WSO - January 2024 - 107

Lin et al.
107
test, or Fisher exact test with correction, as appropriate. A
standard mean difference was calculated between the two
groups. The OR for the association of the NOTCH3 R544C
variant with self-reported neuropsychiatric diseases and
relevant family histories was then calculated using the
epitools package (version 0.5.10.1) in R.13
We then divided the NOTCH3 R544C group into carriers
with and without stroke histories. Chi-square and
unpaired two-sided t-tests were performed to analyze categorical
and continuous cardiovascular risk factor variables,
respectively. Finally, a multivariable logistic regression
model was constructed to assess the independent risk factors
for stroke in the NOTCH3 R544C group, with the stepwise
selection of categorical variables based on a p-value
of <0.10 for entry and of <0.05 for stay, with age and sex
forced into the model. All analyses were performed using R
(version 4.1.2).
Polygenic risk score analysis
In order to assess the risk associated with NOTCH3 variants
compared to common stroke variants, the data from the
carriers and matched noncarriers were compared using the
PRS model.14 We utilized a PRS constructed by a metaanalytic
strategy for ischemic stroke (metaGRS) that captures
the totality of information from five stroke outcomes
and 14 stroke-related phenotypes GWAS summary statistics,14
and analyses this score in the Taiwan Biobank data.
Taiwan Biobank samples were genotyped using the Axiom
Genome-Wide TWB array and imputed using SHAPEIT4
(version 4.1.2) and IMPUTE2 (version 2.3.2) with a merged
reference panel consisting of data from an East Asian population
from 1000 Genomes Project phase 3 (n = 504) and
Taiwan Biobank whole-genome sequencing (n = 1451). The
weighted PRS was calculated using PLINK (version 1.9).
All relevant SNPs were included in the calculation, regardless
of their imputation quality. On a case-control basis, the
predictivity of each PRS unit for stroke was calculated
between stroke (defined as a binary trait) and standardized
PRS via logistic regression. We compared the odds ratio
(OR) for stroke in individuals carrying NOTCH3 variants
versus noncarriers, as well as the OR for one standard deviation
(SD) increase in the ischemic stroke PRS to assess the
risk associated with NOTCH3 variants. The Z-score was
then calculated to estimate the equivalent PRS standard
deviations of carrying the NOTCH3 R544C variant.15
Standard protocol approval, registration,
and patient consent
This study was approved by the National Taiwan University
Hospital Research Ethics Committee B (approval number
202103031RINB). Relevant data were acquired from the
Taiwan Biobank (TWBR11004-02) after endorsement from
an independent research ethics committee. Participant
consent for data use was obtained by the Taiwan Biobank at
enrollment.
Results
Characteristics of R544C carriers and
matched controls
Of the 132,715 individuals who completed the questionnaire
and 114,599 who underwent genotyping, 114,282 had
both a complete questionnaire at enrollment and genotypic
SNP results for NOTCH3 R544C. Of them, 1080 (0.95%)
participants carried the NOTCH3 R544C variant. Through
propensity score matching, 10,800 controls were matched
from the 113,202 noncarriers. The flowchart of the current
study is presented in Figure 1.
NOTCH3 R544C carriers (n = 1080, mean age
49.8 ± 10.9 years) were paired with noncarriers (n = 10,800,
mean age 49.7 ± 10.9 years). The NOTCH3 R544C carriers
and matched noncarriers exhibited similar age, sex, education
levels, alcohol consumption, smoking status, medical
history, and biochemical laboratory data, with an absolute
standardized mean difference of mostly under 0.1 (Table 1).
Systolic blood pressure was slightly higher in the matched
control group; on the other hand, serum uric acid levels
showed marginal elevations among the carriers.
Phenotypic presentations of NOTCH3
R544C carriers
The proportion of participants with histories of stroke
(1.48% vs. 0.59%) and dementia (0.28% vs. 0.009%) was
higher in the carriers than in the noncarriers (Table 2). The
propensity score-adjusted OR (95% confidence interval
(CI), p value) for stroke in carriers compared to matched
noncarriers was 2.52 (1.45, 4.37), p < 0.001), but that for
dementia was 30.1 ((3.13, 289.43), p < 0.001). Between the
two groups, Parkinson's disease, migraine, schizophrenia,
and other neuropsychiatric disorders did not differ significantly
(p > 0.05).
After stratification by age at 10-year intervals, we noted
a serial increase in adjusted OR (95% CI) for stroke in the
R544C carriers; it was 1.11 ([0.14, 8.83), p = 0.99), 2.8
((1.20, 6.54), p = .023), and 3.2 ((1.43, 7.16), p = 0.009) in
the 40 to 49, 50 to 59, and 60 to 69 years age groups, respectively.
Overall, the OR (95% CI) was 2.54-fold ((1.46,
4.42); ptrend = 0.0006, Cochran-Mantel-Haenszel test) for
every 10-year increase in age.
NOTCH3 R544C variant carriers were more likely than
their matched noncarrier counterparts to have a family history
of stroke (adjusted OR: 2.06, 95% CI (1.80, 2.36),
p < 0.001) and dementia (adjusted OR: 1.28, 95% CI (1.02,
1.61), p = 0.034) (Table 2). The OR (95% CI) for the family
history of stroke was higher in second-degree relatives than
in first-degree relatives (second-degree: 2.48 (1.85, 3.34);
p < .001 vs. first-degree: 1.98 (1.72, 2.27); p < .001).
International Journal of Stroke, 19(1)
WSO - January 2024

Table of Contents for the Digital Edition of WSO - January 2024
