Article Text

Download PDFPDF

Steal physiology is spatially associated with cortical thinning
  1. Jorn Fierstra1,2,
  2. Julien Poublanc1,
  3. Jay Shou Han1,
  4. Frank Silver3,
  5. Michael Tymianski2,
  6. Adrian Phillip Crawley1,
  7. Joseph Arnold Fisher4,
  8. David John Mikulis1
  1. 1Department of Medical Imaging, University Health Network, Toronto Western Hospital, Toronto, Ontario, Canada
  2. 2Department of Neurosurgery, University Health Network, Toronto Western Hospital, Toronto, Ontario, Canada
  3. 3Department of Neurology, University Health Network, Toronto Western Hospital, Toronto, Ontario, Canada
  4. 4Department of Anesthesia and Pain Management, University Health Network, Toronto General Hospital, Toronto, Ontario, Canada
  1. Correspondence to Dr D J Mikulis, Medical Imaging, University Health Network, Toronto Western Hospital, Department of Medical Imaging, McLaughlin Pavilion, 3rd Floor Room 431, 399 Bathurst St, Toronto, M5T 2S8 ON, Canada; mikulis{at}


Background The physiological impact of severely impaired cerebral autoregulatory vascular reactivity on cortical integrity is unknown. The purpose of this study is to determine the relationship between severe impairment of autoregulatory flow control associated with steal phenomenon and its impact on cortical thickness.

Methods 250 blood oxygen level dependent (BOLD) MRI cerebrovascular reactivity (CVR) studies were reviewed in order to identify subjects with severe unilateral exhausted cerebrovascular reserve demonstrating steal physiology but with normal appearing cortex on fluid attenuated inversion recovery imaging. 17 patients meeting the inclusion criteria were identified. A reconstructed inflated cortical surface map was created for every subject using Freesurfer software ( The region of interest (ROI) reflecting the steal physiology was determined by overlaying the subject's CVR map on to the cortical surface map. This ROI was compared with the corresponding area in the healthy hemisphere which provided control cortical thickness measurement in each subject.

Results The hemisphere with steal physiology showed an 8% thinner cortex (2.23±0.28 mm) than the corresponding healthy hemisphere (2.42±0.23 mm) (p=0.0005).

Conclusions Our findings indicate that a spatial correspondence exists between impairment of autoregulatory capacity with steal physiology and cortical thinning.

  • Brain mapping
  • cerebral blood flow
  • cerebrovascular
  • cerebrovascular disease
  • neuroradiology

Statistics from


Investigation of cerebral blood flow has revealed three important adaptations of the flow control mechanism. The first is a pressure responsive mechanism that maintains constant blood flow in the microcirculation through a wide range of perfusion pressures.1 The second is a response to metabolic vasoactive molecules such as CO2, O2 and endothelin. The third is a vasoactive response to local neuronal activity.2 3 The final common pathway in these adaptive perfusion mechanisms is smooth muscle tone in the arterial system, predominantly at the level of pre-capillary arterioles.4

Patients with severe large artery stenosis or occlusion, associated with exhausted autoregulation and steal phenomenon, are at higher risk of ischaemic injury than those with intact autoregulation.5–7 This may be because lowering of blood pressure reduces perfusion distal to the stenosis. Also, in the case of vascular steal, increases in arterial Pco2 dilate normal vessels, resulting in a redistribution of blood flow away from the vascular impaired brain areas. Blood pressure variation and hypercarbia are common events in day to day living, particularly during sleep. These ‘normal’ decreases in blood pressure can be a potential risk in patients with vascular stenoses risking hypoperfusion in areas supplied by the stenotic vessel. We considered that repetitive transient non-lethal ischaemic events may, over time, lead to neuronal damage and loss.8–10

The purpose of this study was to determine the relationship between severe impairment of autoregulatory flow control sufficiently severe to cause steal phenomenon and cortical thickness, an indicator of neuronal density. We identified 17 subjects with steno-occlusive cerebrovascular disease (but normal appearing cortex on conventional MRI) and unilateral paradoxical cerebrovascular reactivity (CVR) to hypercarbia (steal phenomenon). We then compared the cortical thickness in the impaired area to the corresponding region in the opposite hemisphere in which autoregulation was preserved.



After approval from our institutional ethics review board, 250 consecutive blood oxygen level dependent (BOLD) MRI CVR studies, conducted between March 2003 and October 2008, were reviewed to identify subjects with areas of unilateral paradoxical CVR but with normal appearing cortex on fluid attenuated inversion recovery (FLAIR) imaging. ‘Steal physiology’ is defined as a reduction in BOLD signal during hypercarbia. The reduction in BOLD signal has previously been shown to represent blood flow reduction.11 Subjects with lacunar infarcts in the white matter were excluded (loss of axons). Subjects with small foci of T2 hyperintensity in the white matter were included in the study. Although these hyperintense foci represent ischaemic demyelination, the axons are still preserved, as is cortical integrity. Seventeen patients with a mean age of 42.3 years (range 12–75, 9 women) meeting the inclusion criteria were identified (table 1).

