Sex × Gender and Sexual Orientation in Relation to Stress Hormones and Allostatic Load

Diurnal Cortisol

Stress hormones can be measured diurnally to capture naturalistic variation. Upon awakening, the cortisol awakening response (CAR) represents a normal surge in cortisol levels reaching maximal concentrations approximately 30 minutes upon awakening. ²² This surge is followed by gradually declining cortisol concentrations throughout the day as pulsatile secretion decreases in amplitude and frequency. ²³ The nadir usually occurs around midnight, and cortisol levels start to rise again in the early morning hours. ²⁴

Repeated measures spread throughout the day can help establish the circadian rhythm of the HPA axis. Five samples provide sufficient information to establish the slope and amplitude of the diurnal curve. ²⁵ Since the CAR is marked by a single pulsatile secretion, repeated assessment over the course of the first hour after awakening is a more reliable marker of the circadian peak and the basal reactivity of the axis because it is sensitive to subtle HPA axis dysregulations. ^26,27 Note, however, that morning cortisol and the CAR may represent an HPA axis mechanism that is distinct from diurnal variation thereafter. ²³ For example, flattened diurnal cortisol patterns have been linked to increased evening levels that are related to distress and fatigue. ²⁸

Diurnal HPA axis functioning can be used to identify disease vulnerabilities (Figure 1). A meta-analysis of 62 studies concluded that while the CAR is positively associated with workplace stress and general life stress, it is negatively associated with symptoms of burnout, fatigue, and exhaustion. ³⁰ Hypocortisolism is a phenomenon that occurs in approximately 20% to 25% of patients suffering from stress-related diseases like chronic fatigue syndrome, fibromyalgia, post-traumatic stress disorder, burnout, ^31,32 and atypical depression to name a few. ³³ By contrast, increased HPA axis functioning during the afternoon and evening has likewise been associated with depressive symptoms. ^34,35

OPEN IN VIEWER

Figure 1.

Hypothetical diurnal cortisol profiles in normal and pathophysiological conditions. ²⁹

Identifying distinct cortisol profiles therefore requires multiple diurnal time points. To demonstrate this, one study found that low overall cortisol concentrations and small differences between morning and evening concentrations predicted depression 2 years later. ³⁶ This underlines the need to assess afternoon and evening levels in addition to morning concentrations and the CAR. Finally, since individual pulses in response to stress can occur at any time during the day, a single sample is essentially useless for the assessment of baseline regulation of the HPA axis for either diurnal or reactive paradigm. ²⁴

Reactive Cortisol

Numerous stress induction paradigms elicit social evaluative threat to trigger stress responses. ³⁷ This occurs in laboratory situations that diminish one’s control as well as where the prospect of being negatively evaluated, rejected, and shamed is contextually manipulated. ^{37 -39} The motivation toward preserving one’s social-self vis-à-vis others’ evaluations involves complex interactions among cognitive appraisals, affective processes, and physiological responses. ^{40 -42} Those who experience social-evaluative threat more saliently fundamentally differ in their sensitivity to relative stressors and toward their need to preserve their self-esteem that can be deleterious to an individual’s health and well-being when it preoccupies them too much. ⁴³

In concert with the autonomic nervous system, short-term adaptations during stress responses are dynamic biological processes that ensure survival. Stress responses are therefore adaptive in the short term, while long-term activations can result in physiological dysregulation. The reactivity hypothesis ⁴⁴ proposes that exaggerated physiological and behavioral reactivity to stressors are risk factors for stress-related diseases like cardiovascular disease (CVD) among others. ^45,46 In this perspective, pathophysiological reactivity is discernable by examining the magnitude of stress responses. Beyond focusing solely on increases from baseline to stressor exposure, researchers have been urged to also assess the duration and prolongation of stress responses after the stressor. ^47,48 Indeed, the reactivity hypothesis has been criticized for often ignoring or dismissing physiological recovery, a period after exposure that is characterized by much individual variability. ^{49 -51}

Noninvasive measures of salivary cortisol are advantageous in stress reactivity paradigms for similar reasons as those discussed for diurnal collection strategies. ⁵² Investigating stable and situational factors might strengthen our understanding of adaptive and maladaptive response patterns. ^53,54 Figure 2 illustrates stress reactive cortisol and dynamic calculations used to capture responses to a popular laboratory-based psychosocial stressor. While stress induction paradigms attempt to understand disease mechanisms, it is important to highlight that responding to a stressor is adaptive and as a preserved mechanism has ensured our survival throughout evolution.

