Increased Cerebral Small Vessel Disease Burden With Renal Dysfunction and Albuminuria in Patients Taking Antithrombotic Agents: The Bleeding With Antithrombotic Therapy 2.

Background The aim of this study was to determine the associations of cerebral small vessel disease (SVD) burden with renal dysfunction and albuminuria in patients taking oral antithrombotic agents. Methods and Results Patients who newly started or continued taking oral antiplatelets or anticoagulants were enrolled in a prospective, multicenter, observational study. Obligatorily acquired multimodal magnetic resonance imaging at registration with prespecified imaging conditions was assessed for cerebral microbleeds, white matter hyperintensities, enlarged basal ganglia perivascular spaces, or lacunes, and an ordinal SVD score was calculated (range, 0-4). Multivariable adjusting covariates were age, sex, hypertension, diabetes, dyslipidemia, current smoking, drinking, and estimated glomerular filtration rate (eGFR). Of 5324 patients (1762 women; median age, 73 years), 4797 (90.1%) patients were taking oral antithrombotic agents for secondary stroke prevention. Cerebral microbleeds were present in 32.7%, confluent white matter hyperintensities in 51.8%, extensive basal ganglia perivascular spaces in 38.9%, and lacunes in 59.4%. Median SVD score was 2. Compared with eGFR category G1 (eGFR ≥90 mL/min per 1.73 m2), adjusted odds ratios for SVD score increment were 1.63 (95% CI, 1.11-2.39) at category G4 (eGFR 15-<30 mL/min per 1.73 m2) and 2.05 (95% CI, 1.33-3.16) at G5 (eGFR <15 mL/min per 1.73 m2). Corresponding odds ratios relative to urinary albumin-to-creatinine ratio (ACR) category A1 (ACR <30 mg/g) were 1.29 (95% CI, 1.12-1.49) for category A2 (ACR 30-<300 mg/g) and 1.37 (95% CI, 1.05-1.77) for A3 (ACR ≥300 mg/g). When combined eGFR and ACR categories were assessed, risks for SVD score increment generally increased as eGFR decreased and ACR increased. Conclusions Both reduced eGFR and albuminuria were independently associated with increased cerebral SVD burden in patients requiring oral antithrombotic medication mainly for secondary stroke prevention. Registration URL: https://www.clinicaltrials.gov; Unique identifier: NCT01581502; URL: https://www.umin.ac.jp/ctr; Unique identifier: UMIN000023669.

urinary albumin‐to‐creatinine ratio

Bleeding with Antithrombotic Therapy

enlarged basal ganglia perivascular spaces

cerebral microbleed

small vessel disease

white matter hyperintensity

Clinical Perspective

What Is New?

The BAT (Bleeding with Antithrombotic Therapy) 2 study aims to provide a precise risk model for antithrombotic‐associated bleeding, taking the cerebral small vessel disease burden into account, and multimodal brain magnetic resonance imaging was acquired at baseline for all patients under prespecified imaging conditions.

BAT2 also collected data of estimated glomerular filtration rate as well as albuminuria.

The objective of this study was to determine the associations of cerebral small vessel disease burden with renal dysfunction and albuminuria in patients who newly started or continued taking oral antithrombotic agents, using the baseline data from BAT2.

What Are the Clinical Implications?

Albuminuria and reduced estimated glomerular filtration rate were independently associated with increased cerebral small vessel disease burden in patients who newly started or continued taking oral antithrombotic agents mainly for secondary stroke prevention.

BAT2 will provide novel risk‐stratification models for antithrombotic‐associated bleeding risk in association with cerebral small vessel disease and other biomarkers, including the chronic kidney disease measures.

Among the magnetic resonance imaging (MRI) biomarkers for cerebral small vessel disease (SVD), cerebral microbleeds (CMBs) are known predictors of intracranial hemorrhage especially in patients on antithrombotic therapy post stroke. , , , Cerebral SVD is also associated with noncerebral problems as a marker of systemic SVD including chronic kidney disease (CKD), which also seems to increase risks for bleeding events. , A total SVD score combining individual MRI features of SVD offers potential for more accurate stratification of the cerebral SVD burden than the use of individual features separately, , and the association of total SVD score increment with reduced estimated glomerular filtration rate (eGFR) was reported in an analysis of 1080 patients with ischemic stroke or transient ischemic attack from the Oxford Vascular Study, with the associations attenuating at older ages. Albuminuria was related to the increased SVD score in 1037 hypertensives from the Investigating Silent Strokes in Hypertensives: a Magnetic Resonance Imaging Study, in which concurrently investigated eGFR showed no significant association with the SVD score.

