Introduction

Pilocytic astrocytoma (PCA) is the most frequent pediatric low-grade glioma and is also not uncommon in young adults, accounting for 15.3% among children and adolescents [1, 2]. It is WHO grade I circumscribed glioma classified under other astrocytic tumors in the current 2016 WHO classification [3]. It commonly occurs in the cerebellum and the optic pathway but can occur throughout the neuroaxis [4]. It is predominantly sporadic but is the most common familial syndrome associated glial tumor, especially in neurofibromatosis 1 (NF1). PCAs are of favorable prognosis with > 95% of 10-year overall survival [1].

PCAs are mediated by constitutive activation of mitogen-activated protein kinase (MAPK) pathway through the in-frame fusion between oncogene BRAF and KIAA1549 genes (resulting from breakpoint at multiple sites of KIAA1549 and BRAF generating different fusion variants), through oncogenic BRAFV600E mutation and uncommonly through alterations in fibroblast growth factor receptor 1 (FGFR1). KIAA1549 Exon 16-BRAF Exon 9 fusion is the commonest followed by KIAA1549 Exon 15-BRAF Exon 9 and KIAA1549 Exon 16-BRAF Exon 11 [5,6,7]. Other uncommon reported fusion variants are Exon 18-10, 19-9, 16-10, 15-11, 17-10 of KIAA1549 and BRAF [5, 6]. Rare non-KIAA gene fusion variants such as SRGAP3-BRAF [8], FAM131B:BRAF [9, 10], GIT2I-BRAF [11], TMEM106B-BRAF fusions [12], and BRAF gain of function mutation are also reported [13]. While BRAFV600E mutation results in constitutive activation of BRAF even in its monomeric form and also induces the same effect on MAPK pathway activation as in BRAF fusions [14]. FGFR1 alterations also induce MAPK/ERK pathway upregulation through intragenic duplications of the FGFR1 tyrosine kinase domain resulting in FGFR1 autophosphorylation. However, this is relatively frequent in grade II diffuse gliomas (24%) than PCAs and LGGs (7.4%) [15].

Although both the BRAF (fusions and V600E) alterations exert their effect of tumorigenesis through the MAPK pathway, but their frequency pattern varies according to age of diagnosis and location. KIAA1549-BRAF fusions are more common in younger age and in cerebellar location, whereas BRAFV600E is more common in non-cerebellar location [13, 16, 17]. KIAA1549-BRAF fusion was reported in around 60–70% of all PCAs, whereas BRAFV600E mutation in < 10% of both pediatric and adult cases [17,18,19].

This is a single institutional study aimed at evaluating the frequency of BRAFV600E mutation and KIAA1549-BRAF (16-9, 15-9, and 16-11) fusions in PCAs and to correlate it with clinicopathological features.

Materials and methods

Tumor samples

The study is approved by our institutional review board, comprised of 358 consecutive (2011–2018) histologically diagnosed cases of PCA (including pilomyxoid astrocytomas (PMA)). Cases with syndromic association (especially NF1) and inadequate tissue were excluded. Formalin-fixed paraffin-embedded (FFPE) tumor tissues were used for the study, which were essentially primary samples at diagnosis; 205 (pediatric 157/276, adult 48/82) were in-hospital and 153 (pediatric 119/276, adult 34/82) were referrals. Clinical and demographic details were collected from electronic medical records.

Immunohistochemistry

p53 (Dako, D07, 1:50 dilution)and ATRX protein (1:750; Polyclonal; Sigma; USA) expression was performed on 4 μm representative FFPE sections using the Ventana Benchmark XT autoimmunostainer by polymer detection kit.

p53 protein expression was graded as positive (> 50% tumor cells with strong intensity nuclear staining), focal positive (10–50% with strong nuclear staining/> 10% tumor cells with moderate intensity staining) and negative (absence of any staining or staining in < 10% tumor cells). ATRX expression was interpreted as retained (nuclear staining in tumor cells), loss (no staining in tumor cells with reactivity of native cells like endothelial and/or neurons as an internal control), and non-contributory (no staining in internal control).

Molecular analysis: BRAF alterations

Selected FFPE blocks were subjected for detection of BRAF mutation by Sanger sequencing and KIAA1549-BRAF fusion transcript detection by Reverse transcriptase PCR (RT-PCR) for three different fusion transcripts; KIAA1549-BRAF 16-9 (124 bp), KIAA1549-BRAF 15-9 (159 bp) and KIAA1549-BRAF 16-11 (Table 1: primer sequences). See Appendix for the detailed methodology of BRAFV600 sequencing and BRAF fusion RT-PCR.

