Low frequency somatic variants in intracranial aneurysms

Until recently, genetic causes of IAs have largely focused on hereditary gene mutations, or common (high frequency) alleles identified through GWAS. We know that the risk of IAs is higher among family members with a history of IA or individuals with certain hereditary genetic disorders, such as autosomal dominant kidney disease or Ehlers–Danlos syndrome.13 High frequency variants identified through GWAS have included SOX17 (rs9298506 and rs10958409), EDNRA (rs6841581), CDKN2B-AS1 (rs10757278), COL1A2 (rs42524), COL3A1 (rs1800255), HSPG2 (rs3767137), SERPINA3 (rs4934), and VCAN (rs251124 and rs173686).13 14 Although many of these common variants have been identified, there has been little overlap with development of IAs.

Over the past decade, more evidence has suggested that low frequency variants may have a large role in common disease phenotypes.15 16 There are currently no variants described that overlap with the high frequency variants described by GWAS, although we anticipate that as more low frequency variants are discovered, there will likely be similar genes and pathways targeted because familial and sporadic aneurysms have the same pathogenesis and morphologically are indistinguishable. The few non-germline (somatic), low frequency variants known to lead to intracranial aneurysm formation are reviewed in this study and summarized in figure 2. Somatic variants have been described in certain vascular wall associated genes such as PPIL4 11 , EDIL3 12 , RNF213, 9 and ANKRD17. 10 These genes are involved in angiogenesis and vascular wall formation, although the mechanism by which these variants lead to IA pathogenesis is unclear. On the other hand, certain variants in cancer related genes have been more clearly associated with the pathogenesis of IAs. Two cancer related somatic variants have been detected within vascular smooth muscle cells (vSMCs). Lai et al 7 found sporadic mutations in SVIL in 17% of saccular IAs and demonstrated the role of SVIL in the modulation of vSMC phenotype to the synthetic phenotype, which is characterized by increased protein synthesis, proliferation, and migration, and is thought to contribute to aneurysm pathogenesis and rupture.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Somatic variants in the pathogenesis of intracranial aneurysms and their associated cellular signaling pathways.

Perhaps the most well described low frequency somatic variant involved in the pathogenesis of IAs is the cancer related variant in the platelet derived growth factor beta gene (PDGFRB). PDGFRB encodes a conserved transmembrane receptor tyrosine kinase that has an essential role in vascular progenitor cell signaling.17 Loss of function mutations in PDGFRB have been associated with a variety of vascular diseases, such as hereditary hemorrhagic telangiectasia and chronic eosinophilic leukemia. A somatic variant in PDGFRB, p.(Tyr562Cyst), was first identified in an individual presenting with IFAs at the internal carotid, vertebral, radial, and coronary arteries ipsilateral to cutaneous vascular malformations.6 Interestingly, this mutation occurs in the activation loop of PDGFRB, disrupting the auto-inhibitory sites of the receptor and resulting in constitutive activation. In another study by Hao et al,8 four different somatic mutations in PDGFRB were identified in four fusiform aneurysms. All mutations also caused constitutive activation of PDGFRB.

While the somatic variant in PDGFRB in IAs has been well described, the exact pathogenesis of the PDGFRB mutation in aneurysm formation is just beginning to be understood. Mutant forms of both PDGFRB and AHNAK lead to constitutively activate ERK, nuclear factor κB, and JAK2 signaling, enhanced cell motility, and induced inflammation related gene expression in vitro.8 One in vivo model of the PDGFRB somatic mutation inducing a fusiform aneurysm phenotype has been developed in zebrafish.8 The model demonstrated hemorrhage and altered vascular morphogenesis. This was the first confirmation that activating PDGFRB variants can drive the formation of fusiform-like vessel dilation in an animal model. This study also showed that the effects of the PDGFRB variant could be reversed in vitro and in zebrafish using ruxolitinib, a JAK inhibitor. Future studies are needed to confirm these results and apply them to other in vivo models.

