16/04/2022
Molecular characterization of histone 3
The molecular assessment of the H3K27M mutation by Sanger sequencing was performed in 24 LGG of the midline. From these, all 22 amplifiable cases were wild type for H3F3A, HIST1H3B and HIST1H3C genes.
Additionally, 20 HGG were also assessed for mutations in histone genes by Sanger sequencing, where 2 cases did not amplify. From the remaining 18, all 9 HGG of the midline were positive for H3K27M mutation (8 in H3.3 and 1 in H3.1 isoforms) while 2 glioblastomas harbored the H3G34R mutation; the remaining 2 glioblastomas, 4 diffuse astrocytomas and 1 anaplastic oligodendroglioma were wild type for the histone genes (Fig 3A–3C). It is important to mention that both H3G34R mutant glioblastomas were located in brain hemispheres. Seeking to correlate results from Sanger sequencing and IHC detection of the H3K27M mutation, all 9 midline diffuse glioma H3K27M mutant cases, rendered as mutated by sequencing, were assessed by IHC with a H3K27M mutation specific antibody and, independently, with an antibody specific for tri-methylated Lys 27 (Fig 3D and 3E). Absolute congruence was observed in all of these cases when comparing the three detection methods; however, it was surprising to note that the only case of H3K27M, mutated on histone isoform 3.1, showed a particular pattern of signal-loss with the trimethylated-Lys27 antibody. While all the other 8 cases, mutated on histone 3.3 isoform, showed a complete loss-of-expression with trimethylated-Lys27 antibody (Fig 3E), the 3.1 mutated isoform showed a heterogeneous or mosaic loss-of-staining (Fig 3F). As negative controls for IHC, the 2 cases with H3G34R variant in histone 3.3 isoform were also assessed with both antibodies. As expected, they were negative for the H3K27M mutation and both cases retained the tri-methylation status of Lys27. Kapplan-Meier analysis demonstrated that the H3K27M mutation was associated with decreased PFS (Log-rank (Mantel-Cox) Test, P = 0.0124) (Fig 3G) and OS (Log-rank (Mantel-Cox) Test, P = 0.006) (Fig 3H).
thumbnail Download:
PPTPowerPoint slide
PNGlarger image
TIFForiginal image
Fig 3. Histone 3 mutation analysis.
A) Representative Sanger sequencing chromatogram showing double peaks in c.83A>T (p.K27M), indicating a H3K27M heterozygous mutation in isoform H3.3. B) Representative Sanger sequencing chromatogram showing double peaks in c.103A>G (p.G34R), indicating a H3G34R heterozygous mutation in isoform H3.3. C) Representative Sanger sequencing chromatogram showing double peaks in A>T (p.K27M), indicating a H3K27M heterozygous mutation in isoform H3.1. D) Representative immunohistochemistry for positive midline diffuse glioma H3K27M mutant case. Scale bar represents 50 μm at 400x magnification. E) Representative immunohistochemistry for H3K27me3 in the H3.3 mutant isoform depicts a complete loss-of-staining in tumor cells, while endothelial cells retain a positive staining status (arrows). Scale bar represents 50 μm at 400x magnification. F) Immunohistochemistry for H3K27me3 in the H3.1 mutant isoform case depicts a heterogeneous mosaic loss-of-staining. Scale bar represents 50 μm at 200x magnification. G) Kaplan-Meyer progression free survival analysis in HGG. H) Kaplan-Meyer overall survival analysis in HGG.
https://doi.org/10.1371/journal.pone.0266466.g003
Discussion
Tumors of the central nervous system are still the leading cause of childhood cancer-related deaths. Despite the many advances in surgical and adjuvant therapy, which have increased survival rates, complete resection is not usually possible for inaccessible, critically-located or infiltrative tumors, which have a worse outcome than superficial lesions. In these particular cases, the identification of specific molecular markers aids in establishing prognosis and individualize treatment strategies that could result in unnecessary sequelae in a developing child. Moreover, molecular markers now assist the histological diagnosis, enabling the classification of different subset of tumors, actually recognized by WHO [2], and providing new insights, risk stratification and treatment opportunities for almost every type of pediatric brain tumor [18, 19].
