Decreased cerebral metabolic rate of oxygen (CMRO₂) is associated with increased tau deposition in Alzheimer's disease

Abstract

Introduction

Transgenic Alzheimer's (AD) mouse models and post mortem studies have shown that tau pathology disrupts the mitochondrial electron transport chain and results in mitochondrial damage and synaptic dysfunction. We assessed whether magnetic resonance imaging (MRI)‐derived measures of oxygen metabolism, oxygen extraction fraction (OEF), and cerebral metabolic rate of oxygen (CMRO₂) were associated with tau deposition in vivo.

Methods

We assessed tau using ¹⁸F‐MK‐6240 positron emission tomography and CMRO₂/OEF using multi‐echo gradient‐recalled echo with QQ processing, as well as pseudo‐continuous arterial spin‐labeling MRI in cognitively normal (CN) elderly and subjects with mild cognitive impairment/AD (N = 42). Associations among imaging measures were evaluated across cortical lobes, including comparisons between regions with and without tau deposition.

Results

Across all subjects, higher tau standardized uptake value ratio (SUVR) was associated with lower cerebral blood flow (CBF) and CMRO₂. Tau SUVR was negatively associated with Montreal Cognitive Assessment scores, whereas OEF, CBF, and CMRO₂ were positively associated. Within subjects, CMRO₂ was lower in tau‐positive temporal regions in both CN and MCI/AD subjects, and CBF was lower in tau‐positive temporal regions in CN subjects.

Discussion

MRI‐derived CBF and CMRO₂ were associated with regional tau deposition, particularly in the temporal lobe, which suggests they have potential as non‐invasive markers of metabolic alterations accompanying tau pathology.

Highlights

• Regional tau deposition is associated with reduced CBF and oxygen metabolism across the AD spectrum.

• MRI‐based CMRO₂ shows spatial colocalization with tau pathology, particularly in temporal brain regions.

• CBF and CMRO₂ are negatively correlated with tau deposition in high‐tau ROIs selected by visual radiologic assessment.

• Lower cerebral perfusion and oxygen metabolism correlate with worse cognitive performance.

1. BACKGROUND

The accumulation of Amyloid β (Aβ) plaques and neurofibrillary tangles of hyperphosphorylated tau proteins in the brain has long been recognized as the key pathological hallmarks underlying Alzheimer's disease (AD). Early tau deposition has been shown to be more closely associated with cognition than Aβ plaques, ¹ , ² , ³ and tau PET maps regional tau deposition in vivo, allowing us to address fundamental questions about the disease mechanism. Tau deposition on positron emission tomography (PET) has also shown to be associated with glucose hypometabolism and brain atrophy. ⁴ , ⁵ , ⁶ Concordance between the distribution of tau PET‐derived stages with the distribution of neuropathologically derived Braak stages confirms the utility of tau PET to stage AD even at the preclinical stage. ⁷ , ⁸ The second‐generation tau tracer ¹⁸F‐MK‐6240 exhibits high affinity and selectivity to neurofibrillary tau tangles and suitable pharmacokinetics and has been widely used to quantify tau deposition in vivo. ⁹ , ¹⁰

Tau pathology has been found to contribute to abnormal mitochondrial dynamics and oxygen hypometabolism, as the accumulation of tau protein impairs mitochondrial distribution and mitophagy, and oxidative stress induces tau hyperphosphorylation and accumulation. ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ Reduced cerebral blood flow (CBF), or hypoperfusion, has also been associated with tau aggregation across the AD continuum, including across stages in which clinical symptoms may not yet be apparent. ¹⁶ , ¹⁷ However, a specific spatial‐temporal pattern of these reductions remains unclear. ¹⁸ , ¹⁹ As hypoperfusion limits oxygen delivery and mitochondrial dysfunction impairs oxygen utilization, we may hypothesize that both processes converge on a state of oxygen hypometabolism.

