A New Micro–Computed Tomography–Based High-Resolution Blood–Brain Barrier Imaging Technique to Study Ischemic Stroke
Background and Purpose—Micro–computed tomography (mCT) offers high-resolution images, but it suffers from low contrast sensitivity and poor soft tissue contrast. We introduce a new mCT imaging technique with improved sensitivity for the dynamic spatial and temporal characterization of poststroke blood–brain barrier (BBB) dysfunction in small animals in vivo.
Methods—Transient middle cerebral artery occlusion was induced for 1 hour in 10- to 12-week-old C57BL/6 mice (n=35). At 4, 24, and 48 hours after ischemic stroke, serial in vivo mCT imaging was performed 5 minutes after intravenous infusion (n=3) or intracarotid infusion of iopromide (240 μL) for 5 minutes (n=32). After intravenous injection of 2% Evans blue, we performed ex vivo near-infrared fluorescent imaging of parenchymal Evans blue leakage, visual assessment of poststroke parenchymal hematoma, triphenyltetrazolium chloride staining of the brain tissue, and quantitative mapping of stroke-related brain lesions.
Results—Infarct-related BBB dysfunction could be demonstrated with intra-arterial but not with intravenous infusion of iopromide. Iopromide leakage across the dysfunctional BBB showed a monophasic (not biphasic) course for 48 hours after ischemic insult in both the parenchymal hematoma (n=5) and the non–parenchymal hematoma (n=24) groups, with relatively severe leakiness and greater hemispheric midline shift in animals with hemorrhage. Parenchymal staining on in vivo mCT overlapped with ex vivo fluorescent staining because of Evans blue. Multivariable analyses showed that midline shift and the amount of iopromide leakage at each of the 3 time points predicted the final infarct size at 48 hours.
Conclusions—The new mCT BBB imaging technique, based on the intra-arterial infusion of clinically available iopromide, allows serial quantitative visualization of poststroke BBB dysfunction in mice, with high resolution and in a sensitive manner.
Restoration of blood flow to ischemic brain can cause reperfusion injury that compromises the blood–brain barrier (BBB),1 which has been reported to have a biphasic pattern of dysfunction2: initial increase and then a significant reduction in BBB permeability at 24 hours after reperfusion, followed by a subsequent repeat of increased leakiness. The leaky BBB can cause cerebral edema and hemorrhagic transformation, which can do more harm to the patient than the initial ischemia.3
Micro–computed tomography (mCT) systems for small animal imaging are self-contained without requiring architectural radiation shielding to be installed. mCT offers high-resolution images4 and rapid data acquisition capabilities.5 However, it suffers from low contrast sensitivity and poor soft tissue contrast. To counter these limitations, we hypothesized that intra-arterial infusion of a clinically available iodinated x-ray contrast agent would better visualize BBB permeability than intravenous administration.
In this study, we develop a new in vivo mCT imaging technique to visualize the leakage of iopromide infused intra-arterially for the assessment of BBB integrity in a mouse model of transient focal cerebral ischemia.
Detailed methods are available in the online-only Data Supplement.
Transient focal cerebral ischemia was induced as previously reported (Methods and Movie in the online-only Data Supplement) by occluding the middle cerebral artery (MCA) of 10- to 12-week-old C57BL/6 mice (n=32) for 1 hour (transient MCA occlusion, tMCAO).
To visualize the BBB dysfunction induced by tMCAO, mCT imaging was performed with a mCT small animal imager serially after intravenous (n=3) or intra-arterial (n=29) infusion of iopromide (Ultravist; Bayer HealthCare Pharmaceuticals, Berlin, Germany) for 5 minutes at 4-, 24-, and 48-hour time points. At the last 48-hour time point, double imaging was performed with an intravenous infusion followed immediately by an intra-arterial infusion with separate imaging for each injection in the 3 animals that had underwent mCT imaging with prior intravenous infusions at 4 and 24 hours.
After the last mCT imaging session at 48 hours, 2% Evans blue was injected intravenously. After examining ex vivo brain tissues (six 2-mm-thick coronal slices) for the presence of poststroke edematous hemorrhagic transformation (parenchymal hematoma [PH]; see Methods in the online-only Data Supplement), each slice was imaged using a near-infrared fluorescent imaging machine to detect fluorescence because of Evans blue leakage into the brain parenchyma. Then, these slices were stained with 2% triphenyltetrazolium chloride solution to delineate white/pale infarcted areas.
Quantitative lesion mapping6 using a template set (Figure I in the online-only Data Supplement) and measurement of hemispheric midline shift7 were performed (Methods in the online-only Data Supplement).
Infarct-Related BBB Dysfunction Could Be Demonstrated With Intra-Arterial But Not With Intravenous Infusion of Iopromide
After tMCAO for 1 hour in C57BL/6 mice (n=3), serial mCT imaging was performed with intravenous infusion of iopromide at 4, 24, and 48 hours (Figure 1). As shown in the representative animal, the venous structures were opacified at each of these time points, but no parenchymal leakage of contrast agent was observed. At 48 hours, the intravenous infusion was followed by a second intra-arterial infusion of the contrast agent. After the intra-arterial infusion, brain parenchymal staining was clearly seen in the left MCA territory.
BBB Leakage Showed a Monophasic (Not Biphasic) Course for 48 hours After Ischemic Insult
As was demonstrated in serial mCT images of representative animals (Figure 2) and in accumulation lesion maps with quantification of all data (Figure 3; Figure II in the online-only Data Supplement), the reported phenomenon of biphasic BBB opening2 was not observed in our animals. The data showed a monophasic increasing pattern of leakage in both the PH group (n=5) and the non-PH group (n=24; see Results in the online-only Data Supplement).