The importance of bleeding complications associated with antithrombotic therapy, including an increased bleeding risk with dual antithrombotic use and the association of high blood pressure levels in the outpatient clinic with later intracerebral hemorrhage occurrence, was clarified in a multicenter registry, the BAT (Bleeding with Antithrombotic Therapy) study. , Responding to the subsequent prevalence of antithrombotic medication, such as development of direct oral anticoagulants and new P2Y12 receptor blockers and improvement of dual antiplatelet therapy for stroke, , , , , , we newly organized the BAT2 study. BAT2 aims to provide a precise risk model for antithrombotic‐associated bleeding, taking the cerebral SVD burden into account, and multimodal brain MRI was acquired at baseline for all patients under prespecified imaging conditions. BAT2 also collected data of eGFR as well as albuminuria. Given the association of bleeding risk with reduced eGFR and albuminuria, and in particular the stronger association with albuminuria, simultaneous analyses of the 2 CKD measures with total SVD score in patients on antithrombotic therapy are relevant.

The objective of this cross‐sectional study was to determine the associations of cerebral SVD burden with renal dysfunction and albuminuria in patients who newly started or continued taking oral antithrombotic agents, using the baseline data from BAT2.

Methods

Data supporting the findings of this study are available from the principal investigator of BAT2 (Toyoda) on reasonable request.

Study Design and Participants

The BAT2 study was an investigator‐initiated, prospective, multicenter, observational study involving 52 hospital sites across Japan from the Network for Clinical Stroke Trials (Table S1). BAT2 was designed to determine the incidence and details of bleeding complications in patients treated with oral antithrombotic agents. The study was registered with ClinicalTrials.gov (NCT02889653) and the University Hospital Medical Information Network clinical trial registry in Japan (UMIN 000023669). The overall protocol has been published elsewhere. All study procedures were reviewed and approved by the ethics committee of the participating sites. The investigators obtained written informed consent from patients or their family members before registration.

Patients with cerebrovascular or cardiovascular diseases (either symptomatic or asymptomatic) who newly started or continued taking oral antiplatelets or anticoagulants were enrolled from October 2016 through April 2019. Brain MRI was mandatory for all patients at registration and contraindication to MRI was an exclusion criterion of this study.

Clinical Data Acquisition and Management

At registration, baseline clinical information and blood test and urinalysis results were collected. The Research Electronic Data Capture system was used for the collection and management of data from each participating site through a secured network connection with authentication. The eGFR (mL/min per 1.73 m2) was estimated based on serum creatinine level using the equation of Japanese Society of Nephrology. CKD severity was staged by the glomerular filtration rate categories according to the NKF‐KDOQI (National Kidney Foundation‐Kidney Disease Outcomes Quality Initiative) guideline. Albuminuria was assessed by urinary albumin‐to‐creatinine ratio (ACR, mg/g) using a spot urine sample and categorized also according to the NKF‐KDOQI guideline.

Acquisition and Management of Brain MRI Data

Brain MRI of magnetic field at 3 or 1.5 Tesla was obtained parallel to the anterior commissure‐posterior commissure line or the orbitomeatal line. MRI was allowed to be performed from 90 days before to 14 days after registration. MRI sequences included T1‐weighted, T2‐weighted, fluid‐attenuated inversion recovery, and T2*‐weighted imaging. T1‐weighted and T2‐weighted images represent water content in low and high intensities, respectively. Fluid‐attenuated inversion recovery images have similar characteristics to T2‐weighted imaging, but the signal of cerebrospinal fluid is suppressed. T2*‐weighted images can detect hemorrhagic changes with high sensitivity. Three‐dimensional time‐of‐flight magnetic resonance angiography was performed.