¹⁵O‐PET can map oxygen metabolism, including oxygen extraction fraction (OEF), a measure of the ability of tissue to use oxygen and the balance between oxygen consumption and delivery reflected by perfusion. Similarly, cerebral metabolic rate of oxygen(CMRO₂), which reflects regional oxygen consumption, has been gaining interest as a potential biomarker for brain tissue viability in neurodegenerative diseases. ²⁰ Global and regional CMRO₂ reduction in AD and cognitive impairment has been demonstrated on ¹⁵O‐PET. ²¹ However, PET is limited in availability and affordability, and the 2‐min half‐life of 15O tracer makes ¹⁵O‐PET hardly accessible for research and inaccessible in clinical practice. Alternatively, various magnetic resonance imaging (MRI) techniques have been quantitative blood oxygenation dependent, calibrated functional MRI, T2‐relaxation‐under‐spin‐tagging (TRUST), quantitative susceptibility mapping (QSM), and QSM+qBOLD (QQ). ²² Recent advancements in MRI technology enable the measurement of OEF and CMRO₂ using readily available sequences in clinical settings, reducing reliance on difficult ¹⁵O‐PET. MRI‐based oxygen metabolism measurements have been used to study cognition, ²³ , ²⁴ vascular and genetic risks, ²⁵ , ²⁶ , ²⁷ , ²⁸ white matter hyperintensities, ²⁷ , ²⁹ and cerebrospinal fluid (CSF)‐based AD pathology. ²⁷ While global oxygen hypometabolism in AD has been reported, regional variations remain unclear.

This study aims to use regional analysis of cerebral metabolic parameters to better understand the link between oxygen metabolism and tau pathology. Participants spanning the continuum from normal aging to symptomatic AD were included to assess associations across a broad range of pathological burden. We hypothesize that regional tau accumulation in the brain on PET is associated with reduced blood flow and cerebral metabolic rate of oxygen. In this study, we analyzed the spatial distribution of CBF, OEF, CMRO₂ changes, and their association with tau burden at both the subject and voxel levels, elucidating the potential role of altered oxygen metabolism in the progression of tau‐related neurodegeneration.

2. METHODS

2.1. Participants

A total of 42 elderly subjects who underwent tau PET (¹⁸F‐MK‐6240) and 3T MRI (T1W, ASL, mGRE) as part of prospective National Institutes of Health‐funded AD studies were retrospectively analyzed in this study. The study was approved by our Institutional Review Board, and all subjects provided a signed written informed consent form. All subjects underwent standardized evaluations by a cognitive neurologist, consisting of clinical and neurological exams, interviews with the subject and study partner, Clinical Dementia Rating Scale, ³⁰ Montreal Cognitive Assessment (MoCA), ³¹ and the National Alzheimer's Coordinating Center (NACC) Uniform Data Set V3.0 telephone cognitive battery. ³² MRI and PET images were reviewed by a board‐certified radiologist with subspecialty certification in neuroradiology and more than 15 years of experience reading brain PET scans. All subject data were reviewed to confirm final diagnosis in a multidisciplinary consensus conference and diagnosed as cognitively normal (CN), MCI, or AD in accordance with NACC criteria.

RESEARCH IN CONTEXT

1. Systematic review: The authors reviewed prior literature focused on tau PET imaging, CBF, mitochondrial dysfunction, and cerebral oxygen metabolism in AD. Previous studies reported global reductions in perfusion in aging and AD and associations between tau pathology and mitochondrial dysfunction. The relevant citations are appropriately cited.

2. Interpretation: This study demonstrates that higher regional tau burden measured by ^{1 8}F‐MK‐6240 PET is associated with lower CBF and reduced oxygen consumption, with spatial colocalization particularly evident in temporal regions. These findings link tau pathology to regional brain oxygen hypometabolism across the AD spectrum and support MRI‐based CMRO₂ as a marker of tissue viability and tau pathology.

3. Future directions: Longitudinal studies are needed to determine the temporal sequence between tau accumulation and oxygen hypometabolism, assess stage‐dependent effects, and evaluate whether CMRO₂ predicts cognitive decline or therapeutic response.

