Arterioectatic Spinal Angiopathy of Childhood: Clinical, Imaging, Laboratory, Histologic, and Genetic Description of a Novel CNS Vascular Pathology

DISCUSSION

This report describes 4 cases of early-onset childhood myelopathy, each expressing a similar pattern of clinical and neuroimaging features and each associated with an unusual form of arterioectatic spinal angiopathy (AESA). All children initially presented with hypotonia and impaired ambulatory motor skills during the first 2 years of life. Most showed rapid progression to quadriparesis, and half died from disease-related complications within 6 years of onset. The clinical and neuroimaging features of AESA have striking overlap with those expressed in venous congestive myelopathy due to spinal cord AVM/AVF. Clinical recognition of the former and differentiation from the latter are critical to prevent misguided and potentially harmful vascular interventions. Our literature review confirms that AESA is a currently unrecognized form of spinal cord vascular pathology. Perhaps the most important factor contributing to the lack of prior recognition is the similarity that AESA bears to childhood spinal cord AVM/AVF. All of the cases presented in this series were initially misdiagnosed as AVM/AVF. It is likely that other cases were also misdiagnosed and perhaps even treated as AVM/AVF.

Every child in the current series showed ectasia and tortuosity of the spinal cord vasculature on MR imaging. Each case also demonstrated T2 signal hyperintensity in the central and anterior spinal cord, as well as cord tumefaction. This combination of MR imaging findings is common in children with intradural spinal AVF and adults with spinal dural AVF, though spinal dural AVF has not been convincingly reported in children.¹³ Notably, gold standard catheter-directed DSA excluded AVF in all our patients. The overlapping neuroimaging features of AESA and spinal cord AVF of childhood may indicate that some aspects of pathogenesis are shared. The pattern of spinal cord signal abnormality and tumefaction demonstrated in our patients suggests vasogenic edema and cord swelling, as seen in children with an intradural spinal AVF. In both conditions, the spinal cord experiences an imbalance between arterial inflow and venous outflow, resulting in vascular congestion. In the former, spinal cord arterial inflow is augmented, and in the latter, spinal cord venous drainage is diminished. The entire spinal cord and ventral medulla are shown to be affected in our patients. Clinically severe (lethal) cases also showed variable cord enhancement, indicating cord–blood barrier disruption.

In contrast to children with intradural spinal AVF, patients with AESA demonstrate cord hyperemia causing early, dense parenchymal staining on DSA. AESA is further characterized by brisk, diffuse angiographic filling of spinal cord veins draining hyperemic spinal cord capillary beds. This pattern of circulation is a characteristic finding on which the diagnosis of AESA may be based. The angiographic features of spinal cord parenchymal venous drainage in AESA contrasts with those seen in cases of an AV shunt with medullary venous hypertension (ie, spinal dural AVF) in which there is an absent or delayed angiographic appearance of the veins serving the affected spinal cord segments.¹⁴ Marked arterial ectasia, out of proportion to venous ectasia, is another diagnostic hallmark of AESA. In our series, catheter-directed DSA shows ectasia and tortuosity of the anterior spinal artery and corresponding central sulcal perforators. Most cases show relatively mild enlargement of posterolateral spinal arteries, suggesting that ectasia is proportional to the size of the vascular territory within the cord. This finding loosely correlates the magnitude of angiopathic changes with blood-flow demand.

AESA as defined by the clinical and neuroimaging features reported here has not been previously described in the peer-reviewed, indexed medical literature. Two cases satisfying the diagnostic criteria for AESA set forth here were described in a book chapter on spinal cord AVM.¹⁵ These patients underwent catheter-directed DSA for further evaluation of myelopathy, one at 2 years of age and the other at 5 years of age. In both cases, spinal DSA revealed ectasia of the anterior spinal artery and central sulcal perforators with corresponding cord hyperemia.

The etiology of AESA and its precise relationship to myelopathy are presently unknown. Analysis of blood, CSF, and RMA tissue from our patients showed no evidence of inflammation. Furthermore, glucocorticoid steroid therapy showed no beneficial effect in 1 patient. Early in our experience, the observation of spinal cord hyperemia and edema led us to hypothesize that neurologic impairment might be the consequence of abnormally increased arterial perfusion of the spinal cord, unbalanced by venous drainage. We thus speculated that controlled reduction of spinal cord hyperemia by surgical ligation of the dominant RMA might reduce cord edema and improve neurologic function. Unfortunately, this intervention was not beneficial, suggesting that cord hyperperfusion is not a primary driver of myelopathy.

