Abstract
Background:
Cerebral microbleeds (CMBs) are a common vascular pathology associated with future intracerebral hemorrhage. Plasma biomarkers of amyloid, tau, and neurodegeneration may provide a screening avenue to identify those with CMBs, but evidence is conflicting.
Objective:
To determine the association between plasma biomarkers (Aβ40, Aβ42, t-tau, p-tau181, p-tau217, neurofilament light chain (NfL)) and CMBs in a population-based study of aging and whether these biomarkers predict higher signal on Aβ-PET imaging in patients with multiple CMBs.
Methods:
712 participants from the Mayo Clinic Study of Aging with T2* GRE MRI and plasma biomarkers were included. Biomarkers were analyzed utilizing Simoa (Aβ40, Aβ42, t-tau, NfL) or Meso Scale Discovery (p-tau181, p-tau217) platforms. Cross-sectional associations between CMBs, plasma biomarkers and Aβ-PET were evaluated using hurdle models and multivariable regression models.
Results:
Among the 188 (26%) individuals with≥1 CMB, a lower plasma Aβ42/Aβ40 ratio was associated with more CMBs after adjusting for covariables (IRR 568.5 95% CI 2.8–116,127). No other biomarkers were associated with risk or number CMBs. In 81 individuals with≥2 CMBs, higher plasma t-tau, p-tau181, and p-tau217 all were associated with higher Aβ-PET signal, with plasma p-tau217 having the strongest predictive value (r2 0.603, AIC –53.0).
Conclusion:
Lower plasma Aβ42/Aβ40 ratio and higher plasma p-tau217 were associated with brain amyloidosis in individuals with CMBs from the general population. Our results suggest that in individuals with multiple CMBs and/or lobar intracranial hemorrhage that a lower plasma Aβ42/Aβ40 ratio or elevated p-tau217 may indicate underlying cerebral amyloid angiopathy.
INTRODUCTION
Cerebral microbleeds (CMBs) are a cerebral small vessel pathology associated with hypertension and cerebral amyloid angiopathy (CAA) pathology. CAA, caused by the deposition of amyloid-β(Aβ) in the cortical and leptomeningeal blood vessels, is associated with intracerebral hemorrhage (ICH), morbidity, and cognitive decline in the elderly [1]. Predicting which individuals may have CAA and be at risk of subsequent ICH is an area of active research.
Currently, neuroimaging features including multiple lobar CMBs or lobar ICH on MRI are key diagnostic features of CAA based on the modified Boston Criteria [1, 2]. However, the search for more sensitive biomarkers of CAA in the form of Aβ-PET imaging and particularly fluid biomarkers are of growing interest due to the potential for more widespread deployment in screening and diagnosis with fluid biomarkers [1, 3–5].
Plasma Aβbiomarkers of CAA are of significant interest in CAA research due to the invasiveness of CSF analysis. Current studies of plasma biomarkers in CAA are limited and conflicting. Some studies have reported positive associations between plasma Aβ40, Aβ42, Aβ42/Aβ40 ratio, tau, and CMBs. However, others have reported no or negative associations [6–12]. Some of these inconsistencies are likely due to small sample sizes and differing methodologies used in the measurement of plasma biomarkers because more sensitive methods such as single molecule array (Simoa) and meso discovery scale (MSD) platforms have recently become more widely available, differences in study populations, and different methodologies for grading CMBs [6, 8]. Additionally, prior studies have not adjusted for presence of renal dysfunction, which has recently been associated with increased plasma amyloid and tau biomarkers [13]. Variability of CAA burden in prior studies may also contribute to differences in results.
Given the inconsistent results in prior studies of plasma AD biomarkers and CMBs we had two primary aims of this study: 1) to determine the association between CMBs and plasma amyloid, tau, and neurodegeneration biomarkers (Aβ40, Aβ42, t-tau, p-tau181, p-tau217, neurofilament light chain (NfL)) in a large population-based study; and 2) to assess whether plasma biomarkers in participants with multiple CMBs predict higher signal on Aβ-PET imaging.
