Regional resting state perfusion variability and delayed cerebrovascular uniform reactivity in subjects with chronic carotid artery stenosis

12nd Department of Radiology, Medical University of Gdansk, Gdańsk, Poland; 2Department of Economic Sciences, Faculty of Management and Economics, Gdansk University of Technology, Gdańsk, Poland; 3Department of Human Physiology, Medical University of Gdansk, Gdańsk, Poland; 4Department of Cardiovascular Surgery, Medical University of Gdansk, Gdańsk, Poland; 51-st Department of Radiology, Medical University of Gdansk, Gdańsk, Poland; 6Department of Neurosurgery, Medical University of Gdansk, Gdańsk, Poland; 7Department of Diagnostic Imaging, Medical University of Warsaw, Warsaw, Poland; 8Faculty of Health Sciences, Pomeranian University of Slupsk, Słupsk, Poland


Highlights
• Cerebral blood flow is low in resting state in frontal, parietal and temporal cortex grey matter in subjects with carotid artery stenosis • A particularly low cerebral blood flow is evident in the occipital cortex • High resting-state cerebral blood flow is observed in the insula and cerebellum • Acetazolamide challenge yields relatively uniform response in both hemispheres in the analysed population

INTRODUCTION
Brain stroke is currently the second most common cause of death.It has been estimated that one in ten European people over 50 years of age and one in two over 70 dies for this reason (Wintermark et al., 2002).Carotid artery surgery remains one of the therapeutic options for this patient population (MRC, 1991;NASCET, 1991).However, there is accumulating evidence that vascular stiffening, rather than carotid luminal narrowing, adversely influences cognitive aging and provides a potential target for ameliorating age-related cognitive decline (Wardlaw et al., 2016).It is becoming clear that further improvement in outcomes for patients with carotid stenosis requires a better understanding of the pathophysiology of this disease (Abbot, 2016).
The population suffering from chronic carotid artery stenosis that we investigated represents an interest-ing pathophysiological state.It is characterized by low global cerebral blood flow (CBF), low global cerebral blood volume (CBV) and lack of significant differences at baseline regarding mean transit time (MTT), CBF and CBV between the ipsilateral and contralateral hemisphere to stenosis (Frydrychowski et al., 2013;Szarmach et al., 2016;Szarmach et al., 2017a,b).Thus, the subjects from the described population are most likely very close to blood supply breakdown leading to stroke.Nevertheless, they still present relatively well-compensated perfusion, although at the very low end of the acceptable range.This population has already been described regarding misery perfusion and reduced cerebral metabolism.However, data related to both the regional pattern of baseline perfusion and regional cerebrovascular reactivity (CVR) in these subjects is missing.
The acetazolamide test is the procedure of choice to investigate cerebrovascular resistance (CVR) in subjects with chronic artery stenosis (Vorstrup et al., 1986;Murakami et al., 2002;Hartkamp et al., 2012).Acetazolamide is a reversible inhibitor of the enzyme carbonic anhydrase, and while the exact mechanism has yet to be unveiled, the vasodilatory effect on the brain vasculature is ascribed to carbonic acidosis (Vorstrup et al., 1984).Vasodilatory change in cerebral perfusion is usually measured 15 to 20 minutes before and after a challenge (Vorstrup et al., 1986;Hokari et al., 2008).However, the response to the acetazolamide test seems to be delayed in stenosed territories (Murakami et al., 2002;Hartkamp et al., 2012).One of our recent studies suggests that morphometric changes in the diameter of arteries originating from the circle of Willis are present even 60 minutes after the acetazolamide test (Szarmach et al., 2017a).Therefore a delayed response to vasodilatory challenge appears to provide important pathophysiological information.
Thus the aim of this study was to assess regional perfusion with the use of CTP at baseline and regional CVR in response to delayed acetazolamide challenge in subjects with chronic carotid artery stenosis.We hypothesized that in the described population there is marked heterogeneity in spatial baseline perfusion and that the delayed acetazolamide test may unveil significant differences in CVR characteristics.In particular, we expected higher CBF in cortical and deep GM and lower CBF in WM.We speculated that WM and structures fed by the vertebrobasilar circulatory system might be characterized by a relatively low CVR.

