Gonadal Hormone Contributions to Individual Differences in Eating Disorder Risk

Abstract

Purpose of review:

Females experience eating disorders at substantially higher rates than males. While sociocultural factors have traditionally been thought to underlie this sex disparity, accumulating evidence implicates differential exposure to gonadal hormones early in life. Gonadal hormones also impact within-sex variability in disordered eating, helping to explain why not all women develop an eating disorder, and some men do. We review recent findings regarding these gonadal hormone effects and their implications for the etiology of eating disorders.

Recent findings:

Males are exposed to significantly higher testosterone levels than females perinatally, and this exposure appears to protect against later binge eating in males relative to females. Within-sex, higher estradiol levels among females and higher testosterone levels among males appear to be protective. Progesterone exhibits minimal direct phenotypic effects on disordered eating, but appears to counteract the protective effects of estrogen in adult females. Importantly, gonadal hormone effects may be moderated by psychosocial factors.

Summary:

Evidence suggests gonadal hormones play a critical role in the etiology of disordered eating. Overall, higher testosterone and estrogen appear to be protective across development. Additional research is needed to identify mechanisms underlying these effects and further explore interactions between hormonal and psychosocial risk.

Introduction

Eating disorders are among the most sexually dimorphic mental illnesses, with females affected at 3-10 times the rate of males in community samples [1]. Historically, sociocultural factors (e.g., greater pressure on women to conform to a thin ideal) have been theorized to underlie this disparity [2]. However, purely sociocultural explanations are complicated by the fact that similar sex disparities in some core eating disorder behaviors (e.g., binge-like eating of palatable food (PF), or foods high in fat and sugar) are observed in non-human animals not subject to Western beauty ideals [3,4**]. Further, sex differences in binge eating phenotypes become significantly more pronounced during puberty in animal models [4**], the same developmental phase in which prevalence of eating disorders increases sharply among girls [5]. Together, these cross-species findings point to the role of biologically-based developmental processes that differ between males and females in differential risk for eating disorders.

Accumulating evidence over the past two decades has implicated gonadal hormones (e.g., estrogen, progesterone, testosterone) as key contributors to between-sex differences and within-sex variability in disordered eating [5,6]. Though gonadal hormones are best known for their role in fertility and reproduction, they also have wide-ranging effects on brain development and functioning, including the patterning of sexually differentiated behavior [7-9]. In particular, gonadal hormones influence food intake [10-12] and neurological (e.g., serotonergic and dopaminergic [13-15]) and neuroendocrine (e.g., the hypothalamic-pituitary-adrenal axis [16]) systems thought to be altered in eating disorders [17,18].

In this article, we review recent evidence for gonadal hormones as critical mechanisms underlying both between-sex and within-sex variability in disordered eating. We take a developmental approach to this literature, highlighting different ways gonadal hormones may influence risk for disordered eating across different life stages. In particular, we draw on an established framework positing that hormones can have both organizational and activational effects [8,19]. Organizational effects occur during sensitive developmental periods (e.g., prenatally, during adolescence), and shape the structure and function of biological systems in a way that persists even when the organizing hormones are no longer present. Importantly, organizational effects may not become phenotypically apparent until later in development (i.e., late puberty, adulthood). By contrast, activational effects are more transient, and reflect the immediate influence of gonadal hormones on physiology and behavior. Organizational hormone effects “set the stage” for later activational effects, which may be disrupted if normative organization fails to occur.

When possible, we incorporate findings from both human and animal studies, as research with non-human animals provides important corroborating evidence for biological effects independent of potentially confounding sociocultural influences. We conclude by discussing future directions that will help further clarify how gonadal hormones mechanistically impact risk for eating disorders, and for whom these effects are most pronounced.

Gonadal hormones and the female predominance of eating disorders

Males and females are exposed to distinct hormonal milieus perinatally, with males experiencing substantially higher levels of the testicular hormone testosterone. These differential exposures have been shown to shape sexually-differentiated behaviors in non-human animals [8,19], including behaviors relevant to eating disorders such as food intake [20,21] and preference for sweet tastes [22]. While human behavior is impacted by social/cultural as well as biological factors, the extensive animal literature on perinatal hormone effects provides a rationale for hypothesizing that between-sex differences in disordered eating may be due in part to differential testosterone exposure early in life.

