WSO December 2023 – Issue 3 - 1221

Moutzouri et al.
1221
Thrombosis and Hemostasis criteria and was independently
adjudicated according to the Swiss-AF main protocol.17,18
Primary aim: to investigate the association between statin
exposure, defined as yes or no at baseline and outcomes
as described above. The secondary aim focused on the
association of the abovementioned primary outcomes with
blood lipid levels (primarily low-density lipoprotein (LDL)
but also total cholesterol (TCHOL), triglycerides (TRG),
and high-density lipoprotein (HDL)) at baseline as continuous
variables.
Statistical analysis
Baseline characteristics are presented separately for statin
users (participants with any statin use at baseline) and nonusers.
Association of statin use or lipid levels (TCHOL,
LDL-C, HDL-C, and TRG separately, each as a continuous
variable per 1-unit increase) with the prevalence of CMBs
at baseline, as well as the progression of CMBs at followup,
was assessed using logistic regression models. The
results are presented as odds ratios (OR, 95% confidence
intervals (CI)). For the occurrence of ICH during followup,
we used flexible parametric survival models. The
results are presented as hazard ratios (HR) with 95% CI.19
All analyses were adjusted for confounding factors potentially
associated with both the exposure (statin and blood
lipid levels) and the outcome (CMBs) using directed acyclic
graph (DAG) models based on prior knowledge of
association with CMBs: sex, age, history of hypertension,
smoking status, body mass index (BMI), history of diabetes,
history of stroke or transient ischemic attack (TIA),
coronary heart disease, antiplatelet use, anticoagulant use,
and education. Due to the low number of events, flexible
parametric survival models for ICH were only adjusted for
sex and age.
In the main analysis, participants were categorized as
follows: baseline statin use versus no use. Further analyses
were done for high-dose baseline use versus low- and
medium-dose use versus no use (high, medium, low statin
doses as defined from ACC/AHA).20 All medications were
assessed at baseline and at years 1 and 2. We performed
time-updated sensitivity analyses by change in statin (stop/
start) and change in statin intensity during follow-up.
Supplemental analyses were performed focusing on
lobar CMBs, based on previous observations suggesting
that statin and LDL levels might be primarily associated
with a higher bleeding risk in the lobar brain regions.21
Additional analyses were performed by restricting the
investigation to patients under oral anticoagulants.
Power calculation is reported in the supplemental material.
All data analyses were carried out using STATA version
16.0. No correction was performed for multiple testing.
The study is reported following the STROBE guidelines for
cohort studies.22
Results
Patients characteristics
Of the 2415 patients enrolled in Swiss-AF, 667 did not have
brain MRI at baseline. The main reason for a missing brain
MRI was the presence of a cardiac device (N = 461, 69%).
Other reasons included further contraindications to perform
an MRI or claustrophobia of the patient. Fifty-five patients
had brain MRI without CMB data. A total of 1693 (468
[27.6%] females, mean [SD] age 72.5 years [8.4], 90.1%
under oral anticoagulants) having CMB data at baseline
MRI were included (Figure 1).23 Statin users (N = 802) had
more hypertension (79.7% vs 59.9%), more diabetes
(23.6% vs 8.6%), more often a history of stroke/TIA (29.1%
vs 11.1%), and other known cardiovascular disease (44.4%
vs 9.2%). The percentage of patients under oral anticoagulation
was 92.1% in the statin and 88.3% in the non-statin
user group. The use of vitamin K antagonist (VKA) compared
with direct oral anticoagulants (DOACs) was higher
under statin users (Table 1). The cohort represented an
intermediate bleeding risk population.24 A total of 72% of
statin users versus 74% of non-statin users had 3.0 T
scanner.
At 2 years, 1188 of the 1693 identified patients had follow-up
MRI (Figure 1). The reasons for declining 2 years
MRI included the following: the patient developed contraindications
for MRI (e.g. pacemaker), patient was no
longer able to come to the hospital on his own (follow-up
visit carried out by telephone), patient developed claustrophobia
or had unpleasant experience from MRI at baseline,
patient did not want to come for other reasons, and patient
withdrew or was deceased. For example, at the study site of
Bern (N = 368), a telephone visit was performed in 15% of
the participants at 2 years follow-up. Eighty (3.3%) patients
had no 2-year follow-up visit.15 The characteristics of
patients who had baseline and follow-up MRI were comparable
(Supplemental Tables S1 and S2).
Cross-sectional analysis
At baseline, 213 (26.5%) statin users and 163 (18.3%) nonstatin
users were found to have at least one CMB. After
adjustment for history of hypertension, smoking status,
BMI, history of diabetes, history of stroke or TIA, coronary
heart disease, antiplatelet use, anticoagulant use, and education
no significant association was observed (adjOR,
1.10 (95% CI = 0.83-1.45)) (Table 2). Similar were the
results for lipid levels (Supplemental Table S3). Covariates'
association with CMBs prevalence is shown in Supplemental
Table S4. No association for lobar CMBs at baseline was
observed, neither for statin use (Table 2) nor for LDL levels
(Supplemental Table S5). Also, no association was observed
by restricting the analysis to patients under oral anticoagulants
(Supplemental Table S6).
International Journal of Stroke, 18(10)
WSO December 2023 – Issue 3

Table of Contents for the Digital Edition of WSO December 2023 – Issue 3
