KRAS/BRAF mutations in brain arteriovenous malformations: A systematic review and meta-analysis

Abstract

Introduction

Somatic KRAS mutations have been identified in the majority of brain arteriovenous malformations (AVM) specimens. The aim of our study was to evaluate the prevalence of Kirsten rat sarcoma (KRAS)/murine sarcoma viral oncogene homolog B1 (BRAF) mutations in brain AVM.

Methods

A systematic literature review was performed in November 2019. We reviewed MEDLINE/PubMed, Cochrane Library, and ClinicalTrials.gov for citation or ongoing trials from January 2010 to March 2020.

Results

6 studies were identified as meeting the inclusion criteria of this review. The total frequency of KRAS mutations in 1726 patients with AVM was 55%. The prevalence of BRAF mutation was 7.5%. The prevalence of AVMs with grade 2 was the most (39%). Frontal and parietal lobes were the commonest sites of AVMs (21%). the most prevalent presentation of patients with AVM was hemorrhage (62%).

Conclusion

Our findings support a high prevalence of somatic activating mutations in KRAS and less commonly, BRAF in the overwhelming majority of brain AVMs. Practically and importantly, this pathway homogeneity in CNS arteriovenous malformations also supports the development of targeted therapies with RAS/RAF pathway inhibitors. However, more studies are needed to confirm this hypothesis.

Introduction

Brain arteriovenous malformations (BAVMs) are high flow vascular abnormalities that occur in the brain in 15 cases per 100,000 people and cause hemorrhagic stroke in children.^1–4 These malformations are abnormal tortuous and morphological vascular channels between the arteries and veins and lack a mediating capillary network, which causes high-pressure of arterial blood to enter the veins directly from the feeding arteries. The underlying cause of cerebral arteriovenous malformations is not known, but similar malformations are found in rare genetic syndromes such as hereditary hemorrhagic telangiectasia (a set of disorders caused by inactivation of germinal mutations in β-SMAD-regulating growth factor regulator)^5–7 and also in capillary malformation-arteriovenous malformation syndrome (CM-AVM) (RASopathy due to inactivation of RASA1 or EPHB4 mutations).^8,9 Based on this pattern, genetics may be the cause of cerebral arteriovenous malformations, but most of these malformations occur as sporadic malformations in people without a family history of the disease ² Hemorrhagic stroke, or intracerebral hemorrhage, refers to the rupture of an artery in the brain. Rare cerebral arteriovenous malformations with a family history may occur in connection with terminal mutations as hereditary hemorrhagic telangiectasia (caused by a mutation in ENG, ACVRL1, and SMAD4) or capillary malformation-arteriovenous malformation (caused by RASA1 mutation as well as EPHB4).^3,8,10–12 However, majority of intracranial arteriovenous malformations are non-hereditary and sporadic. Somatic mutations in KRAS have been reported in most cases of cerebral arteriovenous malformations.^13–15 The pathogenetic mechanisms that happens as a result of KRAS-initiated activation of the MAPK–ERK pathway are mediated by angiogenic genes (e.g., VEGF A/C) and proteins in the notch signaling pathway. ¹⁶ This study aimed to investigate the prevalence of KRAS/BRAF mutation in cerebral arteriovenous malformations.

Materials and methods

Data sources and search strategy

A systematic literature review was performed in November 2019. We reviewed MEDLINE/PubMed, Cochrane Library, and ClinicalTrials.gov for citation or ongoing trials from January 2010 to March 2020. The search criteria were limited to human studies published in English language. The Medical Subject Heading terms used for the search in PubMed were ‘KRAS’, ‘BRAF’, ‘arteriovenous malformations’ or ‘AVM’, ‘brain’ and ‘somatic mutation’.

Selection criteria

To be included in this review, an article had to fulfill the following inclusion criteria: performed in patients with AVM; report results of genetic tests for KRAS mutation status (containing BRAF mutational status or not) from resected specimens. We included studies published after 2010 in order to reflect more updated results regarding recent surgical techniques. Duplicate articles or articles that were on the basis of the same patient cohorts were not included: if inclusion of another published cohort was specified, we included the article with the higher number of patients, if not specified, we included the older article.