Multiple findings suggest that AESA is a manifestation of metabolic myelopathy due to mitochondrial disease. Among these findings are elevated CSF lactate and a cerebral MR spectroscopy lactate peak in 1 case, systemic lactic acidosis and encephalopathy in a second case, and elevated CSF pyruvate in a third. Transaminasemia, premature cataracts, and microcytic anemia in our cohort further support a mitochondrial disease origin.^16,17 Notably, none of our patients demonstrated evidence of myopathy (cardiac or skeletal muscle). Numerous metabolic disorders including mitochondrial diseases presenting with myelopathy in early childhood manifest MR imaging features that resemble those reported here.^18,19 In our cohort, longitudinally extensive spinal cord lesions spanning >3 vertebral bodies were observed. This feature is characteristic of myelopathy in children with primary mitochondrial disease.¹⁹ Moreover, in our series, MR imaging revealed T2 signal hyperintensity and tumefaction involving the central and anterior cord. This MR imaging pattern of myelopathy has also been reported in a subset of children with primary mitochondrial disease.¹⁹ Additionally, cervical cord involvement with extension into the medulla is a common feature of childhood mitochondrial myelopathy seen in our cases.¹⁹

Pathogenic mutations that cause mitochondrial disease can occur in the nuclear genome or the mitochondrial genome.²⁰ Notably, 1 patient in this series was heterozygous for a dominant pathologic mutation of a transferrin receptor gene (TFRC). Transferrin receptors are believed to play a vital role in the regulation of normal mitochondrial function and fragmentation.²¹ Additionally, mouse models show mitochondrial dysfunction in heterozygous transferrin receptor mutants.²² Incidentally, abnormalities of transferrin receptor function may also explain associated anemia and childhood-onset cataracts.^23⇓⇓-26 Although 1 other patient in this series underwent extensive testing of nuclear and mitochondrial DNA without identification of a pathologic mutation, previous studies have shown that such testing yields a positive result in only 10%–20% of children with clinical mitochondrial disease.^27⇓-29

Arteriopathy as a primary or secondary manifestation of mitochondrial disease is now widely recognized.³⁰ Microvascular and macrovascular forms are well-described.³⁰ In patients with mitochondrial disorders, secondary arteriopathies are manifestations of the vascular effects of metabolic dysfunction in distinct organ systems such as the liver or pancreas. One model of secondary arteriopathy in AESA could be based on an abnormally increased metabolic demand in the diseased spinal cord imposing an increased blood-flow requirement that promotes adaptive or maladaptive expansive remodeling of spinal cord arteries. If the AESA phenotype requires the combined effects of 2 mutations, one controlling mitochondrial function and the other controlling arterial adaptation to blood flow demand, it would account for the extreme rarity of AESA. Notably, primary mitochondrial arteriopathy is often characterized by arterial ectasia and aneurysm formation.³⁰ Diagnostic markers of primary mitochondrial arteriopathy include abnormal expression of succinate-dehydrogenase or cyclooxygenase in vascular tissue or abnormal cristae formation in mitochondria from vascular smooth-muscle cells and pericytes.³⁰ The former relies on immunohistochemical staining of fresh-frozen tissue, and the latter relies on electron microscopy of glutaraldehyde-fixed tissue.

Immunohistochemical and electron microscopy studies of affected vascular tissue may improve our understanding of AESA pathogenesis in the future. Abnormal ectasia of the aorta and cerebral arteries has been reported as a primary feature of clinical mitochondrial disorders.³⁰ Mitochondrial arteriopathies involving retinal vessels, cochlear vessels, cervicocerebral vessels, brachial vessels, iliac vessels, skeletal muscle vessels, and cutaneous vessels have been reported.³¹Asymptomatic generalized cerebral arterial ectasia was present in 1 patient from the current series, suggesting that AESA may be part of an angiopathy spectrum that variably affects the entire CNS. Notably, 2 of our patients showed cerebral parenchymal volume loss, and a third experienced progressive encephalopathy, suggesting a diffuse CNS process.