Patients.
A total of 16 symptomatic patients (ten males) aged 70.94±7.71fulfilling the enrolment criteria were included into the study (Table 1).This study was carried out in accordance with the recommendations of Helsinki.The experimental protocol and the study were approved by the ethics committee of Medical University of Gdansk.All volunteers gave written informed consent to participate in the study.
The following inclusion criteria were defined: 1. age > 18 years, 2. written consent to participate in the study, 3. carotid artery stenosis ≥ 90% on the ipsilateral side and ≤ 50% on the contralateral side.The degree of internal carotid artery stenosis was assessed using Doppler ultrasonography examination and digital subtractive angiography (DSA) according to the method documented in the North American Symptomatic Carotid Endarterectomy Trial (NASCET, 1991), 4. complete circle of Willis and normal vertebral arteries (confirmed by DSA).
Angiography.Digital subtractive angiography was carried out before carotid artery stenting to confirm the degree of stenosis.Selective angiography of the target vessel was then performed to reveal the size of the carotid lesion, degree of stenosis, morphology of the internal carotid artery and its contribution to the circle of Willis.
Doppler ultrasound.Carotid arteries were examined with the patient in the supine position.The posterolateral transducer positions were used to examine the carotid arteries in the long and short axis.All the examinations were performed by the same operator and the same ultrasonography machine (Logiq S8, GE Healthcare) with a Doppler angle of 60°.On grey scale presence or absence and location of plaque were evaluated.On Doppler study peak systolic velocity (PSV) and end diastolic velocity (EDV) of common carotid artery (CCA), internal carotid artery (ICA) and external carotid artery (ECA) were evaluated.NASCET criteria were used for measuring the percentage of stenosis.
Computed tomography protocol.The imaging protocol consisted of Non-Contrast Computed Tomography (NCCT) and dynamic perfusion CT (CTP).In all subjects, CTP examinations were carried out twice: before and 60 minutes after an injection of 1.0 g Diamox i.v.(Mercury Pharmaceuticals, London, UK).All examinations were performed in transverse plane using a 64-MDct Light Speed VCT XT scanner (GE Healthcare Technologies, Chicago, Illinois, US).
Technical procedures.All subjects underwent nonenhanced plain brain CT scans (140 kV, 335 mAS, 0.9 second rotation time, number of images=56, total exposure time=6.3s, and CDTIvol between 50 and 60 mGy) to exclude intracranial haemorrhage and distinguish between acute and old ischaemic stroke.
CPT studies were conducted on 40 mm thick transverse axial sections.A 40 mL bolus of non-ionic contrast media (Optiray 350 Mallinckrodt, St. Louis, Missouri, USA) was administered into an antecubital vein using a power injector (Dual Shot Alpha 7, Nemoto Kyorindo Co., Ltd, Japan) with an injection rate of 4 mL/s.After contrast injection, a bolus of 40 ml of physiological saline was administered.CT scanning was performed 5 seconds after the start of the injection with the following acquisition parameters: a sine mode series, 80 kV, 150 mAs, 5 mm slice thickness, 25-cm field-of-view, rotation time=1/s, number of images per rotation=8, cine time between im-ages=0.5 s, image matrix=512x512, time interval between reconstructed images=0.5s, interval=0 mm.
A total of 360 slices was obtained with a total scan time of 45 seconds.CDTIvol was approximately 390 mGy per examination.Scanning was performed at the level of the upper part of the cerebellum (over a distance of 40 mm), with the gantry angle set parallel to the orbitomeatal line to avoid direct exposure of the eye lens to radiation.For the second study in each patient, the same general slice locations and identical technical parameters were used.
