Arterial Transit Awesomeness

See also the article by Di Napoli et al in this issue.

Dr Greg Zaharchuk is a professor of radiology at Stanford University in the Division of Neuroimaging and Neurointervention. He obtained his MD degree from Harvard Medical School, his PhD degree from the Harvard–Massachusetts Institute of Technology Health Sciences and Technology program, and did his clinical training at University of California, San Francisco. His research focuses on advanced medical imaging techniques and algorithms (including artificial intelligence) with the goal of alleviating the burden of neurological disease.

Why do we image patients with recent symptoms related to acute stroke or transient ischemic attack? There can be only one reason—to help neurologists and other providers treat the patient appropriately to minimize their risk of stroke in the future. To that end, many tests have been suggested to provide insight into this question, including vascular imaging for carotid stenosis, evaluation of the morphologic characteristics of the carotid artery vessel wall and plaque (if present), diffusion-weighted imaging, perfusion imaging, and cerebrovascular reserve (ie, a measurement of the ability of the brain to increase cerebral blood flow in response to a vasodilatory challenge). In this issue of Radiology, Di Napoli et al (1) identified a potentially new sign for patients with presumed carotid stenosis: the presence of arterial transit artifact (ATA) on noncontrast arterial spin labeling (ASL) perfusion images.

To understand what ATA is, it is useful to review how ASL works. Water protons in the blood are labeled magnetically in the cervical arteries leading to the brain, usually for a duration of several seconds. After this, a pause (called the “postlabel delay”) in the sequence occurs to allow the labeled water to move from the neck to the capillary bed of the brain, where it is extracted into the parenchyma. The length of the pause is a compromise: it must be long enough for the labeled water to reach the capillaries but not so long that their signal starts to disappear, which occurs at a rate determined by the T1 of arterial blood (1.75 seconds at 3.0 T). For these reasons, the postlabel delay chosen is typically between 1 and 2 seconds (a recent International Society of Magnetic Resonance in Medicine white paper suggested 2 seconds as a good choice for clinical applications) (2).

This timing works well for the vast majority of patients, but there are some people for whom the pause is not long enough. These include older patients, because in humans cerebral blood flow peaks in adolescence and declines relentlessly thereafter (a somewhat depressing thought!) (3). Patients with arterial stenosis are another group for whom this pause is not long enough; this can be for several reasons, including simply slowing of the flow rate through the stenosis or the time it takes for blood to arrive via collateral pathways (4). Regardless of the reason, if the postlabel delay is not long enough for the patient’s labeled blood to reach the capillary bed, a mix of high and low ASL signal intensity, often in a punctate pattern, will appear in the affected vascular territory. This is because two phenomena are occurring. The high signal intensity is because of the presence of the labeled blood in the feeding arteries at the time of imaging. The surrounding low signal intensity reflects more distal parenchyma that the labeled blood will eventually perfuse, but has not yet reached at the time of signal acquisition. This mixed pattern was identified early in the evaluation of ASL in humans by Alsop and Detre (5) and Detre et al (6), who called it “arterial transit artifact.”

However, one person’s artifact is another person’s biomarker. Over the years, investigators and clinicians have realized that this sensitivity to arterial arrival times can be a helpful diagnostic sign. ASL is a sequence that can be prone to many artifacts, including those from incomplete labeling, arrival time variations, through-plane blurring, and cerebrospinal fluid pulsation (7). For this reason, it is most useful when normal because this is evidence that cerebral hemodynamics are not compromised. By this logic, the presence of ATA in a vascular territory suggests that something is not quite right and should prompt the investigation of the upstream circulation for a cause. For example, Yeom et al (8) identified such ASL abnormalities more commonly in patients with neurofibromatosis type 1, in whom large artery vasculopathy is commonly present. In another example, de Havenon et al (9) showed that the presence of ATA immediately after acute ischemic stroke was a predictor of more favorable outcome, representing the presence of good collateral vessels. For reasons such as this, I like to think that the letters ATA should in fact stand for arterial transit awesomeness!

This awesomeness is again demonstrated by Di Napoli et al. Di Napoli et al performed 3.0-T carotid plaque imaging and ASL in 44 patients with 50% or greater carotid stenosis from two registries. They sought to understand what imaging markers were associated with recent symptoms, such as transient ischemic attack or stroke, and specifically looked at degree of stenosis, carotid plaque morphologic structure, presence of intraplaque hemorrhage, and ATA on ASL images. They found ATA to be the only predictive feature of whether a patient was recently symptomatic and hypothesized that this was because of the effect of the stenosis on cerebral hemodynamics. In fact, in all patients with ATA, the stenosis was at least 70% or greater, suggesting that if isolated ATA is found on brain images without an intracranial stenosis, it might make sense to screen the patients for a potential high-grade carotid lesion.

Certainly, some limitations remain. For one, this is a small, single-center study that used a specific 3.0-T MRI scanner and slightly nonstandard ASL parameters. Unfortunately, ASL as a technique is implemented in vastly different ways on scanners from different manufacturers, leading to strikingly different appearances. Also, ASL image quality is markedly better at 3.0 T compared with 1.5 T and the conspicuity of ATA may also differ because of changes in the blood T1 at these two field strengths. For these reasons, it will be useful to follow up the results in larger cohorts scanned with different systems to confirm the generalizability of these findings. Another key point is that the association between recent symptoms and ATA was studied, rather than what is arguably more important: ATA’s relationship with future symptoms. Only at confirmation of this latter association would ATA be valuable for recommending potential interventions such as carotid endarterectomy or stent placement. Finally, it was a bit surprising that no significant relationships were found between the carotid plaque characteristics and recent symptoms, given previous literature to the contrary. However, this might be because of the relatively small sample size and lack of use of a dedicated carotid coil.

In summary, Di Napoli et al pointed out an interesting and easy-to-collect piece of imaging information that can be obtained with most clinical MRI scanners, that does not require contrast material administration, and that may be related to recent and possibly future ischemic symptoms. Abnormalities on ASL images, such as ATA, may yield new insights into the synergistic roles of embolism and hypoperfusion in the etiologic cause of acute stroke because they can be evaluated in large numbers of patients by using routine MRI (10). Future studies evaluating this sign and its prognostic value will help us to truly understand the potentially awesome value of ATA in patients with carotid stenosis.