Participants

We included participants from two studies, which were largely comparable in study procedures, allowing for pooling of data. Patients with sCAA were from the BIONIC (BIOmarkers for cogNitive Impairment due to Cerebral amyloid angiopathy; www.radboudumc.nl/BCS) and the FOCAS (Follow-up in sporadic CAA Study) studies. The BIONIC study is a cross-sectional cohort study on CSF biomarkers for sCAA situated in the Radboud University Medical Center (RUMC), Nijmegen. Patients with probable sCAA according to the modified Boston criteria (version 1.5) [11] were consecutively recruited from the neurology and geriatrics outpatient clinics from the RUMC between December 2018 and July 2023, and from the FOCAS study, a longitudinal study on the disease course of sCAA with comparable inclusion criteria, situated in the Leiden University Medical Center (LUMC) [12]. The diagnosis was verified by a senior vascular neurologist (FHBMS, CJMK, MJHW). Exclusion criteria were contra-indications for lumbar puncture or MRI, and recent (<3 months) symptomatic stroke. In FOCAS, CSF sampling was not performed in all participants; for the current study we selected participants based on the availibility of CSF. Four individuals from the FOCAS study were ultimately diagnosed with mixed CAA; three individuals had one deep microbleed, one individual had three deep microbleeds (signs of concomitant of deep perforating arteriopathy).

Control participants were recruited in the context of the CAFE (Cerebral Amyloid angiopathy Fluid biomarkers Evaluation) study. They were recruited at the RUMC from partners and family of the patients with sCAA. In addition, we recruited individuals via the Dutch Brain Research Registry [13]. Inclusion criteria were age ≥55 years, a Montreal Cognitive Assessment (MoCA) score >28 or a modified Telephone Interview of Cognitive Status (mTICS) score of ≥35 [14, 15]. Additional exclusion criteria for the controls included self-reported (subjective) cognitive decline, and a history of major brain pathology such as spontaneous parenchymal intracerebral hemorrhage, ischemic stroke, neurodegenerative disease, brain tumors, brain infection or inflammation. The controls were age and- sex-matched to the patients with sCAA from the BIONIC study. For details on the BIONIC/CAFE protocol, see the Methods section in the supplement.

CSF analysis

All participants underwent a lumbar puncture according to local protocols (see supplement). At both participating hospitals, the CSF was collected in polypropylene tubes, centrifuged, aliquoted, and stored in polypropylene tubes at −80°C.

All CSF analyses were performed at the RUMC. Patient and controls samples were randomly analyzed to avoid bias. CSF Aβ40, Aβ42, tau phosphorylated at threonine 181 and total tau levels were quantified using the Lumipulse chemiluminescent immunoassay (Fujirebio, Ghent, Belgium). The samples were analysed in different batches; however, we adhere to strict guidelines under the ISO15189 guidance to control that inter-assay variation is kept within predefined limits of variation for each assay.

We stratified participants into individuals with a biomarker profile indicative of concomitant AD (sCAA-AD+) and those without a biomarker profile indicative of concomitant AD (sCAA-AD-). A biomarker profile indicative of concomitant AD (A+T+N+ or AD+) was defined as a combination of the following: amyloid deposition (A+; a decreased CSF Aβ42 concentration), tau accumulation (T+: an increased CSF phosphorylated tau concentration), and neurodegeneration (N+: an increased total tau concentration) [10] We used the following predefined local cutoff values: (CSF Aβ42 (A+): <659pg/ml; phosphorylated tau (T+): >64pg/ml; total tau (N+): >400pg/ml).

Neuropsychological assessment

The following cognitive domains were investigated: Episodic memory (Rey Auditory Verbal Learning Test (RAVLT) and recall) [16], Working memory (mean of Wechsler’s Digit Span Test forward and backward trials), Processing speed (Stroop card I, Stroop card II, Trail Making Test A (TMT-A)), Verbal Fluency (one minute animal naming), Visuoconstruction (Rey-Osterrieth’s Complex Figure Test – Copy trial), Executive functioning (Stroop interference score, TMT interference score).

