BRAF V600E mutation and 9p21: CDKN2A/B and MTAP co-deletions - Markers in the clinical stratification of pediatric gliomas

Tumors of the Central Nervous System (CNS) account for 20–25% of all pediatric cancer diagnoses in the developed world, being the most prevalent group of cancers in children, after leukemia. Although uncommon, pediatric brain tumors represent the leading cause of cancer-related mortality in children and adolescents aged 20 years and under; and the third leading cause of cancer-related death in young adults aged 20 to 39 years. In addition to their usual high mortality, CNS tumors are associated with extensive morbidities such as pituitary dysfunction, growth hormone deficiency, epilepsy, vision loss, impaired motor skills, memory dysfunction, attention and behavioral disorders and reduced intelligence quotient (IQ). So, about 90% of survivors are left with long term cognitive and psycho-social deficits [1].

Glial and neuro-glial derived tumors are the most frequent CNS tumors in pediatric population [1, 2]. They form an heterogeneous group of neoplasia which are categorized by the World Health Organization (WHO) [2, 3] into various entities based on their cellular, genetic and clinical characteristics.

In what concerns their clinical features, gliomas are clustered in two major groups: low-grade gliomas (LGG) and high-grade gliomas (HGG). LGG comprise: i) the non-diffuse/non-infiltrative pilocytic astrocytoma (PA) categorized as WHO grade I tumor due to its predominantly favorable outcome, which, depending on their localization, can be cured by surgery alone: ii) the diffuse/ infiltrative gliomas which are associated with a less favorable clinical outcome, namely, recurrence after initial resection and a higher probability to malignant transformation. The most frequent histological entities ascertained to the former group are: the diffuse astrocytoma grade II, gangliogliomas, the angiocentric gliomas and pleomorphic xanthoastrocytomas.

Conversely, HGG comprise other histological entities that invariably recur and show progressive disease after the initial surgical resection. In the HGG are included all WHO grade III and IV neoplasms and the most frequent histologic categories are: anaplastic astrocytoma, anaplastic pleomorphic xanthoastrocytoma and glioblastoma (GBM).

In pediatric low-grade gliomas, the analysis of the molecular and genetic mechanisms underlying gliomagenesis allowed for an understanding of MAPK/ERK pathway activation as fundamental for their development. This occurs at high frequency by activation of the BRAF oncogene; and in lower frequencies: by MYB and MYBL1 amplifications and rearrangements, and FGFR1 rearrangements and mutations [4–10].

Two different mechanisms may lead to BRAF activation in pediatric brain gliomas: chromosomal rearrangements and point mutations. The most common BRAF rearrangement is the one resulting in BRAF/KIAA1549 fusion protein in which the N-terminus of the protein encoded by KIAA1549 gene is fused with the C-terminus of the protein encoded by BRAF gene, preserving the BRAF kinase domain [10, 11]. BRAF activating rearrangements were reported to be present in 70% of the pilocytic astrocytomas, in 15% of other low-grade gliomas, and have only been punctually observed in high-grade gliomas [9]. Studies performed by Hawkins et al. (2011) [12], Horbinski et al. (2010) [13], and Jones et al. (2008) [14], showed that BRAF rearrangements were an independent favorable prognostic factor in both supra-tentorial and posterior fossa low-grade gliomas.

The vast majority (> 90%) of BRAF mutations in pediatric gliomas are BRAF V600E mutations, a somatic mutation causing the substitution of the amino acid valine by glutamic acid at residue 600 of exon 15. BRAF V600E mutations have been described in a wide variety of lesions: 80% of pleomorphic xanthoastrocytomas 33% of the gangliogliomas, 23% of the diffuse astrocytomas, 10% of the glioblastomas being more frequent in tumors located in the cerebral cortex [15]. Only rarely BRAF V600E mutation occurs in conjunction with a BRAF/KIAA1549 rearrangement in the same tumor [4].

