Cerebrovascular reactivity (CVR) indicates the capacity of the cerebrovascular autoregulation to maintain cerebral blood flow (CBF) and blood O2 partial pressure (PaO2) during a vasoactive stimulus, like the increase of carbon dioxide (CO2) via breath-holding (BH). For the first time during neurosurgery, we determinate intraoperative CVR with Blood Oxygen-Level Dependent (BOLD) fMRI measurements with three cycles of apnea (mimicking BH) in mechanically ventilated neurosurgical patients. Intraoperative end tidal PETCO2 monitoring would have needed extra equipment limiting comfort and intraoperative feasibility.
BH is clinically straightforward, but remains challenging to interpret as ongoing oxidative metabolism and restricted respiratory outwash result in accumulation of CO2 and only a gradual and slow arterial PaCO2 increase. Therefore, the boxcar function regressor needs to be adjusted to the progressive response of the BOLD signal during BH, and the response delay needs to be adjusted voxel-wise. Finally, an “optimal” physiological-based BH regressor in accordance with patient specific oxidative metabolism is presented.BOLD fMRI datasets of five neurovascular patients with unilateral hemispheric hemodynamic impairment were processed with various BH CVR analysis methods. Temporal lag (Phase), percent BOLD signal change (CVR) and explained variance (Coherence) maps were calculated using three different Sine models and two novel “Optimal Signal” model-free methods. We compared a whole brain average lag between CO2 regressor and BOLD signal (Global-delay Sine) and a voxel-wise lag (Voxel-wise delay Sine) in patients with unilateral hemodynamic impairment. Secondly, we parametrize the Sine model for different BH and resting periods durations. Finally, subject’s physiology-specific responses to BH were modeled by the BOLD time series of the sagittal sinus – the brain’s major venous outflow. We hypothesized that it should describe the most robust patient specific cerebral positive BOLD response to BH and be least influenced by potential negative CVR. This “Sagittal Sinus” model was compared to the “Unaffected” Hemisphere model.All models showed significant differences for CVR and Coherence between the affected -- hemodynamic impaired -- and unaffected hemisphere. Voxel wise Phase determination significantly increases whole-brain CVR (0.60±0.18 versus 0.84±0.27; p<0.05). Incorporating different durations of breath hold and resting period in one Sine model (two-task) did increase Coherence in the unaffected hemisphere, as well as eliminating (physiologically impossible) negative Phase commonly obtained by one-task frequency models.
Both novel model-free “Optimal Signal” methods based on the unaffected hemisphere and the sagittal sinus fMRI signal time series, respectively, explained the BOLD MR data similar to the two task Sine model.Our CVR analysis demonstrates an improved CVR and Coherence after implementation of voxel-wise Phase and frequency adjustment. The novel “Optimal Signal” methods provide a robust and feasible alternative to the Sine models, since both are model-free and independent of compliance. Here, the sagittal sinus model may be advantageous, since it is independent of hemispheric CVR impairment.
CVR impairment affect in reality both hemispheres, the combination of positive and negative time series even within one hemisphere neutralizes the mean time series of the hemisphere declared as “unaffected”, and would results in erroneous CVR interpretation.Our analysis methods make the intraoperative determination of CVR possible, and increase sensitivity and reproducibility of BH derived BOLD fMRI CVR.