2.2. Tau PET acquisition

¹⁸F‐MK‐6240 data were acquired from 90 to 120 min after rapid bolus injection of approximately 185 MBq and reconstructed to a 400 × 400 × 109 matrix of 1 × 1 × 2 mm³ voxels with six frames (6 × 5 min) for the calculation of the standard uptake value ratio (SUVR). The SUVR was calculated using the summed PET image with cerebellar cortex as reference region. ³³

2.3. MRI data acquisition

The scans were conducted on a 3T MRI scanner (Siemens Prisma 3T) using a 64‐channel head‐neck coil. T1w was used for co‐registration, and segmentation was acquired using magnetization‐prepared rapid gradient‐echo sequence with repetition time/ echo time (TR/TE) = 2400/ 2.96 ms, flip angle = 9°, field of view (FOV) = 25.6 × 25.6 cm, matrix size = 256 × 256 × 208 sagittal slices, and voxel size = 1 × 1 × 1 mm³. A three‐dimensional (3D) multiecho gradient echo (mGRE) imaging technique used for OEF and CMRO₂ mapping was acquired with parameters including voxel size = 1 × 1 × 1 mm³, acquisition matrix = 256 × 208 × 144, 10 equally spaced echoes (TE1/ΔTE = 6.1/4.0 ms), TR = 50 ms, bandwidth = 260 Hz/pixel, flip angle = 15°. Pseudo‐continuous arterial spin labeling (PCASL) for CBF quantification was acquired with an M0 pre‐scan and a post‐labeling delay (PLD) = 1800 ms, voxel size = 2.5× 2.5 × 2.5 mm³, acquisition matrix = 96 × 96 × 57, TE = 33.8 ms, TR = 4000 ms, flip angle = 120°, with both anterior‐posterior and posterior‐anterior direction phase images for eddy‐current distortion correction.

2.4. OEF, CBF, and CMRO₂ reconstruction

CMRO₂ mapping was obtained by calculating the product of the OEF maps, reconstructed from the mGRE data, and the CBF maps, reconstructed from the PCASL acquisition after the top‐up distortion correction ³⁴ , ³⁵ (Figure 1). OEF maps were reconstructed from combining quantitative susceptibility mapping (QSM) and quantitative blood oxygenation level‐dependent (qBOLD) mGRE magnitude modeling using the QQ ([QSM+qBOLD] with temporal clustering, tissue composition, and total variation) algorithm. ³⁶ QSM was reconstructed using morphology enabled dipole inversion with automatic uniform CSF zero reference algorithm and projection onto dipole field (PDF) for background field removal. ³⁷ , ³⁸ , ³⁹ CBF maps were reconstructed using the ASLtbx. ⁴⁰ CMRO₂ maps were then generated using the equation CMRO₂ (µmol/100 g/min) = CBF (mL/100 g/min) × OEF × [H]a, where [H]a = 7.377(µmol/mL). Both tau SUVR and CMRO₂ maps were co‐registered with FreeSurfer space T1w images of each subject using FMRIB Software Library flirt command. ⁴¹ Cortical frontal, temporal, occipital, and parietal lobes were segmented using FreeSurfer version 7.1 ⁴² for each subject, and all regions of interest (ROIs) were eroded by one voxel to limit spill‐in from adjacent white matter and CSF, thereby reducing the partial volume effects.

FIGURE 1.

The pipeline of CMRO₂ generation. QSM was reconstructed using the MEDI algorithm. OEF was reconstructed using QQ combined modeling of mGRE magnitude and phase. CBF was calculated from PCASL data. CMRO₂ maps were reconstructed by obtaining the product of the OEF and CBF maps.

2.5. Statistical analysis

Student's t‐test was used to assess for significant group differences in continuous variables, and the chi‐squared (χ²) test was used to compare categorical variables between the CN and MCI/AD groups. The Shapiro‐Wilk test was used to assess the normality of the data distribution. Using a cut‐off of ¹⁸F‐MK‐6240 SUVR > 1.3, we determined tau‐positive (tau+) and tau‐negative (tau−) voxels in each lobar region. We then performed paired Wilcoxon rank‐sum tests or t‐tests separately in CN and MCI/AD groups to compare OEF, CBF, and CMRO₂ values in tau+ and tau− voxels. We performed age‐ and sex‐controlled linear regressions between mean tau SUVR and OEF, CBF, or CMRO₂ in the brain regions to show correlation across the whole cohort. Age‐ and sex‐controlled linear regressions were performed between MoCA scores as the outcome variable and tau SUVR, OEF, CBF, or CMRO₂ in the brain regions as the predictor variables to assess the relationship between hemodynamic or metabolic parameters and cognitive status across the whole cohort. Subjects with fewer than 1% tau+ voxels in each lobe were excluded from the corresponding regional analysis to avoid noise‐driven variability, ensure biological relevance of regional tau burden, and improve the robustness of statistical estimates. For 16 subjects exhibiting prominent tau PET hot spots, a radiologist with 25 years of experience (I.K.) identified one high‐tau region in each subject by selecting a circular ROI (diameter = 10 mm, area = 76 mm²) on the slice with the highest tau SUVR. OEF, CBF, and CMRO₂ were age‐ and sex‐controlled linearly regressed against mean tau SUVR in the high‐tau regions across the 16 subjects. The threshold for statistical significance was set to α = 0.05 for all statistical tests and p values were adjusted for multiple comparisons using a false discovery rate‐controlling procedure. ⁴³ , ⁴⁴