Table 1

Clinical characteristics of the subjects

MRI data acquisition

An axial three-dimensional T1 weighted (anatomical acquisition FSPGR) volume (voxel size 0.78 × 0.78 × 2.2 mm) was acquired for spatial co-registration of the BOLD signal variation associated with CO2 induced changes in blood flow using BOLD-EPI (echoplanar) acquisitions (EPI gradient echo with TR 2000, TE 25, 3.75 × 3.75 × 5 mm voxels on a 3.0 T HDX MRI system in 15 cases and EPI gradient echo with TR 2000, TE 40, 3.75 × 3.75 × 5 mm voxels on a 1.5 T HD MRI system in two cases (GE Healthcare, Milwaukee, Wisconsin, USA). During the BOLD signal acquisitions, CVR studies were performed using a change in end-tidal partial pressure of CO2 (Petco2) as the vasoactive stimulus. Subjects were fitted with an airtight sequential gas delivery mask12 (Hi-Ox-80; Viasys Healthcare, Yorba Linda, California, USA). A custom built, computer controlled gas blender (RespirAct; Thornhill Research Institute, Toronto, Canada) was programmed using the algorithms from Slessarev et al13 to deliver set concentrations of O2, N2 and CO2 to the mask in order to attain and clamp targeted levels of Petco2. Subjects underwent two prospective iso-oxic14 pseudo-square wave increases in Petco2 of 10mm Hg from a baseline of 40 mmHg.15 The first increase lasted 45 s followed by a return to baseline for 90 s, then a second increase for 130 s. All transitions were completed by the second breath. All plateaus were maintained within 1 mmHg.16

MRI data analysis

The acquired anatomical acquisition FSPGR volume was analysed for cortical thickness using Freesurfer software (, a method for automated surface reconstruction and accurate cortical thickness measurement.17 Briefly, software reconstruction of the brain creates an inflated three-dimensional brain surface image18 19 which facilitates interpretation of functional MRI data across the entire cortical surface without the interference from cortical folding.20

The MRI and Petco2 data were imported into, and analysed with, AFNI software.21 The BOLD-MR signal was regressed against the Petco2 on a voxel by voxel basis. The slope of the regression of the percentage change in MR signal intensity versus mmHg change in Petco2 is referred to as CVR. CVR was colour coded on a spectrum from grey to red for positive correlations and from grey to blue for negative correlations, and overlaid voxel by voxel on the anatomical scan to generate a CVR map (figure 1a). The CVR map was then overlaid on the inflated cortical surface map to determine a region of interest (ROI) encompassing the region of negative BOLD response (area showing steal physiology) within the affected hemisphere. The corresponding region on the hemisphere showing normal CVR was then identified (figure 2). Mean cortical thickness between the left and right hemisphere ROIs for each subject was calculated by Freesurfer software. Cortical thickness was then compared using a paired t test.

Figure 1

Axial image of blood oxygen level dependent (BOLD) MRI cerebrovascular reactivity (CVR) map and fluid attenuated inversion recovery (FLAIR) image of subject No 10. (a) The axial BOLD CVR map of subject No 10. The negative change in BOLD signal during hypercarbia, related to steal physiology, is delineated in blue. (b) Axial FLAIR image of the same subject is normal in appearance and has no signs of infarction or volume loss.

Figure 2

Relationship between the spatial extent of impaired cerebrovascular reactivity (CVR) and the spatial extent of cortical thinning in subject No 10. (a) Axial view of the same blood oxygen level dependent CVR map as seen in figure 1a, overlaid on the inflated cortical surface map in three dimensions. Unilateral steal physiology is delineated in blue. The region of interest (ROI) was drawn around the blue area (pink dotted line, left hemisphere). The light blue dotted line outlines the ROI in the corresponding area of hemisphere with normal CVR. (b) Cortical thickness map overlaid on the inflated cortical surface map in three dimensions, with the same ROI projected on top of it. Light blue designates the thinnest cortex and red the thickest. The scale is in millimetres.


Nine of the 17 subjects showed steal physiology in the left hemisphere. Mean cortical thickness was 2.23±0.28 (SD) mm in the ROI encompassing paradoxical CVR (figure 2a), and 2.42±0.23 mm in the corresponding contralateral side (p=0.0005) (table 2), a difference of 8%. One subject, with steal physiology in the left hemisphere, had a thinner cortex in the hemisphere showing normal CVR.

