BRAF‐V600E immunohistochemistry in a large series of glial and glial–neuronal tumors

Abstract Introduction Some glial–neuronal tumors (GNT) (pleomorphic xantho‐astrocytoma [PXA], ganglioglioma [GG]) display BRAF‐V600E mutation, which represents a diagnostic clue to these entities. Targeted therapies against BRAF‐V600 protein have shown promising results in GNT. The aim of this study was to assess the utility of BRAF‐V600E immunohistochemistry (IHC, clone VE1) in daily practice in a series of 140 glial, and GNT compared to molecular biology (MB) techniques. Methods We performed BRAF‐V600E IHC on all 140 cases. We used Sanger sequencing and allele‐specific quantitative PCR (ASQ‐PCR) to detect BRAF‐V600E mutation when sufficient amount of materiel was available. Results BRAF‐V600E immunostaining was detected in 29.5% of cases (41/140 cases; 61.5% GG/GC/AGG (32/52), 33% PXA, 6.6% pilocytic astrocytomas). In 47 cases, MB could be performed: Sanger sequencing and ASQ‐PCR in 34 cases, ASQ‐PCR only in 11 cases, and Sanger sequencing only in two cases. In initial tumors, Sanger sequencing identified BRAF‐V600E mutation in 19.5% tumors (seven of 36 tested cases). ASQ‐PCR showed mutation in 48.5% tumors (17/35 tested cases). In six cases (5 GG, one PXA), the results were discordant between IHC and MB; the five GG cases were immunopositive for BRAF‐V600E but wild type with both MB techniques. In another 7 GG, the percentage of mutated (ganglion) cells was low, and Sanger sequencing failed to detect the mutation, which was detected by IHC and ASQ‐PCR. Conclusions In tumors with few mutated cells (e.g., GG), anti‐BRAF‐V600E IHC appears more sensitive than Sanger sequencing. The latter, although considered as the gold standard, is not to be used up‐front to detect BRAF mutation in GG. The combination of IHC and ASQ‐PCR appears more efficient to appraise the indication of targeted therapies in these glioneuronal tumors.

Circumscribed glial and mixed glial-neuronal tumors (GNT) most often develop in children and young adults and are usually low-grade (grade I or II) according to the 2016 World Health Organization (WHO) classification (Louis et al., 2007).
The detection of a BRAF rearrangement has first diagnostic implications as diffuse gliomas do not usually display such an anomaly.
Molecular biology (MB) techniques are expensive and not yet widely available. A BRAF-V600E antibody has recently been commercialized Colomba et al., 2013;Ritterhouse & Barletta, 2015), and it is sensitive and specific in detecting BRAF-V600E mutation in cutaneous melanomas Colomba et al., 2013;Long et al., 2013). In contrast, few studies have assessed the reliability of anti-BRAF-V600E immunohistochemistry (IHC) in CNS tumors (Behling et al., 2016;Chappé et al., 2013;Long et al., 2013). Some groups have found that BRAF-VE1 IHC was suboptimal in characterizing brain tumor tissue and that MB techniques are required for a reliable clinical assessment. The aim of this study was to assess the utility of BRAF-V600E IHC compared to MB on a large series of glial and GNT.

| MATERIALS AND METHODS
One hundred and forty formalin-fixed paraffin-embedded samples  (Louis et al., 2007). The entire cohort is described in Table 1. For 131 of 140 patients, tumor material from the initial surgery (biopsy or resection) was available.
For nine patients (cases no. 11, 14, 27, 41, 68, 114, 133, and 140; see Table 1), the initial sample was not available (exhausted material, surgery at an outside institution), and only the sample obtained at recurrence was reviewed. The protocol and procedures employed were reviewed and approved by the appropriate institutional review committee. The patients were first treated in the Department of Neurosurgery of Angers University Hospital and those who required adjuvant treatment were followed in the pediatric Oncology Department or at the Western Cancer Institute of Angers (Institut de Cancérologie de l'Ouest [ICO]).

| Clinical and radiological data
One hundred and forty patients were included. The mean age at initial diagnosis was 16.2 years (standard deviation 14 years, range 7 months to 74 years). There were 68 males and 72 females (sex ratio M/F: 0.94). The tumor was located in the cerebellum in 32 cases, in the opto-chiasmatic region in 23 cases, in the cerebral hemispheres in 60 cases, in the basal ganglia in 11 cases, in the brainstem in six cases, in the region of the third ventricle in four cases, in the spinal cord in four cases, and in the pineal region in one case. The entire cohort is described in Table 1.