Although many other gene fusions, involving several genes, have been described as critical driver events in pediatric LGG, the KIAA1549-BRAF fusion is one prominent molecular marker, most frequently detected and assists in the diagnosis of pilocytic astrocytoma [20]. In our series, 43/64 (67%) pilocytic astrocytomas contained this fusion, a higher proportion of cases to that previously described [15, 20, 21]. Moreover, Kurani et. al. detected the fusion, by means of RT-PCR and sequencing, in only 41% of their pediatric cohort; although this discrepancy in the percentage of positive cases could be due to the difference in sensitivity of the employed techniques. Similar to that observed in other studies, regardless of the histological classification, we disclosed a significant association between KIAA1549-BRAF gene fusion and its anatomical location to the cerebellum for LGG [21]. While we found the fusion in 61.3% of LGG, others detected it ranging from 59.1 to 89.7% of LGG [14, 15, 20, 22].
Regarding the prognostic value of the fusion and its association with OS and PFS, some controversy has been reported in the literature. While Yang et. al. described a better PFS and OS for KIAA1549-BRAF gene fusion in LGG, the authors also state that this was not an independent prognostic factor in the multivariate analysis, which suggests that this marker could be influenced by another factor [15]. In particular, the series described by Yang et. al. contained a high proportion of pilocytic astrocytomas, which intrinsically have a better prognosis than other LGG, and could have influenced their observations. Conversely, in our series the gene fusion was not associated with a modified outcome, either in LGG as a whole, nor in pilocytic astrocytoma as similarly described by Faulkner et. al. [23] and by Penman et. al. in a review article [24]. Given the presence of the KIAA1549-BRAF gene fusion in over 67% of our pilocytic astrocytoma cases, it could be more of a useful diagnostic tool for this entity, than a prognosis related marker.
Similarly, the BRAF V600E mutation is also a useful molecular marker to estimate evolution and prognosis. Moreover, a recent report described the regression of BRAF mutated tumors in response to the BRAF inhibitor dabrafenib [25], an observation of particular importance in not resectable tumors. Overall, in LGG, we detected the mutation in 12.6% of the cases; however this proportion rises to 50% when only considering gangliogliomas, a proportion in accordance to that described by previous studies which found the BRAF V600E mutation in between 20–60% of cases [9, 26–30]. Particularly in gangliogliomas, which could compromise deep structures of the brain, preventing complete surgical resection and achieving adequate oncological control, BRAF V600E mutation could represent a druggable target for specific inhibitors such as vemurafenib or dabrafenib. Although these treatment options are still in trial and lack a pediatric formal indication, the International Consortium on Low Grade Glioma (ICLGG) of the International Society of Pediatric Oncology (SIOP) recommend their use in these cases [31]. In pilocytic astrocytoma, we detected the mutation in 3/62 (4.8%) of cases, again in accordance to previous reports using direct Sanger sequencing [21, 32]. These results further prove that KIAA1549-BRAF gene fusion was the most common BRAF alteration in pilocytic astrocytoma, while BRAF V600E was the most predominant in gangliogliomas. In all cases, BRAF V600E was assessed by IHC (VE1 clone) and by Sanger sequencing. Results obtained by both methods were congruent with each other, which suppose an advantage for pathology labs without access to a sequencing facility.
In our hands, when only considering pilocytic astrocytoma cases, the BRAF V600E was associated with a worse outcome for both PFS and OS. However, when considering the entire series of LGG, BRAF V600E was only associated with a worse OS, but not with PFS; a fact that could be due to the inclusion of only the subgroup of LGG that fulfilled the 5 year period of follow-up. Controversy still arises when considering BRAF V600E as a prognostic factor for pediatric LGG. While some researchers reported it to be a useful prognostic marker for PFS and OS [15], others reported it to be a useful prognostic factor for PFS, but not OS [11]. Our results are contradicting with this latter study since we reported BRAF V600E to be associated with a worse OS but not with PFS, something that could be due to the differences in age, tumor location and extent of surgical resection between both series of patients. Moreover, and to the best of our knowledge, no other report has explored its prognostic association with PFS or OS exclusively in pediatric pilocytic astrocytomas. Altogether, alterations in BRAF gene do not only constitute a valuable diagnosis tool and prognostic factor for some cases of LGG, particularly pilocytic astrocytoma, but also represents an interesting candidate for targeted therapy aimed at reducing the constitutively activated MAPK pathway.
Regarding the contribution of BRAF V600E mutation as a prognostic factor in HGG, a recent study on pediatric HGG identified the mutation in 5/42 cases, three of which were LGG that progressed to HGG [33]. From this observation, the authors suggested that LGG should be assessed for BRAF V600E mutation and those mutant tumors should be closely followed-up. In our series we detected the mutation in only one case of HGG, which rapidly progressed and died within 28 months from diagnosis.