Data extraction

Data extraction was conducted independently according to the Preferred Reporting Items for Systematic Reviews and Meta Analyses. Two authors selected studies following the previously described inclusion criteria. After this first selection, if needed, another author was consulted to achieve a shared decision. A predefined protocol for data extraction was used to retrieve data of each study, including first author name, year of publication, journal of publication, study period, sample size, study design, demographic profile, KRAS mutational status, BRAF mutational status, surgical interventions, other procedures and significant variable at univariate and multivariate analyses.

Statistical analysis

Meta-analysis was carried out using the inverse variance method, in which the specific weight of each study among the pool is calculated as the reciprocal of the squared standard error. The heterogeneity among the studies was evaluated using the Cochran Q and of the I² statistics. Fixed and random effects were reported.

Results

Literature search results

The electronic search provided a total of 306 results. After screening, 23 were eliminated because they were duplicates and 73 were removed because they were not in English language; 187 were excluded because they were reviews, editorial letters, case reports, no full text available or no quality. At the end of the review process, 6 studies were identified as meeting the inclusion criteria of this review. These articles constitute the study population (Figure 1). Study characteristics are shown in Table 1 and Figure 1. The Cochrane database of systematic review was then cross-checked to ensure that no similar systematic reviews had been undertaken.

Figure 1.

PRISMA flow diagram.

Table 1.

Characeristics of included studies.

Author/year	Mean age	M/F	N	MKRAS	MBRAF	Location					presentation
						T	P	O	F	C	seizure	hemorrhage	headache
Hong et al. 14	28.6	0.4	21	76.2	4.8%	28.5%	28.5%	14.2%	14.2%	14.2%	N/A	N/A	N/A
Nikolaev et al. 13	N/A	N/A	72	62.5%	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Goss et al. 17	19	1.4	22	40.9%	9%	13.6%	13.6%	13.6%	9%	N/A	3	5	3
Goss et al. 18	15.5	0.7	16	50%	12.5%	31.2%	25%	25%	18.7%	0.0	5	0	10
Oka et al. 19	28.5	1.1	38	65.7%	N/A	13.1%	23%	16.4%	42.6%	4.9%	14	32	N/A
Priemer et al. 15	32	0.7	21	28.5%	N/A	19%	23.8%	9.5%	38%	4.7%	4	10	5

MBRAF: prevalence of mutant BRAF; MKRAS: prevalence of mutant KRAS; M/F: male to female ratio; N: sample size; T: temporal lobe; P: parietal lobe; O: occipital lobe; Frontal lobe; C: cerebellum; N/A: not available.

Meta-analysis of frequency of KRAS and BRAF mutations

Based on the random effect model, the total frequency of KRAS mutations in 1726 patients with AVM was 55% (95% confidence interval [CI]:48%–62%, I² = 81.8%) (Table 2, Figure 2(a)). The prevalence of BRAF mutation was 7.5% (95% confidence interval [CI]:0.8%-14.1%) (Table 2, Figure 2(b)).

Table 2.

Meta-analysis results of the prevalence of KRAS and BRAF mutations in BAVMs.

Somatic mutations and patient characteristics		95% Conf. interval
		Low	Up	ES	Weight
KRAS and BRAF mutations prevalence	KRAS	0.48	0.62	0.55	100
	BRAF	0.008	0.141	0.075	100
Spetzler-Martin grade	1	0.14	0.36	0.25	100
	2	0.26	0.51	0.39	100
	3	0.18	0.41	0.30	100
	4	0.016	0.057	0.021	100
Location of AVM	T	0.126	0.137	0.131	100
	P	0.143	0.291	0.217	100
	O	0.082	0.208	0.145	100
	F	0.143	0.283	0.213	100
	C	0.008	0.112	0.060	100
Presentation	Seizure	0.157	0.322	0.240	100
	Hemorrhage	0.53	0.71	0.62	100
	Headache	0.19	0.25	0.210	100

Figure 2.