Our review of the literature has not revealed prior cases of mitochondrial disease expressing AESA as a trait.^{30⇓⇓⇓⇓-35} It is possible that the cases described in this report represent the extreme end of a spectrum in which milder forms are more prevalent but difficult to detect. Mild forms of AESA may not be detected on MR imaging, just as perimedullary vascular flow voids may be inapparent in patients with spinal dural AVFs.^36,37 CNS mitochondrial disorders are known to express a wide range of phenotypic heterogeneity.³⁸ Subclinical mitochondrial vasculopathies revealed by specialized, nonclinical tests have been found in numerous studies.²⁴ It is also possible that mild forms of AESA are overshadowed or preempted by other complications of mitochondrial disease such as heart failure, respiratory failure, and encephalopathy.³⁹ Aortic root ectasia manifest on echocardiography was only recently recognized in children with mitochondrial cardiomyopathy, even though echocardiography was performed in this population for decades prior.⁴⁰ Patients with severe forms of AESA likely succumb to complications of AESA before other features of mitochondrial disease manifest. Recognition of AESA as a feature of mitochondrial disease would thus be prevented. Because severe forms of AESA bear a striking similarity to spinal cord AVM/AVF, it is likely that most cases are misdiagnosed because many centers lack the expertise needed to differentiate AESA from AVM/AVF. Moreover, absence of a pre-existing diagnostic framework that accommodates AESA thwarts recognition.

Endothelial dysfunction is believed to be an important feature of mitochondrial arteriopathies, and observational data suggest that intravenous L-arginine infusion may have therapeutic benefits, particularly in patients with mitochondrial encephalopathy experiencing strokelike episodes.^41,42 It is possible that L-arginine therapy may be helpful in children with AESA. One patient in this series failed to respond to carnitine supplementation. While carnitine supplementation is commonly administered to patients with mitochondrial disorders because of theoretic biochemical effects, therapeutic efficacy has not been shown outside a subset of patients with mitochondrial myopathy.^5,43 Mitochondrial augmentation and transplantation strategies are currently being investigated as a therapeutic approach to mitochondrial diseases.²⁰

An alternative, though a less likely possibility, is that AESA is a transitional form of spinal cord vascular malformation whose developmental progression toward the AVM phenotype has been at least temporarily arrested at an early stage. In animal models, initiation of the brain AVM phenotype begins with angioectasia in precapillary and capillary regions of the developing cerebral circulation, exposing the cerebral microcirculation to an atypical arterial pattern of flow.⁴⁴ Transition to the brain AVM phenotype occurs when exposure to this anomalous flow signal triggers aberrant developmental programing. It is possible that AESA fails to progress toward the AVM phenotype because precapillary angioectasia is unaccompanied by the aberrant genetic programming needed for the transition. While clinical experience and animal models suggest that the window of developmental vulnerability for the phenotypic transition to AVM is limited to early childhood, longer-term follow-up is needed to determine whether AESA can transform to AVM, given the young age of patients in known cases.⁴⁴ Although we did not identify genetic markers of dysfunctional vascular biology in our AESA cohort, novel pathogenic mutations in regulatory regions of DNA or copy number mutations cannot be excluded.

Cerebral proliferative angiopathy (CPA), which has been reported as an atypical cerebral AVM phenotype, has some features that overlap with those of AESA.⁴⁵ The original description of CPA emphasized the presence of “capillary angioectasia” on catheter-directed DSA. The term denotes abnormal enlargement of capillaries. In the setting of catheter-directed DSA, capillary angioectasia is indirectly inferred by the presence of an unusually dense parenchymal capillary blush, accompanied by the brisk appearance of parenchymal draining veins. Thus, capillary angioectasia may be regarded as a feature of AESA. While direct evidence of vascular proliferation with cell-labeling methods was not shown in the original CPA case series, “angioproliferation” was presumed on the basis of transdural vascularization of AVMs and perinidal brain parenchyma. Lasjaunias et al⁴⁵ suggested that angioproliferative vascular changes in CPA were induced by cerebral ischemia engendered by AVM blood-flow steal. Similarly, in AESA, it is possible that angiopathy is the result of altered blood-flow demand attributable to metabolic derangement within the CNS parenchyma.

In contrast to CPA however, AESA is not associated with transdural vascularization or angioproliferative changes by extension. Moreover, published examples⁴⁵ of CPA feature a discrete focal intraparenchymal nidal type AV shunt zone demonstrable on MR imaging and catheter-directed DSA. This aspect of CPA sharply differentiates it from AESA, which conspicuously lacks a focal intraparenchymal nidal AV shunt zone where blood circulates directly from arteries to veins without passing through parenchymal capillaries. Moreover, in CPA, catheter-directed angiography shows the vasculature within the intraparenchymal nidal AV shunt zone to be disproportionately ectatic relative to the extraparenchymal feeding arteries, which are frequently small and stenotic. In contrast, diffuse arterioectasia⁴⁵ involving the extraparenchymal vasculature is the most striking feature associated with AESA.