Lacunes, Microinfarcts, and Vascular Dysfunction in Cerebral Amyloid Angiopathy

Abstract

Objective

To analyze the relationship of lacunes with cortical cerebral microinfarcts (CMIs), to assess their association with vascular dysfunction, and to evaluate their effect on the risk of incident intracerebral hemorrhage (ICH) in cerebral amyloid angiopathy (CAA).

Methods

The count and topography of lacunes (deep/lobar), CMIs, and white matter hyperintensity (WMH) volume were retrospectively analyzed in a prospectively enrolled CAA cohort that underwent high-resolution research MRIs. The relationship of lacunes with CMIs and other CAA-related markers including time to peak (TTP) of blood oxygen level–dependent signal, an established measure of vascular dysfunction, was evaluated in multivariate models. Adjusted Cox regression models were used to investigate the relationship between lacunes and incident ICH.

Results

The cohort consisted of 122 patients with probable CAA without dementia (mean age, 69.4 ± 7.6 years). Lacunes were present in 31 patients (25.4%); all but one were located in lobar regions. Cortical CMIs were more common in patients with lacunes compared to patients without lacunes (51.6% vs 20.9%, p = 0.002). TTP was not associated with either lacunes or CMIs (both p > 0.2) but longer TTP response independently correlated with higher WMH volume (p = 0.001). Lacunes were associated with increased ICH risk in univariate and multivariate Cox regression models (p = 0.048 and p = 0.026, respectively).

Conclusions

Our findings show a high prevalence of lobar lacunes, frequently coexisting with CMIs in CAA, suggesting that these 2 lesion types may be part of a common spectrum of CAA-related infarcts. Lacunes were not related to vascular dysfunction but predicted incident ICH, favoring severe focal vessel involvement rather than global ischemia as their mechanism.

Sporadic cerebral amyloid angiopathy (CAA) is a common small vessel disease in the elderly, pathologically characterized by the accumulation of β-amyloid protein in cortical and leptomeningeal vessels.^1,2 CAA can result in lobar intracerebral hemorrhage (ICH) and lobar cerebral microbleeds (CMBs) as well as ischemic lesions such as white matter hyperintensities (WMH) and cortical cerebral microinfarcts (CMIs).³ Lacunes, defined pathologically as a small infarct with cavitation, resulting from occlusion of a penetrating branch of an artery,⁴ are also common in patients with cerebral small vessel disease.⁵ It was recently demonstrated that lacunes located in lobar regions were more prevalent in patients with CAA-related ICH compared to patients with hypertensive deep ICH (HTN-ICH).^6,7 Lacunes and CMIs are discrete brain lesions presumed to be related to ischemia, but their association and mechanisms have not been investigated in patients with CAA. Potential mechanisms might include discrete tiny vessel occlusions or global vascular dysfunction in CAA; the latter can be quantified using the blood oxygen level–dependent (BOLD) signal obtained from fMRI.^8,9

In this study, we hypothesized that lacunes and CMIs share similar mechanisms, therefore patients with CAA with lacunes are more likely to have CMIs. We also hypothesized that vascular dysfunction, a major contributor to global brain ischemia in CAA, is not associated with lacune and CMI formation, which we see as being preferentially due to focal vessel pathologies. Finally, we explored the potential effect of the presence of lacunes on the risk of incident ICH in patients with CAA.

Methods

Study Participants

We retrospectively analyzed data that were obtained from an ongoing single-center prospective longitudinal study on sporadic CAA. The patients were enrolled between May 2006 and May 2018. The patients with a diagnosis of probable CAA according to the Boston criteria were prospectively included in the study.¹⁰ The inclusion criteria were age ≥55 years; multiple macro/microhemorrhages restricted to lobar, cortical, or corticosubcortical regions; and absence of other cause of hemorrhage. Patients with CAA presented with symptomatic ICH or with other symptoms such as transient focal neurologic symptoms, gait problems, or other neurologic problems. Patients who presented with ischemic stroke, or those who had mild cognitive impairment, dementia, or any other neurodegenerative condition, were excluded with appropriate clinical and imaging evaluation. Patients underwent multimodal high-resolution neuroimaging including both quantitative structural MRI and physiologic MRI, as detailed in the following. The study was performed in accordance with the guidelines of the local institutional review board and informed consent was obtained from all participants.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was performed with the approval of and in accordance with the guidelines of the institutional review board of Massachusetts General Hospital. All participants provided written informed consent. There were no photographs, videos, or other information of any recognizable person.

