Abstract
Alzheimer’s disease (AD) is the leading cause of dementia in individuals over 65 in the U.S. Prevalence is projected to double by 2050, but current treatments cannot stop the progression of AD. Treatments administered before severe cognitive decline may be effective; identification of biomarkers for preclinical/prodromal stages of AD is therefore imperative. Cerebral blood flow (CBF) is a potential early biomarker for AD; generally, older adults with AD have decreased CBF compared to normally aging peers. Characterization should include the relationship between CBF and AD risk factors and pathologies. We assessed the relationships between CBF quantified by arterial spin labeled MRI, hypertension, APOE ε4, and amyloid and tau PET in 78 older adults: cognitively normal, subjective cognitive decline, and mild cognitive impairment. Decreased regional CBF was associated with increased tau pathology, and the relationship between CBF and amyloid pathology differed by hypertension status. Our findings suggest a complex relationship between risk factors, AD pathologies, and CBF that warrants future studies of CBF as a potential early biomarker for AD.
1. Introduction
Alzheimer’s disease (AD)1 is the leading cause of dementia; over 6 million older adults in the U.S. have AD, and this number is projected to double over the next thirty years (Alzheimer’s Disease Facts and Figures, 2021). The results of multiple studies and clinical trials suggest that an effective treatment for AD will likely need to be preventative (Jessen et al., 2014; Crous-Bou et al., 2017). Although there are some promising exceptions, treatments targeting amyloid and tau pathology administered after the emergence of clinical symptoms have generally been unable to restore cognitive function (Lu et al., 2020; Congdon and Sigurdsson, 2018). Therefore, it is imperative to find reliable and accurate biomarkers for the earliest, preclinical stages of AD in order to identify individuals who would benefit from preventative treatment. One such potential early biomarker is quantitative cerebral blood flow (CBF).
CBF is known to be decreased in individuals with AD compared to healthy individuals of the same age (Alsop et al., 2000). This decrease has also been shown to be more than would be expected given the amount of brain atrophy in these individuals, suggesting that altered CBF is itself a pathologic phenomenon in AD (Binnewijzend et al., 2013). These results have led to the development of the two-hit hypothesis of AD, which posits that vascular risk factors and amyloid accumulation exacerbate one another and eventually lead to tau aggregation, neurodegeneration, and cognitive decline (Zlokovic, 2011; Korte et al., 2020). This differs from the longstanding amyloid cascade hypothesis, which posits that amyloid deposition triggers the onset and propagation of the complex pathologies and symptoms that comprise AD (Hardy and Higgins, 1992). Evidence for the two-hit model of AD includes the fact that vascular risk factors like hypertension, diabetes mellitus, and obesity are also risk factors for AD (Gorelick, 2004). Chronic hypertension, in particular, has been shown to cause cerebral microvascular damage, which both worsens and is worsened by amyloid beta (Aβ) aggregation. It has been suggested that altered CBF may occur very early in the course of AD, before cognitive symptoms and potentially even prior to amyloid and tau aggregation (Iadecola, 2004). Given that this early alteration of CBF may play a role in initiating the cascade of pathological changes culminating in AD, it is an intriguing potential early biomarker.
While altered CBF has been described as one of the earliest changes in AD, its typical spatial and temporal patterns throughout the disease course are not yet fully characterized. It has been shown that CBF increases in some brain regions in individuals on the AD spectrum, especially in preclinical and prodromal AD, known as subjective cognitive decline (SCD) and mild cognitive impairment (MCI) (Beishon et al., 2017; Dai et al., 2009). Altered CBF is largely characterized by hypoperfusion by the time individuals have some level of cognitive decline (Wierenga et al., 2014; Swinford et al., 2022). Increased CBF in early, preclinical stages may be compensatory in response to other AD-related pathologies, like amyloid or tau aggregation. Indeed, global CBF is positively correlated with cognitive test scores (Boles et al., 2006). Vascular risk factors such as hypertension and genetic risk factors including the apolipoprotein E (APOE) ε4 allele may play a role in whether individuals have a compensatory increase in CBF and for how long that compensation can be maintained despite pathology and vascular dysfunction. It is also important to elucidate the spatial and temporal relationships between altered CBF and other AD-related pathologies, including amyloid and tau aggregation. CBF tends to be negatively correlated with both amyloid and tau pathology (Albrecht et al., 2020; Rubinski et al., 2021; Visser et al., 2020; Mattsson et al., 2014; Rodell et al., 2016; Tosun et al., 2016); however, some studies also report positive correlations between CBF and these pathologies, mostly in individuals with preclinical AD (Codispoti et al., 2012; Stomrud et al., 2012; Fazlollahi et al., 2020; Tosun et al., 2014; Sojkova et al., 2008). It has been established that both hypertension and APOE ε4 positivity are related to increased Aβ aggregation and are associated with cerebrovascular dysfunction and decreased CBF (Solis et al., 2020; Bell et al., 2012; Wiesmann et al., 2016). In this study, we aimed to assess the relationships between CBF and amyloid and tau pathologies in older nondemented adults and to determine whether the presence of the APOE ε4 allele and/or hypertension affects those relationships.
