Sickle Cell Cerebrovascular Reactivity to a CO2 Stimulus Is Both Too Little and Too Slow

Background: By age 30, over 50% of sickle cell disease (SCD) patients have suffered a cerebral infarct. In response to anemia and the reduction in oxygen-carrying capacity, cerebral blood flow (CBF) increases to match metabolic demand. Increased velocity of CBF in major cerebral arteries is a strong risk factor for stroke in SCD children and adolescents. Despite generally increased CBF, silent cerebral infarcts (SCI) can still occur in patients receiving optimal transfusions. This suggests that the increased CBF does not meet metabolic demand and that vasodilatory response is compromised.

Hypothesis: In adult patients with SCD, cerebrovascular reactivity (slope of the vasodilatory response to CO₂ (CVR) and the steady-state CVR (amplitude) and speed of the vasodilatory response (tau) to a standardized vasodilatory stimulus CO₂), are reduced compared to normal subjects. We also explored for possible associations with clinical characteristics.

Methods: Functional brain imaging performed as part of routine care in adult (≥18) SCD patients (any phenotype) at the University Health Network Comprehensive Sickle Cell Center (Toronto, Canada) between 2017 and 2018 were reviewed. Patients with known cerebral vasculopathy were excluded. CVR was calculated as the change in CBF measured as the blood oxygenation level dependent (BOLD)-MRI signal, in response to a standard vasoactive stimulus of CO₂ (delivered by RespirAct^TM). To calculate the dynamic (tau) and steady-state CVR (amplitude) components of the BOLD signal response, the PET CO₂ waveform was convolved with an exponential decay function. The tau corresponding to the best fit between the convolved CO₂ and BOLD signal was defined as the speed of vascular response. The slope of the regression between the convolved CO₂ and BOLD signal was defined as amplitude.
CVR, amplitude and tau were normalized voxel-wise relative to the mean and standard deviation of the same metric in the corresponding voxels of a previously generated atlas of 42 healthy controls (Z scores). These Z scores were averaged over the vascular territories of the brain for both grey (GM) and white matter (WM). Fisher exact and Pearson correlations were performed to identify possible associations between CVR metrics and SCD comorbid conditions, laboratory parameters, and use of disease-modifying therapy. Associations with univariate P <0.20 were included in the multiple linear regression model. Multi-collinearity was assessed.

Results: Fifteen patients were included in the study. The median age was 27 [IQR22-35]. 5/15 (33.3%) were male. 9/15 (60%) were SS or S/b⁰ and 5/15 (33.3%) were SC. 4/15 (26.7%) were on transfusion. MRI/MRA uncovered Moya moya in 1 patient. SCI were present in 3/15 (21.4%). Compared to the reference atlas of normal subjects, CVR and amplitude were reduced both in GM and WM (mean Z-score for CVR -0.52 [-1.8 – 0.28] and -0.63 [-2.31 – 0.66]; amplitude -0.26 [-2.61 – 0.66] and -0.28 [-2.70 – 0.60] respectively). Tau was lengthened in GM and WM (mean tau Z-score +0.90 [-0.49 – 3.32] and +0.76 [-0.66 – 2.78] respectively). These abnormal metrics were observed with varying severities in all 3 main vascular territories (Figure). CVR decreased linearly with decreasing hematocrit (Hct) (r=0.59, p=0.03). There was also a trend towards lower CVR in SS or S/b⁰ patients (t=-1.41, p=0.18, d=0.76) and was highly collinear with Hct. Hematocrit was the only significant independent predictor of CVR metrics on multivariable regression.

Conclusions: All three measures of cerebrovascular health (CVR, amplitude and tau) in SCD patients were abnormal compared to normal controls. Hematocrit appears to be the strongest independent predictor of these measures. The protocol we applied for measuring CVR provides a standardized reproducible vasodilatory stimulus, enabling comparison against a population of healthy individuals for more accurate assessment of CVR in individual subjects. Furthermore, the stimulus protocol produces rapid changes in arterial CO₂ levels within one breath that can be used to measure the speed of response of the vasculature representing a novel metric of vascular performance postulated to represent vessel compliance and functional endothelial integrity. These findings show that CVR methodology represents a promising tool to assess disease state, stroke risk, and therapeutic efficacy in sickle cell patients and merits further investigation.