We calculated the Stroop interference score by dividing the Stroop part III score by part II. We calculated TMT interference score by dividing TMT-B by TMT-A. Global cognition was assessed using the MoCA version 7.1 [17].

We scored education level using seven categories in accordance with the Dutch educational system, (the Verhage Score) [18], which is comparable to the UNESCO international classification of education levels [19].

We converted all raw test scores into Z-scores corrected for age, sex and education level for each participant based on a large normative dataset from the Advanced Neuropsychological Diagnostics Infrastructure (ANDI; www.andinorms.nl). The ANDI database includes data from up to 26,000 healthy individuals from all ages [20, 21]. Next, we averaged Z-scores of cognitive tests that reflected the same cognitive domain to get a composite Z-score per cognitive domain. If one test of a particular domain was missing, the domain score was based on the remaining tests of that domain. To correct for extreme performances, Z-scores <-3 and >3 were converted to -3 and 3, resulting in a scoring range of -3 to 3 (reflecting the full range of severely impaired to extremely superior performances). Furthermore, we defined cognitive impairment on a domain as a composite Z-score of <−1.5 [22]. We defined global cognitive impairment as a MoCA Z-score of <−1.5. Single domain cognitive impairment was defined as a Z-score of <1.5 on a single domain, multi domain cognitive impairment was defined as a Z-score or of <-1.5 on more than one domain.

Neuropsychiatric assessment

Apathy was assessed using the Dutch version of the self-reported Apathy Starkstein Scale (AS) (performed at RUMC only) and informant version of AS (both RUMC and LUMC) [23]. This questionnaire consists of 14 items with four possible answers ranging from 0 to 3 points, with higher scores indicating more severe apathy (range: 0-42 points). A score of at least 14 points is indicative for the presence of apathy. We considered the informant version more valid, since the presence of (mild) cognitive impairment may influence the validity of self-reported apathy [24]. However, we decided to report both the self-reported and informant AS scale, if available.

Neuropsychiatric symptoms were investigated using the Dutch version of the Neuropsychiatric Inventory Questionnaire (NPI) [25] in participants at the RUMC, and by using the shortened version, the NPI-Questionnaire (NPI-Q) [26] in participants at the LUMC. The NPI is a retrospective caregiver-informant interview with screening questions covering 12 neuropsychiatric symptom domains: delusions, hallucinations, agitation/aggression, dysphoria/depression, anxiety, euphoria/elation, apathy/indifference, disinhibition, irritability/lability, aberrant motor behaviours, nighttime behavioural disturbances, and appetite/eating disturbances. Informants are asked to answer “yes” or “no” in response to each screening question, and to rate the severity of the symptoms present in the last 4 weeks if the answer is “yes” on a three-point scale: (1-mild, 2-moderate, 3-severe). In addition, caregiver burden (6-point scale ranging from “0-not at all” to “5-extreme”) is determined.

The NPI and NPI-Q are highly similar [26]; the main difference is that in the NPI, but not in the NPI-Q, the frequency of a symptom is also rated. To pool the total scores of the questionnaires, we excluded the frequency scale from our analysis. The total NPI and NPI-Q severity score represents the sum of individual symptom scores (presence x severity) and ranges from 0 to 36.

