Investigating the non-linearity of the BOLD cerebrovascular reactivity response to targeted hypo/hypercapnia at 7 T

Cerebrovascular reactivity (CVR) is a mechanism responsible for maintaining stable perfusion pressure within the brain via smooth muscle mediated modulations of vascular tone. The amplitude of cerebral blood flow (CBF) change in response to a stimulus has been evaluated using Blood Oxygen Level Dependent (BOLD) MRI, however the relationship between the stimulus and the measured signal remains unclear. CVR measured invasively in animal models and using blood-velocity based measurements in humans has demonstrated a sigmoidal relationship between cerebral blood flow and CO2 partial pressure. Using an ultra-high magnetic field strength (7T) MRI scanner and a computer controlled gas delivery system, we examined the regional and voxel-wise CVR response in relation to a targeted progressively increasing hypo- to hypercapnic stimulus. The aim of this study was to assess the non-linearity/sigmoidal behavior of the CVR response at varying arterial CO2 (PaCO2) levels. We find that a sigmoidal model provides a better description of the BOLD signal response to increasing PaCO2 than a linear model. A distinct whole-brain and gray matter BOLD-CVR signal plateau was observed in both voxel-wise and regional analysis. Furthermore, we demonstrate that a progressively increasing stimulus in combination with a sigmoidal response model can be used to obtain CVR values and provides additional physiologically relevant information (such as linear and non-linear response domains, and maximum response amplitudes) that may be more difficult to obtain from blocked CVR experiments. Considering these results, we propose an alternative way in which to define CVR based on the derivative of the BOLD-CVR response curve, which can potentially be used to differentiate between healthy and diseased vascular states.