resection. (E and F) MRI 3 weeks postresection, at the time of clinical deterioration during an unsuccessful dexamethasone wean, and before initiation of targeted therapy. Images demonstrate increased diffuse enhancement. (G and H) MRI 9.5 weeks after initiation of targeted therapy. Images demonstrate a decrease in size of the residual brainstem and left medial temporal lobe tumor, which have considerably less enhancement compared with prior studies. (I-N) MRI at 6, 9, and 12 months of targeted therapy demonstrates stable disease. (O and P) MRI at 15 months shows a newly emerging dominant enhancing mass involving the posterior left temporal lobe with associated mass effect and deviation of the midbrain.

Microscopically, the tumor showed a biphasic growth pattern with an astrocytic component arranged in nests and cords surrounded by anastomosing bands of mesenchymal tissue (Fig 2). The astrocytic component expressed glial fibrillary acidic protein (GFAP) and showed mild-to-moderate cellular atypia, low mitotic rate, and Ki-67 labeling index of 6% to 7% without palisading necrosis or glomeruloid endothelial proliferation, which is consistent with a low-grade glial neoplasm. The mesenchymal component was cytologically bland, showed no mitotic activity, and did not express GFAP. The mesenchymal elements, but not the astrocytic elements, showed positive reticulin and collagen stains. These features were consistent with a diagnosis of gliofibroma, which consists of both mesenchymal and glial elements and has a low proliferation index (Fig 2).

FIG 2.

Tumor histopathology reveals a gliofibroma with BRAF^V600E mutation. (A) Hematoxylin and eosin (H&E) –stained section of tumor showing biphasic growth pattern with nests of glial cells (black arrow) surrounded by bands of mesenchymal tissue (white arrow). (B) The biphasic nature of the tumor is more easily visualized using Masson’s trichrome, which stains the collagen fibers bright blue. (C) The mesenchymal component stains for reticulin (white arrow), whereas the glial component does not (black arrow). (D) The proliferative index as measured by Ki-67 was relatively low. Nuclei not in G₀ phase are positive for Ki-67 (DAB, brown), whereas nuclei in G₀ phase are negative (hematoxylin, blue). (E and F) In contrast to the trichrome and reticulin stains, glial fibrillary acidic protein (GFAP) is expressed in the glial component of the tumor (DAB, brown) and not in the mesenchymal component (hematoxylin, blue). (G) Retained nuclear INI1 expression (DAB, brown). (I) H3K27me3 immunohistochemistry (IHC), a surrogate marker for the H3F3A K27M mutation, showed nuclear retention indicating that the mutation was not present. (J and K) Large areas of the glial components of the tumor were positive for the BRAF^V600E-mutated protein (DAB, brown, IHC), although some regions did not express it. All of the mesenchymal components of the tumor were negative for BRAF^V600E protein. A BRAF^V600E mutation was confirmed by molecular methods in the tissue blocks that were positive by IHC. (L and M) p16 protein is retained in vascular endothelial cells (DAB, brown, IHC), but not expressed in the tumor cells. This is consistent with molecular analysis, which demonstrated deletion of CDKN2A and CDKN2B in the tumor. (N-R) At recurrence, the tumor showed high-grade regions that were invasive into the lower-grade regions (N), were highly cellular (P), and had abundant mitotic figures (Q; panels correspond to the dotted regions in panel P). The tumor continued to express the mutant BRAF^V600E in many of the high-grade regions, although as in the original low-grade tumor, there were areas that were negative for this mutation (R).

Significant clinical improvement followed the initiation of dexamethasone, partial resection, and relief of hydrocephalus; however, two careful attempts to wean the corticosteroids were unsuccessful. During the second attempt, the patient experienced severe neurologic deterioration, which necessitated hospitalization and reinitiation of corticosteroids. Magnetic resonance (MR) spectroscopy at that time demonstrated high-grade features—for example, prominent choline—in the residual mass (Fig 3).

FIG 3.