METHODS
Participants
Participants in the Mayo Clinic Study of Aging (MCSA), a population-based study evaluating risk factors for cognitive impairment in Olmsted County, Minnesota residents who underwent MRI with gradient recalled-echo (GRE) sequences and who had stored plasma corresponding to the date of MRI from October 2011 through November 2017 were eligible for inclusion in this study. MCSA participants with amyloid PET and MRI were prioritized for plasma biomarker analyses [13, 14]. All patients without medical contraindication are invited to participate in neuroimaging studies. Medical conditions (hypertension, diabetes, chronic kidney disease) were abstracted from the detailed medical records included in the Rochester Epidemiology Project medical records–linkage system while demographic information and body mass index were determined at the in-person clinical examination [15]. Ever versus never smoking was obtained via self-report. Apolipoprotein E (APOE) genotyping was obtained from a blood sample. MCSA patients with and without biomarker analyses have been previously shown to be of similar age and number of medical comorbidities [13].
Neuroimaging
All MRI examinations were performed at 3T (GE Healthcare) MRI scanners. The complete details of the acquisitions were previously published [16]. A T2* GRE MRI was performed with the following parameters: (repetition time/echo time = 200/20 ms; flip angle = 12°; in-plane matrix = 256×224; phase field of view = 1.00; slice thickness = 3.3 mm; acquisition time, 5 min) [17]. CMBs were identified by trained image analysts and confirmed by a cerebrovascular neurologist or radiologist. CMBs were considered lobar, deep/infratentorial or cerebellar in location [17, 18].
Aβ-PET imaging was performed with Pittsburgh Compound B [19]. PET images were acquired using a PET/CT scanner (GE Healthcare). Details of PET acquisition have been previously published [20]. AβPET images were analyzed with our in-house, fully automated image processing pipeline wherein image voxel values are extracted from automatically labeled regions of interest (ROIs) propagated from an MRI template [21]. Global Aβ-PET SUVr was calculated as the median uptake in the prefrontal, orbitofrontal, parietal, temporal, anterior and posterior cingulate and precuneus ROIs normalized to the median cerebellar crus gray matter.
Plasma biomarker assays
Details of plasma biomarker collection and assays have been previously published [13, 22]. Briefly, participants’ blood was collected in-clinic after an overnight fast, centrifuged, aliquoted, and stored at –80°C. Plasma Aβ40, Aβ42, T-tau, and NfL were measured on the Quanterix HD-1 analyzer using the Simoa® Neurology 3-Plex A (N3PA) (tau, Aβ42, Aβ40) (catalog #101995) or Simoa® NF-light (catalog #103186) Advantage kits. Plasma Aβx40 and Aβx42 were quantified using a common detection antibody (6E10), which binds to the amyloid-βpeptide at an RHD sequence at residues 5–7. Unique C-terminal antibodies were used for Aβ40 (2G3) and Aβ42 (H31L21) [23, 24].
Plasma t-tau was quantified using a capture antibody that binds to the proline-rich P2 region in the mid-domain of the tau protein [13]. The tau calibration curve was generated using a recombinant human tau 381 isoform with a single N-terminal insert and three microtubule binding domain repeats (3R/1N). In the NF-light kits, both the capture and detector antibodies (Uman Diagnostics article #27016-100, #27017-100) bind to the conserved rod domain of the NfL protein [13].
Both p-tau181 and p-tau217 levels were measured on the Meso Scale Discovery (MSD) platform by electrochemiluminescence using proprietary assays developed by Lilly Research Laboratories as previously described and detailed methodology has been previously published [22, 25]. P-tau181 used Biotinylated-AT270 (mIgG1) as the capture and p-tau217 used Biotinylated-IBA493 (mIgG1) as the capture. In this study, both assays used SULFO4G10-E2 (anti-tau monoclonal antibody developed by Lilly Research Laboratories) as the detector [22]. Each assay was calibrated using a unique synthetic p-tau peptide coupled with a polyethylene glycol linker to a second tau peptide matching amino acid 111–130 according to the Tau441 sequence numbering [22].
Standard protocol approvals, registrations, and patient consents
The study was approved by the Mayo Clinic and Olmsted Medical Center Institutional Review Boards. Informed consent was obtained from all participants and/or their proxies for participation in this study.
Statistical analysis
Continuous demographic and clinical data are summarized using means and standard deviations; categorical variables were summarized as counts and percentages. Group comparisons were made with t-tests or Chi Squared tests. Distributions of continuous variables were examined for approximate symmetry and normality using plots; non-normally distributed data were log-transformed. PiB SUVr was reported as means±SDs with the original values for group comparisons while regression was performed using log transformation.
