Blood Oxygenation Level Dependent (BOLD) imaging in combination with vasoactive stimuli can be used to probe cerebrovascular reactivity (CVR)[1]. Characterizing healthy, age-related changes in the BOLD-CVR response can provide a reference point from which to distinguish abnormal CVR from the otherwise normal effects of ageing. In this study, we examine age-dependent differences in grey and white matter (GM/WM) BOLD-CVR response to progressive hypercapnia between young and elderly subjects. Furthermore, we incorporate differences in baseline T2* information to broaden our interpretation of the BOLD-CVR response.

This study was approved by the Medical Ethics Committee of our institution and informed consent was obtained from all subjects. Dual-echo pseudo-continuous ASL (pCASL) data (flip angle: 90, reconstructed resolution: 3 x 3mm², slice thickness: 7mm, slice gap: 1mm, TR: 4 s, TE1/TE2: 13.84/36.28 ms, FOV: 240 x 240 x 87 mm3, slices: 11, volumes: 140, label duration: 1650 ms, post-label delay: 1550 – 2185 ms ) was acquired on a Philips 3T system. Scans were performed in 16 young (28 ±3 years) and 30 elderly subjects ( 66 ±4 years). The computer controlled breathing challenge (RespirAct, Thornhill Research, Toronto) was as follows: (1) 2min baseline; (2) 75s hypercapnic ramp; (3) 100s plateau; (4) 2min baseline [2]. Since the interest of this work lies in the BOLD signal/T2* response to progressive hypercapnia, the ramp data was isolated for data analysis T2* maps were generated via mono-exponential fitting of the multi-echo data. Motion correction was performed (FSL-MCFLIRT) on the first echo data and the resulting transformations were applied to second echo and T2* data. GM/WM masks were created as outlined in Fig 1. BOLD data was computed using a running averaging which combined label/control dual-echo data. BOLD data was detrended using a linear fit to the pre/post-stimulusbaseline signal. PetCO₂ traces were temporally aligned with BOLD data (normalized to baseline BOLD signal) and the resulting BOLD-CVR curves were fit to the following sigmoidal model [3]:

$$ \%\Delta BOLD= Initial\ Signal\ Amplitude + \frac{Sigmoid\ Span}{1+e^{\frac{-(PetCO_2-Sigmoid\ Midpoint)}{Linear\ Portion\ About\ Midpoint}}} $$

Inter-subject normalization was done by calculating the mean response curve as a function of PetCO₂ increases from baseline (fig 2B/C) for the young and elderly groups respectively [4]. Sigmoidal fitting and alignment of T2* timeseries data was performed in a similar manner .

The range of ΔPetCO₂ values surveyed were 6.9±2.2 mmHg and 6.1±1.3 mmHg for young and elderly, respectively (no significant difference; p = 0.13). The GM and WM BOLD-CVR response began displaying non-linear above approximately 4mmHg PetCO₂ from baseline (fig 3B). The initial slope of the GM response appeared steeper in young subjects (fig 3B), while the point at which GM CVR was most sensitive to changes in PetCO₂ was shifted to higher PetCO₂ values (from baseline) in elderly subjects (fig 4B). Overall GM CVR was reduced amongst elderly (0.19±0.06 %ΔBOLD/mmHg) as compared with young subjects (0.26±0.07 %ΔBOLD/mmHg) (p<0.002), while that in WM showed no significant differences (0.04±0.02 and 0.05±0.03 %ΔBOLD/mmHg for young and old, respectively (p = 0.12)). GM/WM baseline T2* values were 48.0±2.2/49.7±3.6 ms and 50.9±3.0/ 54.6±3.2 ms (GM/WM significantly different: p<0.002/p < 0.0001) for young and elderly subjects, respectively. The difference between elderly GM and WM T2* values did reach statistical significance (p< 0.0001), while those between young subjects did not (p = 0.12). Absolute T2* under hypercapnia increased in both subject groups (Fig 3A).Our main findings were the amplitude of the BOLD-CVR response to progressive hypercapnia was reduced in elderly subjects. More novel, was the observation that the shape of the GM BOLD-CVR response curve showed age dependent changes relating to PetCO₂ sensitivity, while that in WM did not (fig 3). Finally, baseline T2* values were higher in elderly subjects, suggesting increased partial volume effects with CSF, higher venous oxygen saturation or B₀ related susceptibility effects. Elevated baseline T2* translates into a brighter baseline BOLD signal that may lead to an apparent reduction in the BOLD-CVR response upon normalization [5]. Baseline T2* can be modulated by increased partial voluming between tissue and CSF resulting from atrophy of the cortex and enlargement of peri-vascular spaces. Inhomogeneous tissue loss and variable decreases in regional microvascular density may further explain the differences seen in the age-dependent GM/WM CVR response. Finally, morphological changes in the composition of control vessels may restrict the sensitivity of the vasculature to changes in PetCO₂. These include thickening of the basement layer, reductions in smooth muscle content and increases in fibrous tissue within arterial walls [6]. The combined effect of these changes is stiffening of the vasculature, which will modulate the CVR response.