2. Key Takeaways
• Using genetic data to predict response to statins, the study revealed that
higher genetically predicted low-density lipoprotein (LDL) response to
statins was associated with steeper LDL lowering and lower risk of
atherosclerotic cardiovascular outcomes, including myocardial
infarction and peripheral arterial disease.
• A higher genetically predicted LDL response to statins was associated with
a higher risk of intracerebral hemorrhage (ICH) only among statin users.
No association was found with those not taking statins.
• The results imply that more aggressive lowering of LDL may increase risk
of ICH and should be balanced against statin benefits in intensive statin
treatment.
• Results support the idea that genetic data of drug response may be
leveraged to investigate side effects of drugs to understand observational
data.
3. Why This Matters
• The effect of statin-induced LDL reduction on ICH risk has been
debated for almost two decades and remains a concern among
medical professionals.
• Previous studies have shown that genetic variants responsible
for lower LDL and higher high-density lipoprotein
cholesterol levels are associated with an increased risk for ICH
over time. However, the association of shorter-term,
pharmacologic LDL-lowering with statin risk on ICH has not
been answered.
• With genetic analysis, study results provide insights about a
personalized response to statin intake and the role of
pharmacologic LDL-lowering in the pathogenesis of ICH.
4. Study Design
• In this longitudinal population-based prospective study, phenotyping
assessments, biochemical assays, genome-wide genotyping data, and
longtiduinal follow-up data from the UK Biobank were accessed.
• From the UK Biobank, 231,336 participants were selected for whom biochemical,
lipidomic, and primary care data were available. Prescriptions data from 1978
until 2019 were used, allowing detailed assessment of duration and dose of statin
intake. Individuals received at least one statin prescription between 1990 and the
end of follow-up prescriptions for dates after 1990.
• Among these patients, 15,137 individuals for whom genetic data were missing
and 155 individuals who had a history of ICH at baseline were excluded.
• Using Mendelian randomization, genomic data from randomized trials allowed
development of a polygenic score from 35 single-nucleotide polymorphisms of
on-statin LDL response, which was tested in the UK Biobank.
• Effects of the genetic score were validated on longitudinal LDL measurements
with generalized mixed models, and variable associations with incident ICH were
explored.
5. Key Results
• The average age of study participants was 57 years, and 55% were
women.
• In the study, statins were prescribed at least once to 31% of study
participants.
• Among statin users, LDL decreased by an average of 3.45 mg/dL per year.
• A higher genetic score of statin response (differing by one standard
deviation) was associated with significant reductions in LDL levels (-.05
mg/dL per year) and showed lipidomic effects on other lipid traits.
• A higher genetic score of statin response was associated with lower risk of
incident myocardial infarction (hazard ratio [HR] per SD increment, 0.98)
and peripheral artery disease (HR per SD increment, 0.92).
• A higher genetically predicted statin response among statin users was
associated with higher ICH risk (HR per SD increment, 0.92).
• There was no association with ICH risk among patients who did not use
statins (P = .89).
6. Limitations
• The constructed genetic score was associated not only with on-statin LDL
lowering but also with off-statin baseline LDL levels.
• In the study, residual confounding due to subthreshold effects on baseline LDL
could not be excluded.
• It may be necessary to interpret the findings cautiously, owing to the fact that the
incidence of ICH in the study population was lower than the worldwide incidence.
This may be due to the healthier profile of the persons in the UK Biobank.
• The study was performed mainly in people of European ancestry, and therefore
the results cannot be generalized.
• Adherence to drug intake was not factored into models.
• Statins were first introduced in 1988, but it was not until 1995 that 90% of
practices were computerized, which may have led to an underestimation of statin
doses used in models.
• A collider bias for atherosclerotic endpoints may have occurred by limiting the
cohort to statin users.