Investigate relations between hypercapnic and hyperoxic BOLD responses and slow baseline BOLD signal fluctuations.Slow fluctuating (< 0.1 Hz) BOLD signals during baseline conditions or ‘resting-state’ have seen interest in numerous studies, both in healthy and disease. Whereas temporal coherence of these fluctuations between regions have been linked to neuronal functional networks¹, the amplitude of these fluctuations have been used in normalization procedures of BOLD responses^2,3 (aiming at reducing inter-subject variability) as proxy for vascularization (CO2 vessel reactivity), but also a proxy for neuronal activity in studies on age effects and disease^4,5,6,7. The aim of this study was to shed light on the origins of these slow fluctuating BOLD signals in terms of CO2 vessel reactivity and baseline cerebral blood volume contributions. Healthy volunteers (n=45; age range=25 – 76 yrs) were scanned at 3T (Philips) with no previous history of cerebrovascular disease, other brain disease or visible structural lesions on MR imaging. These subjects were also included in a previous study⁸. BOLD measurements were obtained using dual-echo pCASL MRI using the following acquisition parameters: multi-slice single-shot EPI, SENSE=2, TR/TE₁/TE₂=4000/13.8/36.3ms, voxel size=3x3x7mm³, slice gap=1mm, FOV=240x240x87mm³, label duration=1650ms, postlabel delay=1550ms, volumes=135, scan time=18min40s. The breathing paradigm consisted of two normoxic hypercapnic periods in which end-tidal CO₂ (EtCO₂) was targeted to progressively increase by approximately 10mmHg above resting baseline EtCO₂ over a period of 75 seconds. The hypercapnic periods were interleaved with a single normocapnic hyperoxic period during which EtO2 was targeted at 300mmHg (See Figure 1). Gasses were administered using a computer-controlled rebreathing approach (RespirAct^TM, Thornhill Research Inc.). BOLD time series (computed by optimally combining the two echoes) were motion corrected (FLIRT), linearly detrended (AFNI), and spatially smoothed with a 5mm FWHM Gaussian kernel. Amplitude maps of slow fluctuating BOLD signals were computed by estimating the mean square-root of the power in the 0.01 – 0.08 Hz frequency band for each voxel during the baseline period (first 200s). Amplitude maps of slow fluctuating BOLD were subsequently normalized by the mean signal averaged over the baseline period and all voxels in the brain mask. Maps of hypercapnic and hyperoxic BOLD signal changes (%) were computed using the EtCO₂ and EtO₂ traces as regressors in a general linear model (Feat, FSL). Hypercapnic %BOLD data were normalized by each subject’s pEtCO₂ (yielding %BOLD/mmHg). The hyperoxic BOLD signal changes were used for obtaining baseline venous cerebral blood volume (vCBV) information. Finally, all maps were registered to common MNI space for comparison purposes in cortical gray matter areas.Maps of BOLD (%) hypercapnic cerebral reactivity (HC-CVR), hyperoxic BOLD reactivity (≈baseline cerebral venous blood volume, HO-CVR) and amplitude of baseline slow fluctuating BOLD signals (SFBS) are shown in Figure 1 (averaged over all subjects). Corresponding density scatter plots maps (cortical gray matter mask in MNI space) shows a high correlation between all three measures; ^R2 = 0.70 (HC-CVR vs. SFBS) , R² = 0.51 (HO-CVR vs. SFBS), R² = 0.56 (HC-CVR vs. HO-CVR). A substantial amount of the information imbedded in the slow fluctuating BOLD signal (SFBS) during baseline conditions is heavily weighted by the baseline cerebral venous blood volume (vCBV). Taking the hyperoxic %BOLD signal changes as a proxy for baseline vCBV we observed that 51% of the spatial variation in SFBF can explained by baseline vCBV– we hypothesize this is largely due to the distribution of large vessels. When normalizing the SFBS by total amplitude across all frequencies (fractional amplitude of low-frequency fluctuation (ALFF), supposedly to reduce large vessel contributions⁹) still 39% of variation is explained by vCBV (maps not shown here). Further, 70% of the spatial variation in SFBF is explained by BOLD HC-CVR variation (of which 56% is subsequently explained by baseline vCBV). This study indicates that baseline vCBV is dominant modulator of the BOLD signal. Care should be taken when attributing BOLD signal changes/fluctuations to solely neuronal activity or as a proxy for CO2 vessel reactivity, especially in disease cases where cerebral blood volume can be compromised.