The assessment of perfusion parameters was carried out from the level of the upper part of cerebellum to the middle part of cerebral hemispheres included in the examination, in the area supplied by the middle, anterior and posterior cerebral artery.
Image post-processing.Post-processing was performed by two investigators twice in each case (before and after acetazolamide administration) using a dedicated diagnostic workstation (AW 4 GE Healthcare Technologies, Wisconsin, USA) equipped with a professional post-processing software package to generate colour overlay maps of dynamic cerebral enhancement data (CT Perfusion version 4 (v 4.3.1),GE Healthcare Technologies, Wisconsin, USA).The readers were blinded to clinical and radiological information.
The standardized elliptical mirrored regions of interest (ROIs) were drawn manually.Each ROI (between 1 cm 2 to 4 cm 2 each) was determined at all analysed levels (Figs 1, 2 and SF1) in the region of the investigated structures.The large vessels, bones and cerebrospinal fluid were automatically excluded via brain perfusion software.Native scans are provided as supplementary Fig. SF1 at www.actabp.pl.
The "neuro brain stroke" mode was used for calculation of the perfusion metrics of Mean Transit Time (MTT), Cerebral Blood Flow (CBF), Cerebral Blood Volume (CBV), and Time to Teak (TTP) for brain stroke assessment using the maximal slope method.
Time-enhancement curves based on the passage of contrast material through the anterior cerebral artery and the superior sagittal sinus yielded the atrial input function (AIF) and the venous output function (VOF), respectively.
The MTT corresponds to the average time that red blood cells spend within a determinate volume of capillary MTT, computed and displayed in seconds.
The CBF is defined as the volume of blood passing through a given amount of brain tissue per unit of time, most commonly milliliters of blood per minute per 100 g of brain tissue (ml/100 g/min).
The CBV is calculated as the ratio of the area under the time-concentration curve for the first contrast material bolus passage through the tissue to the area under the curve for the VOF.The blood volume is the product of the blood flow and the mean transit time: CB-V=CBF*MTT.
The TTP is the time between the onset of the enhancement transient and the peak value of the time curve.Time-to-peak is computed and displayed in seconds, using the raw time curve data.
The absolute values of CT perfusion parameters (MTT, TTP, CBF, CBV) of one hemisphere in the region of anterior, middle and posterior cerebral artery distribution and contralateral mirroring areas in functional maps were measured.
Statistical analysis.To compare the results before and after acetazolamide challenge, differences in absolute CT values (MTT (s), CBV (ml/100 g), CBF (ml/100 g/min) and TTP (s)) and relative values (rMTT, rCBV, rCBF and rTTP -obtained as the ratio of appropriate values from ipsilateral side to contralateral side to stenosis) were analysed.
Raw data from all individual slices were presented as Mean ± S.D and with Median value.Additionally, there were calculated the percentage changes of CT parameters by formula (CT2-CT1)/CT1×100% where CT1 and CT2 meant respectively CT values in the period before and after delayed acetazolamide test.Since the analysed groups were small (N=16) the differences between variables were examined at the same time by paired t-test and alternative nonparametric Wilcoxon's test.The t-test was stronger as it was the parametric test but in this case needed to be satisfied by the normality of paired differences.The W Shapiro-Wilk test revealed that not all of the samples fulfilled this assumption (results not presented).Therefore the best approach was to take into the consideration both tests results.
The level of significance was set at α=0.05.All calculated p-values were for two-tailed tests.All raw data were analysed using the statistical software Statistica 12.5 (StatSoft, Tulsa, OK, USA).