Table 1 Primer sequences
Full size table

Validation of KIAA1549: BRAF fusion transcripts

Sanger sequencing was performed on 10% of the KIAA1549-BRAF fusion-positive samples to validate and confirm the presence of KIAA1549-BRAF fusion transcripts (Appendix). The sequences were aligned and compared with the reference sequences using the BLAST alignment tool.

Statistical analysis

The data was analyzed using SPSS 20.0 software. The BRAF mutation and KIAA1549-BRAF fusion results were correlated with the clinical and histological features using Fisher’s test and p value < 0.05 was considered statistically significant.

Results (Table 2 )

Sample size (n = 358)

214 were males and 144 were females (M:F ratio of 1.5:1). Age range was 1–50 years (interquartile range [IQR] 6–18 years; median10 years) with 276 pediatric cases (≤ 18 years) and 82 adults (19–39 years,77; >39 years, 5).

Table 2 BRAFV600E and KIAA1549:BRAF fusion with clinical and histological features
Full size table

Pediatric group (n = 276)

Clinical

M:F was 1.4:1 and IQR was 5–13 years (0–3 years, 29; 4–14 years, 203; 15–18 years, 44) with median age of 8 years.The most common location was cerebellar (n 123; 44.6%) followed by 3rd ventricular (n 68; 24.6% (suprasellar, 53; 3rd ventricle,11; hypothalamic, 3; pineal, 1)), cerebrum (n 25; 9.1%), spinal (n 17; 6.2%), optic nerve and chiasma (n 15; 5.4%), thalamic (n 14; 5.1%), and brainstem (n 14; 5.1%). Histologically, 270 were PCA while 6 were PMAs.

BRAF alterations

Interpretable BRAFV600E was in 246 cases, while the remaining 30 were uninterpretable (17 could not be amplified for beta-actin gene and 13 showed non-readable sequencing data). BRAFV600E was observed in 22 (8.9%) cases. All cases showed typical heterozygous BRAFV600E substitution replacing adenine in place of thymine nucleotide at codon 600 (Fig. 1).

Fig. 1
figure 1

Representative photomicrographs of pediatric PCA. a Cerebellar PCA (H&E; × 200) with electropherograms showing BRAFV600 and gel picture depicting KIAA1549-BRAF 16-9 fusion positive. b 3rdventricularPCA (H&E; × 200) with electropherograms showing BRAFV600E mutant (black arrow) and gel picture depicting KIAA1549-BRAF fusion negative. c Suprasellar PCA (H&E; × 200) with electropherograms showing BRAFV600 wild type and gel picture depicting KIAA1549-BRAF fusion negative

Full size image

KIAA1549-BRAF fusion PCR could be performed in 239 pediatric cases (37 failed due to degraded RNA), 99 (41.1%) cases showed fusion transcripts. Sixty-six cases were positive for KIAA1549-BRAF 16-9 transcripts (Fig. 1) whereas KIAA1549-BRAF 15-9 and KIAA1549-BRAF 16-11 fusion transcripts were detected in 21 and 12 cases respectively. None of the cases showed the presence of more than one fusion transcript and were mutually exclusive with BRAFV600E.

Of the 276 cases, both BRAF mutation and KIAA1549-BRAF fusion transcript expression could be analyzed in 226 cases with 53.5% (121/226) of cases showing BRAF alterations. 43.8% (n 99) were KIAA1549-BRAF and only 9.7% cases demonstrated the presence of BRAFV600E. All PMA cases within the group were negative for BRAF alterations.

p53 (n = 190) and ATRX protein (n = 272) expression

p53 protein expression was not available in 86 cases, 17 (8.9%) were positive, 47 (23.9%) were focal positive and 126 (66.3%) were negative. All 272 cases showed retained ATRX expression (4 were non-contributory).

Adult group (n = 82)

Clinical

M:F was 1.9:1 and age range was 19–50 years with IQR of 21–28 years (19–39 years, 77; > 39 years, 5), and median age 24 years. Cerebellum was the commonest location (n 27; 32.9%), followed by cerebrum (n 16; 19.5%), 3rdventricular (n 13; 15.9% (suprasellar, 8; 3rd ventricle, 2; hypothalamic, 2; pineal, 1)), spinal (n = 12; 14.6%), thalamic (n 6; 7.3%), brainstem (n 4; 4.9%), and optic nerve/chaisma (n 4; 4.9%).