Relationships between CMBs and plasma biomarkers were evaluated using hurdle models. These models consist of two components: the first estimates the relationship between plasma biomarkers and risk of having a CMB (the “hurdle” component) whereas the second estimates the relationship between plasma biomarkers and the number of CMBs among individuals with at least 1 CMB (the “count” component). Each component takes the form of a regression model with predictors and outcomes. Hurdle models are useful in settings where a large proportion of individuals have no CMBs, and are very flexible, allowing predictors in the hurdle and count components to have different contributions. Within each of these two components, 3 separate sets of predictors were utilized. The first predictor set used the plasma biomarker as a single variable. The second predictor set included the plasma biomarker, age, sex, and APOE status as covariates. Finally, the third predictor set included the plasma biomarker, age, sex, APOE status, body mass index (BMI), chronic kidney disease (CKD), hypertension, diabetes, and a history of smoking as covariates. We subsequently fit negative binomial hurdle models predicting total CMBs, using age, sex, APOE ɛ4, CKD, BMI, hypertension, diabetes, smoking, p-tau181, p-tau217, Aβ42/Aβ40, Nfl, and t-tau as candidate predictors. We formed parsimonious models using backwards elimination, allowing the predictors in the hurdle and count components to differ. Due to comparatively large coefficients with other plasma biomarkers, Aβ42/Aβ40 ratio was scale reduced tenfold and reported as –10Aβ42/Aβ40 throughout the manuscript.
We were interested in evaluating whether there were relationships between plasma biomarkers and AβPiB-PET imaging in those with multiple CMBs, which may be a biomarker of CAA pathology. In this subgroup analysis, only individuals with≥2 CMBs were included with log (PiB) as the outcome variable and with plasma biomarker, age, sex, and APOE ɛ4 status as the predictor variables. Initially these were performed with each biomarker of interest as the predictor. We subsequently performed additional regression analysis using age, sex, APOE ɛ4, p-tau181, p-tau217, Aβ42/Aβ40, Nfl, and t-tau as candidate predictors within a single model. We again formed parsimonious models using backwards elimination to determine the remaining significant predictors. For these regression analyses, the coefficient, p-value, r2 and Akaike information criterion (AIC) were reported. We tested for curvature using quadratic terms for the plasma biomarkers, but only p-tau217 was significant. Significance was set at α< 0.05.
RESULTS
Of the 712 patients in our cohort, 188 (26%) had≥1 CMB. For those with≥1 CMB, the average number of CMB was 2.2±2.9 (mean±SD), with an average of 1.7±2.7 of lobar location and 0.3±0.7 of deep location. Please see Table 1 for full demographic details stratified by presence of having a CMB. Participants with a CMB, compared to those without, were older, more likely to be male and to have hypertension, a history of a stroke, cognitive impairment, and cardiac/metabolic conditions. APOE status and presence of chronic kidney disease did not differ between the groups.
Demographic and plasma biomarker comparisons in individuals with and without cerebral microbleeds
*p-values for differences between groups come from a t-test for the continuous variables and a chi-squared test for the categorical variables. Continuous variables are presented as means (standard deviation). @Summated score indicating the presence or absence of: hypertension, hyperlipidemia, cardiac arrhythmias, coronary artery disease, congestive heart failure, diabetes, and stroke. APOE E4, apolipoprotein E4 allele present; PIB, Pittsburg Compound B; SUVr, Standardized uptake value ratio; MRI, magnetic resonance imaging; p-tau, phosphorylated Tau; Aβ, amyloid-β; NfL, neurofilament light chain. †These were analyzed with a log transformation due to skewness with the distribution.
Plasma Aβ40 (293.0±87.8 versus 266.8±79.5, p < 0.001) and Aβ42 (9.42±3.49 versus 8.82±3.10, p = 0.034) were higher in those with CMBs compared with those without; the Aβ42/40 ratio did not differ between groups. P-tau181 (1.48±0.90 versus 1.33±1.69, p = 0.004) and p-tau217 (0.23±0.18 versus 0.20±0.30, p = 0.001) were both significantly higher in those with CMBs while t-tau was not. Finally, plasma NfL was higher in individuals with CMBs versus without (26.6±16.1 versus 20.7±16.7, p < 0.001).