Magnetic resonance (MR) spectroscopy demonstrates a tumor with high-grade features that improved with initiation of targeted therapy. (A-D) Three weeks postresection, at the time of clinical deterioration during an unsuccessful dexamethasone wean and before the initiation of targeted therapy. MR spectroscopy is concerning for higher grade tumor. Specifically, prominent choline (Cho) relative to creatine (Cr) is generally observed in more proliferative tumors. MR spectroscopy also demonstrates metabolic heterogeneity with markedly different levels of glycine (Glyc) in different parts of the lesion with unclear significance. Lactate (Lac) is observed in both regions of interest at approximately equal levels. Spectra were acquired on a 3T scanner with PRESS (TE, 35 ms). The thin black line is the measured signal. Superimposed is the fit (thick orange line) that is used to quantify metabolite levels. (E and F) Nine and a half weeks after initiation of targeted therapy. Cho is less prominent relative to Cr in the residual tumor. The spectrum was acquired on a 3T scanner with PRESS (TE, 35 ms). (G and H) At 9 months of targeted therapy, the spectrum again demonstrates moderate Cho relative to creatine. The spectrum was acquired with PESS (TE, 35 ms) on a 1.5T scanner. Spectroscopy at 12 months was similar to the one at 9 months (not shown). (I and J) At 15 months of targeted therapy, the spectrum in the region of the newly emerging mass shows significantly increased Cho that is consistent with proliferative tumor. The spectrum was acquired with PRESS (TE, 35 ms) on a 1.5T scanner.

Molecular studies using the OncoScan (Thermo Fisher Scientific, Waltham, MA) chromosomal microarray platform, which has been validated for BRAF^V600E, revealed BRAF^V600E mutation and biallelic deletion of CDKN2A (Table 1). In children, CDKN2A/B loss has been found in gliomas and ependymomas.^1-3 Immunohistochemistry revealed mutant BRAF^V600E in the glial elements of the tumor and absence of p16 (CDKN2A) in both glial and mesenchymal components (Fig 2), which is consistent with the molecular analysis that shows BRAF^V600E mutation and biallelic deletion of CDKN2A/B (Table 1).

TABLE 1.

Chromosomal Microarray Findings of the Patient’s Original Gliofibroma Tumor

The infiltrative nature of the tumor; the inability to wean the patient off dexamethasone; the molecular studies showing a BRAF^V600E mutation and deletion of CDKN2A, which portend a worse prognosis in pediatric glioma patients^1,2,4; and the high-grade features on MR spectroscopy raised concern for the presence of high-grade infiltrative components in the residual tumor despite the low-grade appearance of its resected portions. Furthermore, in view of the poor prognosis of incompletely resected gliofibromas (Fig 4 and Data Supplement), absence of established therapy for them, and an actionable mutation found in the patient’s tumor, targeted therapy was initiated.

FIG 4.

Summary of clinical characteristics of published patients with gliofibroma. Data were collected from published reports queried most recently in March 2018. Our patient is also included in this analysis. Additional details on these patients are in the Data Supplement. Survival status (alive/dead) was not reported for five of the 44 published patients, and duration of survival was not reported for five others for whom survival status was reported. (A) Age of distribution at diagnosis of gliofibroma. (B) Overall survival for the 35 patients with gliofibroma for whom survival data were reported (of a total of 45). (C) Overall survival for patients with gliofibroma reported to have high-grade compared with low-grade histology. Total number of patients with high-grade gliofibroma was nine and low-grade was 36. Duration of patient survival or survival status were not reported for two high-grade and eight low-grade tumors. (D) Overall survival of patients who had complete resection compared with partial resection. Survival data were reported for 35 of 45 patients.

Reports of emerging resistance to BRAF inhibitor monotherapy in patients with melanoma and improved survival with combined therapy prompted us to recommend combining BRAF inhibitor with mitogen-activated protein kinase (MAPK) kinase (MEK) inhibitor.⁵ The patient thus received vemurafenib 550 mg/m² per dose two times per day (a BRAF inhibitor) and trametinib 4 mg/m² per day (an MEK inhibitor). Dexamethasone was successfully discontinued 2.5 weeks later. MRI at 9.5 weeks after starting targeted therapy revealed significant tumor shrinkage (Fig 1). MR spectroscopy demonstrated a pattern consistent with less aggressive tumor in the previously evaluated regions (Fig 3). In particular, choline, a metabolite that is involved in cell membrane metabolism and is generally elevated in more proliferative tumors, was at more moderate levels. Subsequent MRIs at 6, 9, and 12 months (Fig 1) on targeted therapy revealed a stable, smaller tumor. Spectroscopy continued to suggest a lower grade lesion compared with pretherapy spectroscopy, similar to that at 9.5 weeks on therapy (Fig 3 and not shown). Mild elevation of liver transaminases prompted a decrease in vemurafenib to 380 mg/m² per dose two times per day starting at month 7 of therapy. The patient otherwise tolerated therapy well with stable echocardiogram, transient mild hirsutism, mild sun sensitivity, mild pruritus, and pustular furuncles that were successfully managed by dilute bleach baths.

