MEASURING GENETIC SUSCEPTIBILITY
TO PSYCHIATRIC DISORDERS
Gene— environment interactions
It is self-evident that our genes do not wholly
determine our development. Even identical twins,
who share 100% of their genetic make-up, are not
exactly the same in their personality or propen-
sity to develop psychiatric disorders. But how do
our family circumstances, the unpredictable events
that happen in our lives, and our genetic-make-up
interact? Can we meaningfully predict that some
people with a particular genetic predisposition,
at the level of a single gene polymorphism, will
be vulnerable – but
only
if exposed to risky envi-
ronments? For instance, is it true that children
with a particular genetic variant of the monoamine
oxidase A (
MAOA
) gene are much more likely
to develop antisocial behaviour in adulthood if
subject to maltreatment in childhood than if they
did not possess the variant [11]? Should we warn
young people who have a polymorphism of the
catechol
O
-methyltransferase (
COMT
) gene that
they should not smoke cannabis because of a dis-
proportionate increased risk of psychosis [12]?
Do genetic and environmental risk factors com-
bine in ways that lead to a relatively greater proba-
bility of outcome than simply the arithmetic sum of
the individual risks? The controversial argument
from the studies quoted above is that the simple
sum of the risks (gene
+
environmental exposure)
on outcome is much
less
than the observed risk. In
other words, some interaction must have occurred,
between the genetic and the environmental vari-
able that had increased the probability of a delete-
rious outcome disproportionately, and may imply
that these factors had interacted in some way at a
biological level.
Interactions between variables are commonly
modelled to predict outcomes in epidemiological
studies, but many scientists regard gene
×
envi-
ronment interactions as tenuous things, which are
not necessarily biologically real. Such interactions
could instead represent statistical artefacts, and we
may not be correct in assuming we can infer bio-
logical interactions from statistical analyses of this
type [13]. While evidence of non-independence at
a physiological level informs how genetic and other
risk factors should be modelled in epidemiological
studies, the opposite is not true. In other words,
we should not infer a biological mechanism from
epidemiological evidence of a ‘genotype
×
expe-
rience’ interaction. Non-linear summation of risks
may prompt further investigation as to whether a
biological interaction exists, but the observation of
an interaction is not conclusive that there is such
a mechanism operating. This somewhat sceptical
view is reinforced by the observation that many
apparently exciting and novel results in psychiatric
genetics fail to be replicated, for a variety of rea-
sons including over-optimistic data analysis and
publication bias [14]. This issue of non-replication
is discussed in the next section.
Do'stlaringiz bilan baham: |