All MRI examinations were interpreted by a central diagnostic radiology committee consisting of 13 members (Chair: Sasaki) for CMBs, cortical superficial siderosis, white matter hyperintensity (WMH), enlarged basal ganglia perivascular spaces (BG‐PVS), lacunes, and other infarctions according to the criteria of Standards for Reporting Vascular Changes in Neuroimaging. All committee members were blinded to clinical information.

Details for acquisition and interpretation of MRI are described in Data S1 and Tables S2 and S3.

Statistical Analysis

One point for each SVD feature of CMBs (≥1 for any CMBs), confluent WMH, extensive BG‐PVS (≥11), and lacunes (≥1) on MRI was summed as an ordinal SVD score, from a minimum score of 0 to a maximum of 4. Confluent WMH was diagnosed as positive when periventricular hyperintensity grade was 3 or deep and subcortical WMH grade was 2 to 3.

Data were summarized as median (25th percentile, 75th percentile) for continuous variables and as frequency and percentage for categorical variables. Correlations between MRI findings were evaluated with Spearman’s rank order correlation coefficients. We divided patients into 2 groups (lower and higher SVD score groups) using the median SVD score as a cutoff. Statistical differences between these 2 groups were assessed using the Mann‐Whitney U test or the Pearson χ2 test, as appropriate. Proportional odds ordinal logistic regression models were applied to explore risk factors for SVD score increment, using ordinal SVD score as a dependent variable. , , Brant test was used to examine whether the proportional odds assumption was upheld. Binary logistic regression models were also applied to assess risk factors for assignment to the higher SVD score group and the presence of each MRI feature for cerebral SVD. Two multivariable models were created to adjust confounding factors with these logistic models. Model 1 included age and sex. Adjusted covariates for Model 2 included hypertension, diabetes, dyslipidemia, current smoking, drinking, and eGFR categories, as well as age and sex. , , , Each interaction among age, hypertension, eGFR categories, and ACR categories was tested as an addition to logistic models. Stratified analyses by age group (≤74 and ≥75 years), hypertension (presence and absence), eGFR (<30, 30–<60, and ≥60 mL/min per 1.73 m2), and ACR (<30 and ≥30 mg/g) were performed. The reason for including not only eGFR and ACR but also age and hypertension in the stratified analyses was that among the variables studied, age and hypertension have consistently shown strong associations with cerebral SVD. , , For sensitivity analyses, ordinal and binary logistic analyses were conducted for cerebral SVD burden using the same model as in the main analyses on the subgroup in which MRI data were obtained at or after the time of registration.

Missing values were handled using a pairwise deletion method. Statistical significance was set at P<0.05 for all tests. In the present analyses, Stata/MP statistical package (version 16.1; Stata Corp LP, College Station, TX) was used. Correlations between MRI findings were calculated and visualized using Pandas, Numpy, Matplotlib, and Seaborn libraries of Python programming language (3.8.5).

Results

Among the 5378 patients registered, 11 patients with contraindication to MRI, 17 patients without MRI data acquisition for the reasons other than contraindication, 24 patients with incomplete baseline clinical data, and 2 with duplicated registration proved to be ineligible. Thus, 5324 patients (1762 women; median age, 73 years; 5321 Asian) were eligible for the present analyses. Of these, 4797 (90.1%) patients had a history of ischemic stroke or transient ischemic attack at a median of 71 (15, 1428) days for 4371 patients with available data after symptom onset and were taking oral antithrombotic agents as secondary stroke prevention; the remaining 527 (9.9%) patients were receiving antithrombotic therapy as primary prevention of stroke or secondary prevention of cardiovascular diseases.

MRI Findings

Baseline MRI scans were performed at 1.5 T in 3087 (58.0%) cases and 3 T in 2237 (42.0%) cases. T1‐weighted, T2‐weighted, and fluid‐attenuated inversion recovery images were acquired in 4529 (85.1%), 5072 (95.3%), and 4974 (93.4%) cases, respectively. T2*‐weighted imaging was obtained in 4984 (93.6%) patients and susceptibility‐weighted imaging in 125 (2.3%) patients. The median date of MRI performance from the date of registration was −5 (−15, 0) days. The number of patients in whom MRI were acquired at or after the time of registration was 1900 (35.7%). Interrater reliability values of MRI interpretation by the central diagnostic radiology committee expressed as median kappa coefficients were as follows: for deep CMBs, 0.87 (0.72, 0.97); for lobar CMBs, 0.86 (0.74, 0.96); for periventricular hyperintensity grade, 0.68 (0.57, 0.86); for deep and subcortical WMH grade, 0.75 (0.63, 0.81); for BG‐PVS, 0.61 (0.52, 0.81); and for lacunes, 0.75 (0.65, 0.98). Regarding intrarater reliability for these findings, median kappa coefficients ranged from 0.66 to 0.94. Kappa coefficients and concordance rates for MRI findings are shown in detail in Data S1 and Table S4.