Parenchymal Staining on mCT Overlapped With Fluorescent Staining Because of Evans Blue
The areas of iopromide leakage on mCT images overlapped with the areas of EB leakage; and, the iopromide leakage areas and mCT densities on the 48-hour in vivo images correlated with the Evans blue leakage areas and fluorescent images, respectively (Figure IIIA and IIIB in the online-only Data Supplement). The area of iopromide leakage or Evans blue leakage only weakly correlated with the infarct size (Figure IIIC and IIID in the online-only Data Supplement). However, the area of iopromide leakage or mean mCT density in the sensory cortex had a relatively strong correlation with the infarct size on triphenyltetrazolium chloride staining (Figure IV in the online-only Data Supplement).
Hemorrhagic Transformation Occurred Near the Area With the Highest Level of Iopromide Parenchymal Staining in Mice With Relatively Severe Iopromide Leakage
mCT imaging did not predict which animals would undergo hemorrhagic transformation. Compared with the non-PH group, the PH group had wider and denser mCT sites of leakage only on the last in vivo images (Figures 2 and 3; Figure II in the online-only Data Supplement), which were acquired at 48 hours just before confirming the presence of hemorrhagic transformation ex vivo. In addition, infarct size on triphenyltetrazolium chloride staining was bigger in the PH group (mean±SE, 24.0±2.0 mm2) than in the non-PH group (14.7±1.4 mm2; P=0.01).
In the PH group animals, hematomas were located near the core area of infarct, in and around the sites where the mean mCT density at 48 hours was highest (in the middle+lateral portion of the caudatoputamen, compared with other subregions) or relatively high (in the sensory cortex of the PH group, compared with the non-PH group; see also Results in the online-only Data Supplement).
Hemispheric Midline Shift Was Progressively Worse From 4 Hours to 48 Hours and Was Greater in the PH Group Than in the Non-PH Group
As shown in the representative animals and quantitative data for all animals in each group (Figure V and Results in the online-only Data Supplement), midline shift significantly differed between time points (P<0.01). The overall increasing trend of the midline shift for 48 hours did not differ between the PH and non-PH groups (no significant interaction, P=0.30). However, after adjusting for time, the degree of midline shift was relatively prominent in the PH group (P=0.05). In addition, midline shift correlated with parenchymal staining near the midline but not with global iopromide leakage in the entire hemisphere.
Midline Shift and Iopromide Leakage Independently Predicted the Final Infarct Size
Multivariable analyses showed that midline shift and the amount of iopromide leakage at each of the 3 time points were independent predictors of the final infarct size at 48 hours (Results/Table in the online-only Data Supplement).
This is the first report on a high-resolution serial in vivo BBB imaging technique using mCT and a clinically available iodinated contrast agent administered intra-arterially in mice with tMCAO. Using the new BBB imaging technique, poststroke leakage of the x-ray contrast agent could be clearly detected with high resolution and in a sensitive manner, allowing quantitative characterization and serial monitoring of BBB dysfunction.
We have found that (1) it was necessary to use an intra-arterial, rather than intravenous, iopromide infusion paradigm, (2) staining showed a monophasic time course at 4 to 48 hours poststroke, (3) staining correlated well with leakage by an independently detected blood pool agent (Evans blue) and modestly with eventual infarct size, (4) contrast staining and midline shift (surrogate for edema) correlated well, and (5) contrast staining did not predict which animals would undergo hemorrhagic transformation, but sites of highest leakage seemed to correlate to the locations of hemorrhage, when it did occur.
Using this imaging technique, we could characterize the spatial and temporal dynamic nature of poststroke BBB dysfunction. BBB opening began in subcortical periventricular regions and expanded by spreading in and out to nearby basal ganglia and cerebral cortex. In our experimental setting, poststroke BBB opening was monophasic (increase-increase or increase-stay).8 However, poststroke BBB dysfunction has been reported to take a biphasic (increase-decrease-increase) course during the acute period.2 As one of many possible explanations (Discussion in the online-only Data Supplement) for these contradicting results, it may be considered that unlike our study (with limitations described in Discussion in the online-only Data Supplement) most of the previous studies used histological methods, which do not allow serial assessments of BBB permeability in identical animals.
Quantitative brain mapping and multivariable analysis showed that BBB dysfunction was linked with brain edema and final infarct size. Thus, mCT imaging could help predict the final infarct size by reflecting the severity of BBB dysfunction and brain edema. Despite the relatively big infarct size in the PH group and the (weak) correlation between infarct size and iopromide leakage, mCT imaging could not predict the risk of PH, but was helpful in predicting the location of postinfarction hemorrhagic transformation, when it occurred. In the PH group animals at 48 hours after tMCAO, midline-shifting edematous hemorrhagic transformation was observed in the infarct core, near the areas with high levels of iopromide leakage, such as in the lateral portion of the caudatoputamen or in the sensorimotor cortex.
In summary, we have developed a new mCT imaging technique for the serial assessment of poststroke BBB integrity in mice, which cannot be done using conventional (contrast or noncontrast) mCT imaging.
Sources of Funding
This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (2010-0019862) funded by the Korean government.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006297/-/DC1.
- Received May 29, 2014.
- Accepted June 3, 2014.
- © 2014 American Heart Association, Inc.
- Aoki T,
- Sumii T,
- Mori T,
- Wang X,
- Lo EH
- Kim DE,
- Kim JY,
- Nahrendorf M,
- Lee SK,
- Ryu JH,
- Kim K,
- et al