Structural MRI

Patients underwent a research MRI at 1.5T (Siemens Healthineers, Erlangen, Germany) or 3T (Magnetom Prisma-fit; Siemens Healthineers) field strength. The 1.5T and 3T MRI protocols included a 3D T1-weighted multiecho magnetization-prepared rapid gradient-echo sequence, a 3D fluid-attenuated inversion recovery (T2/FLAIR) scan, and a 3D susceptibility-weighted imaging (SWI) scan, all using similar measures as described previously.¹¹

All images were reviewed retrospectively to identify visually the established neuroimaging markers of CAA in these prospectively acquired data. Lacunes were defined based on the STRIVE criteria as round or ovoid lesion with a diameter between 3 mm and 15 mm, hyperintense on T2-weighted images, hypointense on T1-weighted images, and hypointense on FLAIR images with a surrounding rim of hyperintensity.⁵ The lesions matching these criteria were also reviewed on SWI sequences to distinguish them from CMBs (figure 1). The number of lacunes was recorded and classified topographically as following: deep if the lesion was located in thalamus, basal ganglia (BG), internal capsule, external capsule, brainstem, or deep cerebellar regions; lobar if the lesion was located in subcortical nondeep brain regions including centrum semiovale (CSO), frontal, parietal, insular/subinsular, temporal, or occipital lobes.

Figure 1. Lobar Lacune.

Lacune in lobar region is shown as (A) hypointense lesion with a surrounding rim of hyperintensity on fluid-attenuated inversion recovery, (B) hyperintense lesion on T2-weighted MRI, (C) hypointense lesion on T1-weighted MRI, and (D) no corresponding lesion on susceptibility-weighted imaging (SWI).

Cortical CMIs were assessed manually by using MeVisLab (version 2.17, MeVis Medical Solutions AG, Bremen, Germany) according to previously published criteria.¹² Briefly, possible cortical CMIs were identified on T1-weighted images as lesions that were hypointense, <4 mm in diameter, restricted to the cortex, perpendicular to the cortical surface, and distinct from perivascular spaces and ICH. The lesion was excluded if it appeared hypointense on SWI sequences. The remaining possible cortical CMIs were rated as definite cortical CMIs if they were hyperintense or isointense on FLAIR images (figure 2).

Figure 2. Cortical Cerebral Microinfarct.

A cortical cerebral microinfarct is shown in the cerebral cortex as (A) hyperintense on fluid-attenuated inversion recovery and (B) hypointense lesion on T1-weighted MRI.

WMH patterns were grouped into 4 categories according to a previously published study.¹³ These patterns were (1) multiple subcortical spots; (2) peri-BG; (3) large posterior subcortical patches; and (4) large anterior subcortical patches. Enlarged perivascular spaces (EPVS) in the CSO were assessed and graded as previously defined.¹⁴

ICHs and CMBs were identified and counted on SWI sequences, as described previously.¹⁵ The presence of cortical superficial siderosis (cSS) was assessed and classified based on previously published criteria.¹⁶ Average cortical thickness, total supratentorial brain volume (TBV; includes supratentorial gray and white matter volumes), total WMH volume, and estimated total intracranial volume (eTIV) were calculated using FreeSurfer software suite (surfer.nmr.mgh.harvard.edu; version 5.3.0) as previously described.¹⁷ TBV and WMH volumes were expressed as percentage of eTIV (pTBV and pWMH) for the correction of measurements according to the variable head size of the individuals. In patients with ICH, the volume estimates of the ICH-free hemispheres were used and multiplied by 2.