The molecular assessment of the H3K27M mutation by Sanger sequencing was determined in 22 LGG of the midline, all of which were wild type for H3F3A, HIST1H3B and HIST1H3C genes. The diagnostic reasoning behind only including gliomas of the midline, among LGG, for H3K27M assessment was its reported prevalence in this anatomic location together with a disfavorable outcome [2]. Moreover, according to WHO guidelines, those cases histologically characterized as LGG of the midline that contain the H3K27M mutation are reclassified as high grade diffuse midline glioma H3K27M mutant. However, recent reports described that these cases, with low grade histology and H3K27M mutation, may not have such an unfavorable outcome as the high grade diffuse midline glioma H3K27M mutant, or have such a slow progression rate that exceeds the 5 year follow-up periods. Hence, the assessment of H3K27M mutations should be performed in all pediatric midline gliomas independent of their histologic appearance or grade [34, 35]. In line with WHO guidelines, in our series, the only LGG of the midline that harbored the HK27M variant in isoforms H3.3 (coded by H3F3A gene), was reclassified as a high grade diffuse midline glioma H3K27M mutant. However, Yang et. al. reported the H3K27M mutation in only 6.4% LGG, of which 50% were in the midline [15]; probably applying a different criteria for their classification.
Concerning HGG, mutations in genes coding for histone H3 (H3F3A, HIST1H3B and HIST1H3C) assessed by Sanger sequencing revealed that 61% of the cases contained a mutation in one of these genes, either H3K27M in 9 cases or H3G34R in the remaining 2 cases. When only considering the occurrence of H3K27M mutation, 50% of our cases harbored it, a similar proportion to that previously described by Bozkurt et. al. and Huang et. al. [5, 6].
Among those cases with histone H3 mutations, 9 cases were H3K27M mutant (82%), all of which were of midline localization, while the two remaining cases harboring the H3G34R mutation were located in the hemispheres. These latter findings are also in line with previous reports that show a predilection of H3K27M to midline locations and H3G34R to the hemispheres in pediatric HGG [6, 15, 36]. Also in line with the literature, only 1/11 HGG contained the mutation in the HIST1H3B gene (isoform H3.1), while the remaining 10 cases harbored the mutation in the H3F3A gene (isoform H3.3) [33, 37].
Additionally, we observed an absolute congruence between Sanger sequencing and the IHC results in all studied cases for H3K27M. Similar concordance was reported by Huang et. al. when assessing by sequencing a few cases within their studied series [5]. Moreover, the congruency in results was also observed when assessing the loss of tri-methylation status of Lys27 (H3K27me3) in isoform H3.3 and H3.1 by IHC. Regarding the case with H3K27M in isoform H3.1, as detected by sequencing, we described by IHC a patched pattern loss-of-signal with the H3K27me3 antibody. This was a worth mentioning observation, and a similar result was reported by Castel et. al. who described a weak staining with the H3K27me3 antibody in a H3K27M mutant in isoform H3.1 [36]. Additionally this same mosaic pattern of signal-loss was also described for H3K27me3 antibody in cases of malignant nerve sheath tumors [38]. Although there is no clear explanation for this atypical staining pattern, it could be due to the fact that H3.3 is expressed throughout the cell cycle, as well as in quiescent cells, but H3.1 and H3.2 are cell cycle regulated and deposited only during the S-phase and during DNA repair. Given differences in expression and deposition, levels of H3.1 and H3.3 protein isoforms may vary widely in different cell types, tissues, and cell-cycle moments ([7] and references herein). Finally, as previously reported, the H3K27M mutation was associated with a worse outcome, regarding both PFS and OS in HG [5, 33, 37].
Particularly in developing countries where access to high throughput sequencing is still limited, assessing these biomarkers by means of conventional molecular biology techniques is of paramount importance, since they aid in pediatric CNS tumors classification. In fact, this new molecular classification approach aims to reflect the biological immunophenotype of the tumor rather than its morphology, with the ultimate goal of achieving a risk-adapted framework and molecularly targeted therapies which augment or, in some cases, replace conventional therapy.
![De Picciotto et al. [29] found that NMN supplementation was capable of restoring NAD+ levels (by threefold), vascular SI...](https://img4.marketers1.com/799/962/113424597999629.jpg)