Meta-analysis of KRAS (a)/BRAF (b) mutations in BAVMs.

Meta-analysis of Spetzler-Martin grade

Based on the random effect model, the prevalence of grade 2 Spetzler-Martin AVMs was the most(with the prevalence of 39% (95% confidence interval [CI]:026%-51%)).followed by grade3 (with the prevalence of 30% (95% confidence interval [CI]:18%-41%)), grade1(with the prevalence of 25% (95% confidence interval [CI]:14%-36%)) and grade 4 (with the prevalence of 2.1%(95% confidence interval [CI]:1.6%-5.7%)) (Table 2, Figures 3 and 4(a)).

Figure 3.

Meta-analysis of Spetzler-Martin grade 1(a), 2(b), 3(c) and 4(d).

Figure 4.

Radar chart of Spetzler-Martin grade (a) and location (b) in BAVMs.

Meta-analysis of location of AVM

Based on the random effect model, Frontal and parietal lobes were the commonest sites of AVMs (with the prevalence of 21%) followed by Occipital (with the prevalence of 14.5%), Temporal (with the prevalence of 13.1%) and cerebellum (with the prevalence of 6%) (Table 2, Figure 4(b)).

Meta-analysis of presentation of AVM

Based on the random effect model, the most prevalent presentation of patients with AVM was hemorrhage (with the prevalence of 62%) followed by seizure (with the prevalence of 24%) (Table 2, Figure 5).

Figure 5.

Meta-analysis of presentation of AVM (hemorrhage (a) and seizure (b)).

Publication bias

The funnel plot resembled symmetrically which means that there is no publication bias. The size of each circle in the figure is an indicator of the sample size of the corresponding study (bigger circles shows more sample and smaller circles shows fewer sample size).

Meta-regression finding based on mean age of participants and frequency of KRAS mutations

The studies’ meta-regression was according to the association between frequency of KRAS mutations and the mean age of the participants, and the overall rate of KRAS mutations. There was no statistically significant linear trend in univariate meta-regression to explain effect size variation by mean of age of study with coefficient = 0.14 (95% CI –2.17, 2.46), P = 0.88 (Figure 6(b)).

Figure 6.

Begg’s funnel plot for publication bias (a). Meta-regression finding based on the mean of age and frequency of KRAS mutations (b).