Extensive characterization of altered CBF in the context of preclinical AD is necessary to establish it as an effective early biomarker. In older nondemented adults, we assessed the relationships between CBF and global cortical amyloid and medial temporal lobe (MTL) tau pathology using arterial spin labeled (ASL) magnetic resonance imaging (MRI) and positron emission tomography (PET), as well as the interaction effects of APOE ε4 positivity and hypertension on the relationship of CBF with amyloid and tau. We hypothesized that amyloid and tau pathologies would negatively correlate with CBF overall, although localized regions of positive correlation would be present as well, given the early stage of disease in this sample. We also hypothesized that individuals with hypertension or at least one APOE ε4 allele would be less likely to have a compensatory positive correlation between CBF and AD-related pathologies.
2. Methods
2.1 Participants
This study included 78 participants (30 cognitively normal [CN], 25 SCD, 23 MCI) from the Indiana Alzheimer’s Disease Research Center who were at least sixty years of age. All participants completed T1-weighted structural and pseudocontinuous arterial spin labeled (pCASL) MRI scans, amyloid PET scan with either [18F]florbetapir or [18F]florbetaben, and tau PET scan with [18F]flortaucipir. All participants also had a clinical assessment and APOE genotyping. Diagnoses were determined by clinician consensus. SCD was defined as a score of 20 or more on the first 12 items of the 20-item Cognitive Change Index (CCI-20; Saykin et al., 2006), reflecting increased subjective memory concerns, with or without increased levels of informant-based concerns (Jessen et al., 2014), and the absence of a measurable cognitive deficit on objective cognitive assessment. Using standard criteria, an individual was diagnosed with MCI if the participant and/or an informant had a significant complaint about their cognition, and the participant scored 1.5 standard deviations or more below age-, education-, and sex-adjusted normative data on objective tests of cognitive functioning, either in memory or another cognitive domain. Additionally, individuals diagnosed with MCI, by definition, do not show a significant decline in daily functioning. Hypertension was defined by self-report near the time of imaging. Individuals defined as hypertensive reported a previous diagnosis of hypertension. Some hypertensive individuals were taking antihypertensive medication at the time of the study. All procedures were approved by the Indiana University School of Medicine Institutional Review Board. Informed consent was obtained according to the Declaration of Helsinki and the Belmont Report.
2.2 pCASL and T1-weighted MRI
All MRI scans were acquired on the same Siemens 3T Prisma scanner with a 64-channel head/neck radiofrequency coil. After an initial T1-weighted structural scan (sagittal 3D MPRAGE) each participant completed a 3D pseudo-continuous ASL (pCASL) MRI scan. The full MRI protocol consisted of MPRAGE, FLAIR, high resolution hippocampus imaging, DTI, fMRI, HYDI DTI, pCASL, and finally SWI over a total of roughly 58 minutes. The MPRAGE scan had the following parameters: repetition time (TR) 2300 ms, echo time (TE) 2.95 ms, field of view 270 mm, 176 slices, and voxel size 1.1 × 1.1 × 1.2 mm3. The pCASL scan was acquired with Fair QII using a multi-slice interleaved saturation sequence with the following parameters: repetition time (TR) 3840 ms, echo time (TE) 40.7 ms, field of view 240 mm x 240 mm, 54 slices with 2.5 mm thickness, image matrix 96 × 96, and voxel size 2.5 × 2.5 × 2.5 mm3. The label duration was 1500ms and post-labeling delay (PLD) was 1800ms. The total acquisition time was 7min 18sec including a M0 image and 7 pairs of label and control images.