On MRI, CMBs were identified in 32.7% (1671/5116), confluent WMH in 51.8% (2681/5172), extensive BG‐PVS in 38.9% (1998/5135), and lacunes in 59.4% (3118/5247). Overall (n=5324), median SVD score was 2 (1, 3) (Figure S1). Distributions of cerebral SVD scores were similar between patients in whom MRI was performed before registration (median 2 [1, 3]) and those with MRI data acquired at or after registration (median 2 [1, 3]), although the P value was 0.044 (Figure S2). SVD scores were also similar between 1.5‐Tesla (median 2 [1, 3]) and 3‐Tesla (median 2 [1, 3]) MRI scanners (P=0.51). Detailed findings of each SVD marker and its combination are shown in Figures S3 and S4. Relatively strong correlation was seen between deep and lobar CMBs (Spearman’s rho=0.41) and between periventricular hyperintensity grade and deep and subcortical WMH grade (Spearman’s rho=0.78) (Figure S5). Cortical superficial siderosis was observed in 2.1% and nonlacunar infarct in 33.1%. On magnetic resonance angiography, normal or mild stenosis of intracranial arterial stenosis was found in 72.0%, moderate in 10.9%, severe in 7.6%, and occlusion in 9.5%.

Patient Characteristics by SVD Features

Baseline patient characteristics are shown in Table 1. Patients with higher SVD scores (≥3, n=1617) were older, more frequently displayed hypertension and required support in daily life, and had lower eGFR and higher ACR than those with lower SVD scores (≤2, n=3707, P<0.001 each). As comorbidities, ischemic stroke or transient ischemic attack, intracerebral hemorrhage, acute coronary syndrome, and dementia were more frequent and atrial fibrillation was less frequent in the higher SVD score group than in the lower SVD score group (P<0.01 each). Patients with CMBs, with confluent WMH, with extensive BG‐PVS, or with lacunes were older and more frequently had hypertension, lower eGFR, and higher ACR compared with those without each SVD feature. (Tables S5 and S6).