OPEN IN VIEWER

Figure 2.

Schematic of stress reactive cortisol and dynamic calculations. ⁵⁴ The Trier Social Stress Test (TSST) induces social-evaluative threat that modulate physiological responses. Mean salivary cortisol concentrations throughout a testing session (solid black line) can be easily transformed into stress reactivity and stress recovery measures. Here, the area under the curve with respect to increase or AUCi ⁵⁵ is indicated as the area with dotted lines throughout the session). AUCi is a common formula used in psychoneuroendocrinology to summarize dynamics of stress reactive biomarkers. In addition, recovery percent slope (dashed line indicates recovery) represents linear decreases post-TSST.

Concept of Allostasis

Stress responses are adaptive and exemplify the concept of allostasis. Allostasis is defined as adaptive biological processes that preserve “stability through change”. ⁵⁶ The neurobiologist Sterling and the epidemiologist Eyer coined the term allostasis to describe dynamic, multifaceted biological processes that maintain physiological stability by recalibrating homeostatic parameters and matching them appropriately to meet environmental demands. ⁵⁶ Analogous to our understanding of resilient systems that have the capacity to dynamically adjust and stabilize to adapt to perturbations, ⁵⁷ allostatic processes likewise alter metabolic functioning via compensatory as well as anticipatory mechanisms.

Homeostatic models generally define health as an internal milieu whereby all physiological parameters have relatively normal values. Disease, on the other hand, occurs when normal ranges fall into abnormal but pharmaceutically correctable ranges. In this view, Sterling and Eyer (1988) argued that biomedical models are in danger of inadvertently causing deleterious effects such as when correcting one parameter leads to another becoming dysregulated. These iatrogenic effects can arise when restoring one parameter to a normal range causes other parameters to become inappropriately high or low and cause altogether separate health issues. For example, hypertension medications can negatively alter interconnected biological systems (eg, potassium) when lowering blood pressure. In contrast, the allostatic model defines health as a state of responsiveness. A parameter with values outside the normal range is not necessarily considered inappropriate as each parameter is controlled by mutually reinforcing signals. ⁵⁸

Allostasis fundamentally differs from homeostasis because of its emphasis on dynamic rather than static biological set points, considerations of the brain’s role in feedback regulation, ⁵⁹ and its view of health as a whole-body adaptation to contexts. ⁶⁰ According to allostatic theory, pathology itself is a form of adaptation where, for instance, an individual develops hypertension in order to conciliate environmental demands, even if these processes are ultimately pathophysiological. Health in allostatic theory is therefore defined as optimal predictive fluctuation, whereby shifts in the probability of demand should shift the response, and when the prediction reverses, so should the response. ⁶¹ It is when allostasis is chronically dysregulated that pathophysiological processes emerge and contribute to disease.

Allostatic Load Time Course

The AL model proposes that stress-related pathophysiology can be detected and objectively measured by assessing a battery of subclinically relevant biomarkers from multiple nonlinear and interconnected biological systems. ⁶ As illustrated in Figure 3, these contribute to pathophysiological allostatic states that reflect response patterns in which allostatic systems are dysregulated. There are 4 postulated physiological profiles that represent allostatic states: (1) repeatedly activated responses, (2) nonhabituating responses, (3) prolonged responses, (4) and inadequate responses. ⁶⁷ Detecting allostatic states is possible by investigating repeated measures of specific biomarkers like cortisol dynamics. Because allostatic states prompt compensatory remediation among numerous biomarkers, delineating time courses of dysregulation in AL involves the challenge of determining which biomarkers falter first in the cascade. ⁶⁵

OPEN IN VIEWER

Figure 3.

Allostasis, allostastic states, and allostatic load. ²⁹

It has been proposed that under toxic stress, stress hormones like adrenaline and cortisol first become misbalanced and induce an interconnected “domino effect” on interdependent biological systems that collectively collapse as individual biomarkers “topple and trail” toward disease. ⁶⁸ Numerous biomarkers contribute to further AL and eventually lead to disease in a theoretical sequence embodied in our “domino effect” analogy. We shall next explore this theoretical cascade in detail to highlight AL interconnectivity.