Processing steps were performed using Statistical Parametric Mapping 12 (SPM12; Wellcome Department of Cognitive Neuroscience, London, UK) and Matlab R2022a as previously described (Butcher et al., 2021; Wang et al., 2003). For each individual, all pCASL images were aligned to the first pCASL image and averaged. The mean pCASL image was coregistered to the individual’s structural MRI image, and the structural image was coregistered to an averaged brain mask template. Then to obtain gray matter maps, the individual T1-weighted structural MRI images were segmented, and the gray matter images were normalized and smoothed with a 6 mm isotropic full width at half maximum (FWHM) Gaussian kernel. To calculate CBF from pCASL images, pairs of pCASL images were subtracted and then averaged to obtain a mean perfusion image. A standard single-compartment perfusion model (Alsop et al., 2015) was used to calculate mean CBF maps from the mean perfusion images (http://cfn.upenn.edu/perfusion/software.htm). To create a gray matter mask, the gray matter tissue map from the SPM12-segmented structural MRI was coregistered to the mean perfusion image, thresholded (> 0.75), smoothed (isotropic 6 mm FWHM Gaussian kernel), and binarized (> 0.20). This individualized gray matter mask was then applied to the mean CBF map, which was normalized to Montreal Neurological Institute (MNI) atlas space using nonlinear transformation parameters from the segmentation step, resampled to 2 mm/side isotropic voxels, and smoothed with an isotropic 6 mm FWHM Gaussian kernel, with the cerebellum excluded in order to remove individual variability in this region due to the imaging method. CBF was extracted from the resulting masked CBF map from frontal, parietal, temporal, and occipital lobes and limbic regions of interest (ROIs) using Automated Anatomical Labeling (AAL) version 1 and the MarsBaR toolbox (Brett et al., 2002).
2.3 Amyloid PET
Each participant underwent either a [18F]florbetapir (AmyVid; Eli Lilly and Co.) or a [18F]florbetaben (Neuraceq; Life Molecular Imaging, formerly Piramal Imaging) scan. For the [18F]florbetapir scans, 10 mCi ± 10% of [18F]florbetapir was injected intravenously, followed by a 50-minute uptake period. For [18F]florbetaben scans, 8.1 mCi ± 10% of [18F]florbetaben was injected intravenously, followed by a 90-minute uptake period. Participants were then imaged on the same Siemens Biograph mCT PET/CT scanner for 20 minutes with continuous list mode data acquisition. For both types of scans, a computed tomography scan was acquired for scatter and attenuation correction. The list mode data were then binned into four 5-minute frames and reconstructed using parameters defined in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) protocol (http://adni.loni.usc.edu). These parameters include corrections for scatter and random coincidence events, attenuation, and radionuclide decay. Each 5-minute frame was spatially aligned to the participant’s T1-weighted structural MRI using coreg, normalized to MNI space, and smoothed with an 8 mm FWHM Gaussian kernel, all using SPM12. The spatially aligned PET images were averaged to create a single image from the total acquisition time period for each individual (50-70 minutes for [18F]florbetapir and 90-110 minutes for [18F]florbetaben), and standardized uptake value ratio (SUVR) images were generated by intensity normalizing with the average radioactivity from a whole cerebellar region of interest from the Centiloid project (http://www.gaain.org/Centiloid-project/). For this study, we used the global cortical amyloid measure calculated in Centiloid units, a unified metric that allowed us to directly compare data from both scan types. The global cortical ROI was determined by the Centiloid project and is the ROI for which use of Centiloid units is intended. Both types of scans were processed with the Centiloid algorithm at a voxel-wise level as defined by the Centiloid project (Risacher et al., 2021). Global cortical Centiloid data were then extracted from both scan types using the MarsBaR toolbox (Brett et al., 2002). Amyloid positivity was defined as global cortical amyloid >21 Centiloid units.
2.4 Tau PET
All participants underwent a [18F]flortaucipir PET scan. 10 mCi 10% of [18F]flortaucipir was injected intravenously, followed by a 75-minute uptake period. Individuals were then imaged on the same Siemens Biograph mCT PET/CT scanner for 30 minutes with continuous list mode data acquisition. These data were binned into six 5-minute frames, and scans were reconstructed using a standard scanner software program (Siemens, Knoxville, TN) according to the ADNI protocol (http://adni.loni.usc.edu). The 5-minute frames corresponding to 80-100 minutes post-injection were spatially aligned to each participant’s T1 image and normalized to MNI space, averaged to create a single image, intensity normalized by the cerebellar crus (Schwarz et. al., 2016) to create SUVR images, and smoothed with an 8 mm FWHM Gaussian kernel, all using SPM12. For this study, we extracted the MTL SUVR. The constituent ROIs were generated from FreeSurfer 6.0 (Fischl and Dale, 2000; http://surfer.nmr.mgh.harvard.edu/), and the bilateral mean MTL (average of fusiform gyri, parahippocampal gyri, and entorhinal cortex) SUVR values were extracted for each participant using MarsBaR. For interaction analyses, tau positivity was defined as MTL tau SUVR greater than the third quartile value in this sample (1.152). This relative definition of tau positivity was similar to a previous data-driven tau cutoff (Mishra et. al., 2017).