3. RESULTS

3.1. Demographics

Participant demographics, including age, sex, educational level, MoCA scores, and amyloid and tau status, are shown in Table 1. Forty‐two elderly subjects were included in this study, with 23 diagnosed by consensus as CN and 19 as MCI/ AD. There was no significant difference in age or education level between the CN and MCI/AD groups. The majority (78%) of the healthy subjects were female, whereas the cognitively impaired group had a more balanced sex distribution (p = 0.0085), with 54% female participants. The MCI/AD group showed significantly lower MoCA scores (15.31 ± 5.04) than the healthy group (20.23 ± 1.84), as expected.

TABLE 1.

Participant demographics.

Variable	Overall (N = 42)	CN (N = 26)	MCI/AD (N = 16)	P value
Age (years), mean ± SD	68.07 ± 8.10	68.04 ± 7.20	69.38 ± 9.10	0.67
Sex, N (%)				0.038
Male	15 (36%)	6 (23%)	9 (56%)
Female	27 (64%)	20 (77%)	7 (44%)
Education level (years)	16.48 ± 2.40	17.04 ± 1.80	15.56 ± 2.99	0.051
MoCA score	18.36 ± 4.15	20.23 ± 1.84	15.31 ± 5.04	0.000052
Amyloid positive (N, %)	26 (61.9%)	10 (38.5%)	16 (100%)

3.2. Concordance between CBF, CMRO₂ and tau

Figure 2 shows the reconstructed OEF, CBF, CMRO₂ maps, and tau SUVR image of a 59‐year‐old female AD patient overlaid on T1w. The concordance of spatial distribution in elevated tau deposition, CBF hypoperfusion, and CMRO₂ hypometabolism is easily visualized while OEF contrasts across brain regions are more subtle.

FIGURE 2.

OEF, CBF, CMRO₂, and tau SUVR maps overlaid on T1w image of a 59‐year‐old female AD patient. Areas of higher tau deposition (red) correspond to areas of low CBF and CMRO₂ (blue).

3.3. Voxel‐level analyses within subjects

In CN subjects, CBF (p < 0.01) and CMRO₂ (p < 0.01) in tau+ regions were lower than those in the respective tau− regions in the temporal lobe (Table 2 and Figure S1). In MCI/AD subjects, OEF in tau+ regions is statistically higher than in tau− regions in the temporal lobe (p = 0.040) and lower than in tau− regions in the parietal lobe (p = 0.048), yet the mean OEF differences were small (<0.5%) and lower than the typical noise level in OEF maps, suggesting limited practical significance. CMRO₂ in tau+ regions is lower than in tau− regions in the temporal lobe (p = 0.048) (Table 2 and Figure S2).

TABLE 2.

Results of paired t‐tests comparing mean OEF, CBF, and CMRO₂ values in tau+ and tau− regions within subjects.

	Cognitively normal subjects			MCI/AD subjects
	OEF	CBF (ml/100 g/min)	CMRO2 (µmol/100 g/min)	OEF	CBF (ml/100 g/min)	CMRO2 (µmol/100 g/min)
Frontal lobe	p = 0.17	p = 0.80	p = 0.65	p = 0.42	p = 0.15	p = 0.11
Temporal lobe	p = 0.50	** p = 0.0014 MD = −5.92	** p = 0.0010 MD = −13.31	* p = 0.040 MD = 0.0043	p = 0.056	** p = 0.048 MD = −9.67
Parietal lobe	p = 0.064	p = 0.17	p = 0.17	* p = 0.048 MD = −0.0017	p = 0.20	p = 0.14
Occipital lobe	p = 0.41	p = 0.17	p = 0.17	p = 0.12	p = 0.056	p = 0.11

Abbreviations: CBF, cerebral blood flow; CMRO2, cerebral metabolic rate of oxygen; MD, mean difference; OEF, oxygen extraction fraction. Statistical significance is indicated by p values.