BRAF alterations

Ten were uninterpretable for BRAFV600E (7 had suboptimal DNA preservation and 3 showed noisy sequencing data); 9.7% (7/72 (19–39, 7/67; > 39, 0/5)) were BRAFV600E mutant (Fig. 2).

Fig. 2
figure 2

Representative photomicrographs of adult PCA. a Cerebellar PCA (H&E; × 100) with electropherograms showing BRAFV600 and gel picture depicting KIAA1549:BRAF 15-9 fusion positive. b Suprasellar PCA (H&E; × 100) with electropherograms showing BRAFV600E mutant (black arrow) and gel picture depicting KIAA1549:BRAF fusion negative. c spinal PCA (H&E; × 200) with electropherograms showing BRAFV600 wild type and gel picture depicting KIAA1549:BRAF fusion negative

Full size image

KIAA1549-BRAF fusion was interpretable in 70 cases (12 had degraded RNA). Of these, 25.7% (18/70 (19–39, 16/65; > 39, 2/5)) were positive for KIAA1549-BRAF fusion (KIAA1549-BRAF 16-9, 11; KIAA1549-BRAF 15-9, 5; and KIAA1549-BRAF 16-11, 2) (Fig. 2).

Sixty-seven cases were interpretable for both BRAF alterations, of which 25 (37.3%) BRAF alterations (25/67; BRAFV600E, 7; BRAF fusion, 18).

p53 (n = 59) and ATRX (n = 78) protein expression

p53 was positive in 13.6% (8/59) cases, focal positive in 20.3% (12/59) and 66.1% (39/59) negative cases. One case showed a mosaic pattern of ATRX expression (which was negative for BRAF alterations) and 77(98.7%) showed retained ATRX protein.

Validation of KIAA1549: BRAF fusion transcripts

12/117 (10%) of KIAA1549-BRAF fusion-positive cases were confirmed by Sanger sequencing for all three fusion variants and their breakpoints (Fig. 3).

Fig. 3
figure 3

Representative gel images and electropherograms of validated cases of KIAA1549:BRAF 16-9, 15-9 and 16-11 fusion variants

Full size image

Correlation of BRAF alterations with clinicopathological parameters

Clinical

BRAFV00E was detected in the age range of 5–33 years (median age13). No significant difference (p value 0.82) was observed between BRAFV600E and age group (pediatric 8.9% and adult group 9.7%) and also for gender (p value 0.84). Of 117 KIAA1549-BRAF fusion positive cases, 78 were males and 39 were females with a median age of 8 years. There was no statistically significant difference observed for KIAA1549-BRAF fusion and gender (p value 0.15). KIAA1549-BRAF fusions were more common in the pediatric age group as compared with adults and this preponderance in ≤ 18 years (99/239; 41.4%) is statistically significant (p value 0.02) as compared with that in adults (> 18 years 18/70; 25.7%). Within the adult group, two cases were > 39 years (42 years of cerebellar location and 46 years of cervicomedullary location) showed presence of KIAA1549-BRAF 16-9 fusions.

Location

None of the tumor location showed significant statistical correlation for BRAFV600E mutation in pediatric and/or adult age group. None of the optic nerve/chiasma cases showed BRAF mutation. In the pediatric group, BRAF fusions were significantly associated with cerebellar as compared with non-cerebellar location (p = 0.01). It was the commonest location within fusion positive cases. 55/99 (55.6%) fusion positive pediatric cases were from cerebellar location. Among the three variants of fusion transcripts, KIAA1549-BRAF 16-9 correlated significantly with cerebellar location than other two fusion transcripts (p = 0.03) whereas fusion 15-9 transcript showed significant association with non-cerebellar location and negatively correlated with cerebellar location (p = 0.00).Adult cases also showed cerebellar (37.5%) as the predominant location for BRAF fusion but did not show a statistical significant association (p = 0.15).

p53 and ATRX protein expression

8.3% of BRAFV600E were in p53 positive and 8.9% were in focal positive while 11.5% were in negative cases. BRAF fusion transcripts were present in 33.3% p53 positive, 37.7% focal positive and 43.6% in p53 negative cases. No statistically significant correlation was observed. All except one case showed ATRX retained expression and thus no further comparative evaluation was done.