2.5 Statistical Analyses
The demographic and clinical variables, as well as global CBF, total gray matter volume, and amyloid and tau PET measures were compared between diagnostic groups using one-way ANOVA for continuous variables and chi-square test for categorical variables. Stepwise multiple regression model (entry p=0.10, removal p=0.20) with dependent variable regional CBF and independent variables diagnostic group, age, sex, hypertension status, APOE ε4 status, and total gray matter volume (from T1-weighted image gray matter estimates by SPM12 tissue volume utility) were used to determine demographic/clinical variables that correlated with CBF in this sample. Additionally, amyloid and tau PET metrics were treated as the dependent variables in a multiple regression model with age, sex, hypertension status, APOE ε4 status, and total gray matter volume as independent variables to assess correlations between amyloid and tau PET and demographic/clinical variables. Partial Pearson correlations assessed the relationship between regional CBF and amyloid or tau PET metrics, with age, sex, total gray matter volume, hypertension status, and APOE ε4 status as covariates. Analyses including CBF were repeated on a voxel-wise basis using multiple regression with CBF maps as the dependent variable and amyloid or tau PET metrics, age, sex, total gray matter volume, and hypertension status as independent variables or covariates as appropriate. A voxel-wise two-way ANCOVA model assessed interactions of amyloid/tau positivity and hypertension/APOE ε4 status with age, sex, and total gray matter volume as covariates. CBF was extracted from significant clusters and modeled as the dependent variable in linear regression to visualize interactions. Results are listed for clusters present when viewed at threshold p≤0.001, cluster size (k)≥50 voxels (uncorrected), significant at cluster-wise p≤0.05 with Family-Wise Error (FWE) correction, and with peak voxels within gray matter regions. All voxel-wise analyses were completed using SPM12. Other statistical analyses and all graphs were created in IBM SPSS 25.
3. Results
3.1 Demographics
Demographics and clinical variables are given in Table 1. The MCI group had a greater global cortical Centiloid value than the CN and SCD groups (p=0.001). Diagnostic groups also differed by prevalence of hypertension (p<0.001); the SCD group had the highest prevalence of hypertension, followed by MCI, and then by CN. Finally, the groups differed by proportion of women (p=0.05). CN had the greatest proportion of women, followed by SCD, and then by MCI. Global CBF did not differ by diagnostic group (p>0.6).
Demographic and clinical variables by diagnostic group.
3.2 CBF vs. Demographics
Mean CBF from each lobe ROI was entered into a stepwise multiple regression with diagnosis, age, sex, hypertension status, APOE ε4 status, and total gray matter volume as independent variables. Age and hypertension were negatively correlated with CBF in all lobes (Table 2). No correlations were found between CBF and diagnosis, APOE ε4 status, or total gray matter volume. The age by hypertension status interaction was not statistically significant in any of the lobes. Voxel-wise regressions were used to model the spatial relationships between CBF and age (Figure 1A & B) and hypertension (Figure 1C & D) with covariates of sex, gray matter volume, and age where appropriate. Clusters and peak voxels for these correlations are presented in Tables 3 and 4.
Correlation of age and hypertension status with lobar CBF.
Regions of negative association between age and voxel-wise CBF. Peak effects reported in MNI coordinates (mm).
Regions of negative association between hypertension and voxel-wise CBF. p<0.001; k>50. Peak effects reported in MNI coordinates (mm).
Negative correlations between voxel-wise CBF and age (A & B) and hypertension status (C & D). Covariates are age (in the hypertension analysis), sex, and total gray matter volume. Illustrated at p 0.001, k 50.
3.3 Tau and Amyloid PET vs. Demographics
The global cortical Centiloid and the MTL tau SUVR data were used as dependent variables in a stepwise regression model with age, sex, hypertension status, APOE ε4 status, and total gray matter volume as independent variables. The global cortical Centiloid value was associated with APOE ε4 status (β=0.414, t=3.96, p<0.001). MTL tau SUVR was negatively associated with hypertension status (β=-0.223, t=-2.00, p=0.049).