Routine MRI at 15 months while on BRAF inhibitor (vemurafenib) plus MEK inhibitor (trametinib) therapy revealed new growth with imaging features of a high-grade tumor, similar to pretherapy MR spectroscopy (Figs 1 and 3). The patient was asymptomatic. Reresection showed gliofibroma with areas of low-grade morphology which was similar to the original tumor, along with extensive areas that now contained high-grade features (Fig 2). High-grade features included increased cellularity, high mitotic rate, and regions with markedly elevated mitotic index (Ki-67 positive in up to 70% to 80% of nuclei). On the basis of the high-grade features of the recurrent tumor, the patient proceeded to receive intensity-modulated radiation therapy to the tumor.

Molecular studies of the recurrent tumor again revealed the BRAF^V600E mutation (OncoKids next-generation sequencing panel and OncoScan chromosomal microarray) and the CDKN2A biallelic deletion (OncoScan). Similar to the original tumor, immunohistochemistry for BRAF^V600E again showed areas with the mutant protein and areas without it (Fig 2). Regions with high-grade features did express the mutant BRAF. Of importance, using OncoKids, two new mutations that were not present in the original tumor were revealed in the recurrence: PDGFRA [pNM_006206.4:c.2524G>T (p.Asp842Tyr), also known as D842Y] and PTPN11 [NM_002834.3:c.215C>T (p.Ala72Val)].

Four days before completing radiation therapy, the patient developed headaches, nausea, and emesis. MRI showed new extensive subependymal tumor spread and infiltrative tumor in the anterior part of the lateral ventricle in an area outside the radiation field. Per the parents’ request, reresection was performed. OncoKids on this tumor, obtained just 2 months after the first recurrence, showed the prior mutations along with two new mutations in TP53 [NM_000546.5:c.587G>A(p.Arg196Gln)] and PIK3CA [NM_006218.3:c.3140A>G (p.His1047Ar); Table 2]. The patient expired 20.5 months after presentation and 5.5 months after the first regrowth of the tumor.

TABLE 2.

OncoKids Next-Generation Sequencing Cancer Panel Findings From Second Reresection (second recurrence, third tumor sample)

Written informed permission for the use and disclosure of this patient’s protected health information for research purposes was obtained from the family.

DISCUSSION

Gliofibroma was first reported in 1978 by Friede as a “peculiar neoplasia of collagen-forming glia-like cells,” as noted in the title of Friede's article.⁶ As discerned from case reports, these tumors are usually considered to be benign and localized and are only infrequently disseminated or in multiple sites.^7-38

Review of the literature revealed 44 cases of gliofibroma,^7-38 with our patient being the 45th case. Figure 4 and the Data Supplement summarize these gliofibroma cases. Of 45 patients, most were diagnosed in the first decade of life (median age, 9 years). Thirty-four patients (76%) presented at or before age 20 years (Fig 4). There was slight female predominance (males, n = 18; females, n = 27). On the basis of author-reported histopathology, patients with high-grade tumors had significantly worse overall survival compared with low-grade tumors (P < .001; Fig 4), similar to prior assessment.¹⁵ Survival of patients who underwent complete resection was not significantly different from that of patients with partial resection or biopsy (P = .108; Fig 4). To date, only two patients with gliofibroma have been reported to have experienced a response to chemotherapy. Both were infants with unresectable low-grade gliofibromas. One experienced a response to carboplatin and vincristine³⁶ and the other to metronomic vinblastine.³² Analysis of the rare published cases presented here is limited in that it depends on individual case reports that mostly provide relatively short and incomplete follow-up and may represent an unbalanced cohort of patients. Thus, although this summary, to our knowledge, is the most comprehensive review of published gliofibroma cases to date, these limitations must be considered.

Our patient provides four novel findings with regard to gliofibroma: This is the first report, to our knowledge, on gliofibroma with a molecularly proven BRAF^V600E mutation and CDKN2A biallelic deletion; the first patient with gliofibroma reported to receive targeted therapy and to respond to BRAF plus MEK inhibition, albeit only for limited time; the first gliofibroma described to acquire PDGFRA and PTPN11 mutations upon recurrence as a high-grade glioma resistant to BRAF and MEK inhibition; and a rare opportunity to view the rapid molecular evolution of this tumor, with (first recurrence) and without (second recurrence) therapy with BRAF plus MEK inhibition. Of interest, the BRAF^V600E mutation in our patient was only found in the glial elements of the tumor, which suggests that the mesenchymal regions may be reactive to the glial pathology rather than independently tumorigenic. Whereas p16 (CDKN2A) immunohistochemistry was negative in both glial and mesenchymal components, it is not known if this is a result of the deletion of CDKN2A in both compartments or if CDKN2A is deleted in one compartment and its expression is merely low in the other. It is interesting that in another glial tumor, ganglioglioma, BRAF^V600E-expressing cells were mostly those with neuronal capacity or both neuronal and glial capacity.³⁹ In contrast to that shown in our patient, in those gangliogliomas, p16 was robustly expressed.³⁹