Hurdle models
We evaluated the risk of having a CMB as well as the number of CMBs and plasma biomarkers with two separate hurdle models. The hurdle component associations between plasma biomarkers and risk of having a CMB can be found in Table 2, while the count component (i.e., number of CMBs) associations between plasma biomarkers and number of CMBs in those with any CMB can be found in Table 3. There was no association between having a CMB with plasma –10Aβ42/40 ratio. In contrast, in those with at least one CMB, a lower –10Aβ42/40 ratio was strongly associated with having more CMBs in multivariable analyses (IRR 568.5 95% CI 2.8–116,127) along with male sex (IRR 2.1 95% CI 1.04–4.2) and hypertension (IRR 3.1 95% CI 1.2–4.3). Neither Aβ40 nor Aβ42 were associated with risk of having a CMB in multivariable analysis. However, higher Aβ42 in those with≥1 CMB was associated with reduced risk of having more CMBs (IRR 0.875, 95% 0.78–0.983) whereas male sex (IRR 2.465 95% CI 1.21–5.02) and hypertension (IRR 3.445, 95% CI 1.404–8.454) predicted more CMBs in this model while adjusting for age, APOE4, CKD, BMI, diabetes, and history of smoking.∥Similarly, plasma t-tau and NfL were not associated with risk of having a CMB in multivariable analysis. In those with≥1 CMB, higher t-tau (IRR 0.745 95% CI 0.56–0.989) was associated reduced risk of having another CMB whereas both male sex and hypertension were associated with increased risk of another CMB. Higher NfL predicted a reduced risk of having additional CMBs (IRR 0.964 95% CI 0.936–0.992) in those with≥1 CMB in the adjusted model while male sex and hypertension again were associated with higher risk of more CMBs. Neither plasma p-tau181 nor p-tau217 were associated with risk or number of CMBs in any of the models.∥After forming parsimonious models using backwards elimination with all biomarker predictors and variables for risk of having a CMB, only age (OR 1.085 (95% CI 1.060–1.111)) and male sex (OR 1.510 (95% CI 1.047–2.178)) were significant. Parsimonious models utilizing backwards elimination for predicting number of CMBs in those with≥1 CMB largely confirmed the findings of the models with individual plasma biomarker predictors. In this model a lower –10Aβ42/40 ratio again most strongly predicted more CMBs (IRR 535.9 (95% CI 2.493–115197)) with NfL again predicting fewer CMBs (IRR 0.966 (95% CI 0.938–0.994)) while t-tau was no longer significant in the parsimonious model. Age (IRR 1.073 (95% CI 1.017–1.131)), male sex (IRR 2.255 (95% CI 1.151–4.416)) and hypertension (IRR 2.494 (95% CI 1.099–5.659)) all predicted more CMBs in this model as well (Table 3).∥
Relationships between plasma amyloid biomarkers and having a cerebral microbleed
*–10Aβ42/Aβ40 represents a scaled value reduced tenfold due to large coefficients. Bold values indicate significant plasma biomarker associations. OR, odds ratio; BMI, body mass index; CKD, chronic kidney disease; APOE, apolipoprotein E4 allele presence; p-tau, phosphorylated tau; Aβ, amyloid-β; NfL, neurofilament light chain.
Relationships between plasma amyloid biomarkers and number of CMBs in individuals with any CMB
*–10Aβ42/Aβ40 represents a scaled value reduced tenfold due to large coefficients. Bold values indicate significant plasma biomarker associations. IRR, Incidence rate ratio; BMI, body mass index; CKD, chronic kidney disease; APOE, apolipoprotein E4 allele presence; p-tau, phosphorylated Tau; Aβ, amyloid-β; NfL, neurofilament light chain.
Participants with multiple CMBs
81/188 (43%) subjects with CMBs had multiple CMBs (i.e., ≥2 CMBs). Multivariable regression models evaluated relationships between plasma biomarkers and log (PiB) PET signal in patients with multiple CMBs. After adjusting for age and sex, plasma t-tau, p-tau181, p-tau217, and APOE ɛ4 all predicted higher log (PiB) (Table 4). However, increasing p-tau217 (r2 0.603, AIC –53.0) was most strongly associated with increasing log (PiB) with the best fit of all the models in those with multiple CMBs. As an illustration, predicted log (PiB) values for a typical 82-year-old individual increases as p-tau217 increases, and the rate of increase slows as p-tau217 gets larger (Fig. 1). Plasma Aβ40, Aβ42, Aβ42/Aβ40 ratio, and NfL did not predict log (PiB) in individuals with≥2 CMBs. In the single parsimonious model of all plasma biomarkers predicting log (PiB), p-tau217 remained the strongest predictor of log (PiB) while higher NfL and t-tau predicted lower log (PiB) but to a lesser degree than p-tau217 (R2 0.63 AIC –31.0, Table 4).