Synergistic effects of primary mediators exact primary effects on cellular activities (enzyme, receptor, ion channel, genomic) that compromise the physiological integrity of allostatic mechanisms. For example, mitochondrial dysfunction, oxidative stress, and inflammation represent cellular consequences of the hyperglycemic effect that GC induce. Together, this leads to further AL at a cell–molecular level that we refer to as primary outcomes. ⁶⁹ Over time, subsidiary biological systems compensate for the over- and/or underproduction of primary mediators/effect/outcomes and in turn shift their own operating ranges to maintain abated chemical, tissue, and organ functions. This prodromal stage is referred to as the secondary outcomes, whereby metabolic, cardiovascular, and immune parameters reach subclinical levels. Finally, AL progression leads to tertiary outcomes that include a plethora of diseases.

Medical professionals routinely incorporate many related biomarkers in standard blood tests. Unfortunately, clinical attention is largely placed on biomarkers values that reach cutoffs. The AL model proposes that by measuring the multisystemic, reciprocal interactions among primary mediators, primary effects, primary outcomes, and secondary outcomes together, individuals at high risk of tertiary outcomes can be identified. ⁶⁷ This has important implications not only in terms of research but potentially also in clinical practice as detection and prevention strategies. For example, patients with severe mental disorders are prone to developing comorbidities that could be proactively prevented by accounting for subtle multisystemic dysregulations. ⁶¹

Indexing AL

Seeman and colleagues have pioneered the development of AL algorithms that are predictive of numerous tertiary/clinical outcomes and are predicted by diverse psychosocial factors identified in the stress-disease literature. Validation using longitudinal data from the MacArthur Studies on Successful Aging cohort led to a count-based AL index representing the following 10 biomarkers ⁷⁰ : 12-hour urinary cortisol, adrenalin, and noradrenalin output; serum dehydroepiandrosterone sulfate (DHEA-S), high-density lipoprotein (HDL) and HDL to total cholesterol ratio; plasma glycosylated hemoglobin; aggregate systolic and diastolic blood pressures; and waist to hip ratio. Participants’ values falling within high-risk quartiles (clinical and preclinical ranges based on percentiles) with respect to the sample’s biomarker distributions are dichotomized as “1” and those within normal ranges as “0.” Once tabulated, these are summed to yield an AL index ranging from a possible 0 to 10 which can then be used to predict health outcomes.

Beyond these traditional biomarkers, many others have been incorporated in numerous analyses worldwide using similar formulations and/or more sophisticated statistical approaches (eg, multivariate reduction techniques). Notwithstanding, the simplicity of the count-based formulation makes it amenable to both empirical and clinical contexts. The thematic advantages of applying an elevated risk zone system when scoring AL are 5-fold as they represent: (1) early warning signals since cutoffs are anchored at subclinical thresholds, (2) multifinality in that similar AL algorithms predict different tertiary outcomes, (3) flexibility since calculations are based on different biomarker combinations, (4) synergism capturing the cumulative interaction of numerous biomarkers, and finally (5) antecedents that predict individual variation in AL. ⁷¹ In sum, AL algorithms are objective reflections of biological functioning that are intricately interconnected with genetic, neurological, developmental, behavioral, cognitive, and social factors.

Causes and Consequences of AL

By applying an AL index in analyses of the MacArthur cohort, researchers validated that increased AL levels are related to cognitive and physical impairments at baseline and associated declines over 3 years, as well as an increased risk of having CVD. ⁷⁰ At 7-year follow-up, those with the highest AL levels were at greatest risk of cognitive and physical decline as well as risk of CVD and even all-cause mortality. Note that while differences were found when using neuroendocrine (stress hormones and DHEA-S) versus metabolic syndrome biomarkers (glycolipids, blood pressure, adiposity), the multisystemic AL index was better than using only single biomarkers or systemic clusters to predict outcomes. ⁷² In a similar cross-sectional analysis, it was shown that primary mediators contribute independently of cardiovascular and metabolic risk factors in predicting cognitive and functional declines. ⁷³ These results were cross-culturally complemented using a Taiwanese cohort aged 54 to 71. ^{74 -78}

As represented in Figure 4, AL indices correlate either cross-sectionally or longitudinally with a plethora of antecedents (eg, socioeconomic disadvantage, poor social networks, workplace stress, lifestyle behaviors) and consequences (eg, all-cause mortality, CVD, psychiatric symptoms, cognitive decline, physical/mobility limitations, neurological atrophy). Beyond the powerful effect of aging on AL, unique configurations of these antecedents appear to be experienced differently by each sex throughout life span development. As the following will demonstrate, a binary view of sex is insufficient in studies assessing AL.

