This is an observational cross-sectional study that will be conducted at the Leiden University Medical Center (LUMC). This study involves a whole-day visit to the LUMC for each participant and includes the following procedures: a 3T brain MRI scan of 60 minutes, a 7T brain MRI scan of 60 minutes, a neuropsychological assessment, and questionnaires on demographics and vascular risk factors. An overview of the study procedures can be found in Figure 1. The objectives of this study are as follows: (1) to study the association between a more irregular WMH shape and anatomical, hemodynamic, and white matter integrity abnormalities on MRI; (2) to study the association between WMH shape and cognition or other cerebral SVD markers; and (3) to study the association between novel MRI markers of brain clearance and both cerebral SVD markers and cognition.

We will prospectively include 50 outpatients aged >65 years with memory complaints from the memory or geriatric clinic in one of their first visits to LUMC (Leiden, the Netherlands), Alrijne Hospital (Leiden, the Netherlands), or Haga Hospital (The Hague, the Netherlands). All participants provide written informed consent before any study procedures. This study involves 3T and 7T MRI and newly developed sequences and markers, especially for brain clearance, which have not been applied in patient populations before. Therefore, we conducted the sample size calculation based on the WMH shape analysis. To provide a frame of reference, we performed a sample size calculation using data from a previous study [26]. The linear regression on hypertension and convexity, corrected for age and sex, was conducted using data from 71 older adults without dementia. The same WMH shape analysis pipeline was used in the previous study as we aim to use it in this study. Using the data from the significant results of the aforementioned study, the sample size calculation conducted in G*Power [27] (input parameters: effect size=0.1; α=.05) showed a result of 79 participants. As the data used in the calculation were obtained from community-dwelling individuals, we expect higher effect sizes in our study, which comprises a memory clinic population. In this population, the prevalence and severity of SVD and, thus, WMHs will be higher, justifying our assumption of increased power. At the same time, we do not want to risk an underpowered study. For example, in a case-control study comparing patients with type 2 diabetes with healthy controls, an average effect size of 0.37 for a more irregular WMH shape was found [12]. To illustrate this, we calculated the sample size with 2 different effect sizes that are higher than the one used in the initial calculation (0.2 and 0.3; α remained at .05) but still lower than 0.37 (as found by de Bresser et al [12]). These calculations resulted in a sample size of 42 or 29 participants, which was obtained using the effect sizes of 0.2 and 0.3, respectively. Therefore, 50 participants should provide the necessary statistical power to overcome biological and clinical variation.

To be eligible to take part in this study, a participant must meet all the following criteria: receipt of care in the outpatient memory clinic or the geriatric clinic at LUMC or the memory clinic at Alrijne Hospital or Haga Hospital, age of >65 years, and eligibility for MRI. Moreover, the participants must have native-level proficiency in Dutch due to the requirements of the neuropsychological assessment.

A potential participant who meets any of the following criteria will be excluded from this study: (1) claustrophobia; (2) contraindications for MRI, such as metal implants or pacemakers; (3) regular use of benzodiazepines; (4) initiation of treatment with antidepressants less than 6 weeks before inclusion; (5) inability to provide written informed consent (assessed by the treating physician); (6) individuals who have been declared mentally incapacitated; (7) other severe neurological disease outside of the dementia spectrum; (8) cognitive impairment due to known other neurological diseases; and (9) previous brain surgery.

We will collect basic demographic information, including age, sex, and educational level. Information about medical history, psychiatric comorbidity, medication, and current blood values are extracted from the clinical files of the participants. Furthermore, a cardiovascular risk factor questionnaire will be used to gather information about hypertension, diabetes, arrythmia, alcohol consumption, smoking, physical activity, and medical history. We will also assess sleep habits using an adapted version of the Pittsburg Sleep Quality Index [28]. The sleep questionnaire will be included because of the influence of sleep on the brain clearance system.