Functional MRI Acquisition and Analysis

BOLD fMRI, using a block design with a visual stimulus, was performed during the research scans at both 1.5T and 3T in 78 patients. The patients without fMRI did not differ for baseline characteristics and CAA-related imaging markers from the fMRI cohort (p > 0.2 for all comparisons). fMRI was acquired at both field strengths with a traditional single-shot echoplanar image covering the occipital lobe and the following 1.5/3T measures: echo time 38 ms/30 ms, repetition time 1,500 ms/2,000 ms, 17/29 slices, 128/96 dynamics, and total scan duration of 192 seconds. Briefly, the visual stimulus consisted of 16 blocks of an 8-Hz flashing radial black and white checkerboard pattern for 20 seconds alternated with 28 seconds of gray screen. Functional volumes were preprocessed as previously described in detail.⁹ Using an in-house program developed in Python (Python Software Foundation, Python Language Reference, version 3.7; python.org), an average block for each participant was then calculated representing the BOLD response of active voxels. In order to address the possible role of CAA-related tissue damage on the fMRI response, a relatively stringent uncorrected threshold of p < 0.0001 was used to exclude nonresponding voxels for each individual. By doing so, we excluded nonresponding voxels (p > 0.0001) that might occur in the occipital lobe, principally in regions of occipital ICH, thereby mitigating any effect that ICH could introduce into fMRI analyses. Time to peak (TTP) was calculated using a nonlinear least squares curve fitting algorithm (scipy.optimize.curve_fit, Python) and defined as the time from the beginning of the block (t = 0 seconds) to the model-determined peak.⁹

Follow-Up

The patients enrolled in this prospective study consented to be contacted by phone every 6 months, during which documentation is systematically obtained regarding whether they sustained the prespecified outcome measures of new ICH, as extensively described in previous publications.^18,19 If a patient could not be reached, health care proxies were contacted. Such follow-up data were available for 106 patients. Baseline characteristics and CAA-related markers were not different between these 106 patients included into the longitudinal cohort and the remaining 16 patients without follow-up data (p > 0.2 for all comparisons). The presence of incident ICH was defined as a symptomatic stroke with neuroimaging evidence of ICH. For patients who had multiple ICHs during follow-up, the first ICH was used as the time point of outcome event.

Statistical Analysis

Patients with and without lacunes were compared for demographics, vascular risk factors, and neuroimaging markers. Discrete variables were presented as count (%) and continuous variables as mean (±SD) or median (interquartile range [IQR]), as appropriate based on their distribution. Categorical variables were analyzed using Fisher exact test and continuous variables using the independent-samples t test (for normal distributions) or Mann-Whitney U test (for non-normal distributions). A multivariate logistic regression model was used to assess the independent association between the presence of lacunes and CMIs. The relationship between lacune and CMI counts was evaluated with bivariate correlation analysis. The association of TTP with lacunes and CMIs as well as with pWMH volume was assessed using both univariate and multivariate models. All multivariate regression models included variables that showed an association with a p value < 0.2 in univariate analyses as well as those considered to be relevant based on previous studies (age, sex, presence of hypertension, presence of ICH, and MRI field strength). We checked the assumptions required for multiple linear regression models using scatterplots, residual analysis, and variance inflation factor values and these analyses showed linear relationship between the outcome variable and the independent variables, normal distribution of residuals, and no multicollinearity. We also performed sensitivity analyses after excluding the only patient who had a deep lacune and all of the reported results were unchanged.

In the follow-up cohort, the mean follow-up time was calculated and the incidence rates of ICH per 100 person-years of follow-up were determined. We used univariate Cox proportional hazards models to calculate the crude hazard ratios for the risk of incident ICH of each potential predictor including demographics, vascular risk factors, and neuroimaging markers. Adjusted Cox regression models were performed for ICH occurrence to evaluate differences between patients with and without lacunes. The Cox model included the variables with a p value < 0.2 in univariate analyses as well as those variables considered to be relevant. For the Cox models, we tested the proportional hazard assumption using graphical checks and Schoenfeld residuals–based tests. Statistical analyses were performed using SPSS (Statistical Package for Social Sciences for Windows version 23 software). A p value less than 0.05 was considered statistically significant and all significance tests were 2 tailed.