Recently, research on perinatal testosterone has been extended to binge eating phenotypes. While male rats eat and weigh more than females on average [11], female rats are 2-6 times more likely than males to be prone to binge eat on PF in adulthood [3] (i.e., consume a large amount of PF in a short period of time), mirroring sex disparities in binge eating in humans [1]. To examine whether lower perinatal testosterone exposure in females accounts for this increased risk, Culbert et al. compared rates of binge eating prone (BEP) phenotypes in male rats, female rats, and female rats treated with testosterone shortly after birth [4**]. They found that untreated female rats showed higher rates of BEP phenotypes in mid/late puberty and adulthood than male rats or testosterone-treated female rats, which were not significantly different from each other. Interestingly, the effects of perinatal testosterone did not become apparent until mid/late puberty. This suggests a protective organizational effect of perinatal testosterone that does not manifest behaviorally until adolescence, the developmental period when sex differences in rates of disordered eating emerge in humans.

Though perinatal hormones cannot be experimentally manipulated in humans, opposite sex (OS) twin pairs provide an informative natural experiment. Females from OS twin pairs are thought to be exposed to higher levels of testosterone in utero than a typical singleton female due to the presence of their male co-twin [23]. If prenatal testosterone protects against later disordered eating, these female twins would be expected to be less prone to disordered eating in adolescence and adulthood than females without a male twin. Some studies have reported findings consistent with this theory, demonstrating lower disordered eating symptoms (i.e., as measured by a composite symptom score assessing body dissatisfaction, weight preoccupation, binge eating, and compensatory behavior) among women from OS twin pairs than women from same sex twin pairs and/or women raised with a non-twin brother [24,25]. However, conflicting findings exist [26-28], and studies of index/ring finger length ratios (i.e., 2D:4D ratios), another proxy for in utero testosterone exposure, have shown mixed results [29-33].

There are several possible reasons for contradictory findings. One possibility is that the effects of perinatal testosterone on dysregulated eating are non-linear, and even girls exposed to relatively high testosterone levels prenatally do not reach the threshold necessary for a consistent protective effect. There may also be differences in relationships between prenatal testosterone and specific eating disorder symptoms or diagnoses [30]. In particular, there is some evidence that while bulimia nervosa (BN) is associated with lower prenatal testosterone exposure in women, anorexia nervosa (AN) may be associated with higher prenatal testosterone exposure [30]. This could indicate differential influences of prenatal testosterone on binge eating and restricting phenotypes.

Culbert et al. explored another possibility – namely, that the protective effects of prenatal testosterone vary by developmental stage [34]. Consistent with this hypothesis, Culbert and colleagues observed that girls and women from OS twin pairs and with more masculinized 2D:4D ratios had lower scores on a composite measure of disordered eating symptoms in early adolescence and young adulthood, but not in late adolescence [34]. The authors posited that the biologically protective effects of prenatal testosterone are lessened during periods of elevated psychosocial and/or neurobiological risk for eating disorders, such as late adolescence. This possibility highlights the need for additional research exploring moderation of gonadal hormone effects on disordered eating by psychosocial factors that may differ across developmental stages and sociocultural environments.

Gonadal hormones and within-sex variability in eating disorder risk

While females are at higher risk for eating disorders than males as a group, there is also significant variability in disordered eating within sex. Gonadal hormones appear to play a role not only in different rates of disordered eating between sexes, but also in determining who is at greatest risk for disordered eating within each sex. Notably, gonadal hormone influences on eating behavior after the perinatal period are somewhat sex specific [22,35]. This may be because higher perinatal testosterone exposure organizes the male central nervous system (CNS) to be more responsive to androgens later in development, while a relative lack of perinatal testosterone organizes the female CNS to be more responsive to estrogen and progesterone. Thus, we focus our discussion of within-sex effects on estrogen and progesterone in females and androgens in males. Because the majority of this research has focused on females, we begin our review there, followed by discussion of emerging findings in males.

Within-sex effects in females

During gonadarche (the phase of puberty in which secondary sex characteristics develop due to increased hormone production from the gonads), genetic influences on a composite measure of disordered eating symptoms [36] and binge eating in particular [37] increase significantly in girls (from 0% heritability in pre/early gonadarche to ~50% heritability in mid/late gonadarche and adulthood [36,37]). However, recent evidence indicates that not all girls experience this etiological shift to the same extent. Instead, increases in heritability may be driven primarily by girls who experience relatively low levels of estrogen during gonadarche. In the first large-scale study to examine ovarian hormones as a moderator of genetic effects during puberty, genetic influences on binge eating were minimal in girls with relatively high estradiol levels (2% heritability) during gonadarche, but much greater in girls with relatively low estradiol levels (69% heritability), even after accounting for age, BMI, and stage of pubertal development [38**]. Some genetically-based sources of risk for binge eating may therefore not be behaviorally expressed unless a girl is exposed to a risky hormonal milieu (i.e., low estrogen) during gonadarche.