MRI acquisition

Participants at the RUMC (patients with sCAA and controls) underwent brain MRI on a 3.0 Tesla MRI scan (Siemens Magnetom Prisma, Siemens Healthineers, Erlangen, Germany) using a 32-channel head coil. Participants were examined using a comprehensive protocol (Supplementary Table S1). For the current study, 3D T1-weighted sequence using Magnetization Prepared 2 Rapid Acquisition Gradient Echoes (MP2RAGE), the 3D T2-weighted sequence, the 3D fluid-attenuated inversion recovery (FLAIR), and the 3D multi-echo gradient echo T2*-weighted sequence were analyzed. Magnitude and phase data from the multi-echo gradient sequence was processed to a SWI using the “Contrast-weighted, Laplace-unwrapped, bipolar multi-Echo, ASPIRE-combined, homogeneous, improved Resolution SWI” (CLEAR-SWI) method [27]. Participants at the LUMC were scanned on a 3.0 Tesla MRI scanner (Philips Healthcare, Best, the Netherlands) with a 32-channel head coil. This protocol included SWI, T2 and FLAIR sequences. Further details are described in previous reports [12].

Cerebral small vessel disease markers

We assessed the number and distribution of cerebral microbleeds (CMBs) [28], presence and extent of cortical superficial siderosis (CSS) [11], presence and extent of enlarged perivascular spaces (EPVS) in the centrum semi-ovale (CSO; using a dichotomized classification: high (≥21 EPVS) or low (≤20 EPVS)) [29] and white matter hyperintensities (WMH) according to the Fazekas Scale [30, 31]. Using these four parameters, we calculated a summary score of SVD markers in sCAA [29], referred to as CAA-related SVD burden score, ranging from 0 to 6 points. Two trained readers rated the different SVD makers on the MRI; in case of disagreement, a senior vascular neurologist (FHBMS) or neuroradiologist was consulted to reach final consensus.

Hippocampus volumetry

The total intracranial volume and hippocampal volume were estimated using the ‘SAMSEG’ function within Freesurfer (version 7.3.2) [32]. Hippocampus segmentation was manually corrected if needed. To account for head size, we computed the ratio of the bilateral hippocampal volumes to the total intracranial volume: normalized hippocampal volume (nHV) = [(right + left hippocampal volumes)/ intracranial volume]. The nHV is available for the RUMC participants only.

Data analysis

Data are represented as mean ± standard deviation (SD) or median with interquartile range (IQR). We compared demographic, clinical, radiological, neuropsychological test data, and neuropsychiatric questionnaire data between sCAA-AD+ versus sCAA-AD- participants, and between sCAA-AD- participants and controls. We compared sCAA-AD- patients and controls to study the association of sCAA alone with cognition and neuropsychiatric symptoms. Differences were analyzed with a Student’s t-test or Mann-Whitney U test as appropriate. Depending on group size, differences in proportions were compared by a chi-square test, or a Fisher Exact test. In case of more than two categories in a row, the Fisher-Haller-Freeman test was used to assess differences in proportions.

Since prior symptomatic ICH may influence both cognitive function and neuropsychiatric symptoms, we repeated the analyses after exclusion of sCAA patients with prior ICH.

Since CSF Aβ42 is also decreased in sCAA, and p-tau is the most specific for AD, we also analyzed the effect of the individual CSF biomarkers of AD on cognitive performance and neuropsychiatric symptoms. We used binominal logistic regression analyses to analyze the association of separate CSF AD biomarkers, the p-tau/Aβ42 ratio and nHV; with (I) global cognition (impairment vs normal), (II) episodic memory (impairment vs normal), and (III) neuropsychiatric symptoms (1 or more symptoms vs no symptoms) in participants with sCAA, all adjusted for history of ICH. Odds ratios were standardized, and 95% confidence intervals (CI) were calculated. Episodic memory was analyzed since this domain is considered to be specifically impaired in patients with AD [33]. Because cognitive scores were already adjusted for age, sex, and education level, these variables were not additionally included in the models. In the model for neuropsychiatric symptoms, we adjusted for age [33].

Ethical statement

We obtained written informed consent from all participants. All participants underwent lumbar puncture in the context of studies on biomarkers for CAA, which were approved by the local medical ethics committee of the RUMC (NL 63298.091.17, NL 62669.091.17) and LUMC (NL63256.058.17).

Data Availability

Anonymized data not published within this article will be made available by request from any qualified investigator.