3.4 Tau PET vs. Regional CBF
Partial correlations were completed for regional CBF with MTL tau SUVR, with age, sex, and total gray matter volume as covariates. There were no significant correlations for any of the ROIs between CBF and MTL tau SUVR. The addition of hypertension status and APOE ε4 status as covariates did not change the results. Using voxel-wise regression models with age, sex, and total gray matter volume as covariates, MTL tau SUVR was negatively correlated with CBF in the left hippocampus and left parahippocampal gyrus (p≤0.001, k≥50; Table 5). The significant cluster and beta maps of the negative correlations between MTL tau SUVR and CBF are illustrated by Figure 2, and the partial correlation in the significant cluster (R2=0.256) is plotted in Figure 3. In order to determine whether APOE ε4 status or hypertension status affected the relationship between amyloid or tau PET measures and CBF, a voxel-wise two-way ANCOVA was used, with CBF as the dependent variable, tau positivity and APOE ε4 or hypertension status as independent variables, and age, sex, and total gray matter volume as covariates. Tau positivity was defined as MTL tau SUVR greater than the third quartile value in this sample (mean: 1.110, SD: 0.110, range: 0.706, third quartile: 1.152). There were no effects of tau positivity by hypertension interaction or tau positivity by APOE ε4 positivity interaction on CBF.
Negative correlations between voxel-wise CBF and MTL tau SUVR. p<0.001; k>50. Peak effects reported in MNI coordinates (mm).
Voxel-wise CBF rendering (A) and beta maps (B-D) of negative correlations between voxel-wise CBF and MTL tau SUVR. Parahippocampal/hippocampal CBF are lower in those with increased MTL tau. Age, sex and total gray matter volume were the covariates. Rendering (A) displayed at p 0.001, k 50.
Scatterplot of negative correlations between voxel-wise CBF and MTL tau SUVR. Parahippocampal/hippocampal CBF are from the significant cluster in the voxel-wise analysis. Blue = CN, green = SCD, red = MCI. R2=0.256.
3.5 Amyloid PET vs. Regional CBF
Partial correlations were completed for regional CBF with global cortical Centiloid, with age, sex, and total gray matter volume as covariates. There were no significant correlations for any of the ROIs between CBF and global cortical Centiloid. The addition of hypertension status and APOE ε4 status as covariates did not change the results. Using voxel-wise regression models with age, sex, and total gray matter volume as covariates, there were no statistically significant correlations between CBF and global cortical Centiloid at the p≤0.001, k≥50 level. In order to determine whether APOE ε4 status or hypertension status affected the relationship between amyloid or tau PET measures and CBF, a voxel-wise two-way ANCOVA was used, with CBF as the dependent variable, amyloid positivity and APOE ε4 or hypertension status as independent variables, and age, sex, and total gray matter volume as covariates. There was a positive interaction of the amyloid positivity by hypertension interaction on cluster-wise CBF at p≤0.001, k≥50 (Table 6). Spatial patterns of the positive association between the interaction variable and CBF are illustrated by beta maps in Figure 4. CBF in the left visual motor area was positively associated with global cortical Centiloid in hypertensive individuals (R2=0.200) and negatively associated with global cortical Centiloid in normotensive individuals (R2=0.167; Figure 5). Each participant’s CBF values from the significant cluster were extracted and entered into a regression model as the dependent variable with global cortical Centiloid, hypertension status, age, sex, total gray matter volume, and hypertension status by global cortical Centiloid interaction as independent variables. The interaction variable was positively associated with the cluster CBF (β=0.624, t=4.16, p<0.001), while global cortical Centiloid and hypertension were both negatively associated with the cluster CBF. There was no amyloid positivity by APOE ε4 positivity interaction effect on CBF.
Amyloid positivity by hypertension status interaction effect on voxel-wise CBF. p<0.001; k>50. Peak effects reported in MNI coordinates (mm).
Voxel-wise CBF rendering (A) and beta maps (B-D) of the amyloid positivity by hypertension status effect in parietal, temporoparietal, temporal, and occipital regions. Age, sex and total gray matter volume were the model covariates. Rendering (A) displayed at p 0.001, k 50.