OPEN IN VIEWER

Figure 4.

Transdisciplinary summary of the allostatic load literature. ^29,68 In accordance with earlier review articles, ^66,68,79 black triangles represent the identified antecedents of allostatic load, white triangles represent the biological systems that have traditionally been used to index allostatic load, and the gray circles represent health outcomes correlated or predicted by allostatic load.

Epigenetics Considerations

Biological sex at a genetic level (eg, XY and XX chromosomes) is shaped by numerous factors spanning preconception, gestation, childhood, adolescence, and beyond. Moreover, developmental processes are shaped by organizational milestones as well as (de)activation effects of sex hormones throughout the life cycle. In terms of the individual differences that transcend genomes, genetic sex and sex hormones are influenced by social and experiential circumstances that further alter the phenotypic expression. ⁹³ This has influenced sex differences throughout evolution. For example, both genetic males and genetic females have sex hormone receptors throughout their brains that modulate structure and function. ⁹⁴ These brain regions also respond differently to stressors with minimal overlap except to a subset of genes in critical stress regulatory regions like the hippocampus. ⁹⁵ Interestingly, even identical twins show divergent epigenetic effects due to unique experiences throughout life. ⁹⁶ In the context of stress responsivity, different genomes result in differences in how male and female genomes will generally respond to specific stressor as well as sex differences in these endophenotypes. ^95,97 In considering epigenetics, it is important to emphasize that men and women (including gender diverse people) often process cognitive–affective information similarly, but they often use different neural circuits to do so. ⁹⁸ Indeed, epigenetic regulation provides an insurance program that gonadal sex and brain sex are coordinated independent of genetic sex. ⁹⁹

Sex × Gender Roles and AL Among Workers

Gender-based factors might influence changing demographics of wealth and health. The benefits that some women have gained in terms of socioeconomic independence as workers over the last decades ^{123 -125} have been countered by arguments stipulating that gender role conflicts can negatively affect health and well-being. For example, “masculine” working women like managers ¹²⁶ may have had to struggle to achieve their status independently of their feminine identity. In the same vein, the greater implication of men into the organization of family life may not be well received by peers as this represents an appropriation of femininity. ¹²⁷

In Canada and other industrial nations, differential exposures to workplace stressors are partially explained by an unequal sex distribution among occupations that often render women of diverse ages and social strata ¹²⁸ more vulnerable to psychiatric disorders. ^{129 -131} These sex differences may be due, in part, to divergent gender roles and ranks held within occupational hierarchies. ¹³² In general, men are driven more by achievement motivations, while women are driven more by interpersonal motivations. ¹³³ This parallels our understanding of masculine “agency” and feminine “affiliation” in the gender roles literature that have yet to be investigated in relation to AL.

In a study of 204 psychiatric hospital workers, we assessed the synergistic influences of sex and gender roles on AL and mental health while accounting for essential covariates like HPG axis functioning. We tested a popular psychological theory in gender studies developed by Sandra, Lipsitz, and Bem (1944-2014) in relation to AL. Our analysis assessed whether androgyny, that is, high self-endorsement of both masculinity and femininity, would be psychobiologically adaptive and related to lower AL. In contrast to androgyny, those scoring low on both masculinity and femininity are called undifferentiated and believed to be the most maladapted. Previous research had shown that gender roles are predictive of health outcomes, but not necessarily independently from one’s biological sex. ^{134 -136} Because women consistently report poorer health, ^{137 -139} visit physicians more often while taking more sick days off from work, ¹⁴⁰ and are greater consumers of medications ¹³⁹ compared to men, it could be presumed that women are more vulnerable to AL.

Our findings revealed that sex and gender roles interact in association with AL. ¹⁴¹ Specifically, androgynous men with both high masculinity and femininity showed low AL compared to undifferentiated men. In addition, undifferentiated men showed higher AL than undifferentiated women. Independent of sex, androgynous individuals showed better mental health as evidenced by higher self-esteem, well-being, and less depressive symptoms than undifferentiated men and women. ¹⁴¹ Future research is needed to identify who undifferentiated individuals are and how they develop in comparison to other gender role types. On a positive note, further understanding the salubrious nature of psychological androgyny—especially among men—could allow insights into protective pathways toward resilience. Finally, women as a group showed lower AL than men; however, AL did not differ among women by gender roles.

