Hemodynamic and metabolic changes during hypercapnia with normoxia and hyperoxia using pCASL and TRUST MRI in healthy adults

Blood oxygenation level-dependent (BOLD) or arterial spin labeling (ASL) MRI with hypercapnic stimuli allow for measuring cerebrovascular reactivity (CVR). Hypercapnic stimuli are also employed in calibrated BOLD functional MRI for quantifying neuronally-evoked changes in cerebral oxygen metabolism (CMRO2). It is often assumed that hypercapnic stimuli (with or without hyperoxia) are iso-metabolic; increasing arterial CO2 or O2 does not affect CMRO2. We evaluated the null hypothesis that two common hypercapnic stimuli, ‘CO2 in air’ and carbogen, are iso-metabolic. TRUST and ASL MRI were used to measure the cerebral venous oxygenation and cerebral blood flow (CBF), from which the oxygen extraction fraction (OEF) and CMRO2 were calculated for room-air, ‘CO2 in air’ and carbogen. As expected, CBF significantly increased (9.9% ± 9.3% and 12.1% ± 8.8% for ‘CO2 in air’ and carbogen, respectively). CMRO2 decreased for ‘CO2 in air’ (−13.4% ± 13.0%, p < 0.01) compared to room-air, while the CMRO2 during carbogen did not significantly change. Our findings indicate that ‘CO2 in air’ is not iso-metabolic, while carbogen appears to elicit a mixed effect; the CMRO2 reduction during hypercapnia is mitigated when including hyperoxia. These findings can be important for interpreting measurements using hypercapnic or hypercapnic-hyperoxic (carbogen) stimuli.

The conversion of pCASL measured ∆M to CBF can be considered as a scaling problem, when applying a single-compartment kinetic model (Buxton et al. 1 ). Here, the scaling factor 'C' depends on the familiar tissue-specific parameters: T1 and BAT and 'global' parameters: blood T1,a, bolus duration τ, inversion efficiency α, blood/tissue water partition coefficient λ, and the voxel specific calibration M0,b. Note that the apparent T1' dependency on CBF (1/T1' = 1/ T1,tissue + f/λ with f = CBF in mL/g/s) in the kinetic model makes it deviate from a pure scaling problem, however, the effect of CBF on the apparent T1 is negligible. The latter can be seen by the very slight broadening of the CBF normalized kinetic curve for different simulated CBF values with respect to inversion time TI (see Supplemental Figure 2 below).
A range of kinetic curves was generated for a range of CBF values (0 -1.5 ml/g/s  0 -90 mL/100 g/min) and subsequently normalized by the CBF value. Tissue specific values where T1,GM = 1.3s 2 , T1,WM = 0.84s 2 , BATGM 3 and BATWM 3 . The resulting, 'normalized kinetic curves', give the conversion from CBF to ∆M with respect to the inversion time TI. The scaling factor C, used to convert ∆M to CBF, is then obtained by taking the reciprocal of the normalized kinetic curve at the pCASL sequence's TI. In Supplemental Figure 2 below, CBF normalized kinetic curves for GM and WM based values, tissue T1 and BAT, are depicted in blue and red, respectively. To convert measured ∆M to CBF (in mL/g/s) using GM based values one gets a scaling factor of CGM ≈ 1/0.5 ≈ 2 for the pCASL sequence inversion time used here: TI = τ + PLD = 1.5s + 1.7s = 3.2s. To convert measured ∆M to CBF using white matter based values, one gets a scaling factor of CWM≈ 1/0.42 ≈ 2.4 at the same inversion time.
From this exercise, we observe that for WM regions, when using GM based values in CBF quantification, the CBFWM is underestimated by a factor of CWM/CGM ≈ 2.4/2 ≈ 1.2, which is about 20%. This factor can be used to correct the CBF values in the WM ROI, i.e. increasing WM CBF by ~20%, yielding corrected global (whole-brain) CBF estimates and thus CMRO2 estimates. We found that the CMRO2 findings did not change notably when incorporating the WM correction on global CBF, the numerical values did change.

2/6
Supplementary Figure 1: Flow (CBF) normalized kinetic curves for GM and WM based values (tissue T1 and bolus arrival time, BAT) are depicted in blue and red, respectively. These curves give the conversion from CBF to ∆M with respect to the inversion time TI. The scaling factor C, to convert ∆M to CBF, is then obtained by taking the reciprocal of the normalized kinetic curve at the pCASL sequence's TI. To convert measured ∆M to CBF (in mL/g/s) using GM based values one gets a scaling factor of CGM ≈ 1/0.5 ≈ 2 for the pCASL sequence inversion time used here: TI = τ + PLD = 1.5s + 1.7s = 3.2s. To convert measured ∆M to CBF using white matter based values, one gets a scaling factor of CWM ≈ 1/0.42 ≈ 2.4 at the same inversion time. As a result, when using GM based values in CBF quantification, the WM CBF is underestimated by a factor of CWM/CGM ≈ 2.4/2 ≈ 1.2, which is about 20%. This factor can be used to correct the CBF values in the WM ROI, i.e. increasing WM CBF by ~20%, yielding corrected global (whole-brain) CBF estimates and thus CMRO2 estimates.

Supplementary Figure 2.
Sensitivity analysis of the effect on hyperoxic arterial blood water T1,HO on CBF and CMRO2 quantification. The T1,HO value used for carbogen (1.49s) and variations therein (δ) were plotted against the modelled percentage change (δ) in CBF and CMRO2 quantification. The used T1,NO for room-air (1.65s) is depicted by the dashed blue line. Note CBF and CMRO2 are directly proportional by the arteriovenous O2 difference. The resulting percentage change in CBF and CMRO2 were plotted for a range of possible T1,HO values, for absolute values and the percentage difference with respect to the reference T1,HO (=1.65s) for carbogen. CBF = cerebral blood flow; CMRO2 = cerebral metabolic rate of oxygen, T1,HO = arterial blood water T1a during hyperoxia, T1,NO = arterial blood water T1a during normoxia. Figure 3. A) CBF, B) ∆CBF, C) CMRO2 and D) ∆CMRO2 results for different arterial blood water T1a scenarios for normoxic (T1,NO) and hyperoxic (T1,HO) conditions. Although the absolute CBF and CMRO2 results change in value, the impact on the ∆CBF and notably the ∆CMRO2 changes, which is the topic of this study, did not change significantly for the different T1 scenarios. For the hyperoxic (carbogen) condition, a paO2 of 460 mmHg was assumed, and for scenarios II and IV we incorporated the reported hyperoxic T1a relativity by Ma et al. 4 (see Methods 'Effect of arterial blood water T1a on CBF and CMRO2 quantification'. CBF = cerebral blood flow; CMRO2 = cerebral metabolic rate of oxygen; paO2 = partial pressure of arterial O2, T1,HO = arterial blood water T1a during hyperoxia, T1,NO = arterial blood water T1a during normoxia. The boxplots show the minimum, maximum, median and interquartile range, open circles denote outliers. pEtCO2 = end-tidal partial pressure of CO2, CBF = global cerebral blood flow, Yv = venous blood oxygenation, OEF = oxygen extraction fraction, CMRO2 = cerebral metabolic rate of oxygen, std = standard deviation, †note this is the fractional change in percentage in Yv, i.e. not percentage points.