Data Availability

Any data not published within the article are available by request from a qualified investigator.

Results

The study cohort consisted of 122 patients with a mean age of 69.4 ± 7.6 years; 89 (73%) were men. Regarding vascular risk factors, hypertension was present in 64 (52.5%), hyperlipidemia in 51 (41.8%), coronary artery disease in 12 (9.8%), atrial fibrillation in 12 (9.8%), and diabetes in 8 (6.6%) patients. Eighty-four (68.9%) of 122 patients presented with ICH; the remaining 38 (31.1%) presented with lobar CMBs only, all fulfilling the Boston criteria for a diagnosis of probable CAA. Of all patients, 76 (62.3%) underwent 1.5T MRI and the remaining 46 (37.7%) 3T MRI scanning. Demographics and vascular risk factors were not statistically different between patients who underwent 1.5T MRI and those who underwent 3T MRI.

Lacunes were present in 31 (25.4%) patients, all but one located in lobar as opposed to deep brain regions. The only patient with a deep lacune presented with a lobar ICH and 6 strictly lobar microbleeds and was included in all cross-sectional analyses but lost to follow-up and thus not included in longitudinal analyses presented below. None of the patients had any lacune in the posterior fossa region including brainstem and cerebellum. The count of lacunes ranged between 1 and 3. Among 76 patients who underwent a 1.5T MRI, 22 had (28.9%) lacunes, as compared to 9 of 46 patients (19.6%) who underwent a 3T MRI scan (p = 0.289).

Association of Lacunar Infarcts With Risk Factors and Structural Neuroimaging Markers

The patients with lacunes and the patients without lacunes did not differ in terms of demographics and vascular risk factors (table 1). Thirty-five patients (28.7%) had cortical CMIs, ranging between 1 and 15 CMIs. Among 76 patients who underwent a 1.5T MRI, 22 (28.9%) had CMIs, as compared to 13 of 46 patients (28.3%) who underwent a 3T MRI scan (p = 1.0).

Table 1.

Comparison of Demographics and Vascular Risk Factors Between Patients With Lacunes and Those Without Lacunes

Patients who had one or more lacunes were more likely to have cortical CMIs (51.6%) when compared to patients without lacunes (20.9%, p = 0.002). There were more patients with ICH among the ones with lacunes, although this association did not reach statistical significance (80.6% vs 64.8%, p = 0.120). Presence of lacunes was not associated with the number of CMBs, the presence of cSS, the degree of EPVS in CSO, the pattern of WMH, pWMH volume, pTBV, or cortical thickness (table 2). The association between presence of lacunes and cortical CMIs remained significant in a multivariable model adjusting for age, sex, hypertension, MRI field strength, and the presence of ICH (odds ratio 4.3, 95% confidence interval [CI] 1.7–10, p = 0.002). In bivariate correlation analysis, lacune count was correlated with cortical CMI count (ρ = 0.308, p = 0.001).

Table 2.

Comparison of Markers of Cerebral Amyloid Angiopathy Between Patients With Lacunes and Those Without Lacunes

Association of fMRI Measures of Vascular Dysfunction With Lacunes and Cortical CMI

We calculated TTP of the BOLD response from 78 patients who received fMRI (43 underwent 1.5T and 35 underwent 3T MRI). As presented in the Methods, there was no difference in any of the baseline characteristics between patients who had or did not have fMRI. There was no statistically significant difference in TTP between patients with 1.5T and those with 3T MRI (9.7 ± 2.7 ms vs 8.9 ± 2.1 ms, p = 0.151). TTP did not show statistically significant differences when compared between patients with and without lacunes or patients with and without cortical CMIs (all p > 0.2). Multivariate regression models did not suggest an association between TTP and presence of either lacune (p = 0.898) or cortical CMI (p = 0.965). Longer TTP response was independently associated with higher pWMH volumes, even after adjustment for age, sex, hypertension, presence of ICH, and MRI field strength (β = 0.334, p = 0.001), confirming previously reported associations.^8,9