The following tests will be included in the neuropsychiatric assessment: (1) Mini-Mental State Examination [29], (2) clock drawing test [30], (3) 15-Word Verbal Learning Test (immediate and delayed) [31,32], (4) visual association test [33], (5) Stroop Color and Word Test (40-item version) [34], (6) Trail Making Test parts A and B [35,36], (7) letter digit substitution test [37], (8) animal fluency test [38], (9) Hospital Anxiety and Depression Scale [39], and (10) Informant Questionnaire on Cognitive Decline in the Elderly [40].

All MRI scans will be conducted at the LUMC using a 3T Philips Ingenia Elition and a 7T Philips Achieva MRI scanner (Philips Healthcare). The MRI scan protocols are shown in Table 1.

Conventional (3T) brain MRI scans will be used to determine global and functional markers of cerebral SVD, such as WMH volume and presence of lacunes, microbleeds and superficial siderosis (on a 3D T1-weighted, 3D fluid-attenuated inversion recovery [FLAIR], susceptibility-weighted imaging, and diffusion-weighted imaging scan), and hemodynamics (flow territory mapping [41]). SVD markers will be assessed according to the second iteration of the Standard for Reporting Vascular Changes on Neuroimaging consensus criteria [42]. WMH shape will be calculated in the 3D space using an in-house developed pipeline, as described previously [18]. In brief, WMHs are segmented automatically from FLAIR MRI scans. Lateral ventricles are segmented from T1-weighted scans, and these ventricle masks are inflated by 3 mm and 10 mm. The inflated ventricle masks represent regional references for categorizing WMHs into periventricular, confluent, or deep subtypes. Examples of WMH segmentations and the shape markers are shown in Figures 2 and 3. Convexity, solidity, concavity index, and fractal dimensions will be determined based on the segmentations of the periventricular or confluent WMHs [18]. A lower convexity and solidity and higher concavity index and fractal dimension indicate more irregular shapes. For deep WMHs, fractal dimensions and eccentricity will be calculated [18]. Lower eccentricity indicates a rounder shape, and a higher fractal dimension indicates a more irregular shape of deep WMHs. The formulas of the WMH shape markers are shown in Figure 3. Moreover, total WMH volume, as well as the volumes of periventricular, confluent, and deep WMHs, will be calculated in milliliters.

White matter structural integrity will be measured using a magnetic resonance fingerprinting (MRF) sequence [43] and an inhomogeneous magnetization transfer (ihMT) scan [44]. A gradient-spoiled MRF [43] sequence with an optimized flip angle pattern for T1 and T2 quantification [45] was used. The MRF sequence is sensitive to both T1 and T2 relaxation effects in a time-efficient manner. This allows for quantitative T1 and T2 measurements and quantification of underlying magnetization fractions of the tissue present in the brain [46,47]. A 3D gradient echo–based ihMT sequence was acquired in which we used interleaved off-resonance radio-frequency saturation pulses to isolate dipolar effects, which have been linked to myelin [44]. These 2 sequences quantify underlying white matter tissue properties within and outside of the WMHs to better understand their pathological processes in vivo and see whether these are related to a more irregular WMH shape.

A high temporal resting-state functional MRI sequence will be obtained, with its first obtained slice strategically positioned below the fourth ventricle. This will allow for probing of the inflow dynamics of CSF into the brain and correlating these with gray matter BOLD signals [48]. Altered BOLD-CSF coupling might signal impaired drivers of brain clearance activity, such as a decreased driving force of CSF motion due to arteriolar vasomotion, especially when found in relation to lower cognitive scores or more severe SVD burden [49,50].

Phase-contrast sequences will be used, first, to quantify arterial blood flow in the feet-head direction of blood flowing into the brain (Q flow) and, second, as angio-surveys to facilitate the placement of the labeling plane perpendicular to brain-feeding arteries and measure the offset distance between arteries (ie, left-right and anterior-posterior) for flow territory mapping. Q flow can also be used to measure blood flow velocities. Flow territory mapping is achieved through arterial spin labeling MRI. It can demarcate perfusion territories by changing the labeling efficiency within the labeling plane in the right-left and anterior-posterior directions over sequential arterial spin labeling measurements.