RESULTS
Resting state CBF (ml/100 g/min), CBV (ml/100 g), MTT (s) and TTP (s) parameters and respective delayed changes in response to acetazolamide are presented in Figs 3-10.Detailed tabular results are provided in the supplementary section of the manuscript (ST1-ST9).All responses to acetazolamide test were statistically significant.There were no statistically significant differences between hemispheres.
Summing up, resting state CBF in subjects with carotid artery stenosis was low in all analysed structures with the exception of insula, thalamus and cerebellum.All values were provided in millilitres of blood per minute per 100 g of brain tissue (ml/100 g/min).
TTP responses to the acetazolamide test were similar across the various brain regions.

DISCUSSION
To the best of our knowledge, this is the first study describing regional cerebral perfusion pattern in subjects with chronic carotid artery stenosis.There are several new findings to report.Resting state CBF was low in frontal, parietal, temporal and particularly in occipital cortex grey matter in subjects with carotid artery stenosis.The CBF observed in insula and cerebellum was relatively high.The relatively uniform response to acetazolamide challenge seems to confirm well-developed collateral blood flow in subjects with chronic carotid artery stenosis.As expected grey matter was more reactive to acetazolamide challenge than white matter or brainstem.

Resting state
Our study demonstrates substantial spatial nonuniformity in all CTP parameters in the resting state, both cortically and subcortically, independent of the hemisphere (ipsi-or contra-lateral to stenosis).In this respect, our study is consistent with an earlier arterial-spin labeling (ASL) report in healthy subjects of various ages (Chen et al., 2011).However, that examination was limited to cortical and subcortical regions, which is why we also assessed white matter, brainstem, and cerebellum.
Radiation exposure is unavoidable in CTP and represents one of the main disadvantages of this imaging technique.For this reason, the inclusion of a group of healthy control subjects is ethically impossible.Nevertheless, CTP has been shown to correlate well with positron emission tomography (PET) data, the gold standard for brain per- fusion measurement (Kudo et al., 2003).Also, other radiological neuroimaging techniques such as ASL (Kamano et al., 2013;Zhang et al., 2014) and single photon emission computed tomography (SPECT) (Matsumoto et al., 2013) have been validated against PET.Therefore, some limited parallel comparisons can be made with caution.
The CBF in cortex grey matter (frontal, parietal and temporal) in our population was relatively low, particularly on the ipsilateral side.In Chen and coworkers (2011) ASL study in healthy subjects the mean CBF value across the entire cortical grey matter volume was 52.6±9.3,52.0±10.7,and 42.7±8.8ml/100 g/min in the young (age<40 years), middle-aged (40≤age≤60 years) and older adult groups (age≥60 years), respectively.Similar results in healthy volunteers were obtained by Pantano and coworkers (1984) in PET study.The MTT in the frontal, parietal and temporal cortices grey matter was within the normal range (3-5 s), suggesting relatively balanced circulation and perfusion pressure (Schumann et al., 1988;Murphy et al., 2011).The moderate CBF decline in subjects with chronic carotid stenosis is in line with other reports by our team and others (Szarmach et al., 2017a;Szarmach et al., 2017b;Niesen et al., 2004;Trojanowska et al., 2006).
The very low CBF in the occipital cortex was quite striking.The CBF in the occipital cortex can be altered even in cognitively intact and nondemented subjects in the seventh decade of age (Trojanowska et al., 2006).However, the hypometabolic occipital cortex is suggested to be prognostic for the development of mild cognitive impairment (Pikkarainen et al., 2009).Ashraf and coworkers (2015) assessed glucose metabolism with PET.Although PET of tissue metabolic activity based on regional glucose uptake does not directly measure CBF, glucose metabolism and CBF are closely correlated and often show similar results (Baron et al., 1982).Also, an MTT in occipital cortex grey matter above 5 s may indicate impaired circulation haemodynamics (Schumann et al., 1988;Murphy et al., 2011) in subjects with chronic carotid artery stenosis.
The role of the cerebellum in non-motor functions is increasingly recognized.Its involvement in cognitive functions is supported by the clinical finding that lesions confined to the cerebellum produce cerebellar cognitive affective syndrome (reviewed by Tedesco et al., 2011).The cerebellum contains more than half of all the neurons in the brain, and exhibits strong interconnections with the contralateral cerebral hemisphere both in feedforward and feedback directions (Tedesco et al., 2011).However, it remains to be elucidated how changes in cerebral cortex grey matter may affect the cerebellum.Due to the parallel characteristics of cerebello-cerebral connections (Van Overwalle et al., 2017) it is tempting to speculate that the high cerebellar CBF observed in our study is compensatory in nature.In particular, the MTT in the cerebellar cortex remained low, indicating well-balanced circulation.
Changes in biochemical markers suggestive of underlying inflammation or glial activation have been demonstrated in the insula and thalamus in subjects with obstructive sleep apnoea, a disease characterized by intermittent episodes of nocturnal hypoxia.Importantly, the greatest biochemical alterations accompanied the lowest oxygen saturation (Yadav et al., 2014;Sarma et al., 2016).It is very likely that such alterations are also present in subjects with chronic carotid artery stenosis due to misery perfusion, prolonged hypoxia and oxidative stress (Szarmach et al., 2017b).Therefore the high CBF present in our patient population might be due to inflammatory changes in these structures.Alternatively, both the insula and thalamus are involved in sympathetic brain responses triggered by oxidative stress and the activated renin-angiotensin-aldos-terone system (Winklewski et al., 2015;Winklewski et al., 2016).Sympathetic activation may add functional component that further explains augmented CBF.The low MTT suggests well-balanced circulation in these structures.
The brainstem represents an evolutionarily old structure that is responsible for the maintenance and integration of vital homeostatic and life support functions (i.e., cardiovascular, respiratory, etc., Noble et al., 1984).In the extreme conditions of complete anoxia in cats, the survival time of cerebral structures appeared to be related to the CBF, with older structures surviving 25-40 s and the cerebral cortex surviving only 14-15 s (Sugar & Gerard, 1938).Therefore, in our study, the preserved CBF in phylogenetically old brain regions such as the brainstem, cerebellum, thalamus, and insula may be explained by their high reactivity to persistent hypoxia that is likely present in the described subject population.