Association of Lacunes With Outcome Measures During Follow-Up

Of 106 patients who had follow-up, 25 patients (23.1%) had lobar lacunes. Nineteen patients (17.6%) experienced ICH during a median follow-up of 30 months (IQR 14–55 months). All incident ICHs were in lobar locations typical for CAA. Lobar ICH occurred in 34.6% of patients with lacunes and 12.5% of patients without lacunes (p = 0.017). Incident ICH was in the same hemisphere as the baseline lacunes in 4 of 9 patients who had lobar lacune at baseline and lobar ICH during follow-up. Of these patients, 2 had incident ICH and lobar lacune in the same lobar region, while they were in different lobar regions in the remaining 2 patients. In univariate Cox analyses, presence of lobar lacunes and presence of cSS were significantly associated with incident ICH risk (table 3). Using multivariate Cox regression analysis, the presence of lobar lacunes was independently associated with increased ICH risk (table 4). As the inclusion of 8 relevant potential predictors can result into overfitting of a Cox model with 19 events, we repeated the same analysis using an automated variable selection method (backward stepwise) and the presence of lacunes was again a significant predictor of ICH occurrence during follow-up (hazard ratio 1.68, 95% CI 1.89–15.4, p = 0.002). Figure 3 shows the survival curves for the occurrence of ICH in patients with lacunes and those without lacunes.

Table 3.

Univariate Cox Regression Analyses of Predictors for Occurrence of Intracerebral Hemorrhage (ICH)

Table 4.

Incidence Rates and Hazard Ratios for the Occurrence of Intracerebral Hemorrhage (ICH) During Follow-Up

Figure 3. Survival Curves for the Occurrence of intracerebral hemorrhage (ICH).

Unadjusted survival curves of patients with lacunes and those without lacunes for the occurrence of ICH.

Discussion

The main findings of our study are that lacunes are associated with the presence of cortical CMIs and neither lacunes nor cortical CMIs are associated with vascular dysfunction measured by fMRI. Moreover, the presence of lacunes independently predicted the risk of subsequent ICH in CAA. We also confirmed the findings of a previous study reporting that lacunes are almost exclusively found in subcortical white matter in CAA as opposed to deep gray structures.

Lacunes are believed to be the consequences of ischemia caused by occlusion of a penetrating branch of a cerebral artery.⁴ Although they have been recognized as one of the imaging markers of arteriolosclerosis, recent studies of the topographic distribution of lacunes in different types of small vessel diseases have suggested that they may be features of CAA as well as arteriolosclerosis.^6,7 Our results that show that lacunes are almost exclusively found in lobar (nondeep) brain regions confirm the previous findings obtained in different cohorts. Our cohort included not only patients with CAA-related ICH but also patients who presented with CMBs only (about one third of the cohort) and using this as a binary variable did not change any of the results obtained in multiple regression models. This finding raises the important possibility that lobar lacunes might be candidate markers for diagnosing CAA in the absence of (or prior to) ICH. This hypothesis will need to be tested in larger cohorts with lobar microbleed only CAA that also include non-CAA small vessel diseases and other elderly populations.