Discussion

BAVMs are of important causes of intracerebral hemorrhage. In addition to the clinical signs associated with hemorrhage, BAVMs() have symptoms such as seizure or headache, and the rest of these patients (at least 15%) are asymptomatic. ³ In this study, in line with previous studies, the most common symptoms of BAVMs were hemorrhage (62%), seizure (24%), and headache (21%). These findings were consistent with a study by Schaller et al who reported hemorrhage as the most common symptom of BAVMs with a prevalence of 50%, as well. ²⁰ In most cases, BAVMs are sporadic, but familial cases are rare, and BAVMs have been reported in several cases of congenital syndromes.^21,22 For example, the prevalence of BAVMs has been reported in 4 to 13% of patients with hereditary hemorrhagic telangiectasia.^23,24 In this systematic review and meta-analysis, the prevalence of mutant KRAS and BRAF genes in patients with BAVMs were assessed. Based on the results of preliminary studies, the total prevalence of mutant KRAS gene in this study was 55% in patients with BAVMs.Out of 6 studies included in the present meta-analysis 4 of them reported the presence of KRAS mutations in more than half of their cases which may indicate the pathogenic role of KRAS mutations. KRAS-activated mutations play a pathogenic role in development of some tumors, and the presence of KRAS mutant genes is an independent prognostic factor in a number of tumors.^25–27 KRAS gene mutation is mostly associated with tumor formation and cancer development. These mutant genes activate growth factor signal transduction pathways, thereby resulting in cell change and genome instability.^16,28–30 The BRAF mutations are of other gene mutations that are considered as prognostic factors in some types of cancer. In their systematic review and meta-analysis, Ny et al. concluded that malignant melanoma patients with mutant type of BRAF genes had a poorer prognosis and were more at risk compared to patients with wild type BRAF. ³¹ In contrast, Vuong et al in their systematic review and meta-analysis reported that BRAF mutations had a favorable prognostic impact in gliomas and their prognostic value might be dependent on patient’s age and tumor grade. ³² In another study, Xing et al. reported that the prevalence of mutant BRAF genes in patients with papillary thyroid carcinoma was 45%. ³³ The prevalence of BRAF in this study was 7% in patients with BAVMs. All of the mutations found in AVM specimens including BRAF and KRAS mutations has been reported to be oncogenic activators and cancer growth stimulators which is in contrast with the presence of these mutations in BAVM specimens, because AVMs are not malignant vascular growth. This indicates the context-dependent and tissue-specific role of the RAS/RAF pathway in vascular tissue or it might demonstrate that there is a requirement for multiple additional genetic hits in order to develop cancer which is opposed to a monogenetic nature of arteriovenous malformations.^16,34The results of frequent and common KRAS mutations in BAVMs, the pathogenic nature of those mutations, the lack of KRAS mutations in cerebrovascular malformations other than BAVMs ¹³ and the direct development of arteriovenous malformations by induction of mutation in an animal model ³⁵ suggest that KRAS mutations occur in arteriovenous malformations due to endothelial proliferation, angiogenic signaling, or vascular remodeling process. Also, the lack of relationship between the frequency of mutations with age in this study is inconsistent with the hypothesis of passenger mutations, in which KRAS/BRAF mutations increase randomly over time. In human specimens of Brain AVMs, activated form of NF-kB can be found in some types of endothelial and inflammatory cells. ³⁶ On the other hand, studies conducted about VEGF illustrated that it plays a great role in different pathogenic sets such as causing age-related macular degeneration by angiogenesis stimulation and raising the permeability of vessels as well as activating intracellular signaling in a RAS-dependent manner.VEGF-C and VEGF-A both are recognizable using immunohistochemistry in specimens of patients with brain AVM. ³⁷ There are also other studies indicating the high prevalence of VEGF expression in patients with brain AVM compared with the normal cerebral cortex.^38,39 It has been reported that the induction of KRAS mutations in endothelial cells has performed as an exacerbating factor of brain AVM which can be explained by the fact that KRAS mutation can increase the endothelial cell migratory activity and It can also induces other genes which promote angiogenesis. These results are also in line with the results of other studies demonstrating a correlation between the presence of KRAS mutations and Spetzler-Martin grade of brain AVMs. ¹⁹ Realizing that a high proportion of BAVMs have KRAS or BRAF mutations, in potential clinical trials, the treatment can be initiated without prior tissue sampling and genetic confirmation, which is not possible without surgery in BAVMs.In a recent case report, the therapeutic response to trametinib which is an anti-cancer drug was evaluated in a patient with mutant KRAS AVM, and showed a significant decrease in blood flow to the AVM, with a 75% reduction in arterial inflow after 6 months of trametinib therapy. ¹⁵ these results suggests that KRAS function may contribute to the pathobiology of BAVM and can become a therapeutic target.

Limitations

The meta-analysis does not show possible differences between the different types of KRAS and BRAF mutations because KRAS mutations were not distinct in these studies and therefore no information was extracted for analysis. Therefore, due to the lack of information in the studies, it was not possible to compare the rate of BAVMs and grading between the mutant group and the wild type.

Conclusion

The findings of this study show the high prevalence of somatic activating mutations in KRAS and BRAF in most cases of brain arteriovenous malformations. Practically and more importantly, the pathway homogeneity in the central nervous system (CNS) arteriovenous malformations indicates the development of targeted therapies with Ras/Raf pathway inhibitors. However, more studies are needed to confirm this hypothesis.