Empiric Treatment of Metameric Cerebrofacial Arteriovenous Malformation Syndrome with Trametinib-Mediated MEK Inhibition

This report describes our experience with empiric treatment of a metameric brain AVM by MEK inhibition, uninformed by lesion genotype. The AVM was not considered curable, and conventional treatment approaches were regarded as having an unfavorable risk: benefit profile. Given the presence of high-risk features such as flow-related aneurysms, the long-term risk of hemorrhagic stroke was expectedly high. Encouraged by a prior report of MEK inhibition in a patient with spinal metameric AVM, we started our patient on a trial of trametinib. In contrast to prior reports of MEK inhibition for AVM, our therapy was not guided by AVM genotype since there were no cutaneous lesions safe for biopsy and sequencing of somatic mutations^2,3. We reasoned that metameric AVMs might share somatic activating mutations clustered in specific gene families, and that the positive therapeutic responses (i.e. reduction in pain, nidal size and nidal flow) achieved by MEK inhibition in the previously reported case of spinal metameric AVM would be possible in our patient. The unanticipated adverse event experienced by our patient contrasts with experience previously reported in patients receiving genotype-directed therapy. It is unclear if differences in treatment response and clinical outcome are related to genetic, anatomical or other factors. In the future lesional genotyping should be prioritized to better contextualize drug response.

Although trametinib therapy was associated with a marked decrease in AVM blood flow and regression of flow-dependent high-risk features within a few months, these changes were attributable to progression of a focal intracranial LICA stenosis rather than involution of AVM nidus. Consequently, total AVM blood flow as reflected by straight sinus flow, initially decreased but then increased, likely as a result of AVM flow demand being redistributed to alternative arterial feeders. The focal stenosis of the intracranial LICA and associated alterations of cerebral perfusion were not reversible after discontinuation of trametinib on short-term follow up. Longer term follow up will be needed, however. Although the left AchA aneurysm showed a remarkable decrease in size in response to trametinib-induced hemodynamic changes, there was no significant change in the left SHA aneurysm. These differences may relate to differential proximity of the aneurysms to the focal intracranial LICA stenosis and associated flow jet. Notably, although the position of the AchA aneurysm in relation to the stenosis was a modulating factor, the primary driver of aneurysmal involution was the reduction in aneurysmal flow accompanying the progression of upstream arterial stenosis.