At variance with BRAF rearrangements, the role of BRAF V600E mutation in the glioma’s evolution and patient’s follow-up is far from being fully understood and some contradictory results are found in literature. Accordingly, while Horbinski et al. (2012) [13] showed that in their cohort of pediatric low-grade gliomas, BRAF V600E mutationended to a worse progression-free survival when compared to wild-type tumors, Mistry et al. (2015) [16] showed that this mutation was associated with a prolonged latency to malignant transformation and, consequently, with a better overall survival when compared to wild-type pediatric low-grade gliomas.

Moreover, Korshunov, et al. (2015) [17] described a subgroup of glioblastomas, exclusive to the pediatric population, that was characterized by the BRAF V600E mutation and CDKN2A deletion. Although these tumors had a better overall survival, they still had a high recurrence rate (67%).

The CDKN2A gene is mapped at the chromosome 9p21 region and encodes the p16^INK4A and p14^ARF proteins. p16^INK4A protein has a key role as negative regulator of the proliferation of normal cells, controlling the progression through G1 into the S phase of the cell cycle. According to Raabe et al. (2011) [18], the worst outcomes associated with CDKN2A gene deletion could reflect a failure to induce senescence or an escape from the induced tumor senescence in BRAF driven CDKN2A⁻ tumors.

In order to further understand the interplay between BRAF V600E mutation and the chromosomal region abrogating the CDKN2A gene: 9p21, a cohort of 100 pediatric gliomas was retrospectively analyzed.

Incidence and interplay of the presence of BRAF V600E mutation and 9p21 region deletion was evaluated in a series of 78 LGG and 22 HGG aiming to determine the role of these two alterations in recurrence and gliomas’ malignant transformation. There was also interest in verifying if they could be used in a clinical set for stratifying patients for existing target therapies and a tailored strict surveillance.

As to the incidence rate of BRAF V600E mutation results corroborated previous data [4, 15, 17], showing that this mutation occurred in all spectra of glioma types, although being more frequent in gangliogliomas and diffuse astrocytomas. Accordingly, we identified BRAF V600E mutation in 43% of the gangliogliomas, in 33% of the diffuse astrocytomas in 13% of the pilocytic astrocytomas, but also in one anaplastic astrocytoma and one glioblastoma.

In our population we observed a recurrence incidence of 8.9% and a 2,6% incidence of pediatric LGG, which undergo malignant transformation into HGG.

Statistical evaluation revealed no correlation between the presence of BRAF V600E and recurrence or progression in our series.

Mistry, et al. (2015) [16] by analysis of a cohort of 886 patients reported a similar incidence rate of pediatric LGG transformation to sHGG - 2,9%, but at variance with our data they showed BRAF V600E mutation to be associated with prolonged latency periods in pediatric low-grade gliomas. Our data suggests that tumor progression is independent of the presence of BRAF V600E mutation since progression free survival rates were very similar between the two groups: BRAFwt and BRAFmut.

In our series 9p21 deletion was rare in grade I (12.2%, p = 0.0178) but frequent in grade IV gliomas (62.5%, p = 0.0087). Moreover, this deletion also correlated with the histologic categorization of tumors, a p = 0.0012 was found between 9p21 deletion and glioblastoma histology.

In accordance with previous studies performed by Horbinski et al. (2012) [13], Mistry et al. (2015) [16] and Roy et al. (2016) [20] results led to the conclusion that homozygous or heterozygous deletions of the 9p21 chromosomal region were associated with a shorter overall survival and with a worse progression-free survival.

Interestingly, the two pediatric LGG (LGG81 and 82) that underwent malignant transformation into HGG (HGG 23 and 9 respectively), both presented a 9p21 region deletion. One, LGG81, also had a BRAF V600E mutation ab initio and the observed latency period until the development of the HGG was two years. LGG82 did not present ab initio a BRAF V600E mutation and had a much longer latency period until the development of the HGG (13 years). Although the number of cases is recognizably small, these findings suggest that in LGG81, a cooperative interplay might have occurred between BRAF mutation and 9p21 homozygous loss, which enhanced malignant transformation.

