Introduction

Cerebral metabolic rate of oxygen (CMRO₂) denotes the amount of O₂ that the brain consumes. Due to the scarcity and complexity of techniques to measure CMRO₂ in humans (e.g., ¹⁵O-PET), longitudinal changes in CMRO₂ have not been fully characterized. Using an MRI-based global CMRO₂ technique, the present study reports CMRO₂ changes in older adults from a total of 544 measurements, the largest CMRO₂ dataset to date. We examined longitudinal and cross-sectional CMRO₂ levels in the aging brain, as well as their dependence on the apolipoprotein E4 (APOE4) genotype. We also performed the analysis on two related physiological parameters, cerebral blood flow (CBF) and cerebral oxygen extraction fraction (OEF). In addition, we examined the difference in CMRO₂ between cognitively normal and impaired participants.

Methods

Experimental procedure: The participants in this study were part of the Biomarkers for Older Controls at Risk for Dementia (BIOCARD) cohort. A total of 174 participants with normal cognition and 55 cognitively impaired individuals (either mild cognitive impairment or dementia), aged 50 to 92 years old, were included in these analyses. Each participant was scanned on a 3T MRI (Philips) approximately every two years. On average, participants had 2.4 scans (range = 1-4) collected over 3.3 years of follow-up. Figure 1 shows the illustration of CMRO₂ estimation from the arterio-venous difference in oxygen content^1-2, i.e., CMRO₂ = CBF*(Y_a-Y_v)*C_h, where Y_v was the venous oxygenation measured by T₂-Relaxation-Under-Spin-Tagging (TRUST)^3-4, CBF was measured with phase-contrast MRI applied on the four feeding arteries of the brain⁵. Y_a, the arterial oxygenation, was assumed to be 98%. C_h is the oxygen-carrying capacity of the blood. OEF was calculated as (Y_a−Y_v)/Y_a.
Data analysis: Linear mixed effect (LME) models were used to study cross-sectional and longitudinal alterations in aging.

Results and Discussion

Table 1 summarizes the demographic and clinical information of the participants. Table 2 shows the results of the LME model, examining changes in CMRO₂ among individuals with normal cognition. Longitudinally, there was an increase in CMRO₂ with time (p<0.001), suggesting that the brain is “working harder” to compensate for its lower efficiency. There was also an effect of baseline age on CMRO₂ (p=0.003). From the fifties to the eighties, there was a general effect of decreasing CMRO₂ cross-sectionally, which may be related to the brain running out of compensatory capacity and showing signs of neurodegeneration. Note that the longitudinal follow-up time is on the order of 2-6 years, which is much shorter than the cross-sectional age range (~30 years). Therefore, the cross-sectional and longitudinal observations are not necessarily contradictory, because they are reporting CMRO₂ changes at different time spans.

The above findings suggest that CMRO₂ changes in aging are bell-shaped, rather than monotonic. To further investigate this question, we examined baseline age as a predictor of the slope of the CMRO₂ change (in μmol/100g/min/year). We found an age-dependence of the CMRO₂ slope (p=0.02). Specifically, older adults less than the age of 75 years old tend to show a time-increase in CMRO₂ whereas those at more advanced age tend to show a time-decrease in CMRO₂ (Figure 2).

Next, we studied the effect of APOE4 on brain physiological parameters, including CMRO₂, OEF, and CBF among cognitively normal participants. The number of APOE4 alleles (coded 0,1,2) was associated with OEF (coefficient=-1.5, p=0.02), but not with CMRO₂ or CBF (both p>0.05). When only including noncarriers and participants with one APOE4 allele (i.e., heterozygotes), we did not observe any association between APOE4 carrier status and the three parameters. However, when including noncarriers and participants with two E4 alleles (i.e., homozygotes), the homozygotes had lower OEF and higher CBF compared to noncarriers (p=0.02, p=0.04, respectively), indicating that brain physiological alterations by APOE4 are most pronounced in homozygotes. These findings also suggest that the decrease of OEF in APOE4 carriers is mainly caused by the increase of blood supply rather than an alteration of oxygen consumption.

With 55 cognitively impaired participants included in the LME model, the diagnosis category (coded 0 for normal cognition and 1 for cognitive impairment) was negatively associated with CMRO₂ (coefficient=-7.3, p=0.02). In other words, CMRO₂ was lower in cognitively impaired participants than those with normal cognition. This finding indicates that CMRO₂ is sensitive in detecting neurodegeneration.

Additionally, from a basic physiology point-of-view, we aimed to understand the relationship between CBF and OEF. We examined the correlation between CBF and OEF (Figure 3). A strong negative association was observed (R=-0.53, p<0.001).

Conclusion

This study suggests CMRO₂ may follow a “bell-shaped” curve in the aging brain, and indicates that CMRO₂ may be a useful biomarker of neural efficiency in neurodegenerative diseases.

Acknowledgements

No acknowledgement found.