Discussion

PCAs are characterized by constitutive activation of the MAPK pathway through two major BRAF gene alterations—BRAF fusion and BRAFV600E mutation. The BRAF gene located at chromosome 7 (7q34) encodes for serine/threonine RAF kinase which acts as an intracellular transducer of the RAS/RAF/MEK/ERK pathway where activated RAS induces dimerization and activation of BRAF leading to MAPK activation, which causes tumorigenesis [20, 21]. The most common mechanism of alteration of MAPK pathway in PCAs is tandem duplication of approximately 2 MB 7q34 region with subsequent fusion resulting in KIAA1549-BRAF fusion variant [5]. The event of KIAA1549-BRAF fusion results in the replacement of the N-terminal regulatory region of BRAF with the N-terminal end of KIAA1549 resulting in constitutive activation of BRAF kinase domain and upregulation of the MAPK pathway. Another mechanism of constitutive MAPK activation is BRAFV600E mutation, the most common hotspot mutation observed across various cancers including glial tumors and is a missense substitution in exon 15 of BRAF gene at nucleotide 1799 replacing valine to glutamic acid at codon 600 within the activation segment coding region resulting in constitutive activation of BRAF in its monomeric form [20, 22]. BRAF fusion is the most common genetic event and characteristic of PCAs whereas BRAFV600E is uncommon. The latter is more frequent in non-PCA other astrocytic tumors like pleomorphic xanthoastrocytomas and gangliogliomas [17].

Detection of these BRAF alterations has a diagnostic value and has become an integral part of routine diagnostic practice in neuro-oncology. KIAA1549-BRAF fusion transcripts were detected using the most popular RT-PCR due to its ease to perform, cost-effectiveness, easy interpretation. However, RT-PCR requires good-quality RNA and a different set of primers for different variants with well optimization of each set [23]. The other method of detection is fluorescent in situ hybridization (FISH), this method is challenging especially in FFPE section. In view of the low cellularity of these tumors compounding with nuclear truncation usually result in variability in the signal pattern leading to difficulty in interpreting of fusion signals and is also endowed with inability to identify definite fusion variant [24]. However, a well-optimized FISH assay can also be used as an alternative method for BRAF fusion detection in diagnostic setting with skilled personnel for interpretation.Recent more accurate and sensitive techniques have also been developed based on droplet digital PCR, NanoString technology and Pyromark system but use of these techniques are restricted due to unavailability of these platforms in routine setups [25,26,27]. BRAF mutation can also be detected with high resolution melting curve analysis, allele-specific SNaP shot and by immunohistochemistry (IHC) with BRAF mutation-specific VE-1 antibody. The detection of BRAFV600E with VE-1 antibody is a robust, cheaper, and less time-consuming method and also showed high concordance with Sanger sequencing, but in some instances, it may yield non-specific staining necessitating the confirmation with sequencing [19, 28]. This study employed RT-PCR and Sanger sequencing for detection of limited BRAF-KIAA fusions (16-9, 15-9, and 16-11) and BRAFV600E in pediatric and adult groups.

The frequency of fusion transcripts within the PCA group varies from 50 to 80% across various studies (Table 3). The frequency of BRAF fusions in the present study was 41.1% in pediatric cases which is comparatively lower than reported, even in comparison with the three fusion variants evaluated. This possibly reflects the inherency of this population or probably due to the lesser sensitivity of the RT-PCR though only the cases with adequate representation of the tumor were selected [23, 24]. BRAF fusions were observed to be less frequent in adult PCAs and the frequency was observed to be 25.7% which is similar to previous studies [16, 29].

Table 3 Frequency of BRAF alterations in PCAs across previous studies
Full size table

KIAA1549-BRAF fusion transcripts are predominantly seen in PCAs and PMAs, although the presence of these fusion transcripts are also reported in few non-PCA/PMA tumors like in 70–80% of low-grade glioneuronal tumors, diffuse leptomeningeal glioneuronal tumors [30,31,32], pediatric oligodendroglial tumors [33], and few reports of pediatric diffuse gliomas, NOS [17, 34,35,36,37]. However, none (although the number is small) of the PMA cases in the present study showed the presence of BRAF fusion.