BRAF^V600E oncogenic mutation is reported in 17% of pediatric low-grade gliomas.^1,4,40-42 To date, BRAF^V600E mutation has not been reported in gliofibromas. One case report stated that BRAF mutations are present in approximately 50% of gliofibromas⁷; however, the authors’ case did not have a BRAF mutation and no reference was provided to support the statement. In pediatric patients with low-grade gliomas who received chemotherapy and irradiation, BRAF^V600E was associated with worse outcomes. Whereas 10-year progression-free survival in patients with non-BRAF^V600E gliomas was 60.2%, in those with mutant BRAF^V600E it was only 27% (P < .001).⁴ Moreover, incomplete resection or CDKN2A deletion independently contributed to poor outcomes among children with BRAF^V600E mutant low-grade gliomas.⁴ The clinically aggressive features of our patient’s original tumor fit this pattern of BRAF^V600E mutation combined with CDKN2A biallelic mutation.

An infant with gliofibroma that harbored KIAA1549-BRAF fusion was described by Schroeder et al.³² KIAA1549-BRAF fusions also activate the MAPK pathway and are observed in more than 50% of incompletely resected pediatric low-grade gliomas.⁴³ Their prognosis is better than fusion-negative tumors—5-year progression-free survival is 61% ± 8% for fusion-positive tumors versus 18% ± 8% for fusion-negative tumors.^43,44 Compared with KIAA1549-BRAF fusion-positive pediatric low-grade gliomas, BRAF^V600E mutant tumors have significantly worse survival.⁴ As fusion-positive gliomas treated with BRAF inhibitor monotherapy may paradoxically increase the activation of the MAPK pathway but may respond to MEK inhibitors,⁴⁵ careful molecular analysis for both point mutations—for example, BRAF^V600E—and fusions—for example, BRAF-KIAA1549—is needed before initiation of targeted therapy. This can be achieved best by a next-generation sequencing panel that covers both molecular alterations, such as the Children’s Hospital Los Angeles–developed OncoKids panel.⁴⁶

Considering the poor response of incompletely resected BRAF^V600E-mutated and CDKN2A-deleted pediatric low-grade gliomas to chemotherapy and irradiation,⁴ as well as the aggressive behavior of our patient’s tumor at presentation, we elected to treat him with the BRAF^V600E inhibitor vemurafenib combined with the MEK inhibitor trametinib. This was based on excellent response of BRAF^V600E-mutant gliomas and melanomas to combined BRAF and MEK inhibition compared with the high rate of resistance on BRAF-inhibitor monotherapy after initial response.^47-50 Mechanisms responsible for resistance of BRAF^V600E tumors to BRAF inhibitor monotherapy include an increase in BRAF dimers, BRAF splice variants, neuroblastoma RAS viral oncogene homolog (NRAS) and kirsten rat sarcoma viral oncogene homolog (KRAS) mutations, BRAF amplifications, MEK mutations, and others.^49,51,52 These render extracellular regulated kinase signaling insensitive to BRAF inhibitors while the downstream MEK remains a viable target. Combining a BRAF inhibitor with an MEK inhibitor is thought to help retain tumor responsiveness and to have lower adverse effects compared with BRAF monotherapy.^{47,49,50,53-55} Trametinib and other MEK inhibitors are in clinical trials for pediatric patients. Toxicity studies in young rats showed decreased bone growth, delayed sexual maturation, and corneal dystrophy.⁵⁶ Thus, monitoring for these should be considered in treated children. Whereas preclinical research suggested a benefit to combining a BRAF inhibitor and/or MEK inhibitor with a cyclin-dependent kinase 4/6 inhibitor in tumors with both BRAF^V600E and CDKN2A deletion,^57,58 peer-reviewed clinical data on the safety and efficacy of such combination were not available when the patient presented.