Associations between plasma biomarkers and log (PIB) PET in individuals with multiple CMBs
*–10Aβ42/Aβ40 represents a scaled value reduced tenfold due to large coefficients. AIC, Akaike information criterion; APOE, apolipoprotein E4 allele presence; PIB, Pittsburg Compound B; SUVr, Standardized uptake value ratio; MRI, magnetic resonance imaging; p-tau, phosphorylated tau; Aβ, amyloid-β; NfL, neurofilament light chain.
Predicted log(PiB) versus p-tau217in individuals with≥2 CMBs.
DISCUSSION
The main finding of our study is that in individuals with≥1 CMB, a lower Aβ42/Aβ40 ratio was associated with more CMBs and had a more robust effect than age, male sex, hypertension, or any other plasma biomarker. In contrast, Aβ42/Aβ40 ratio did not predict the presence of a CMB . These findings are consistent with pathology studies showing the greater the number of CMBs, the more likely CAA pathology was present at autopsy [26]. This suggests a lower Aβ42/Aβ40 ratio may be an antemortem biomarker for underlying CAA in individuals with multiple CMBs. However, it is well-established that a lower plasma Aβ42/Aβ40 ratio is associated with a higher amyloid cortical burden. In this case, is likely that lower Aβ42/Aβ40 ratio reflects cerebral amyloidosis in general rather than being specific vascular amyloid pathology of CAA [27]. This also explains the lack of association between lower Aβ42/Aβ40 ratio and risk of having a CMB. Individuals with a single CMB are more likely to have an alternative cause of their CMB such as hypertension or lacunar infarct whereas those with more than one CMB are more likely to have CAA.
Our results agree with results from the majority of other studies which have also found an inverse relationship between Aβ42/Aβ40 ratio and CMBs [6, 9]. In the Framingham study, the finding was likely driven by plasma Aβ40 because the authors also found an association between higher plasma Aβ40 and lobar CMBs [6]. Both the Framingham and the current study utilized a Simoa platform, but participants were much younger in the Framingham cohort. Our cohort may have more comorbid AD pathology and/or white matter disease burden related to CAA or small vessel disease than the Framingham cohort, possibly explaining differences between studies. Only one study has reported a higher Aβ42/Aβ40 ratio has been reported in those with CMBs compared with those who did not have CMBs, although this study did not model relationships between CMB number and Aβ42/Aβ40 ratio [28].
Another possible contributor to the difference from prior studies is that others have utilized antibodies targeting the full-length protein (Aβ1–40 and Aβ1–42) which are more specific for AD whereas our study utilized antibodies to Aβx40 and Aβx42. Theoretically, antibodies to Aβx40 and Aβx42 may bind more vascular amyloid than antibodies targeting the full-length protein although studies targeting this hypothesis directly would be needed to answer this question [6, 29]. Hereditary cerebral hemorrhage with amyloidosis-Dutch type (HCHWA-D) represents a more “pure” form of CAA and may provide additional insight as to how plasma Aβbiomarkers behave in relationship to CMBs. Plasma Aβ40 and Aβ42 are both lower in presymptomatic HCHWA-D mutation carriers compared with non-mutation carriers. Further, Aβ40 more than Aβ42 decreases over longitudinal follow-up in mutation carriers [8]. Unfortunately, this study did not report Aβ42/Aβ40, ratio so whether a lower Aβ42/Aβ40 is seen in HCHWA-D carriers requires further evaluation.
The relationship between plasma phosphorylated tau and CMBs is not well-studied. We found that p-tau181 and p-tau217 were higher in those with CMBs, although we did not find an association with the presence or number of CMBs. However, in individuals with multiple CMBs, increasing plasma t-tau, p-tau181, and p-tau217 along with APOE ɛ4 status predicted increasing PiB-PET, with p-tau217 being the only factor associated with increasing PiB-PET in the model adjusting for all plasma biomarkers This is consistent with prior studies showing both plasma p-tau181 and p-tau217 are elevated in individuals with positive PiB-PET scans [30]. As more lobar CMBs increase the probability of underlying CAA, p-tau181, and p-tau217 could be a potential non-invasive biomarker of cerebral amyloid deposition in individuals with multiple CMBs but is not specific to CAA [26]. In other words, among individuals with multiple CMBs, higher p-tau181 and p-tau217 could suggest CAA as the etiologic factor of the CMBs in these individuals. However, as we have shown that pathologic CAA does not significantly contribute to PiB-PET when accounting for Aβplaques, our results may simply reflect co-morbid AD pathology [3, 29]. Additional studies of higher CMB burden or ICH may provide additional insight into the relationships between phosphorylated tau and PiB-PET in these patients.