Somatic variants based on aneurysm morphology

IA morphologies, saccular and fusiform, are phenotypes that have previously been assumed to have separate genetic causes.18 Mutations in SVIL 7 and PPIL4 11 have been described in intracranial saccular aneurysms (ISAs) only. The variants in PDGFRB were initially described solely in intracranial fusiform aneurysms (IFAs).6 8 In a series of aneurysm specimens, activating variants in PDGFRB were found in 4/6 IFAs but in 0/38 ISAs.6 This variant has been implicated in 14 patients in other reports,19 including a series of 12 patients with activating PDGFRB mutations, 7 of which had aneurysms involving the cerebral, coronary and renal arteries.20

One preliminary study21 involved whole exome sequencing of 20 ISAs of which 11 (55%) had 48 detectable somatic mutations, highly enriched in cancer related genes. A p.Tyr562Aps somatic mutation was detected in the PDGFRB gene and was predictably deleterious. This mutation is within the same codon as mutations detected in IFAs.

One recent study however, found somatic variants within PDGFRB common to both ISAs and IFAs. Shima et al 5 conducted whole exome sequencing on the largest series of 54 ISAs and 11 IFAs, and found multiple variants in PDGFRB in both tissue types. Additionally, somatic mutations in five other genes (AHNAK, OBSCN, CACNA1E, OR5P3, and RBM10) were also detected. These results suggest possible overlapping genetic risk factors in both ISA and IFA phenotypes. However, the frequency of somatic PDGFRB mutations was different among ISA and IFA tissue, at 4% and 36%, respectively. Other variants described in this review, such as mutations in EDIL3, RNF213, and ANKRD17, did not specify IA morphology. Given the results in the Shima et al 5 study, we advocate for future publications on somatic variants of IAs to specify the aneurysm morphology and the different frequencies of the variants described among the two different morphologies.

Mosaic effect of somatic variants

The mechanism for genetic heterogeneity within an individual leading to a unique cell lineage is called mosaicism. There are two mechanisms for genetic heterogeneity within an individual. Mosaicism occurs at any point following the first replication of the zygote with single nucleotide errors in DNA replication occurring 10^-9 errors per cellular division.22 This results in significant genetic variation across all tissues, increasing as we age, and this can result in beneficial changes in the development of neuronal diversity,23 evasion of germline inherited mutations,24 and enhancing the adaptive ability of tissues.24 However, many of these variants are also implicated in various human diseases, including vascular diseases.25

Sequencing of a cohort of syndromic and non-syndromic patients with IFA and other fusiform aneurysms revealed PDGFRB variants, and these variants were not detected in peripheral blood.4 This suggests a mosaic expression associated with fusiform aneurysm formation.6 A mosaic mechanism of this variant is an important characteristic because of the therapeutic implications if this pathway is targeted with a drug. Only a subset of cells will express this variant resulting in the pathologic phenotype; therefore, systemic delivery of a drug targeting this pathway is both unnecessary and has potentially harmful off-target effects.

Personalized therapeutics: opportunities and challenges

Kinase inhibitors may theoretically alter the progression of PDGFRB associated disorders. Overactivity of the PDGFRB protein can be partly corrected by tyrosine kinase inhibitors (TKIs), such as sumatinib,26 which have successfully treated patients with PDGFRB activating variants.20 Patients with PDGFRB variants causing myofibromas have been treated with imatinib with robust responses.27 Thus direct inhibition of activated receptor tyrosine kinases may be useful in targeting IAs, particularly IFAs. One study5 found that virus mediated overexpression of a mutant PDGFRB induced a fusiform-like dilatation of the basilar artery in mice, and this phenotype was blocked by systemic administration of sunitinib, a TKI. Chenbhanich et al 19 reported a patient with the PDGFRB p(Tyr562Cys) variant with multiple progressively enlarging IFAs and coronary artery fusiform aneurysms treated with a TKI (sorafenib), one of which subsequently ruptured. It is unclear whether the rupture was related to the therapy or the natural course of the disease.

The possibility of genetic treatment for intracranial aneurysms is currently unknown. There has been no large series to date reporting TKIs to treat aneurysms associated with PDGFRB variants. One pitfall of this variant directed treatment is the possibility for mutation mediated resistance. Resistance may be encountered in the p.Asp850Tyr variant although kinase inhibitors targeting downstream targets (eg, AKT, ERK, and JAK2) may be an alternative, personalized strategy.6 One example of targeting a downstream signaling pathway of PDGFRB was demonstrated by Hao et al 8 in which the JAK2-STAT pathway was successfully inhibited using the JAK specific inhibitor ruxolitinib. Inhibiting p-JAK2 effectively reversed the PDGFRB driven inflammatory phenotype in vSMCs, with implications for treatment of IFAs.