Cortical CMIs are pathologically defined as microscopic regions of cellular death or tissue necrosis, sometimes with cavitation.³ One question that we aimed to address was whether lobar lacunes and cortical CMIs tend to co-occur in CAA, an association that would support the possibility that they represent lesions of different sizes related to the same pathologic pathway. In recent years, there has been growing interest in cortical CMIs in older individuals as well as in CAA, as CMIs are more commonly detected in vivo using high-resolution MRI scans.¹² The exact pathophysiologic origin of CMI is uncertain but there are likely many factors that can cause CMIs, including occlusive vascular disease, microembolism, hypoperfusion, focal injury related to oxidative stress or inflammation, and disruption of the blood–brain barrier.^12,20 Both radiologic and postmortem neuropathologic studies have shown that cortical CMIs are common in patients with CAA.^11,12,21,22 The association of cortical CMIs with the presence of any subcortical infarct or specifically with lacunar infarcts has been reported previously in memory clinic patients and population-based cohorts.^23,24 Our results extend these prior observations to patients with CAA. We find that patients with CAA with lobar lacunes are significantly more likely to have at least one CMI on MRI. The lack of association between these 2 discrete brain lesions and conventional vascular risk factors as well as measures of vascular dysfunction support our hypothesis that lacunes and CMIs arise via similar mechanisms in patients with CAA, via pathways that may not involve global ischemia or end results of classical risk factors. Previous studies have yielded conflicting data regarding the association between CMIs and vascular risk factors in different populations, some suggesting associations with hypertension and diabetes and some not showing any such relationship.^12,21,24,25 The absence of association between either lacunes or cortical CMIs and conventional vascular risk factors in our study suggests that these lesions can occur through other CAA-specific mechanisms. Further studies that include non-CAA small vessel diseases and healthy controls are needed to compare the association of lacunes and cortical CMIs between patients with CAA and these comparator groups, in order to minimize collider bias.

CAA has been demonstrated to cause impaired reactivity of cerebral vessels to physiologic stimulation.⁹ Delayed TTP to visual stimulation has been observed in multiple CAA cohorts and has been applied as an outcome marker in a phase II randomized trial.^8,9,26,27 Our results confirm previous reports of association between WMH and vascular dysfunction and support the notion that impaired vascular dysfunction may at least partly mediate CAA-related global ischemic injury.^8,9,28 However, the absence of relationship between vascular dysfunction and lacunes or CMIs suggests that more focal vascular pathologies might be responsible for these smaller discrete ischemic lesions as opposed to global ischemia that might be the underlying mechanism of WMH. This theory of discrete vessel pathologies primarily contributing to CMIs is further supported by recent neuropathologic observations in autopsy cases with CAA demonstrating that vessels involved in cortical CMIs are severely affected by Aβ accumulation.²⁹ The absence of a relationship between presence of lacunes and other global markers of injury such as WMH volume, cortical thickness, and pTBV also support the view that lacunes have an at least partially different physiopathology than global ischemia. Larger and ideally multicenter studies are needed to put the multiple different markers of CAA-related tissue damage into a broader context.

We also compared the risk of incident ICH in patients with and without lacunes and demonstrated significantly higher rate of incident CAA-related ICH among patients who had lacunes. These results suggest that lacunes might be a marker for severe focal vessel pathology that promotes the development of incident ICH. In a previous study conducted in a stroke-free general population, burden of WMH and MRI-defined brain infarcts—mostly lacunes—were found to be associated with incident ICH risk, particularly when co-occurrence of higher burden of WMH and infarcts was present.³⁰ Although lacunes are known to be associated with intrinsic small vessel abnormalities, described as lipohyalinosis or fibrinoid necrosis resulting in ischemia with the focal occlusion of the vessel,³¹ there is also evidence of blood–brain barrier dysfunction in subcortical white matter in patients with lacunar stroke, supporting the idea that there might be other mechanisms that can lead to the development of lacunes.^32,33 Further neuropathologic studies are needed in order to understand the mechanism underlying lobar lacunes and their association with the development of subsequent ICH in patients with CAA.

