|
ANDROGENS AND CARDIOVASCULAR DISEASE: CAN WE MAKE SENSE OF THE
EPIDEMIOLOGICAL DATA?
Yoav Ben-Shlomo
Department of Social Medicine, University of Bristol, Canynge Hall,
Whiteladies Rd., B58 2PR, UK Y.Ben-Shlomo@bristol.ac.uk
There is a long history of clinicians being interested by the potential
therapeutic benefits of androgens. In the early 1940s several case
reports appeared in the literature suggesting that patients with
angina reported an improvement of symptoms after injections of testosterone
propionate, though none of these were randomly controlled trials.
The existing epidemiological literature that has examined an association
between androgens and cardiovascular disease (CVD) is almost solely
based on observational data and is not simple to synthesise. Many
of the reported studies use either a cross-sectional or case control
design. This is problematic. Firstly such studies only recruit living
patients hence cases who have a fatal cardiac event are by definition
excluded. This could result in a survivor bias if the association
between androgens and CVD is stronger for fatal rather than non-fatal
disease. Secondly, they cannot exclude the possibility of reverse
causality so that androgen levels may be altered subsequent to disease
onset. Thirdly, the choice of controls is often problematic and
can result in selection bias so that levels may differ because controls
differ on other confounders, which may be associated with androgen
levels. A review by Alexandersen and colleagues 1 identified 30
cross-sectional studies combining IHD, atherosclerosis and angina
as outcome measures. 18 found reduced levels of testosterone or
DHEA in cases, 11 found similar levels and 1 found an increase in
DHEA. Many studies were small and underpowered.
More informative reports have been based on either prospective studies
where healthy subjects have had their androgen levels measured at
baseline and are then followed up over time or case control studies
that are "nested" within cohort studies or RCTs. In both these designs,
fatal cases can be included as the exposure is measured before disease
onset and we can assume that reverse causality is unlikely to be
important. There are only 12 such publications from 9 studies with
very little data on women. Many of these studies have been reported
as essentially refuting any association between androgens and heart
disease. An editorial on this topic noted "…prospective studies
suggest that whereas low testosterone levels are associated with
an increase in biochemical risk factors of cardiovascular disease,
they are not associated with an increased risk of cardiovascular
mortality or morbidity."2 This is surprising as the data on androgens
and cardiovascular risk factors is far more consistent. If higher
androgen levels are associated with a more favourable risk factor
profile, why is this not translated into a reduced risk of cardiovascular
events?
The first cohort report from the Rancho Bernardo study noted an
inverse association between DHEAS and fatal CHD but this was not
confirmed by a later report using total testosterone though it had
wide confidence intervals. Several "negative" nested case control
studies followed from the MRFIT, US Health Professionals and Helsinki
studies. An intriguing report also came from the long term follow-up
of the Honolulu cohort study. It noted an overall 20% reduction
with increasing DHEAS but this was compatible with chance. A sub-group
analysis reported a significant reduction in risk for fatal CHD
cases (odds ratio 0.45) but no reduction for non-fatal CHD cases
(odds ratio 1.11). The authors assumed that these differences were
real rather than due to sampling variation and concluded that low
DHEAS was a non-specific marker of frailty and mortality.
The Caerphilly Prospective study is a population cardiovascular
cohort of around 2500 men that was set up in 1979. It initially
reported an inverse association between prevalent heart disease
and total testosterone, though a subsequent report looking at incident
cases concluded that there was no association with IHD. However
a recent re-analysis with over twice as many IHD events (320 cases)
did find an inverse association with testosterone though the evidence
for this was modest.3 This is the most statistically powerful study
to date and highlights the problem with small studies that may not
have sufficient power to detect weak effects. There was no evidence
of a differential association between fatal and non-fatal events
refuting the previous conclusions from Honolulu. Stronger associations
were seen using the ratio of cortisol to testosterone as a biomarker
of chronic stress. These associations were markedly attenuated after
adjustment for the components of the insulin resistance syndrome,
highlighting a possible intermediary pathway though it is possible
that insulin resistance is the primary pathophysiological driver.
Recent reports have examined carotid atherosclerosis as a surrogate
outcome. A Dutch cohort found not only that free testosterone was
inversely associated with atherosclerosis in a cross-sectional analysis
but also that it predicted progression over time. In conclusion,
there is a modest inverse association between androgens and CVD
risk in men. It is possible that the observed associations under-estimate
the true risk. Older studies have used crude measures of androgen
activity only measured on one occasion. This will poorly characterise
an individual's life course exposure and hence introduce random
error. Many of the old assays had relatively large coefficients
of variation, which would further attenuate any true associations.
New studies are required with better endocrine measures done repeatedly
over time and utilising sub-clinical intermediary markers such as
carotid IMT, endothelial dysfunction and pulse wave velocity.
The gold standard of epidemiological evidence is the randomised
controlled trial but to show that androgens are useful for primary
prevention would necessitate a large RCT of healthy men receiving
androgens, which presents major logistical and ethical issues. We
must therefore try to improve the quality of observational data.
References
1. Alexandersen P, Haarbo J, Christiansen C. The relationship of
natural androgens to coronary heart disease in males: a review.
Atherosclerosis 1996; 125: 1-13.
2. Rosano GMC. Androgens and coronary artery disease. A sex specific
effect of sex hormones. Eur Heart J 2000;21:868-871
3. Davey Smith G, Ben-Shlomo Y, Beswick A, Yarnell J, Lightman S,
Elwood P. Cortisol, testosterone and coronary heart disease: prospective
evidence from the caerphilly study. Circulation 2005;112:332-340.
|