Table 1

Patient Characteristics and Cerebral SVD Score

	Total (n=5324)	Total SVD score ≤2 (n=3707)	Total SVD score ≥3 (n=1617)	P value
Age, y	73.0 (66.0, 79.0)	71.0 (63.0, 78.0)	76.0 (69.0, 81.0)	<0.001
Female sex	1762 (33.1)	1226 (33.1)	536 (33.1)	0.96
Height, cm	162.0 (155.0, 168.0)	163.0 (155.0, 169.0)	161.0 (153.0, 166.0)	<0.001
Weight, kg	61.0 (53.0, 69.0)	62.0 (54.0, 70.0)	60.0 (52.0, 67.0)	<0.001
Body mass index, kg/m2	23.2 (21.2, 25.5)	23.3 (21.2, 25.6)	23.1 (21.2, 25.3)	0.096
Systolic blood pressure, mm Hg	134.0 (122.0, 148.0)	133.0 (121.0, 147.0)	135.0 (123.0, 149.0)	<0.001
Diastolic blood pressure, mm Hg	77.0 (68.0, 86.0)	77.0 (69.0, 86.0)	77.0 (68.0, 86.0)	0.96
Pulse rate, beats/min	75.0 (66.0, 84.0)	74.0 (65.0, 84.0)	75.0 (66.0, 85.0)	0.007
Modified Rankin Scale score of 0–2	4666 (88.0)	3340 (90.5)	1326 (82.4)	<0.001
Risk factors
Hypertension	4203 (79.0)	2796 (75.4)	1407 (87.1)	<0.001
Diabetes	1483 (27.9)	1021 (27.5)	462 (28.6)	0.43
Dyslipidemia	3453 (64.9)	2433 (65.7)	1020 (63.1)	0.075
Current smoking	781 (14.7)	568 (15.3)	213 (13.2)	0.044
Current drinking (≥8 units/wk)	1615 (30.4)	1179 (31.9)	436 (27.1)	<0.001
Habitual use of nonsteroidal anti‐inflammatory drugs	148 (2.8)	105 (2.8)	43 (2.7)	0.73
Comorbidities
Ischemic stroke or transient ischemic attack	4797 (90.1)	3294 (88.9)	1503 (92.9)	<0.001
Intracerebral hemorrhage	117 (2.2)	43 (1.2)	74 (4.6)	<0.001
Subarachnoid hemorrhage	27 (0.5)	15 (0.4)	12 (0.7)	0.11
Asymptomatic cerebrovascular disease	390 (7.3)	300 (8.1)	90 (5.6)	0.001
Atrial fibrillation	1070 (20.1)	780 (21.0)	290 (17.9)	0.010
Acute coronary syndrome	377 (7.1)	232 (6.3)	145 (9.0)	<0.001
Congestive heart failure	208 (3.9)	138 (3.7)	70 (4.3)	0.29
Peripheral artery disease	123 (2.3)	79 (2.1)	44 (2.7)	0.19
Deep venous thrombosis	95 (1.8)	65 (1.8)	30 (1.9)	0.80
Active malignancy	106 (2.0)	62 (1.7)	44 (2.7)	0.012
Liver disease	54 (1.0)	40 (1.1)	14 (0.9)	0.48
Chronic obstructive pulmonary disease	90 (1.7)	58 (1.6)	32 (2.0)	0.28
Dementia requiring support	170 (3.2)	80 (2.2)	90 (5.6)	<0.001
eGFR *, mL/min per 1.73 m2	64.4 (53.3, 75.8)	65.7 (55.0, 77.1)	61.5 (50.6, 72.9)	<0.001
eGFR categories, mL/min per 1.73 m2				<0.001
G1, ≥90	411 (7.8)	316 (8.6)	95 (5.9)
G2, 60–<90	2797 (53.0)	2045 (55.7)	752 (46.8)
G3a, 45–<60	1406 (26.6)	908 (24.7)	498 (31.0)
G3b, 30–<45	485 (9.2)	307 (8.4)	178 (11.1)
G4, 15–<30	113 (2.1)	60 (1.6)	53 (3.3)
G5, <15	68 (1.3)	36 (1.0)	32 (2.0)
ACR † , mg/g	17.0 (7.0, 51.7) (n=3142)	14.5 (6.1, 45.0) (n=2180)	23.0 (9.9, 78.7) (n=962)	<0.001
ACR categories, mg/g				<0.001
A1, <30	1985 (63.2)	1454 (66.7)	531 (55.2)
A2, 30–<300	926 (29.5)	581 (26.7)	345 (35.9)
A3, ≥300	231 (7.4)	145 (6.7)	86 (8.9)

N (%) or median (25th percentile, 75th percentile).

ACR indicates urinary albumin‐to‐creatinine ratio; eGFR, estimated glomerular filtration rate; and SVD, small vessel disease.

eGFR is estimated as follows: eGFR=194×serum creatinine−1.094×age−0.287[×0.739 if female].

Data for albuminuria are unavailable in 2182 patients.

Figure 1 shows that proportions of advanced age and hypertension increased along with an increase in SVD score. The higher the SVD score, the greater the proportion of advanced eGFR categories. Likewise, the proportion of microalbuminuria as well as macroalbuminuria increased with the SVD score. Note that data for ACR were unavailable in 2182 patients (40.9%). Vascular risk factors were generally more frequent in patients with ACR data than in those without ACR data (Table S7). Patients with higher SVD score more frequently used antiplatelet agents and less frequently used anticoagulants than those with the lower score (Table S8).

Figure 1

Age (A), hypertension (B), estimated glomerular filtration rate (C), and albuminuria (D) by total SVD score.

ACR indicates urinary albumin‐to‐creatinine ratio; eGFR, estimated glomerular filtration rate; and SVD, small vessel disease.