Our patient’s tumor showed clinical and radiographic response to vemurafenib and trametinib for more than 12 months. Molecular studies of the recurrent tumor revealed persistence of mutant BRAF, which indicated the emergence of resistance to BRAF inhibition. Two mutations—PDGFRA-D842Y and PTPN11-A72V—in the recurrent tumor had not been detected in the original tumor. Although it is possible that intratumor heterogeneity was the basis for their absence in the sample tested from the original tumor, their coexistence with the BRAF^V600E mutation and CDKN2A deletion in the recurrent sample suggests that they are newly acquired and may have played a role in the recurrence. It is not known if these new mutations are in tumor cells that express the mutant BRAF protein or in those that stain negative for it (Fig 2R).

Overexpression and mutations in PDGFRA are frequent in pediatric gliomas and in other cancers, with mutations occurring in more than 20% of high-grade gliomas but only rarely in low-grade ones.^59-61 The PDGFRA mutation, in addition to the aggressive nature of our patient’s recurrent glioma, supported the use of radiation therapy for the recurrence.⁶¹ Mutations in D842, located in the activation loop of PDGFRA,⁶² comprise more than one third of PDGFRA substitution mutations reported to date in the COSMIC Catalogue (610 of 1,705), with the majority (95%) being D842V.^63,64 PDGFRA-D842V renders PDGFRA resistant to imatinib in vitro,^64-66 but it remains sensitive to crenolanib, a newer PDGFRA inhibitor that is currently in clinical trials.^63,64 However, PDGFRA-D842Y, the mutation in our patient’s tumor, is moderately resistant to imatinib in vitro and is not as sensitive to crenolanib as the D842V mutant or the wild-type receptor.⁶⁵

PTPN11 (SHP-2) is mutated in the germline in Noonan and Leopard syndromes as well as somatically in cancer.⁶⁷ Across cancers, PTPN11 is mutated in 1,320 (1.9%) of 70,712 cancer samples reported in COSMIC.⁶⁴ PTPN11 mutations are observed in both low-grade and high-grade pediatric gliomas.⁶¹ The PTPN11 Ala72Val (A72V) mutation in our patient’s recurrent tumor is located in the N-terminal SH2 domain of PTPN11 and is predicted to be a gain-of-function mutation.⁶⁸ Of the total 1,285 PTPN11 substitution mutations reported in COSMIC, 180 are in alanine 72, the second most mutated amino acid after glutamine 76. Seventy-eight of the reported 180 Ala72 mutations are A72V. Several PTPN11 inhibitors are now beginning clinical trials in cancer in adults.⁶⁹

In this respect, it is interesting that in mice PTPN11 can mediate glioma formation downstream of PDGFRA overexpression and INK4A (CDKN2A) loss.⁷⁰ It is therefore conceivable that, with this gliofibroma’s PDGFRA-D842Y and PTPN11-A72V activating mutations and the biallelic deletion in CDKN2A, this pathway may have contributed to the recurrent tumor in our patient.

The serial acquisition of new mutations on subsequent recurrent specimens is consistent with clonal evolution of disease. Our findings support those by Mistry et al,¹ who describe BRAF^V600E and CDKN2A deletion as early alterations in pediatric low-grade glioma that are undergoing transformation. Our patient acquired at least two new mutations with each progression. The most recent mutations—in TP53 and PIK3CA—are more deleterious and consistent with the rapid second progression. PIK3CA mutations are found in 21% of pediatric gliomas compared with 17% in adult glioblastomas^71,72 (Table 2).

In summary, to our knowledge, this is the first published report of a gliofibroma that harbors a BRAF^V600E mutation, the first report on a gliofibroma that responded to targeted therapy, and a unique view of clonal evolution of this rare tumor. Although this suggests that molecular screening may benefit patients with gliofibroma, especially those with incompletely resected and/or infiltrative or clinically aggressive tumors, it also indicates that broader drug combinations may need to be explored.

AUTHOR CONTRIBUTIONS

Conception and design: Kristiyana Kaneva, Mark D. Krieger, Anat Erdreich-Epstein

Administrative support: Jaclyn A. Biegel

Provision of study material or patients: Kristiyana Kaneva, Debra Hawes

Collection and assembly of data: Kristiyana Kaneva, Kee Kiat Yeo, Debra Hawes, Stefan Bluml, Mark D. Krieger, Anat Erdreich-Epstein

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST AND DATA AVAILABILITY STATEMENT

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Marvin D. Nelson

Patents, Royalties, Other Intellectual Property: Royalties from a book published in 2012

Expert Testimony: MDNelson,jr

No other potential conflicts of interest were reported.