fpsyg-08-01361
August 7, 2017
Time: 15:7
# 6
Rozenblat et al.
5-HTTLPR and Disordered Eating
FIGURE 1 | Interaction between 5-HTTLPR short (s) allele and the experience
of severe parental physical punishment in predicting EDI-2 Bulimia scores.
to eating pathology. Results support the notion that correlates of
and risk factors for disordered eating symptoms show substantial
variation between males and females (
Lewinsohn et al., 2002
;
Striegel-Moore et al., 2009
). They also support past studies
linking depression (
Puccio et al., 2016
) and emotional control
(
Svaldi et al., 2012
) to ED symptomatology, and are aligned with
theories proposing that individuals may engage in disordered
eating in attempt to better regulate negative affect or other
undesirable emotions (
Stice, 2001
;
Haynos and Fruzzetti, 2011
;
Pearson et al., 2015
). Overall, results highlight the importance of
psychological factors and the role of the social environment in
eating pathology.
The tentative suggestion that
5-HTTLPR may moderate
the relationship between severe, but not mild-to-moderate
parental physical punishment and bulimic behaviors is reflected
in findings from
Rozenblat et al. (2017)
, which reported
moderation of both sexual and physical abuse by
5-HTTLPR
in predicting bulimia-spectrum pathology, with strongest effects
when both types of abuse were experienced. This suggests
that further investigation in larger independent samples in
Hardy–Weinberg Equilibrium (the expected allele distribution
in a given population, deviation from which may compromise
validity of results) may well support moderation of more
extreme forms of adversity by
5-HTTLPR. Further support for
the idea that
5-HTTLPR might moderate more severe forms
of risk for disordered eating come from past studies in the
ED field that found traumatic life events were associated with
bulimic symptoms and disordered eating for individuals with the
5-HTTLPR s-allele (
Akkermann et al., 2012
;
Stoltenberg et al.,
2012
), and is reflected by the focus on traumatic life events and
sexual abuse in the depression field (
Nugent et al., 2011
).
Lack of genetic moderation of depression or emotional control
in predicting drive for thinness and bulimic tendencies is mostly
consistent with previous findings (
Rozenblat et al., 2017
). These
results align with the secondary data analysis in
Rozenblat
et al. (2017)
, and suggest that the null findings for depression
and impulsiveness reported in the secondary data analysis were
likely not due to sample heterogeneity or use of the broad
impulsiveness variable as opposed to examining a personality
construct that is more closely associated with eating pathology,
such as emotional control. However, sample size limitations
mean that the presence of small effects cannot be entirely
ruled out and further investigation in larger samples remains
important.
The lack of genetic moderation reported for depression
does, however, contradict the significant G×E interactions
between depression and
5-HTTLPR identified in two past studies
(
van Strien et al., 2010
;
Mata and Gotlib, 2011
); however, the
sample of
Mata and Gotlib (2011)
(
N = 50) was very low
for investigation of genetic association (
Duncan and Keller,
2011
), while
van Strien et al. (2010)
examined emotional eating,
which differed somewhat from the eating constructs measured
in the present study. One possibility is that as psychological
factors appear to be strong direct predictors of eating pathology,
they may function as risk factors irrespective of
5-HTTLPR
genotype. In contrast, certain environmental factors may have a
more tenuous association with ED symptoms and thus plausibly
could increase risk primarily for individuals with a genetic
susceptibility.
The absence of direct genetic association in this study also
partly conflicts with previous findings (
Calati et al., 2011
;
Chen
et al., 2015
). Direct genetic prediction of ED has been investigated
in several past studies examining clinical populations with mixed
results. Two meta-analyses identified a direct association between
5-HTTLPR and eating pathology (Odds Ratio: 1.35, 95%CI:
1.07−1.71,
Calati et al., 2011
;
Chen et al., 2015
), although they
examined almost entirely the same group of studies, while the
largest and most recent meta-analysis on this topic reported no
association (
Solmi et al., 2016
). Notably, these meta-analyses were
limited by substantial heterogeneity, the inclusion of studies with
very small sample sizes (
N
< 100), and omission of tests for
publication bias. Publication bias is noted to be a major problem
affecting studies of G×E interactions and contributing to false-
positive findings (
Duncan and Keller, 2011
;
de Vries et al., 2016
),
with such issues argued to most strongly affect studies with small
sample sizes (
Ioannidis et al., 2014
).
Strengths and Limitations
Strengths of the present study include use of a homogenous,
high-quality data set, with measurement of drive for thinness and
bulimic tendencies in a community sample. Results are therefore
of key relevance to aiding prevention of the development of
clinical-level eating pathology. A further strength was the fact that
the present study constituted the second largest unified sample
investigating G×E interactions in eating pathology, following
Akkermann et al. (2011)
(
N = 767), with mean sample size of
existing ED G×E studies
N = 288 (
Rozenblat et al., 2017
). This
study was powered to detect direct and interaction effects of
moderate size, which would be of clinical significance if detected.
In light of growing evidence that genetic effect sizes involved in
psychiatric disorders are exceedingly small, even larger samples
are desirable. Methodological issues include use of the di-allelic
Frontiers in Psychology | www.frontiersin.org
6
August 2017 | Volume 8 | Article 1361