Mechanistically, lower levels of estrogen may organize developing neural systems in a way that potentiates latent genetic risk in females. In support of this hypothesis, a recent animal study from our group found significantly increased rates of BEP phenotypes in mid-late adolescence and adulthood following pre-pubertal ovariectomy (OVX) in female rats (Klump KL, Sinclair EB, Hildebrandt BA, et al., unpublished data). In contrast, OVX in adulthood does not change the relative proportions of BEP/BER phenotypes (Klump KL, Van Huysse J, Kashy D, et al., unpublished data). Together, these findings suggest that a female’s tendency to be relatively prone or resistant to binge eating may be set (i.e., organized) by estrogen levels during gonadarche, with higher estradiol levels protecting against genetically mediated risk.

Research in both humans and non-human animals also points to a protective activational effect of estrogen in mid/late adolescence and adulthood. Women experience natural changes in ovarian hormones across their menstrual cycle, allowing for examination of hormonal effects on disordered eating on a monthly basis in adulthood. Studies show that binge eating [39] and emotional eating [40,41] tend to be lowest when estrogen is high and progesterone (which antagonizes the protective effects of estrogen) is low, with stronger effects in women who experience clinically significant binge eating [42]. Animal studies involving experimental OVX broadly corroborate these findings. While the tendency to be relatively prone or resistant to binge eating appears to be organized during puberty (i.e., hormone changes in adulthood do not change these phenotypes – see above), both BEP and BER rats consume more PF following adult OVX [43], consistent with protective, activational effects of estrogen in adulthood. These increases in PF intake following adult OVX are accompanied by neurobiological correlates. In a recent study by Micioni Di Bonaventura et al. [44**], female rats that underwent OVX without hormone replacement showed more binge eating and neural activity in brain regions involved in stress, affective processing, and eating behavior (i.e., the central nucleus of the amygdala, paraventricular nucleus of the hypothalamus, and bed nucleus of the stria terminalis) than OVX animals treated with replacement estrogen. Intriguingly, these effects were most apparent in animals that had a history of food restriction and frustrative non-reward stress, suggesting potential interactive effects between ovarian hormones and known psychosocial/behavioral risk factors for disordered eating (i.e., dieting, stress [45,46]).

Progesterone, in contrast, is associated with greater risk for disordered eating in adulthood. While progesterone is unrelated to genetic influences on binge eating during puberty [38**], adult women with higher progesterone levels show stronger genetic influences on emotional eating than women with lower progesterone levels [47], and genetic influences on emotional eating increase during phases of the menstrual cycle characterized by high progesterone [48]. However, these genetic effects may not be solely due to progesterone, as progesterone appears to influence phenotypic risk for disordered eating indirectly through its effects on other risk or protective factors. As discussed above, progesterone exacerbates binge eating behaviors across the menstrual cycle by countering the protective effects of estrogen, which is evidenced by significant interactions between estradiol and progesterone in studies with direct measures of hormones [40,42]. There is also emerging evidence that progesterone may interact with psychosocial factors to determine risk. For example, the relationship between social pressures (i.e., weight-related teasing, beliefs that popularity is linked to appearance) and body image concerns was recently found to be moderated by progesterone in adolescent girls, such that the association was stronger among girls with higher progesterone levels [49*]. Consideration of interactions between hormones and environmental/psychological risk is therefore necessary to achieve a complete understanding of gonadal hormone influences, particularly because such hormones may exert their effects in part by altering perception of social/emotional stimuli [50,51].

Within-sex effects in males

Emerging evidence suggests that androgens may not only protect against disordered eating during the perinatal period (contributing to between-sex differences in risk), but may also be protective in males at later stages in development (contributing to within-sex differences in risk). While genetic influences on a composite measure of disordered eating symptoms are constant from gonadarche into adulthood in males (~50% heritability) [52], boys experience increases in genetic effects during adrenarche [53**], an early stage of puberty characterized by rising adrenal androgens (i.e., dehydroepiandrosterone, dehydroepiandrosterone-sulfate, and androstendione) that begins prior to gonadarche and observable physical changes [54]. These findings suggest that adrenal androgens may have organizational effects in males that parallel the theorized effects of estrogen during gonadarche in females. Interestingly, while both males and females undergo adrenarche, only males experience increases in genetic influences on disordered eating during this time. This is consistent with the idea that the influence of postnatal androgens on development and behavior may depend on perinatal testosterone exposure. However, additional studies that directly measure hormone concentrations are needed to confirm that changes in genetic effects in boys during adrenarche are related to androgens (rather than other developmental factors), and that higher androgen levels are protective (as appears to be the case for higher estrogen among girls during gonadarche).