KIAA1549-BRAF fusion variant of 16-9 (pediatric 66.7%, adults 61.1%) is the commonest followed by fusion 15-9 (pediatric 21.2%, adults 27.8%) and fusion 16-11 (pediatric 12.1%, adults11.1%), which is concordant with the predominant literature [5, 38]. However, one Egyptian study reported a higher incidence of 15-9 fusion variant [34]. Faulkner et al. and Kondo et al. have shown the presence of more than one type of fusion variant in PCA cases but none of the PCA cases of the present study showed the presence of more than one variant [23, 25]. KIAA1549-BRAF fusions were detected in both pediatric and adult cases ranging from 1 to 46 years of age but were significantly associated with the pediatric age group than adults, which is consistent with previous studies [16, 17, 35].

BRAFV600E was seen in 8.9% and 9.7% in pediatric and adult groups respectively which is consistent with the current reported literature [17, 19, 39]. A recent study has also showed the presence of novel somatic duplication mutation in BRAF gene (p.V504_R506dup) in five PCA cases [40]; however, the current study did not detect non-BRAFV600E mutations. Behling et al. and Schindler et al. have reported a high frequency of BRAF mutation in ≥ 30 years; however, in the present study, BRAFV600E was more common in < 30 years with a median age of 13 years [17, 19].

Schindler et al. [17] in their large cohort (n = 1320) showed an association of BRAF mutation with non-cerebellar origin of tumors whereas the present study did not show any significant difference of BRAF mutation between cerebellar (11/134, 8.2%) and non-cerebellar locations (18/186, 9.7%), while Horbinski et al. reported a higher frequency of BRAF mutation for cerebral hemispheric location [39]; but that was not the case in the present study; however, the number of cerebral hemispheric cases in the present study was very small (n = 25).

In the study under discussion, 55.6% of BRAF fusions in pediatric cases were cerebellar; among the three variants of fusion transcripts, fusion 16-9 was strongly associated with cerebellar location whereas fusion 15-9 showed a significant correlation with non-cerebellar location (16/21; 76.2%). The association of fusion 15-9 with non-cerebellar tumors was also demonstrated by Faulkner et al. in their study of 32 cases of PCAs [23]. Fusion 16-11 did not show any specific pattern for location.

BRAFV600E has been shown to be a negative prognostic marker in pediatric LGGs. Lassaletta A et al. have shown that BRAFV600E positive cases showed significantly poor progression-free survival (27%) as compared with BRAF wild type (60.2%) and showed a higher tumor recurrence rate after conventional chemotherapy [41]. However, BRAF fusion has been shown to be a positive prognostic marker and is dependent on the clinicopathological features such as age of patient and tumor location [42]. In addition to its diagnostic and prognostic significance, it also represents a good candidate for targeted therapy as both BRAF fusion and BRAFV600E constitutively activated the MAPK pathway. In a phase I trial of the MEK inhibitor selumetinib, Banerjee et al. demonstrated a promising effect showing 20% (5/23) of recurrent LGG children with sustained partial response. Among these 5 cases with partial response, authors could perform BRAF alterations in 4 and all were positive for BRAF alterations [43]. BRAFV600E has also been shown as an effective target in progressive optic pathway glioma (OPG) where biopsy and its molecular evaluation is not a routine clinical practice. Upadhyaya et al. (2018) in their case series demonstrated an excellent positive effect of vemurafenib on progressive NF1-negative OPG cases who did not respond to frontline chemotherapy and also highlighted a potential impact of BRAF alterations in OPG cases [44]. In the present study, 5.4% of cases were OPGs and 30.8% showed the presence of BRAF alterations; however, due to the small number of OPGs, this may lead to underrepresentation of the frequency. This biological implication does suggest the need for timely biopsy and the molecular evaluation in these refractory LGG cases to identify the subjects for targeted therapy. Drobysheva et al. and Miler et al. also demonstrated a positive response and improvement in cases with PCA when treated with MAPK inhibitors, thus providing supporting evidence for BRAF alterations as a candidate for MAPK targeted therapy [45,46,47]. The present study is majorly limited by the lack of clinical follow-up; thus, it does not address the issue of prognostic significance.

In summary, this is the largest study from our subcontinent, essentially aimed at determining the frequency and distribution pattern of BRAF alterations in PCAs across all age groups. BRAF alterations were seen in 53.5% pediatric and 37.3% adult PCA cases. In keeping with the previous studies, BRAF fusion was found to be a key event in PCA than BRAF mutation and was strongly associated with PCA and cerebellar location. Interestingly, no significant difference was observed for BRAV600E for location and age group. This study will also serve as a baseline reference for other subsequent subcontinental studies.