^*Statistically significant at p < 0.05, but small MD, likely of no practical clinical significance.

^**Statistically significant at p < 0.05 and likely meaningful MD.

3.4. Subject‐level analysis across the cohort

Among all subjects, OEF is positively associated with tau SUVR in the occipital lobe (p = 0.016, R ² = 0.24) controlling for age and sex. CBF is negatively associated with tau SUVR in all lobes: frontal lobe (p = 0.005, R ² = 0.29), temporal lobe (p < 0.001, R ² = 0.39), parietal lobe (p < 0.001, R ² = 0.35), and occipital lobe (p = 0.001, R ² = 0.34). CMRO₂ is negatively associated with tau SUVR in all lobes: frontal lobe (p = 0.011, R ² = 0.25), temporal lobe (p < 0.001, R ² = 0.38), parietal lobe (p = 0.006, R ² = 0.27), and occipital lobe (p = 0.020, R ² = 0.23) (Figure 3).

FIGURE 3.

Partial regression plots showing association between mean OEF (top row), CBF (middle row), and CMRO₂ (bottom row) and tau SUVR in four brain lobes. Black dots represent cognitively normal subjects, and red triangles represent AD/MCI subjects. The x‐axis displays residuals from regressing OEF, CBF, and CMRO₂ on age and sex, while the y‐axis shows residuals from regressing tau SUVR scores on the same covariates.

3.5. OEF, CBF, CMRO₂, and tau with cognitive function

Linear regressions between MoCA scores and imaging biomarkers were performed controlling for age and sex, as shown in Figure 3. Specifically, tau SUVR was negatively correlated with MoCA scores in the frontal (p < 0.01, R ² = 0.60), temporal (p < 0.01, R ² = 0.61), parietal (p < 0.01, R ² = 0.56), and occipital lobes (p < 0.01, R ² = 0.39) (Figure 3, column 1). OEF was positively correlated with MoCA scores in the frontal (p = 0.02, R ² = 0.22), temporal (p = 0.02, R ² = 0.23), parietal (p = 0.02, R ² = 0.22), and occipital lobes (p = 0.02, R ² = 0.23) (Figure 3, column 2). CBF was positively correlated with MoCA scores in the frontal (p < 0.01, R ² = 0.30), temporal (p < 0.01, R ² = 0.30), parietal (p < 0.01, R ² = 0.30), and occipital lobes (p < 0.01, R ² = 0.26) (Figure 3, column 3). CMRO₂ was positively correlated with MoCA scores in the frontal (p < 0.01, R ² = 0.35), temporal (p < 0.01, R ² = 0.35), parietal (p < 0.01, R ² = 0.35), and occipital lobes (p < 0.01, R ² = 0.30) (last column of Figure 4).

FIGURE 4.

Partial regression plots showing association between mean tau SUVR, OEF, CBF, and CMRO₂ against MoCA scores in four brain lobes. The x‐axis displays residuals from regressing tau SUVR, OEF, CBF, and CMRO₂ on age and sex, while the y‐axis shows residuals from regressing MoCA scores on the same covariates. Statistical significance is indicated by p values in red. Tau SUVR is negatively correlated and OEF, CBF, and CMRO₂ are positively correlated with MoCA scores in all four lobes.

3.6. OEF, CBF, and CMRO₂ in high‐tau regions identified by visual radiologic assessment

Among the 16 subjects with prominent tau PET hotspots, OEF is negatively associated with tau SUVR in the high‐tau ROI (p = 0.048, R ² = 0.47). CBF is negatively associated with tau SUVR in the high‐tau ROI (p = 0.0024, R ² = 0.69). CMRO₂ is negatively associated with tau SUVR in the high‐tau ROI (p = 0.0025, R ² = 0.68) (Figure S3).