| Immunohistochemical study
Anti-BRAF-V600E IHC was performed on all samples (
Sanger sequencing and/or ASQ-PCR could be performed in 47 cases.
For many specimens, the amount of (FFPE) material available was not sufficient to allow for MB analysis. This fact underlines the key role of IHC in detecting biomarker expression in routine practice. One caveat of our study is that we first selected for sequencing the cases with ambiguous or restricted (few ganglion cells only) staining for which IHC could not reliably predict BRAF mutation. So, no firm conclusion can be drawn from the sensitivity and specificity values in our study.
However, the results of BRAF IHC were very good although inferior to those obtained in melanomas (Colomba et al., 2013;Long et al., 2013).
As already mentioned, 38.5% of GG/AGG in our series were immunopositive for BRAF-V600E only in the ganglion cell component.
This observation has been reported in the literature but has not been expanded upon (Koelsche et al., 2013). However, this finding leads to discuss the detection threshold of MB techniques. GG harbor a variable proportion of ganglion cells, from a few scattered cells to an authentic gangliocytoma (tumor composed entirely of ganglion cells with no glial tumor component) (Louis et al., 2007). The glial component is most often predominant in GG. In our study, Sanger sequencing did not detect BRAF mutation in 10 cases of GG/AGG that were immunopositive for BRAF. In three of those 10 cases, the immunostaining was detected only in a few ganglion cells but with a significant intensity. In 7 of 10 cases (including two of the three cases with rare immunopositive ganglion cells), the mutation was detected by ASQ- antibody detects only the BRAF-V600E variant even though cross reactivity with non pV600E mutations has been described (e.g., V600K, V600R) (Ihle et al., 2014). The absence of immunostaining does not rule out another V600 mutation (Ritterhouse & Barletta, 2015). This is also true for ASQ-PCR, which is designed to detect BRAF-V600E mutant. Only Sanger sequencing can detect the different variants (but with a lower sensitivity compared to ASQ-PCR). Of note, we did not identify BRAF mutation other than the V600E variant. Patients with mutation other than BRAF-V600E are eligible to targeted therapies such as vemurafenib or dabrafenib (Lee et al., 2014;Sosman et al., 2012). Because of the therapeutic implications, the relevance of detecting BRAF mutation is already expanding to other types of CNS tumors, therefore requiring robust detection techniques. It would be interesting to evaluate newer MB techniques in GNT such as nextgeneration sequencing or pyrosequencing, which is fast, more sensitive and has a better yield compared to Sanger sequencing (Colomba et al., 2013). Those techniques detect all BRAF mutations but are still expensive and not widely available.
If the presence of BRAF-V600E mutation is a diagnostic clue to a circumscribed low-grade glial or glial-neuronal tumor, this mutation is not specific of GNT; it has been reported in rare cases of low-grade diffuse gliomas in children and in 10% of GB, especially in the epithelioïd variant (which also expresses CD34 more often) (Brastianos et al., 2014;Ichimura et al., 2012;Kleinschmidt-DeMasters, Aisner, & Foreman, 2015;Kleinschmidt-DeMasters et al., 2013). The detection of the mutation can still help to distinguish a GG from the cortical infiltration of a diffuse glioma or a GG from an astrocytoma especially in the cerebellum, where PA rarely exhibits BRAF-V600E mutation (but instead the BRAF-KIAA1549 fusion in 80% of cases).
As expected, the percentage of BRAF-V600E mutant PA in our series was low (6.6%), which is comparable to the results obtained in other studies (Chappé et al., 2013;Schindler et al., 2011). It is interesting to note that the four mutant PA in our cohort were all of supratentorial location (basal ganglia, third ventricle, and optic pathways).
However, extra-cerebellar BRAF-V600E mutant PA remains rare (Roth et al., 2015) leading to wonder whether it is a distinct entity or actually GG whose neuronal component was not sampled.

| CONCLUSION
The detection of BRAF-V600E mutation represents a diagnostic aid in glial and GNT. Targeted therapies against BRAF-V600 mutant proteins have shown promising results. In this context, anti-BRAF-V600E IHC plays a key role in clinical practice, especially as it is a fast, inexpensive, and easily accessed technique. In GG, the presence of the mutation in only scattered neuronal cells reduced the sensitivity of Sanger sequencing. However, ASQ-PCR was able to detect the mutation in few tumor cells. Thus, while waiting for the assessment of newer MB techniques, we should order IHC in addition to ASQ-PCR, which should be preferred to Sanger sequencing in glioneuronal tumors.