Delayed acetazolamide test
Delayed acetazolamide challenge led to a relatively uniform response on the ipsi-and contra-lateral sides to the stenosis.It is likely that all patients in the studied population had well-developed collateral flow as the duration of carotid artery stenosis exceeded 5 years (Barnett et al., 1997;Waaijer et al., 2007).The highest difference between stenotic and non-stenotic territories was visible in MTT response in cerebral cortex grey matter.A higher cortex MTT response on the contralateral side might be due to delayed reactivity on this side, and/or wearing off of the acetazolamide effect on the ipsilateral cortex (Murakami et al., 2002;Hartkamp et al., 2012).
As expected the highest CVR was recorded in cortex grey matter, while the lowest was in the white matter and brainstem.Significant differences in CO 2 reactivity between the internal carotid artery and vertebrobasilar circulation have been proposed in several physiological Doppler studies.According to physiologists lower CO 2 reactivity in the vertebrobasilar system may reflect conservative and relatively stable blood flow to the medulla oblongata to maintain vital systemic functions, while high CO 2 reactivity in the internal carotid artery system ensures proper tolerance to varied blood flow demand in the cerebral cortex (Sato et al., 2012;Skow et al., 2013).Our study confirms that structures fed by the vertebrobasilar system (brain stem, occipital cortex and cerebellum) have a relatively low CVR.Surprisingly, a relatively low CVR was also observed in the insula and thalamus.This suggests that the compensatory reserve of these structures has already been partially exhausted.Such a conclusion is supported by Ito and coworkers (2002), who reported that subcortical structures like the thalamus and putamen received relatively large increases in CBF during hypercapnia in healthy subjects.
Low CO 2 reactivity in white matter is typically explained by its relatively lower degree of vascularization compared to grey matter (Mandell et al., 2008;Bhogal et al., 2014).Mandell and coworkers (2008) demonstrated a CBF "stealing effect" in white matter in response to rapid CO 2 partial tension increases.Interestingly, a selective reduction in CBF to white matter during mild hypercapnia in young, healthy human subjects occurs precisely in those locations where elderly patients develop leukoaraiosis (Mandell et al., 2008).Our results collected one hour after acetazolamide challenge support the concept that white matter CVR is low compared to other structures.Consequently, its ability to respond to variations in CO 2 supply might be deteriorated in older subjects.
Delayed acetazolamide test might be better suited for elderly subjects with several cardiovascular diseases and most likely diminished CVR.To the best of our knowledge our team is the first to show that the acetazolamide effect persists for at least 60 minutes in subjects with chronic carotid artery stenosis (Szarmach et al., 2017a).There is accumulating evidence that CVR should be seen as a process of temporally and spatially defined dynamics rather than a single value based on measurement at the point of the highest response (Bhogal et al., 2014;Bhogal et al., 2015;Bhogal et al., 2016).It has been demonstrated that various white (WM) and grey matter (GM) structures in healthy subjects show substantial differences in CVR to increasing concentrations of breathing carbon dioxide (CO 2 , Bhogal et al., 2014).This pattern of CVR complexity in time and space is further complicated by changes observed in healthy aging (Bhogal et al., 2015;Bhogal et al., 2016;Nöth et al., 2008).Future studies in subjects with chronic carotid artery stenosis may also benefit from dynamic CVR assessment.
Computed tomography imaging and especially computed tomography perfusion (CTP) is still more commonly used than magnetic resonance imaging (MRI) because CTP has many advantages over MRI.CTP has better spatial resolution than MRI perfusion with significantly shorter study time.In this project acquisition time was less than one minute.Campbell et al. (2012) reported that quantitative computer tomography perfusion mismatch classification using two main parameters is similar to a diffusionperfusion mismatch in MRI.Imaging by MRI is very tempting because it does not use ionizing radiation.But magnetic resonance imaging has many contraindications (e.g., pacemakers, neurostimulators, cochlear implants, and claustrophobia), the study takes a long time, and patients have to remain still while scanning.Also, emergency departments are often not equipped with an MRI scanner.These key arguments favour the use of CTP.
Quantitative CTP data is highly dependent on the postprocessing software and the investigators.Software differences are frequently considered the main cause of variability in perfusion results relative to interoperator and intraoperator differences (Kudo et al., 2010;Kamalian et al., 2011).To avoid technical mistakes we used equipment manufactured by one producer and used the same post-processing procedure for all subjects, increasing the reliability of the study.Furthermore, both experienced neuroradiologists independently drew two standardized elliptical mirrored ROIs in the same patients twice (before and after acetazolamide administration; Fig 1 and 2, SF1).Finally, in all subjects, we used a coverage size of 40 mm, which is well suited for detecting perfusion parameters owing to the high density of the scans (Szarmach et al., 2016).
The number of studied subjects involved in the study is relatively small and heterogeneous in terms of medical history, including previous history of stroke, carotid artery endarterectomy and stenting.The study population is nevertheless representative for a group of patients with chronic carotid artery stenosis (Szarmach et al., Szarmach et al., 2016;Szarmach et al., 2017a;Szarmach et al., 2017b).Moreover, the study group is very homogeneous with respect to low CBV and CBF values due to strict inclusion criteria (carotid stenosis >90% on the ipsilateral side, less than 50% on the contralateral side, complete circle of Willis and normal vertebral arteries).
The study raises several important questions from the clinical perspective that should be answered in future research.Is the low resting state CBF in occipital grey cortex related to cognitive impairment in the studied population?Is the high resting state CBF in the cerebellum compensatory in nature?Are the observed results in the insula and thalamus compensatory or inflammato-ry in nature?Finally, how do the described changes affect functional outcome and prognosis?Future research may include combined ASL, BOLD-MRI, and dynamic contrast-enhanced (DCE)-MRI studies to address these questions.Hypoxia and inflammation are increasingly recognized as the pathophysiological background in several neurological and psychiatric diseases.Consequently, in-depth studies in larger groups of subjects suffering from prolonged hypoxia and the chronic inflammatory state may provide benefits to several fields in medicine.