Apart from BOLD-CSF coupling, these advanced 3T MRI sequences are secondary to the main aims and will provide additional information on the tissue and perfusion properties of WMHs.

Ultrahigh-field (7T) brain MRI scans will be used to determine WMH shape [18] (solidity, convexity, concavity index, fractal dimension, and eccentricity) and other markers of cerebral SVD in and surrounding the WMHs, such as enlarged PVS, (cortical) microinfarcts, and microbleeds (on a T1-weighted, T2-weighted, FLAIR, and T2*-weighted scan). We will investigate the effect of a higher resolution and increased contrast on the WMH shape calculations. Moreover, a recently implemented MRI technique with high resolution called CSF-selective T2-prepared readout with acceleration and mobility encoding (CSF-STREAM) provides detailed whole-brain imaging of CSF mobility, from subarachnoid spaces down to the level of PVS as a proxy for brain clearance activity [51]. An example of measurements obtained using this technique can be found in Figure 4. Moreover, we will measure CSF and blood flow at the level of the foramen magnum using phase-contrast MRI. Heart rate and respiratory signals will be measured during the scans (3T and 7T MRI) using standard vendor-supplied equipment, namely, a respiratory belt and pulse oximeter.

Patients are asked to lie as still as possible during scanning. If any discomfort is felt, they may move that body part slightly during breaks in between acquisitions, but we emphasize that they should not move their head, neck, or shoulders in doing so. Between the 3T and 7T scans, an hour break is planned. Furthermore, patients are asked beforehand if they have neck or back pain that prevents them from laying supine for extended periods. If that is the case, they are excluded as the risk of not completing the MRI scans is high.

CSF-STREAM consists of 7 subscans of 2 minutes, 45 seconds, limiting the effect of motion during each subscan. The 7 subscans will be coregistered with the first scan to correct for motion. If high motion occurs during one of the subscans, resulting in strong image artifacts, the corresponding scans will be excluded from the analysis. BOLD-CSF coupling is a very fast BOLD echo-planar imaging sequence and can be motion corrected retrospectively. The ihMT scan consists of 5 acquisitions with varying on- and off-resonance radio-frequency pulses. These 5 images will be motion corrected to each other similar to the CSF-STREAM strategy or excluded if they show high within-image motion. Furthermore, the large voxel sizes will somewhat mitigate motion blurring.

The T2* gradient-echo readout has a navigator that can prospectively mitigate motion during acquisition.

Multivariable regression analyses will be conducted to relate periventricular and deep WMH volume and shape (solidity, convexity, concavity index, fractal dimension, and eccentricity) to brain clearance markers (eg, peak BOLD-CSF coupling magnitude, CSF mobility and flow, and PVS count) and white matter integrity (magnetization transfer ratio and mean T1 and T2 of the white matter maps). Furthermore, the association between WMH shape and brain clearance markers and z-transformed cognitive domain scores (memory, executive function, visuoconstruction, and processing speed) will be studied via multivariable regression analyses.

Secondary analyses could include association analyses with other SVD markers (presence of microbleeds, lacunes, and small recent subcortical infarcts).

Potentially confounding demographic variables such as age, sex, level of education, intracranial volume, and vascular risk factors will be included as covariates for subsequent analyses when appropriate.

Statistical assumptions such as linearity and normality of residuals will be checked using Q-Q plots, histograms, and other distribution measures. If applicable, appropriate nonparametric tests will be used, or the variables will be log-transformed. For the main exploratory analyses, the significance will be set at P<.05. Correction for multiple testing will be performed when appropriate.

The WHIMAS study (NL78641.058.21) was approved by the Medical Ethics Committee Leiden The Hague Delft (P21.114) and is registered on ClinicalTrials.gov (NCT06010511). All participants will provide written informed consent before any experiments. Data will be anonymized and, after completion of data collection, transmitted to a secure repository with safeguards in place to protect confidentiality. Participants will receive travel compensation.

Only one scan of a single patient was used for this paper as an example of the techniques used. In-house developed magnetic resonance imaging sequences can be shared with interested research groups when proper collaboration agreements are in place.