It is remarkable that the largest effect of trametinib on vascular structure and function was isolated to a short focal segment of intradural LICA directly upstream to the AVM. The affected segment of internal carotid artery was shown to have mild narrowing on baseline imaging studies prior to the initiation of trametinib. It is possible that vascular cells in the affected ophthalmic segment of internal carotid artery (ICA) share a common progenitor cell origin with vascular cells giving rise to AVM and that both vascular tissues harbor a shared somatic mutation in the KRAS signaling pathway. Modern theories of vascular dysgenetic neurocristopathies associated with aberrant ICA development, such as the PHACES syndrome, are based on the concept that defective neural crest cells possessing somatic mutations contribute to aberrant vascular development in derivative tissues throughout their migratory pathway¹¹. Concepts about the embryological basis of cerebrofacial metameric syndromes are based on similar mechanisms of defective neural crest cell migration⁹. Cranial neural crest cells contributing to development of the forebrain and retinal vasculature also contribute to the development of the ICA^12,13. Accordingly, it is possible that the drug-responsive segment of intracranial ICA possesses the same somatic activating mutation as diencephalic AVM and that differential effects of trametinib on the ophthalmic segment of ICA and AVM nidus are primarily a function of pharmacokinetics. Together with prior reports, our findings emphasize that the response of vascular malformations and other host tissues to MEK inhibition may be variable and depend not only on the nature of the mutation underlying AVM pathogenesis, but the distribution of the mutation in other host tissues outside of the AVM. Our results suggest that pre-existing arterial stenoses involving a major cerebral artery, within the metameric distribution of the AVM (i.e. internal carotid artery), could be an important biomarker of stroke risk in patients with metameric brain AVMs receiving trametinib. Future studies of MEK inhibition for AVM should rely on a detailed understanding of these factors.
The neuropathological effects and clinical complications of cerebral arteriovenous malformations (AVM) are direct manifestations of the pial hemodynamic derangements caused by arteriovenous shunting. Cerebral arteriovenous shunting 1) markedly elevates vascular wall tension stresses, causing a flow-dependent cumulative mural damage process that leads to the formation of flow-related aneurysms and venous ectasias that rupture, 2) steals perfusion from brain tissue and 3) elevates luminal pressure in regional veins causing cerebral venous hypertension and congestion. The combined effects of perfusion steal and venous dysfunction lead to pathological depolarization of cortical neurons and seizures, as well as infarction and hemorrhage. The severity of vascular mural wall tension, cerebral perfusion steal and cerebral venous dysfunction are all directly related to the magnitude of arteriovenous shunting. Hence, the probability of clinical complications is proportional to the magnitude of blood flow directed through the AVM. This is a well-established guiding principle in the management of brain AVMs. Consequently, the primary endpoint for monitoring treatment response in our patient was quantitative blood flow through the AVM. Since the complex architecture of the AVM precludes direct, non-invasive, quantitative determination of AVM blood flow, we used quantitative 4D flow MRA measurements of internal carotid artery flow and straight sinus flow as surrogates. As the effects of MEK inhibition on AVM structure and function are unknown, and potentially non-uniform in distribution this global approach was favored and has been previously described in studies of trametinib therapy for metameric spinal AVM2. Our initial assessments based on these metrics confirmed that AVM flow was dramatically decreased, and that our treatment goals were being met. Unfortunately, the decrease in AVM blood flow was due to a generalized upstream reduction in flow through the internal carotid artery, which also reduced blood flow to normal brain tissue. Since MRI does not allow detailed evaluation of the structural changes within the AVM vasculature, the absence of changes in the overall size of the AVM was not necessarily an indication of treatment failure. The inability of MRI to accurately depict changes in AVM structure over time is well known, and definitive assessments are based on catheter-directed cerebral angiography. While the progression of focal internal carotid artery stenosis was not initially appreciated on MRI, this may not have compelled cessation of trametinib therapy in the absence of clinical complications. At the time of our patient’s arterial ischemic stroke, the severity of ipsilateral focal internal carotid artery stenosis was not yet understood. Since our therapeutic endpoints were being achieved, it was believed that a trametinib dose reduction would cause a sufficient increase in carotid blood flow to satisfy cerebral blood flow demand. The patient clinically stabilized on the lower dose of trametinib, and quantitative flow metrics also stabilized. When the severity of internal carotid artery stenosis was later revealed by catheter-directed angiography, our consensus was that the potential for further stenosis and recurrent stroke, even on the reduced dose of trametinib, outweighed the potential benefits of further non-specific AVM flow reduction. Notably, the highly focal steno-occlusive changes involving the ophthalmic segment of intracranial internal carotid artery in our patient were not evident on non-invasive vascular imaging studies, perhaps due to the unusually short segment of vessel involved and the widespread decrease in flow-related enhancement upstream to the index steno-occlusive lesion. Since MRA displays signal intensity proportional to flow, the structural nature and location of the underlying steno-occlusive changes were not accurately depicted. Although arterial spin labeling (ASL) sequences were routinely obtained as part of each imaging evaluation performed in our patient, these failed to accurately represent progressive changes in cerebral blood flow over time. Quantitative 4D flow MRA also lacks spatial and anatomical information. On the other hand, catheter-directed angiography and other contrast-based vascular imaging studies clearly demonstrate the nature of underlying focal steno-occlusive changes. Our experience emphasizes the indispensable role of accurate, reliable, high fidelity contrast-based vascular imaging for monitoring and understanding treatment responses in patients receiving MEK inhibition therapy for vascular malformations.
Optimal AVM therapy should achieve a targeted involution of aberrant vessels within AVM nidus. Selective constrictive remodeling of arterial feeders supplying both AVM and brain may promote regression of flow-dependent high-risk AVM features, but this benefit will be counter-balanced by an increased risk of ischemic stroke due to increased competition for blood flow between brain and AVM, as observed in our patient. In such cases, a net benefit of drug therapy may be possible if the reduction in blood flow is sufficient to induce a regression of high-risk AVM features without precipitating cerebral ischemia. Future research directed at illuminating how quantitative changes in dosing can achieve predictable treatment responses may enable reliable dose titration for a net benefit.