Several tumor suppressor genes (TSG) are known to be located at 9p21, namely, the Cyclin Dependent Kinase Inhibitor’s: CDKN2A and B; the Kelch Like Family Member 9 (KLHL9) and the Metylthioadenosine Phosphorilase (MTAP). The role of all these TSG in gliomas has been under investigation. Huilard et al. (2012) [21] by performing functional studies in neural progenitor cell and mouse models evidenced that homozygous CDKN2A/INK4a-Arf deletion in BRAF V600E expressing cells was sufficient for the formation of tumors with histologic features similar to the malignant astrocytomas in humans [21]. Mistry et al. (2015) [16] in an array of comparative genomic hybridization analysis of 886 pediatric gliomas alsodefined that CDKN2A deletions had a central role in low-grade glioma malignant transformation. On the other hand, Roy et al. (2016) [20] in a cohort of 379 adult LGG, conversely observed in their genomic analysis that CDKN2A inactivation was not associated with progression. Indeed, by correlating gene expression with zygosity status, they found that CDKN2A expression levels were normal or even elevated when associated at heterozygous gene loss, and proposed rather that KLHL9 and MTAP would concurrently and in a context-dependent manner promote tumor aggressiveness.

In this study BAC-RP11–14912 was used for FISH analysis. This BAC has a 101,115 bps genomic insert that encompasses not only CDKN2A/B, but also the MTAP-009 gene segment, which encodes a 112 amino acid MTAP protein. The detection of losses of this DNA segment in our cases was indicative that both genes could be involved in the malignant transformation of pediatric LGG. By performing expression analysis by qPCR for both the CDKN2A and MTAP genes in cases referred in our series as: LGG 64; 73; 74 and HGG 3; 18; 23, it was evidenced that in the LGG characterized by heterozygous gene losses there was still a “normal” expression of both genes, but in HGG cases, the homozygous loss was associated with a significant down-regulation of both CDKN2A and MTAP genes. The observation that MTAP is also involved in pediatric gliomas is of foremost importance since in very recent surveys [22, 23] it was revealed that MTAP deleted cancers were rendered therapeutically vulnerable at the methionine salvage pathway. Accordingly, to Mavrakis et al. (2016) [22] and Marjon K et al. (2016) [23] deletions at 9p abrogating the MTAP gene lead to the accumulation of its substrate 5′ methilthioadenosine (MTA) and a metabolic “rewiring” of the tumor cells, which rendered them specifically sensible to drugs that inhibit the methyltransferase axis composed by: Methionine Adenosyltransferase II alpha (MAT2A), Protein Arginine Methyltransferase 5 (PMRT5) and Rio domain containing protein 1 (RIOK1).

As aforementioned, HGG18 case of our cohort (Table 1) was diagnosed as a secondary glioblastoma following medullar transplant and radiotherapy of an ALL-B. Reports on the genetic analysis of secondary CNS tumors in children and their previous ALL are exceedingly rare, and the driver versus passenger mutations that underlie leukemia metastasis and the development of secondary tumors are still to be defined [24–26]. HGG18 was negative for the presence of BRAF V600E mutation but positive for 9p21 deletion (Table 1). No molecular evaluation for these two genetic alterations could be retrospectively performed in the leukemia cells, but the karyotype observed in bone marrow blasts at relapse revealed a hiperdiploid karyotype, with 54 chromosomes and a gain of two copies of chromosome 9. Although, it could not be ruled out experimentally that in this case a specific CDKN2A/B and MTAP deletion or epigenetic silence of these TSGs were not present ab initio and at relapse in the leukemia cells of the patient, it is possible to consider that in the CNS microenvironment, but not in the medullar microenvironment, CDKN2A/B/MTAP loss and down-regulation contributed to expansion of the patients neoplastic clone, therefore reinforcing the view that chromosomal deletions encompassing these genes are important for glioma progression.