Another challenge to initiate TKI therapy for IAs is the significant systemic side effects experienced with the drug. Orally bioavailable TKIs are associated with adverse effects,28 such as skin toxicity, edema, nausea, hypothyroidism, vomiting, and diarrhea. Most TKIs also exhibit hematological side effects, like anemia, thrombocytopenia, and neutropenia, as well as an increased risk of bleeding due to inhibition of vascular endothelial growth factor receptor.29 In one double-blind, placebo-controlled, multicenter, randomized phase III trial studying sunitinib administration among 207 patients with gastrointestinal tumors, 83% of patients experienced an adverse event, of which 20% were considered serious.30

Much of the literature thus far detecting PDGFRB variants is based on tissue harvested from resected IAs at the time of surgery. However, many IAs (including vertebrobasilar IFAs) are often not amenable to microsurgical treatment. Since somatic variants in PDGFRB arise post-zygotically, testing peripheral blood is of limited efficacy. In determining whether an individual has this PDGFRB variant and, in turn, candidacy for personalized TKI therapy, minimally invasive endovascular biopsy techniques yielding cells amenable to such sequencing has been proposed.31

Endovascular biopsy is a technique for harvesting viable vascular endothelium using a deployable device, such as a coil through a microcatheter, to contact the endothelium of the aneurysm.31 The device is then removed and isolated in a buffer, and the endothelial cells are incubated. The length of time that the devices are left in contact with the area of interest is not well reported. Several hundred endothelial cells can be collected in this way, although cell yields vary considerably across studies and devices.32–35 Unfortunately, cell yields may be low particularly in lesions with sparse endothelial cell populations, and studies have not yet identified aneurysm specific variants of endothelial cells obtained through this technique. Despite this limitation, one study detected the PDGFRB variant in endovascular biopsy acquired cells obtained from a patient with cutaneous manifestations, which differed from those in the patient’s iliac artery.36 Variant to allele frequencies were 2.8% and 0.9% for the aneurysm and iliac artery, respectively. This technique paves the way for detection of other somatic variants in IFAs, yielding relevant targetable molecular information for personalized treatment.

Endovascular devices may also provide substrate to enable local drug delivery of chemotherapeutics. Local delivery would be superior for chemotherapeutic delivery to IA tissue for multiple reasons. First, the existence of the blood–brain barrier limits the ability to achieve therapeutic doses of drug to the target site.37 Transient damage to the vascular endothelium may allow enhanced drug delivery, minimizing the blood–brain barrier limitations. Second, local drug delivery reduces systemic concentration of drug, thereby mitigating the non-target systemic exposure and organ toxicity often associated with chemotherapeutics such as TKIs.38 Third, by delivering drugs locally, the pharmaceutical concentration at the target site can be maximized, bypassing the multiple steps of metabolic degradation that occurs when the drug is delivered systemically.38 Finally, there is significant evidence that IAs may develop via somatic mosaic variants and the pathogenesis indicates that the lesion is localized. Therefore, delivery of the drug to all cells is both unnecessary and potentially dangerous.

The field of interventional cardiology has set a blueprint for the introduction of drug eluting stents to intracranial vessels. The same methodologies for surface modification may be applied to our field, where surface modifications of intracranial stents have been developed to reduce thromboembolic complications.39 Surface modification of intracranial stents for local gene therapy (eg, the release of active drugs, such as TKIs) would reduce or eliminate the need for systemic drug therapy.39

This review serves as a guide to understand the very beginning of possible targets for gene therapy to treat IAs. Given the relatively new technology of whole exome sequencing that has enabled the detection of these low frequency variants, there are only a small number of variants reported to date.5–9 11 12 40 Furthermore, obtaining IA tissue for sequencing is another significant limitation of this work. There are other obvious limitations to this review, including the large heterogeneity in the data reported, limited animal models and their unclear clinical relevance, as well as lack of data regarding humans treated with gene therapies such as TKIs.