Table 2

Cortical thickness measurements


To our knowledge this is the first study relating reduced vascular reactivity to thinning of the cerebral cortex, an indication of neuronal loss.

The limit of smooth muscle relaxation response to a decrease in blood pressure is the point of exhausted autoregulation. This can occur in a specific vascular territory in a normotensive subject when a severe vascular stenosis or occlusion results in a drop in downstream perfusion pressure. In this scenario, steal physiology occurs primarily because the drop in flow resistance in the normally supplied tissue exceeds any additional drop in flow resistance in the tissue supplied by narrowed vessels. If these territories are connected by collateral vessels (through the circle of Willis or pial collateral vessels), then steal physiology becomes manifest.11

The normal appearing cortex on FLAIR imaging shows that there is still enough perfusion to maintain cellular neuronal function. However, repeated periods of hypoperfusion may not be sufficient for immediate cell death, but over time may lead to apoptosis or involution of cells, resulting in neuronal fallout. This concept of ‘selective neuronal loss’ is similar to that proposed by others.8–10 The exact nature of this tissue loss requires further investigation.

We carefully selected this cohort of subjects that represent the most severe end of the hypoperfusion spectrum to examine our hypothesis. Because cortical thinning is a gross measure of neuronal number,22 changes in cortical thickness would only be seen with severe neuronal loss. As cortical thickness may vary from person to person, we used the thickness of the cortex in the patient's own unaffected cortex for control values. The retrospective design of this study was necessary as unilateral steal physiology cannot be prospectively identified without a CVR study. The retrospective review method gave a large pool of patients (n=250) that provided a sufficiently large cohort with the specific inclusion criteria17 we required to test our hypothesis.

It is possible that at least part of the thinning of the cortex was due to reduced blood volume and not loss of cells. This however is unlikely as cerebral blood volume should be at a normal level or even elevated (due to compensatory vasodilation) in areas of reduced perfusion pressure with no signs of ischaemia.23 Another confounding mechanism may be the presence of cortical dehydration with loss in cell volume. This is unlikely as a limitation in blood flow is generally associated with decreased cellular energy resources leading to cell swelling. This therefore leaves loss of neuroglial tissue as the most likely explanation for the regional cortical thinning that we observed.

Interestingly, nine of our cases had left-sided steal physiology with eight showing a thinner left hemispheric cortex compared with the right hemisphere. This hemispheric difference is therefore more significant in that it occurred in the typically thicker left hemisphere.24 25 We note that handedness has no significant effect on cortical (hemispheric) asymmetry.26 However, one subject measured a slightly thicker cortex in the hemisphere with the paradoxical CVR, the left hemisphere in this case. Luders et al24 reported that the left hemisphere is thicker (left vs right 2.42 (0.14) vs 2.36 (0.13)); given these robust SD, the cortex on the left side could still have undergone thinning without the overall thickness being reduced below the normally thinner right hemisphere.

The interval from onset of symptoms to time of CVR study was 39.4 (SD±73.9; median 8) months. We attempted to draw a correlation with first report of symptoms and degree of cortical thinning in order to determine the annual rate of cortical thinning in the area of brain with steal phenomenon. The calculated rate was 0.023 mm per annum. However, the reliability of patient recall in determining symptom onset is questionable and the outcome was mainly supported by two patients (subject Nos 8 and 10; table 1). Further data are required for confirmation of this correlation. In addition, the association with symptom onset may not be coincident with the development of steal physiology. These issues preclude accurate determination of the magnitude of cortical thinning per unit time in the small sample of patients studied. Serial CVR studies may allow for more accurate quantification of this result and provide insight as to whether medical intervention can have an impact on cortical integrity. Future research should also explore the relationship between cortical thinning due to steal physiology and neurocognitive functioning.

Our findings indicate that a spatial correspondence exists between impairment of autoregulatory capacity with steal physiology and cortical thinning.


We thank Marat Slessarev and Alex Vesely for their contributions to the development of the breathing apparatus. We thank the Toronto Western Hospital MRI technologists, particularly Eugen Hlasny, David Johnstone and Keith Ta, for their contributions to data acquisition.



  • Linked articles 201459.

  • Funding JP and DJM are supported by the Ontario Research Fund Brain Consortium Grant (RE02-002).

  • Competing interests JAF and DJM contributed to the development of the Respiract. These authors stand to gain financially if the device is successfully commercialised by Thornhill Research Inc, a University of Toronto/University Health Network related company.

  • Ethics approval This study was conducted with the approval of the UHN Research Ethics Board Office, Toronto, Ontario.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Editorial commentary
    Randolph S Marshall