4. DISCUSSION

This cross‐sectional study investigated the relationship between MRI‐based CMRO₂ and PET‐measured tau deposition in an AD‐spectrum cohort by integrating ASL MRI‐based CBF and mGRE‐derived OEF. We demonstrated the regional variations of OEF, CBF, and CMRO₂ in cortical lobes at both voxel and subject levels and associated these variations to tau burden measured with ¹⁸F‐MK‐6240 PET. Our results demonstrated a significant negative relationship between regional CBF, CMRO₂, and tau burden across brain cortical lobes. Specifically, CBF and CMRO₂ were consistently lower in temporal regions exhibiting tau accumulation in both CN subjects and patients with MCI/AD, supporting our hypothesis that elevated tau pathology is associated with impaired oxygen metabolism. Furthermore, cognitive performance measured by MoCA scores correlated positively with OEF, CBF, and CMRO₂ and negatively with tau deposition, underscoring the functional relevance of blood and oxygen metabolism biomarkers.

4.1. Global association of tau burden with lower CBF and CMRO₂ across subjects

CBF and CMRO₂ were both negatively associated with tau SUVR in all lobes across all subjects, indicating that greater tau deposition was related to reduced cerebral perfusion and oxygen consumption. This regional pattern suggests that increasing tau burden is broadly linked to reductions in perfusion and total oxygen consumption. These results were further corroborated by the negative association between perfusion, oxygen consumption, and tau deposition in high‐tau ROIs selected by visual radiologic assessment. In the subjects with prominent tau hotspots, the negative associations between CBF and CMRO₂ and tau level were stronger and might suggest metabolic changes depending on tau deposition spatiotemporal subtypes. Furthermore, OEF was found to be negatively correlated with tau SUVR in the tau hotspots and may reflect tissue damage with higher tau densities. The temporal lobe demonstrated the steepest negative slope between tau SUVR and metabolic/perfusion measures, while the frontal and parietal lobes showed more moderate and similar slopes, and the occipital lobe demonstrated the weakest relationship. These findings are generally consistent with the known spatial distribution of tau pathology in AD, where temporal and parietal regions tend to exhibit greater tau involvement, whereas occipital involvement is typically less pronounced until later disease stages. However, longitudinal studies will be necessary to clarify whether the regional patterns represent any stage‐dependent phenomenon.

4.2. Colocalization of tau deposition with reduced CBF and CMRO₂ in temporal regions

Within subjects, we showed significantly lower mean CBF and CMRO₂ in tau+ regions compared to tau− regions in the cortical temporal lobe, an early region of tau deposition, in CN subjects, suggesting spatial overlap between increased tau deposition and reductions in both blood flow and oxygen consumption and that these associations may occur even in cognitively normal aging. Given the early involvement of the temporal lobe in AD, ⁷ the colocalization of reduced CBF and CMRO₂ with elevated tau burden supports the notion that temporal regions are particularly vulnerable to combined pathological and metabolic alterations. Reduced CBF was no longer observed in the disease cohort in the temporal lobe only and may show perfusion changes that are more widespread and stage dependent. Higher OEF in tau+ regions of the temporal lobe but lower OEF in tau+ regions of the parietal lobe in the disease group were statistically significant, but the mean OEF differences between tau+ and tau− regions were small (< 0.5%), and there is limited practical significance of the observed differences.

4.3. Association between cognitive function, OEF, CBF, CMRO₂, and tau deposition across subjects

Tau SUVR was negatively correlated with MoCA scores across all lobes, indicating that higher regional tau burden was associated with lower global cognitive performance. CBF and CMRO₂ demonstrated positive correlations with MoCA scores across all lobes, suggesting an association of higher cerebral perfusion and oxygen metabolism with better cognitive performance. Reduced CBF ¹¹ , ¹⁶ , ¹⁸ , ¹⁹ , ²⁰ , ²³ , ⁴⁵ and CMRO₂ ^{23, 24} , ²⁶ , ²⁸ in aging and early AD agrees with previous reports and were expected. OEF was also positively correlated with MoCA scores, although the strength of these associations was comparatively weaker. However, prior studies reported mixing results including positive, ²⁵ negative, ²⁶ , ²⁷ or no associations ²³ , ²⁴ between OEF and cognition across studies, which might be due to differences in cohort characteristics in those studies. Participants in the current study were predominantly at an earlier stage of MCI/AD, which may influence the direction of the relationships considering that there might be a complex compensation mechanism for blood flow autoregulation and OEF. In addition, vascular risk factors were not explicitly stratified in this analysis, which may also account for discrepancies. The broadly similar associations between imaging measures and MoCA scores in all brain lobes may reflect the diffuse physiological effects of tau burden and the use of a global cognitive measure, whereas more spatially localized effects were observed in analyses focused on tau+ and tau− regions. These relationships across multiple lobes show that, although tau burden can be local or global, the alterations in cerebral perfusion and metabolism might be global and broadly related to cognitive function and show little regional difference.