CONCLUSIONS
To conclude, we have shown a relatively low resting state CBF in frontal, parietal and temporal cortex grey matter in subjects with carotid artery stenosis.A particularly low CBF was evident in the occipital cortex.The MTT in cortex grey matter tentatively suggests a balanced circulation.The high resting state CBF observed in the insula and cerebellum might be compensatory or inflammatory in nature.The relatively uniform response to acetazolamide challenge seems to confirm well-developed collateral blood flow in subjects with chronic carotid artery stenosis.Grey matter is more reactive to acetazolamide challenge than white matter or brainstem in the studied population.

Figure 2 .
Figure 2. Post-acetazolamide cerebral blood flow (CBF) images in patient with severe right internal carotid artery (RICA) stenosis.After acetazolamide administration, there was a normal response (increase in CBF) to acetazolamide on the left side.The right hemisphere demonstrated "poorer" increase of CBF due to chronic impairment of cerebrovascular reactivity.Brainstem and cerebellum (a), occipital and temporal lobes (b), insula (c), thalamus (d), frontal lobe (e), parietal lobe and white matter of brain hemisphere (f).

Figure 3 .
Figure 3. Resting state CBF and delayed CBF response to acetazolamide test in selected brain regions.

Figure 5 .
Figure 5. Resting state CBV and delayed response to acetazolamide test in selected brain regions.

Figure 7 .
Figure 7. Resting state MTT and delayed response to acetazolamide test in selected brain regions.

Figure 9 .
Figure 9. Resting state TTP and delayed response to acetazolamide test in selected brain regions.