Risk Factors for Increased Cerebral SVD Score

The ordinal logistic regression models consistently showed significant associations of SVD score‐increment with advanced age, hypertension, lower eGFR, and higher ACR (Figure 2). SVD score significantly increased in patients with eGFR category G4 (adjusted odds ratio [OR], 1.63; 95% CI, 1.11–2.39; Model 2) and G5 (adjusted OR, 2.05; 95% CI, 1.33–3.16; Model 2) as compared with G1. Significant SVD score‐increment was also shown in patients with ACR category A2 (adjusted OR, 1.29; 95% CI, 1.12–1.49; Model 2) and A3 (adjusted OR, 1.37; 95% CI, 1.05–1.77; Model 2) as compared with A1. Associations between these CKD measures and SVD burden were relatively evident in patients ≤74 years old and in hypertensive patients (Figures S6 and S7). When combined eGFR and ACR were assessed, risks for SVD score increment generally increased as eGFR decreased and ACR increased (Table 2). The proportional odds assumption was not violated for each risk factor. Binary logistic regression models for SVD score ≥3 versus ≤2 showed associations similar to those seen in the ordinal logistic regression models (Figure 2).

Figure 2

Multivariable models of risk factors for cerebral SVD burden.

Plots showing odds ratios (ORs) and 95% CIs from multivariable models. Adjusting covariates are age categories and sex for Model 1 and age categories, sex, hypertension, diabetes, dyslipidemia, current smoking, drinking, and eGFR categories for Model 2. ACR indicates urinary albumin‐to‐creatinine ratio; eGFR, estimated glomerular filtration rate; and SVD, small vessel disease.

Table 2

Risks for SVD Score by Estimated Glomerular Filtration Rate and Albuminuria

Adjusted odds ratios of SVD score increment*		ACR (mg/g)
		<30	≥30
Estimated glomerular filtration rate (mL/min per 1.73 m2)	≥60	1 (Reference)	Model 1 1.57 (1.31–1.88), P<0.001
			Model 2 1.43 (1.19–1.71), P<0.001
		n=1333	n=558
	30–<60	Model 1 1.22 (1.03–1.46), P=0.023	Model 1 1.49 (1.23–1.79), P<0.001
		Model 2 1.16 (0.97–1.38), P=0.108	Model 2 1.32 (1.09–1.60), P=0.005
		n=631	n=520
	<30	Model 1 1.49 (0.64–3.45), P=0.35	Model 1 2.34 (1.56–3.49), P<0.001
		Model 2 1.54 (0.66–3.62), P=0.31	Model 2 1.96 (1.31–2.95), P=0.001
		n=19	n=77

Adjusted odds ratios (95% CI). ACR indicates urinary albumin‐to‐creatinine ratio; and SVD, small vessel disease.

Ordinal logistic regression models. Adjusting covariates are age categories and sex for Model 1 and age categories, sex, hypertension, diabetes, dyslipidemia, current smoking, and drinking for Model 2. P for interaction=0.69 in Model 1; P for interaction=0.65 in Model 2.

Risk Factors for Each SVD Marker

Both advanced age and hypertension showed significant associations with the presence of any CMBs, confluent WMH, extensive BG‐PVS, and lacunes (Figure 3). Lower eGFR and higher ACR also showed significantly or marginally significantly increased risks of these MRI SVD markers. Higher ACR showed a significant association with the presence of any CMBs.

Figure 3

Multivariable models of risk factors for SVD features.

Plots showing odds ratios (ORs) and 95% CIs from binary logistic regression models. Adjusting covariates are age categories and sex for Model 1 and age categories, sex, hypertension, diabetes, dyslipidemia, current smoking, drinking, and eGFR categories for Model 2. ACR indicates urinary albumin‐to‐creatinine ratio; BG‐PVS, enlarged basal ganglia perivascular spaces; CMB, cerebral microbleed; eGFR, estimated glomerular filtration rate; SVD, small vessel disease; and WMH, white matter hyperintensity.

Sensitivity Analyses With Subgroup With MRI Data Acquired at or After Registration

Because the sample size of this sensitivity analyses (n=1900) was considered to be statistically underpowered for the detailed eGFR and ACR categories as in the main analyses, eGFR was grouped into ≥60, 30 to <60, and <30 mL/min per 1.73 m2, and ACR was grouped into <30 and ≥30 mg/g. The distribution of ORs calculated in the multivariable analyses showed no relevant difference from that of the main analyses (Figure S8).