The present study has several limitations. We did not have comparator groups of non-CAA ICH or healthy controls. We note that the association of lobar lacunes with CAA has been confirmed in 2 larger studies that included patients with HTN-ICH, one from the United States and one from Taiwan.^6,7 We also note that the similar rates of lacune detection in this study that used high-resolution MRI and previous work that relied on clinical-grade MRIs suggest that the detection of lacunes might be fairly consistent across different CAA patient populations regardless of the MRI protocol and scanner characteristics. Future work should indeed include direct comparison of different MRI protocols in the same patients to understand any potential differences in lacune detection. The use of MRIs obtained in 1.5T and 3T scanners can be considered as a potential limitation of the current study. In vivo detection of CMIs was first described in studies using 7T MRI^34,35 and these studies were followed by several other studies using 3T and 1.5T MRIs with different scan protocols.^{21,23,24,36,37} Although no studies compared detection rates of CMIs between 3T MRIs and 1.5T MRIs, it has been proposed that the sensitivity of detection predominantly relies on using high-resolution 3D sequences. In our cohort, all MRI scans had the same protocol that included 3D T1-weighted and 3D T2/FLAIR sequences with isotropic 1 × 1 × 1 mm voxels, and as such the rates of detection of CMIs in 1.5T vs 3T were effectively the same (28.9% vs 28.3%, p = 1). Moreover, we included the field strength to every multiple regression model used in this study, which did not change the significant interrelationship between CMIs and lacunes or any other association. These results suggest that the difference in field strength would not introduce any unmeasured bias that could have skewed our analyses. Nevertheless, the sensitivity of MRI is limited as compared to neuropathologic examination and smaller CMIs remain undetected even using highly customized MRI protocols performed in ultra-high field strength scanners.³⁸ Future work should continue to optimize MRI protocols to maximize detection of CMIs. Future work should also include much larger CAA patient cohorts in order to study the potential effects of classical vascular risk factors on CAA-related brain lesions including lacunes and microinfarcts. Our results are in line with previous studies that suggest CMIs and lacunes fall on the same spectrum of total ischemic lesion burden in the brain.

Lacunes are found almost exclusively in lobar locations in about 25% of patients with CAA diagnosed with or without a symptomatic ICH. Lobar lacunes correlate strongly with the presence of CMIs and neither of these radiologic lesions is related to classical vascular risk factors in patients with CAA. Lacunes and CMIs did not correlate with an fMRI-based measure of vascular dysfunction. Our data thus suggest that lobar lacunes and CMIs are related to focal vascular effects of CAA rather than global ischemia related to widespread vessel dysfunction. This study found that the presence of lobar lacunes independently predicts risk of incident ICH in patients with CAA. If future work reproduces these findings in larger cohorts with appropriate comparator groups, lobar lacunes might serve as an important diagnostic and prognostic marker in patients with CAA.

Glossary

BG

basal ganglia

BOLD

blood oxygen level–dependent

CAA

cerebral amyloid angiopathy

CI

confidence interval

CMB

cerebral microbleed

CMI

cerebral microinfarct

CSO

centrum semiovale

cSS

cortical superficial siderosis

EPVS

enlarged perivascular spaces

eTIV

estimated total intracranial volume

FLAIR

fluid-attenuated inversion recovery

HTN-ICH

hypertensive deep intracerebral hemorrhage

ICH

intracerebral hemorrhage

IQR

interquartile range

pTBV

percent total supratentorial brain volume

pWMH

percent white matter hyperintensity

SWI

susceptibility-weighted imaging

TBV

total supratentorial brain volume

TTP

time to peak

WMH

white matter hyperintensities

Appendix. Authors

Study Funding

Supported by the NIH (NINDS NS083711 [M.E.G.] and NINDS AG26484 [S.M.G.]).

Disclosure

Elif Gokcal, Mitchell J. Horn, Susanne J. van Veluw, Aina Frau-Pascaul, Alvin S Das, Marco Pasi, Panagiotis Fotiadis, Andrew D. Warren, Kristin Schwab, Jonathan Rosand, Anand Viswanathan, and Jonathan R. Polimeni report no disclosures. Steven M. Greenberg reports funding from the NIH (NINDS AG26484). M. Edip Gurol reports funding from the NIH (NINDS NS083711). Go to Neurology.org/N for full disclosures.