Caveats to Acute Imaging for Acute Stroke in the Setting of Venoarterial Extracorporeal Membrane Oxygenation

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is increasingly used for cardiorespiratory support in medically refractory cases. The high rate of neurologic complications, specifically cerebrovascular disorders such as acute ischemic stroke, in VA-ECMO patients frequently prompts the acquisition of neuroimaging studies such as CT with angiography and perfusion imaging. Clinicians must be familiar with the ECMO-related artifacts when interpreting such studies. Here, we describe a case of asymmetric contrast opacification in a VA-ECMO patient with axillary artery cannulation.

PRACTICAL IMPLICATIONS

Neurologic complications, such as stroke, are common in ECMO patients and may prompt urgent acquisition of contrast-enhanced neuroimaging studies. Therefore, clinicians should be cognizant of ECMO vascular configurations that can alter the timing of contrast opacification during neuroimaging acquisition. This is especially important if the neuroimaging findings conflict with the neurologic examination. Ancillary studies such as carotid and transcranial Doppler ultrasound may be helpful in challenging cases.

Case

A 20-year-old woman with a history of antiphospholipid syndrome, heterozygous Factor V Leiden mutation, and nonischemic cardiomyopathy presented with acute chest pain and shortness of breath due to heart failure exacerbation at 27 weeks of gestation. She underwent an emergent cesarean section complicated with cardiogenic shock requiring venoarterial extracorporeal membrane oxygenation (VA-ECMO) support. Initial right femoral artery cannulation resulted in lower extremity ischemia, so the cannulation site was subsequently switched to the right axillary artery. As the cardiac function improved, the arterial cannula flow rate was reduced to 2.32 L/min. There was some degree of cardiac recovery at that point. The ejection fraction at that time was 38%. Shortly thereafter, she developed transient extensor posturing and right gaze deviation for several minutes prompting an acute stroke activation. A noncontrast CT scan showed small bilateral patchy areas of gray-white matter loss of differentiation in the high frontal convexities, more conspicuous on the left (Figure 1A). Head and neck CT with angiography (CTA) did not show contrast opacification within the right brachiocephalic, subclavian, common, internal, and external carotid arteries and the right anterior and middle cerebral arteries (ACA and MCA) as shown in Figures 1A and 2. Delayed phase imaging showed collateral flow to the right MCA and ACA territories from the left anterior circulation. Head CT with perfusion imaging (CTP) demonstrated an increased time to peak with decreased cerebral blood flow and volume in the right MCA and ACA territories (Figure 1A). However, the right carotid pulse was palpable, and the neurologic examination improved spontaneously with no residual focal weakness or gaze deviation. Her arterial cannula flow rate on return from CT was 2.78 L/min. A bedside carotid Doppler ultrasound demonstrated normal blood flow within the right common and internal carotid arteries, suggesting that the CTA and CTP findings may represent artifactual pseudo-occlusion of the carotid artery secondary to the ECMO circuit and configuration. A follow-up transcranial Doppler ultrasound (TCD) 6 hours later confirmed normal velocities and pulsatility indices (PIs) of the bilateral MCAs. Also, a 12-hour follow-up noncontrast CT did not show interval development of a right hemispheric infarct. Continuous electroencephalography showed right frontotemporal rhythmic delta activities, suggesting a postictal phenomenon, and anti-seizure medications were started. The patient was liberated from ECMO 2 days later. Repeat CTA and CTP after decannulation demonstrated normal flow dynamics on the right (Figure 1B). Brain MRI 3 days after the neurologic event showed patchy bilateral (left greater than right) areas of diffusion restriction that predominantly affected the supratentorial cortices, most consistent with hypoxic-ischemic injury rather than a hemispheric stroke (Figure 1C). The patient only had subtle bilateral intrinsic hand muscle weakness during the outpatient follow-up examination.

Figure 1. Demonstration of ECMO-Related Contrast Opacification Artifact With Serial Neuroimaging.

While on ECMO, the CT scan with perfusion imaging and arterial phase demonstrated that the contrast was not captured within the imaging acquisition's normal timing (green arrow). There was decreased blood volume (white arrow) and increased time to peak (orange arrow) despite normal velocities in the right middle cerebral artery (A). After the patient was taken off ECMO a day later, the patient's perfusion imaging and arteriogram were normal with noted capture of contrast (green arrow), symmetric blood volume (white arrow), and no noted increase in time to peak (orange arrow) (B). A follow-up MRI performed 2 days later showed changes consistent with her previously being on ECMO with scattered areas of restricted diffusion in the bilateral cortex (blue arrows) but no large territorial infarction was found in the right hemisphere (C). ADC = attenuated diffusion coefficient; CBV = cerebral blood volume; DWI = diffusion-weighted imaging; ECMO = extracorporeal membrane oxygenation; FLAIR = fluid-attenuated inversion recovery; TTP = time to peak.

Figure 2. Demonstration of Altered Contrast Opacification on Initial CT Angiography While on ECMO.

The arterial extracorporeal membrane oxygenation cannula in the right axillary artery (green arrow) with opacified blood in the aorta (A), but with a notable mixing of unopacified blood exiting the brachiocephalic artery (red arrow). The right carotid artery (B) is without any contrasted blood as seen from the cervical portion and intracranially on the CT angiogram (right). The left carotid artery (C), however, shows contrast opacification.

Discussion

This VA-ECMO via axillary artery cannulation case^1,2 describes a flow dynamics artifact on CTA and CTP imaging, commonly obtained for acute stroke evaluations. Although the exact mechanisms are unclear, variations in the cannulation site and ECMO settings can explain the findings.³ A computation model of axillary artery cannulation in VA-ECMO has shown that brain oxygenation relies on the cannula insertion site and ECMO flow rate. Although flow rates greater than 4.9 L/min are typically required for adequate perfusion of the common carotid arteries during femoral cannulation, flow rates as low as 1 L/min may be sufficient during axillary cannulation.³ Per Bernoulli principle, a high blood flow rate from the right axillary artery cannula results in a lower pressure in the right brachiocephalic artery and its distal branches. It thereby decreases the entry of contrast-opacified (and oxygenated) blood into cerebral blood vessels on the right. However, residual intrinsic cardiac function perfuses the left subclavian artery with contrast-opacified blood, as seen in this case. Several pediatric studies have investigated this question using TCD and near-infrared spectroscopy. A study of 40 VA-ECMO patients (19, 17, and 4 with axillary, femoral, and central artery cannulation, respectively) found a higher blood flow velocity in the right MCA than the left MCA independent of cannula position.⁴ Axillary artery cannulation was associated with a lower PI (0.55 and 0.47 in the right and left MCA, respectively) than femoral artery cannulation (0.94 and 0.79 in the right and left MCA, respectively).⁴ In our patient, the PIs were 0.76 and 0.72 in the right and left MCA, respectively, consistent with previous reports.⁴ In this case, the physical and neurologic examination and ultrasound findings helped confirm the imaging studies' artifactual nature. The accurate diagnosis prevented unnecessary procedures such as catheter angiography for endovascular thrombectomy and wrongful neuroprognostication. As evidenced by the clinical semiology and EEG findings, the event was likely a seizure precipitated by hypoxic cortical injury. She was treated with levetiracetam without recurrence of seizure-like activity. In conclusion, ECMO configuration may confound contrast-enhanced neuroimaging findings and thorough neurologic examination findings, but Doppler ultrasound studies can help.

Appendix. Authors

Study Funding

No targeted funding reported.

Disclosure

The authors report no disclosures relevant to manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.