DECT for Differentiating Intracerebral Hemorrhage From Contrast Extravasation

Background:

Stroke is a leading cause of death in the world and the majority of strokes are ischemic (80%). The first method of treating acute ischemic (AIS) is tissue plasminogen activator (tPA)[1, 2] , which is a thrombolytic medication that breaks up the blockage in the blood vessel. The second is called endovascular thrombectomy (EVT), which is a procedure where the clot is physically retrieved using a catheter[3]. While tPA can only be used within 4.5 hours, EVT can be used 16 hours[4] and even up to 24 hours[5] after a patient develops symptoms.

The go-to modality for imaging AIS is computed tomography (CT)[6, 7]. On a CT scan image, those objects that are brighter are hyperdense and those objects that are darker are hypodense. To determine the location of the blockage and complete the EVT procedure, contrast dye is used in combination with the CT. In patients presenting with delayed AIS, the CT-Perfusion (CTP) modality is also used, which can determine tissue that is receiving less blood flow but has not infarcted yet. Interestingly, CTP can also generate information about the permeability of the blood-brain barrier, which can be used to create a "permeability-surface area product map" (PS). Previous work by our group has demonstrated that using PS, the likelihood (not extent) of hemorrhagic transformation can be predicted[8].

Both tPA and EVT can lead to the complication of intracerebral hemorrhage (ICH). If ICH develops in a patient following AIS, the care-plan and prognosis is very different. Another complication, albeit less dangerous, is contrast extravasation (CE) into the region of the stroke. Contrast disappears after one to two days and does not affect the patient clinically. It is standard of care to perform a CT scan 24 hours after a patient receives AIS intervention to rule out ICH. Conventional, single-energy CT (SECT) uses one x-ray spectrum. Because contrast and blood are both significantly denser compared to surrounding tissue, they appear identical to one another. In dual-energy CT (DECT), two different x-ray spectra are used to create an additional "iodine overlay map" (IOM) and a "virtual-noncontrast" images [9-11]. By using the IOM and VNC images in conjunction with original SECT image, one can differentiate ICH from CE (Figure 1). DECT does not expose patients the higher levels of radiation compared to SECT and can be a useful technique for differentiating objects on CT.

We will perform a study looking at whether DECT can accurately identify hyperdensities caused from hemorrhage versus contrast. We will validate the role of DECT in post stroke care for those patients that undergo EVT and have post procedural hyperdensities.

Hypothesis:

DECT will have a greater sensitivity and specificity for differentiating between ICH and CE compared to SECT in patients with AIS who receive acute intervention.

Objectives:

1. To determine whether DECT can accurately identify hyperdensities caused from hemorrhage versus contrast.
2. To validate the role of DECT in post stroke care for those patients that undergo intervention and have post procedural hyperdensities.

Methods:

Outcome Measures:

Our primary outcome is the sensitivity and specificity of DECT in differentiating ICH from CE.

Our secondary outcomes include: duration of hospitalization, level of disability after discharge from hospital, mechanism of large artery occlusion, type of post-stroke therapy (antiplatelet vs anticoagulant).

Patient Selection: All patients will be selected prospectively. In Manitoba, the Health Sciences Centre (HSC) is the only institution that provides EVT. Therefore, the patient population for our study includes the entire population served by HSC which includes the province of Manitoba, north-western Ontario, and southern Nunavut.

Criteria:

Inclusion criteria for our study are as follows: patients presenting with AIS that are candidates for 1) tPA (presenting within 4.5 hours of symptom onset, no ICH on CT, not on anticoagulation) and 2) EVT (presenting within 24 hours of symptom onset, large vessel occlusion, National Institute of Health Stroke Scale (NIHSS) > 6).