4.4. Implications for tau‐associated physiological alterations

The observed reduction in oxygen consumption colocalized with tau deposition may reflect multiple physiological alterations associated with tau pathology, including reduced metabolic activity secondary to neuronal dysfunction or neurodegeneration. Although the underlying mechanisms cannot be determined from the present imaging measures, the spatial correspondence between tau deposition and reduced CMRO₂ is consistent with prior experimental evidence linking tau pathology to disrupted cerebral metabolism. In mouse models, tau oligomers and neurofibrillary tangles have been shown to cause mitochondrial dysfunction by preferentially disrupting Complex I of the mitochondrial electron transport chain in transgenic mouse models of AD. ⁴⁶ , ⁴⁷ The burden of tau fragments in the mitochondria of post mortem AD brain specimens has been found to correlate with mitochondrial impairment and pathological synaptic changes, resulting in mitochondrial loss. ¹³ , ⁴⁸ Global oxygen metabolism was measured in both amyloid/amyloid precursor protein and tau AD mouse models using MRI techniques, and the tau model showed decreased CMRO₂, whereas the amyloid model did not. ⁴⁹ Reduced oxygen consumption in tau+ regions may therefore reflect a combination of altered cellular energetics, decreased neuronal activity, and reduced metabolic demand associated with tissue injury. Our findings are also consistent with global and regional CMRO₂ reduction in AD and cognitive impairment demonstrated with 15O‐PET. ²¹ Overall, our results quantitatively links oxygen metabolism to tau pathology and demonstrate the potential utility of CMRO₂ for understanding and monitoring tissue viability and disease severity.

4.5. Study limitations

While our study demonstrated robust associations of the evaluated imaging biomarkers, it has several limitations. First, it used a cross‐sectional design, which prevents inference of causality. Longitudinal studies are essential to determine whether tau‐related hypometabolism precedes cognitive decline or reflects downstream effects. Second, the current QQ model for OEF is limited by a signal model of random cylinder dephasing and strong regularization. Future improvement in signal modeling may include realistic venous and lesion geometries and effects of diamagnetic plaques, voxel size, and field strength. Third, the sample size, particularly within the MCI/AD subgroup, and unbalanced sex distribution may limit generalizability. Future research should explore these associations in larger, more diverse populations and investigate whether CMRO₂ changes can predict clinical outcomes or response to therapeutic interventions. Moreover, the inclusion of complementary imaging modalities, such as amyloid PET or functional MRI, could provide a more holistic view of the interplay between amyloid, tau, and metabolic dysfunction in AD.

5. CONCLUSION

The concordance observed between tau deposition, hypoperfusion, and oxygen hypometabolism across key brain regions underscores the utility of QQ‐based CMRO₂ mapping in detecting region‐specific impairments in oxygen metabolism. The observed associations emphasize the role of tau pathology in disrupting brain oxygen metabolism, opening a potential pathway for understanding the neurotoxicity of tau accumulation and development of targeted disease modifiers.

CONFLICT OF INTEREST STATEMENT

Pascal Spincemaille and Yi Wang are co‐inventors on QSM‐related patents owned by Cornell University and have ownership shares in MedImageMetric LLC. Dr. Chiang has served as a consultant for Life Molecular Imaging and Alnylam Pharmaceuticals, and she has received speaker honoraria from Efficient CME and PeerView. The remaining authors have no conflicts of interest to report. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

This study was approved by the Weill Cornell Medicine IRB, and written informed consent was obtained from all participants.

Supporting information

Supporting Information: dad270442‐sup‐0001‐ICMJE.pdf

Supporting Information: dad270442‐sup‐0002‐Figures.pdf

Zhuang H, Zhou L, Wang XH, et al. Decreased cerebral metabolic rate of oxygen (CMRO₂) is associated with increased tau deposition in Alzheimer's disease. Alzheimer's Dement. 2026;18:e70442. 10.1002/dad2.70442