Discussion

The BAT2 study retains data for cerebral SVD based on multimodal MRI under prespecified imaging conditions for 5324 patients who were taking oral antithrombotic agents mainly for secondary stroke prevention. In the present analysis using the baseline data from BAT2, both reduced eGFR and albuminuria were independently associated with increased SVD score.

Among the 2 CKD measures, albuminuria showed significant associations with the SVD score increment at microalbuminuria as well as macroalbuminuria categories, although the clear association between SVD score and eGFR was observed only at the eGFR categories G4 and G5, when eGFR was already severely decreased. Albuminuria not only reflects glomerular damage but also is a sensitive indicator of generalized endothelial dysfunction, , and extensive studies have also been conducted on the role of endothelial dysfunction in the pathogenesis of cerebral SVD. Recently, risk scores for ischemic or hemorrhagic stroke using a component of SVD like CMBs have been given attention. Nonetheless, both albuminuria and reduced eGFR would also increase the risks for ischemic and hemorrhagic stroke. , Our results suggest that, in determining the antithrombotic‐associated bleeding risk in association with the cerebral SVD burden, these CKD measures should be included in the risk models.

Regarding each component of SVD score, lower eGFR was associated with increased white matter lesions and lacunar infarcts in both the present study and the Rotterdam Scan Study involving 484 participants ≥60 years old. In contrast, reduced eGFR was independently associated with the presence of CMBs in a hospital‐based cross‐sectional study involving 162 patients with predialysis CKD but not ours. A positive association of albuminuria with increased risk of any CMBs has been identified both in the present study and in a hospital‐based study of 285 patients with hypertension.

The relationship between the total MRI burden of cerebral SVD and body mass index has not been established. In the present analysis of BAT2, body mass index was not included in adjusting covariates because body mass index was not significantly associated with the severity of cerebral SVD.

Key strengths of this study were, first, the much larger number of registered patients than similar cohort studies and, second, the unified imaging conditions of MRI for all the patients and central diagnosis by experts with sufficient intra‐ and interrater reliability.

This study has some limitations. First, almost all participants were Asian, which might affect generalization of the present results to other ethnicities. A previous pooled meta‐analysis suggested that there were some differences in predominant underlying SVD between East Asian and Western populations. Second, unavailable data on ACR in 40.9% of the overall patient cohort might have contributed to bias in the analysis. Third, the cross‐sectional design of this study precludes investigation of a causal relationship between CKD and cerebral SVD. Last, both the patients who newly started oral antithrombotic agents and those who had been on antithrombotic medication for a certain period of time were included, and there were no data on the duration of the medication for the latter patients.

Conclusions

In conclusion, albuminuria as well as reduced eGFR were independently associated with increased cerebral SVD burden in patients who newly started or continued taking oral antithrombotic agents mainly for secondary stroke prevention. BAT2 will provide novel risk‐stratification models for antithrombotic‐associated bleeding risk in association with cerebral SVD and other biomarkers, including the CKD measures.

Sources of Funding

BAT2 was organized by a central coordinating center located at the National Cerebral and Cardiovascular Center, with funding support from the governmental Japan Agency for Medical Research and Development (18ek0210055h0003, 21lk0201094h0003, 21lk0201109h0002) and from the Japan Society for the Promotion of Science (Grants‐in‐Aid for Scientific Research 19K17023).

Disclosures

All of the following conflicts are outside the submitted work. Yakushiji reports grants from Daiichi‐Sankyo. Koga reports grants from Astellas Pharma, Daiichi Sankyo Company LTD, Nippon Boehringer Ingelheim Co., Ltd., Shionogi, and Takeda Pharmaceutical Company, Limited; and personal fees from Bayer, Daiichi Sankyo Company, and ONO Pharmaceutical Co., LTD. Hirano reports personal fees from Daiichi‐Sankyo, Bayer Yakuhin, Nippon Boehringer Ingelheim, Pfizer, and Bristol‐Myers Squibb. Toyoda reports personal fees from Bayer, Bristol‐Myers Squibb, Daiichi Sankyo, Otsuka, Novartis, and Abbott. The remaining authors have no disclosures to report.

Data S1

Tables S1–S8

Figures S1–S8

Click here for additional data file.