While genetic influences on disordered eating do not change in boys after adrenarche, circulating androgens during/post-puberty may have a protective phenotypic effect. Indeed, boys with higher circulating testosterone levels in mid/late puberty score lower on a composite measure of disordered eating symptoms [55]. Together, these data suggest that androgens may protect against disordered eating through both organizational (during the perinatal period and adrenarche) and activational (during later puberty/adulthood) mechanisms, but additional longitudinal studies involving direct measurement of androgens are needed.

Future Directions: Moderators and Mechanisms of Gonadal Hormone Effects

Research on gonadal hormone influences on disordered eating has significantly advanced etiological models of eating disorders, particularly with respect to binge eating and related phenotypes. Overall, higher levels of testosterone and estrogen appear to have protective effects across development. Nevertheless, important questions remain. Perhaps most notably, the exact mechanisms through which gonadal hormones impact genetic and phenotypic risk for disordered eating are still unknown. Gonadal hormones regulate several neurobiological systems involved in reward processing [14,56], mood [13,15,57], weight/eating [11,21], and the stress response [16], but it is not yet clear which of these systems are most relevant to changes in genetic risk for disordered eating during development and phenotypic and genetic effects in adulthood. Recent data suggest one important intermediary system might be the default mode network (i.e., a neural system involved in self-referential thinking that becomes more active when an individual is not engaged in externally-focused cognitive tasks [58]), as connectivity in this network showed reduced responsivity to changing hormone levels across the menstrual cycle in women with emotional eating [59*]. However, additional longitudinal neuroimaging studies across critical points in development and distinct hormonal milieus (e.g., pre- and post-ovulation) with direct measures of hormones are needed to clarify the neural circuits involved in hormonal regulation of disordered eating. Animal models may also provide important evidence regarding the impact of hormones on gene transcription in candidate neural circuits. For example, female BEP rats show differential neural responses to PF than their binge eating resistant counterparts [60], and future research could examine whether ovarian hormones modulate these effects.

Second, there is accumulating evidence that hormonal effects on disordered eating may be moderated by what have been typically thought of as psychosocial risk factors. Careful consideration of possible interactions between hormonal/biological, psychological, and environmental risk is therefore necessary to understand when and for whom gonadal hormones may be most influential. While studies examining moderation by social pressures and stress/dietary restriction provide an important start, one salient potential moderator that has yet to be explored is emotion/emotion regulation. Neural systems involved in affect and emotion regulation are impacted by gonadal hormones [50,61] and undergo significant maturation during adolescence [62,63] alongside the emergence of phenotypic gonadal hormone effects on disordered eating. Negative affect and difficulties with emotion regulation also strongly predict disordered eating [64,65], but whether affective and hormonal risk factors for disordered eating are additive or interactive is unknown. Examining how these factors may influence each other could enhance affective and biological models of disordered eating, and perhaps provide a way to attenuate hormonal risk through psychosocial intervention.

Finally, research is needed to investigate whether similar hormonal effects are seen across culturally/socioeconomically diverse groups. Prior research has shown that individuals who differ on race/ethnicity show different patterns of associations between environmental/psychosocial stressors and disordered eating [66,67] and differential timing of hormonally-mediated developmental milestones, such as pubertal onset [68]. Thus, it cannot be assumed that the effects of hormones on disordered eating across development are universal. Research that specifically examines how demographic factors may moderate hormone effects is critical to ensure that emerging etiological models of disordered eating are widely applicable.

Conclusion

While research has historically focused on sociocultural contributions to sex differences in eating disorder risk, accumulating evidence implicates gonadal hormones in between-sex and within-sex variability in disordered eating. Greater perinatal testosterone exposure may protect boys from later disordered eating via organizational effects that become phenotypically apparent during and after puberty. Higher levels of circulating testosterone during mid/late puberty may also protect against disordered eating among boys. Among girls and women, estrogen appears to protect against genetic and phenotypic risk for disordered eating, while progesterone may have opposing effects. Additional research is needed to identify the neurobiological mechanisms underlying hormone effects and potential moderation by psychosocial factors, including possible differences among individuals from diverse demographic groups.

Key points:

• Historically, research has focused on sociocultural contributions to sex differences in eating disorder risk; however, a growing body of evidence implicates gonadal hormones in between-sex and within-sex variability in disordered eating.

• Greater testosterone may protect males from disordered eating via organizational and activational effects that become phenotypically apparent during and after puberty.

• Among females, estrogen may decrease genetic influences on disordered eating during puberty (through protective organizational effects) and phenotypic risk during adulthood (through protective activational effects).

• In adult females, progesterone may increase risk for disordered eating by opposing the protective effects of estrogen and exacerbating the effects of other risk factors.

• Further research is needed to identify the mechanisms underlying hormone effects and potential moderation by psychosocial factors.