Previous work has shown both plasma t-tau and NfL are higher in CAA compared with controls and t-tau is associated with CMBs [6, 9]. Plasma NfL also predicted higher risk of recurrent hemorrhage in those with probable CAA [9]. We also found that NfL was higher in those with CMBs than those without. However, in our study higher plasma t-tau and NfL were associated with reduced risk of having more CMBs in individuals with≥1 CMB in multivariable analyses, although in the model with all plasma biomarkers only NfL continued to predict fewer CMBs. This is somewhat counter intuitive as NfL is thought to be a marker of axonal injury and/or neurodegeneration which would be expected to be higher in those with more than one CMB [31]. It is possible that individuals with more advanced stages of dementia or cerebrovascular disease may have not undergone imaging studies due to higher levels of disability, thus did not undergo plasma analyses but would be expected to have higher levels of NfL. As such, our sample may have been biased towards overall lower levels of NfL and influenced these results. Given the limited prior data on the associations between both plasma t-tau and NfL with CMBs, further studies are needed to better evaluate this relationship. Regardless, the impact of Aβ42/Aβ40 ratio on CMB number is order of magnitudes higher than NfL, so our NfL results may reflect statistical, but not clinical significance.
Limitations
We utilized a population-based study of aging of predominantly Caucasian individuals who had both MRI and plasma AD biomarkers available. As such, the number of CMBs in our cohort was relatively low and our results may not be generalizable to individuals with lobar ICH, hereditary CAA, or non-Caucasian individuals. Further, our population with CMBs was older and had frequent markers of co-morbid AD pathology which may have confounded our results and limit comparison with prior studies [6, 33]. However, as CMBs are more common with increasing age and likely to be present in those with co-morbid AD pathology, our population is an accurate representation of the population of interest for plasma AD biomarkers as a marker of CMBs. Few prior studies have shown elevated plasma Aβ42 in individuals with multiple intracerebral hemorrhages compared with a solitary hemorrhage. Elevated plasma Aβ42 in individuals with probable CAA and acute lobar hemorrhage has also been associated with a poorer prognosis [11, 34]. While we did find an elevated plasma Aβ42 in individuals with CMBs, none of our patients had symptomatic ICH so we were unable to evaluate this relationship. It should be noted that in essentially all models involving individuals with any CMB, both hypertension and male sex predicted more CMBs, both of which are well established risk factors for CMBs. These should be considered when evaluating any association between plasma biomarkers and CMBs, but neither of these were as strongly predictive as lower Aβ42/Aβ40 ratio [28, 35]. Other factors associated with CMBs including severity of hypertension, lacunar infarcts and white matter hyperintensities were not systematically assessed in the current study limiting our ability to address these potential confounders. Finally, additional markers of cerebrovascular pathology such as glial fibrillary acidic protein and soluble PDGFRβare of growing interest in CMB research but were not available in the samples utilized in the current study.
Conclusions
Plasma biomarkers, particularly a lower Aβ42/Aβ40 ratio may be a marker of CAA in individuals with multiple CMBs. Given the potential for confounding by co-morbid AD pathology as well as small vessel ischemic disease, the utility of these plasma biomarkers may be less optimal in mild CAA. Plasma ptau181 and ptau217 predicted higher Aβ-PET signal in those with a higher burden of CMBs suggesting these could be a useful biomarker for co-existing cerebral amyloid pathology suggesting CAA, but further confirmatory studies are needed. Future studies should focus on plasma amyloid biomarkers in the setting of greater CMB burden and those with symptomatic ICH to better evaluate the utility of plasma biomarkers in CAA and whether they may have predictive value for future hemorrhage.
Footnotes
ACKNOWLEDGMENTS
Funding for this study was provided by grants from the National Institutes of Health (RF1 AG069052-01A1, U01 AG006786, R37 AG011378, R01 NS097495, R01 AG041851, and P30 AG062677), the GHR Foundation, Elsie and Marvin Dekelboum Family Foundation, Alexander Family Alzheimer’s Disease Research Professorship of the Mayo Clinic, Liston Award, Schuler Foundation and Mayo Foundation for Medical Education and Research. This study was made possible using the resources of the Rochester Epidemiology Project, which is supported by the National Institute on Aging of the National Institutes of Health under Award Number R01 AG034676.