Interaction effect of global cortical Centiloid by hypertension status in visual motor and supramarginal CBF as defined by a significant cluster in the voxel-wise interaction analysis. In visual motor and supramarginal regions, CBF is positively associated with global cortical Centiloid in hypertensive individuals and negatively associated with global cortical Centiloid in normotensive individuals. Orange = individuals with hypertension, blue = individuals without hypertension.
4. Discussion
We found that CBF is negatively correlated with MTL tau aggregation in older adults. There was also an interaction effect of global cortical amyloid and hypertension status on CBF in this sample, where global cortical amyloid and CBF were positively correlated in hypertensive individuals and negatively correlated in normotensive individuals. Negative correlations between CBF and MTL tau aggregation were present in the parahippocampal gyrus and hippocampus.
Previous studies concerning the relationship between tau and CBF have reported negative correlations. Unlike amyloid pathology, tau tracks both spatially and temporally with disease severity and brain dysfunction. Therefore, as tau pathology progresses in its characteristic pattern over the disease course, lower CBF progresses similarly (Gomez-Isla et al., 1997; Ashford et al., 2000). Similar to the amyloid literature, animal models of hypoperfusion develop increased tau pathology (Raz et al., 2019; Zhang et al., 2010), and animal models of tau pathology develop hypoperfusion (Bennett et al., 2018; Jaworski et al., 2011; Lourenco et al., 2017; Park et al., 2020). In human studies including either a mix of diagnoses including clinical AD or only clinical AD, there have been findings of negative correlations between tau pathology and CBF in temporal, parietal, and occipital lobes, and entorhinal cortex (Albrecht et al., 2020; Rubinski et al., 2021; Visser et al., 2020). Okamura et al. (2002) reported that the ratio of (increased) tau in cerebrospinal fluid (CSF) to (decreased) posterior cingulate CBF predicted the progression from MCI to AD with 88.5% sensitivity and 90% specificity. Studies in CN individuals report a mix of results, similar to amyloid pathology studies in preclinical AD. Hays et al. (2020) reported a negative correlation between CSF tau and CBF in the anterior cingulate cortex. Stomrud et al. (2012) reported a negative correlation between CSF tau and CBF in the medial frontal lobe, as well as a positive correlation between tau and CBF in the frontotemporal border zone. Another study reported that asymptomatic individuals with the same amount of tau pathology as individuals with MCI had increased CBF in the MTL and thalamus over time (Codispoti et al., 2012). Although there are fewer studies focused on the relationship between tau pathology and CBF compared to amyloid pathology, it appears that a similar pattern of findings is emerging.
Here, we found that MTL CBF correlates negatively with MTL tau pathology. These findings agree with previous reports of negative correlations between CBF and AD pathology in preclinical AD. They also agree with evidence of the temporal and spatial correlation between tau pathology and brain dysfunction in AD. While we did not see positive correlations between CBF and amyloid or tau pathology in the whole sample, as have also been reported in preclinical populations, this may be due to a limited sample size. There also may be distinct time courses for compensation in response to Aβ and tau pathology that are dependent on other pathologies and risk factors. We also found that hypertension was associated with decreased CBF in this sample. This has been reported previously (Alosco et al., 2013). It has been described that chronic hypertension shifts microvascular regulation in favor of constriction, which leads to decreased CBF in hypertensive individuals (Toth et al., 2017). As the two-hit hypothesis of AD posits, vascular risk factors such as hypertension result in cerebrovascular dysfunction that both aggravates and is aggravated by amyloid and tau pathology.
Interacting relationships among multiple risk factors and pathologies in AD have been previously studied; specifically, the effect of APOE ε4 on relationships between CBF and other AD pathologies or symptoms has been reported. Two papers (Wang et al., 2019; Zlatar et al., 2016) reported a positive correlation between CBF and cognition in APOE ε4-negative CN individuals and a negative correlation between CBF and cognition in APOE ε4-positive CN individuals. Hays et al. (2020) found a negative relationship between tau in the cerebrospinal fluid and CBF in the anterior cingulate in APOE ε4-positive but not APOE ε4-negative individuals. APOE ε4 may act either alongside or synergistically with Aβ to exacerbate neurodegeneration (Khan et al., 2017). These interactions are consistent with the established relationships between APOE ε4 and AD pathologies amyloid and tau. APOE ε4 is clearly associated with increased Aβ aggregation, most likely by reducing the clearance of Aβ from the brain and increasing its production (Donahue and Johanson, 2008; Castellano et al., 2011; Wahrle et al., 2007; Mahley et al., 2006). It has also been reported that APOE ε4 carriers have increased levels of tau in the MTL (Jack et al., 2020; Ossenkoppele et al., 2016).