Exclusion criteria are patients who are not candidates for EVT or tPA. Sample Size: Sample size was calculated using the formula: sample size= p(1-p)(Z/E)2. The confidence interval was set at 95% and therefore, Z was set at 1.96 and the desired margin of error E, was set at 0.05. Previously, it has been proposed that the proportion of patients that develop ICH post-EVT for AIS was more than 68%[12]. Based on our experience, we hypothesize that a hyperdensity could be seen in up to 85% of patients undergoing AIS treatment on their 24-hr post treatment CT scan of their head. Therefore, 0.85 was used as the value for population proportion, p. With this calculation, the ideal sample size for recruitment would be 196. We will aim for a sample size of 200. With this selected sample size, the resulting margin of error would be 0.06, which we found to be acceptable.

Imaging Protocols: The standard of care is to perform a non-contrast CT and CT with contrast when patients first present with symptoms of AIS. If patients receive tPA and/or EVT they undergo a non-contrast CT scan 24 hours post-treatment. We will be performing DECT in addition to the standard-of-care SECT at 24 hours. The amount of radiation generated by DECT is the same if not less than that generated by SECT[13].

In patients who have a hyperdensity, on 24-hour CT, a repeat scan (either CT or MRI) will be done at 72-hours post-intervention. The 72-hour scan will be used as the gold-standard to determine if the hyperdensity on 24-hour CT was indeed ICH or CE.

Imaging Data Collection and Analysis: Images will be reviewed by all trained subspeciality neuroradiologists working at the Health Sciences Center.

The study population will first be separated into two categories: patients 1) with and 2) without hyperdensity seen on post-24 hour SECT scan. Of the patients with hyperdensity, we will further categorize them into four groups: 1) Diagnosed as ICH on SECT and 2) diagnosed as CE on SECT 3) Diagnosed as ICH on DECT 4) Diagnosed as CE on DECT. Finally, of the patients with initial hyperdensity of 24-hour scan, it will be determined if the hyperdensity is still present on 72-hour scan; if it is still present, the diagnosis will be confirmed as ICH.

Statistical Analysis:

Analysis will be performed by SPSS software. Continuous data will be expressed as means ± standard deviations or 95% confidence intervals, and categorical data will be expressed as numbers of patients with percentages, respectively. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of DECT for identifying ICH will be recorded. Statistical significance will be calculated using Fisher exact test. A 2-sided P value less than 0.05 will be considered to indicate a significant difference. We will perform receiver operating characteristic curve analyses to calculate the area under the curve (AUC). The optimal cutoff values for all parameters will be determined by the Youden index, the difference between sensitivity and 1- specificity. Diagnostic performance will be compared by using ROC analysis with the DeLong method.

Expected outcome: This study will determine if DECT is superior to SECT in differentiating ICH from CE and will validate the use of DECT in patients with AIS who receive intervention.

Significance: Stroke is a leading cause of death in the world. In 2013, there were 405,000 individuals in Canada living with the effects of stroke and this number is expected to increase to about 700,000 by 2038[14]. In Manitoba alone, about 2000 patients each year suffer from a stroke and up to 500 will have a recurrent stroke. To decrease the risk of subsequent strokes, physicians target optimal blood pressure, blood sugar, and blood thinner management as soon as is safe for the patient. Thrombolytic thrombectomy therapies have been shown to decrease the disability suffered by Manitobans who present with AIS but a significant proportion of these patients develop ICH. These patients' care is separate from those that do not develop ICH and are aimed toward decreasing the bleeding instead of preventing recurrent stroke. The start of antiplatelet and anticoagulation therapies, which have been shown to decrease the likelihood of subsequent stroke the earlier they are started, must be delayed. Blood pressure goals are also separate for these patients.

If DECT is validated for the use in Manitobans who present with AIS who receive intervention, clinicians will be able to determine more accurately whether a patient truly has an ICH and if their therapy needs to be modified accordingly. Those patients who are determined to just have CE on DECT will not need to have their management needlessly changed and their care can be targeted toward avoiding recurrent stroke. The use of DECT will lead to more diagnostic accuracy for those Manitobans presenting with AIS and more effective, patient-directed care.