Our finding of positive correlations between Aβ aggregation and CBF in hypertensive individuals was the opposite of the interaction effect we expected hypertension to have, but previous findings of positive associations between Aβ pathology and CBF in nondemented older individuals suggest that this finding could be meaningful. To our knowledge, this is the first report of such an interaction. Not only does this suggest that hypertension status may alter the interpretation of altered CBF as a biomarker, but it also suggests that cerebral vascular function is differentially affected by AD pathology in older adults with and without cardiovascular risk factors during preclinical stages of disease. The direction of the interaction was not expected; individuals with the risk factor of hypertension seemed to have more of a compensatory response (increased CBF) to Aβ pathology. This may suggest that the combined presence of risk factors and pathologies does not always favor loss of function in early disease stages. There is not yet a standard time course of when compensation fails to compensate, and it likely differs for each individual based on comorbidities, cardiovascular health, lifestyle choices like diet and exercise, and demographic, genetic, and environmental variables. Another possible reason that our findings were different than expected is because we defined hypertension as presence or absence of the condition at the time of imaging, but midlife hypertension confers the greatest risk for AD in old age (Lennon et al., 2021). It has also been reported that hypotension in old age, especially following midlife hypertension, is a risk factor for AD as well (Dickstein et al., 2010). As shown by the diverse findings reported in the literature, the relationship between amyloid pathology and CBF in older adults on the AD spectrum is very complex.
Surprisingly, we did not find an interaction effect of hypertension and tau on CBF or of APOE ε4 and either PET metric on CBF. It could be that hypertension and Aβ have relatively more independent effects on the cerebral vasculature than the other risk factor and pathology pairs. In support of this, we did find correlations between global cortical amyloid and APOE ε4 and between MTL tau and hypertension status in this sample. This is also in agreement with the two-hit hypothesis of AD, in which cardiovascular risk and Aβ aggregation culminate to initiate and propagate the pathological processes in AD. At a more liberal statistical significance level, we did find brain regions with APOE ε4 by Aβ and tau interaction effects on CBF. With larger sample sizes, these interactions may prove to be meaningful. It is also possible that these interactions would be detectable and meaningful during later disease stages but not in preclinical stages.
This study was limited by its modest sample size. Some regions of associations between voxel-wise CBF and PET biomarkers did not reach statistical significance, and a larger sample size could have led to more statistically significant results suggesting a clearer picture of correlative patterns. Future studies should use larger sample sizes of individuals with preclinical AD so that statistical power is adequate to elucidate the complex relationships between CBF, amyloid, tau, and other AD-related pathologies. Our sample was also homogenous in terms of ethnoracial and educational backgrounds, and it will be important to utilize diverse samples in future studies. Additionally, we used global cortical Centiloid and MTL tau SUVR as our metrics of PET imaging. It would be ideal to retain the spatial information in both the PET and ASL images in order to determine areas of spatial overlap between AD pathologies and altered CBF. Finally, we defined hypertension by self-report rather than by a threshold blood pressure at the time of scanning, and some hypertensive individuals were taking antihypertensive medications. The addition of blood pressure and use of antihypertensives in our models did not change the results, but future studies with a narrower definition of hypertension may be beneficial.
4.1 Conclusions
Our findings suggest that MTL tau pathology is associated with decreased MTL CBF in older adults. They also suggest a positive interaction of amyloid pathology and hypertension status on occipital CBF; there may be a complex relationship between amyloid pathology and CBF that depends partially on cardiovascular health. Altered CBF is likely both an instigator and an outcome of increased amyloid and tau pathology, and these and other factors interact both spatially and temporally in the environment of an at-risk brain. CBF can be used as a potential biomarker in future studies to identify those at risk so that they can enter clinical trials and/or receive treatment, track disease progression in both treated and untreated individuals, and perhaps help elucidate the complex mechanisms that initiate AD and drive its progression.
Data Availability
All data produced in the present study are available upon reasonable request to the authors.
Funding
This study was funded in part by the following grants from the National Institute on Aging: K01 AG049050, R01 AG061788, P30 AG010133, P30 AG072976, R01 AG019771, R01 AG057739, U01 AG024904, R01 AG068193, T32 AG071444, U01 AG068057, and U01 AG072177 and the National Library of Medicine: R01 LM013463. In kind support for tau PET tracer precursor was provided by Avid Radiopharmaceuticals, a subsidiary of Eli Lilly and Company.
Declarations of Interest
Dr. Apostolova receives support in the form of consulting fees from Biogen, Two Labs, IQVIA, NIH, Florida Department of Health, NIH Biobank, Eli Lilly, GE Healthcare, and Eisai; in the form of payment for lectures, etc. from AAN, MillerMed, AiSM, and Health and Hospitality; and in the form of travel and meeting support from Alzheimer’s Association. She also participates on Data Safety Monitoring or Advisory boards for IQVIA, NIA R01 AG061111, UAB Nathan Shock Center, and New Mexico Exploratory ADRC; in leadership roles for Medical Science Council Alzheimer Association Greater IN Chapter, Alzheimer Association Science Program Committee, and FDA PCNS Advisory Committee; stock or stock options Cassava Neurosciences and Golden Seeds, and receipt of materials, etc. from AVID Pharmaceuticals, Life Molecular Imaging, and Roche Diagnostics.
Dr. Saykin has received support for consultation to Bayer Oncology (Scientific Advisory Board), Eisai (Scientific Advisory Board), Siemens Medical Solutions USA, Inc. (Dementia Advisory Board), and Springer-Nature Publishing (Editorial Office Support as Editor-in-Chief, Brain Imaging and Behavior).
Dr. Wang has received book royalties from American Psychiatric Publishing, Inc., and consultant fees for serving on a Data Safety Monitoring Board.
Dr. Clark receives royalties for two UpToDate articles on aphasia and has done paid consulting for Eli Lilly and Company on the topic of amyloid scans.
Dr. Brosch discloses the following: AbbVie [Progressive supranuclear palsy]; AbbVie [Alzheimer disease]; Avanir [Alzheimer disease]; Biogen [Alzheimer disease]; Biogen [Prodromal or mild Alzheimer disease]; Dominantly Inherited Alzheimer Network [Alzheimer disease]; Eisai [Early Alzheimer disease]; Eisai [Dementia with Lewy bodies]; Eli Lilly [Alzheimer disease]; Eli Lilly [Parkinson disease dementia]; Genentech [Alzheimer disease]; Novartis [Alzheimer disease]; Roche [Alzheimer disease]; Suven [Alzheimer disease].
All other authors report no disclosures.
Acknowledgements
John D. West, MSEng, Indiana University School of Medicine, played a primary role in optimizing and automating the pCASL processing pipeline, and the authors are grateful for his contributions to this research which we dedicate to his memory. We are grateful to the research participants and their families without whose generosity and commitment this study would not have been possible.
Footnotes
Author Email Addresses: cgswinfo{at}iupui.edu, srisache{at}iupui.edu, aavosm{at}iu.edu, radear{at}iu.edu, echumin{at}iu.edu, mdzemidz{at}iupui.edu, yucwu{at}iu.edu, sgao{at}iu.edu, mcdonalb{at}iupui.edu, kkyoder{at}iupui.edu, funverza{at}iu.edu, sophwang{at}iupui.edu, mfarlow{at}iupui.edu, jbrosch{at}iupui.edu, clarkdg{at}iu.edu, lapostol{at}iu.edu, simsjb{at}iupui.edu, dannyjwa{at}usc.edu, asaykin{at}iupui.edu
1 Abbreviations
- Aβ
- amyloid beta
- AAL
- Automated Anatomical Labeling
- AD
- Alzheimer’s disease
- ADNI
- Alzheimer’s Disease Neuroimaging Initiative
- APOE
- apolipoprotein E
- ANCOVA
- analysis of covariance
- ANOVA
- analysis of variance
- ASL
- arterial spin labeled
- CBF
- cerebral blood flow
- CCI
- Cognitive Change Index
- CN
- cognitively normal
- CSF
- cerebrospinal fluid
- FWE
- Family-Wise Error
- FWHM
- full width at half maximum
- MCI
- mild cognitive impairment
- MNI
- Montreal Neurological Institute
- MRI
- magnetic resonance imaging
- MTL
- medial temporal lobe
- pCASL
- psuedocontinuous arterial spin labeled
- PET
- positron emission tomography
- PIB
- Pittsburgh compound B
- ROI
- region of interest
- SCD
- subjective cognitive decline
- SPM
- Statistical Parametric Mapping
- SPSS
- Statistical Package for the Social Sciences
- SUVR
- standardized uptake value ratio
References