Why would androgyny not be protective against AL among women? The appropriation of multiple gender roles among working women has not been matched by appropriate social equity. In general, women still retain more “feminine” roles and responsibilities than the average man in both work and home life. ^108,142,143 This is likely subjectively distressing for modern women, but perhaps this does not correspond to objective indicators like AL. Beyond perhaps representing sex/gender differences in subjective versus objective indices, it is possible that changing gender roles might also influence men in ways that have yet to be studied fully using sex- and gender-based approaches. For example, a Swedish study found that paternity leave—which is considered as departure from traditional masculinity—is associated with decreased mortality among fathers. ¹⁴⁴ Also, how we deal with major stressful events in our lives is influenced by sexed/gendered coping strategies that interact with numerous additional factors that were not assessed.

Taken together, these findings force us to conceptually move beyond bipolar views of not only men and women but also masculinity and femininity. In developing her theory that androgynous individuals are more adaptively “flexible” in their gender expression, Bem ¹⁴⁵ proposed that masculine and feminine gender roles are independent entities and therefore not binary. Yet, when interpreting gender roles along continuums, it is easy to resort to simplifications that can make masculine and feminine gender roles seem like oppositional constructs when they are in fact context-specific according to Bem.

Hammarström ¹⁴⁶ has noted that gender role theories have inherently assumed a complementary coexistence, for instance, between “active” men versus “passive” women. We must be cautious with such dichotomizations of stereotyped gender roles as we refine our conceptualization when researching stress phenomena. Indeed, Bem’s theory concerning psychological androgyny stands in stark contrast to oversimplified views of masculine versus feminine behaviors. ¹⁴⁷ Indeed, we did not find marked differences when contrasting masculine and feminine individuals, but instead when comparing androgynous and undifferentiated individuals. This emphasis on the relative combinations of masculinity and femininity, and not their singular effect alone, provides compelling psychobiological support for Bem’s hypothesis.

Sexual and Gender Minority Stress

Sexual minority stress models ²¹ propose that the stress of living in environments that stigmatize minority sexual orientations contributes to later health disparities according to 2 stress processes. First, proximal minority stress processes refer to individual-level experiences like internalized homophobia and concealment of one’s minority identity. ¹⁵ Second, distal minority stress processes refer to objective stressors like discrimination and violence experienced at the interpersonal level. At the macro-level, structural stigma represents the social conditions, cultural norms, and institutional policies that constrain the opportunities, resources, and well-being of the stigmatized. Sexual minority stress also applies to the understudied experiences of gender minorities like transgender and gender fluid people who face unique challenges in societies that often denigrate gender nonconformity. ³¹ Indeed, transgender people are at ultra-high risk of health problems, even more than LGB subgroups.

The accumulated findings over the past decades suggest that sexual and gender minority stress plays a role in the psychological, physical, and behavioral health of LGBT individuals. However, this rich body of psychological research is matched by a paucity of biopsychosocial research. Exceptionally, studies that use biological data have examined the relationship between psychological distress and physical health generally among HIV-positive gay and bisexual men. Beyond the HIV/AIDS pandemic, however, the health of LGB populations has yet to be studied extensively using GCs or AL. This may, in part, be because exploring the biological correlates of sexual orientation has been controversial. ¹⁴⁸

Sexual Orientation and Sex Hormones

The HPG axis was once the topic of great research interest in the context of understanding the biological correlates of sexual orientation. While controversial, it was argued that anomalies in the organizational and activational effects of sex hormones could determine nonheterosexual proclivities. Based on animal sexual behavior, several eminent scientists proposed that prenatal androgen deficits could cause male homosexuality, while prenatal androgen excesses could cause female homosexuality. ¹⁴⁹ During a critical period between 2.5 to 5.5 months’ gestation, it was argued that gestational triggers like maternal stress could expose the human fetus to hormonal imbalances that would sculpt hypothalamic and limbic brain regions. ¹⁵⁰ A central tenant of this neurohormonal hypothesis is that differentially shaped brains of LGBT individuals correlate with reversals in sex hormones and sexual behaviors. ¹⁵¹

Throughout the 1970s, many studies published in prestigious journals focused on these hypothesized sex reversals in HPG axis functioning among sexual and gender minorities. In several studies using analytes extracted from either urine, serum, or plasma that were contrasted to secretory profiles of heterosexual controls, many mixed findings were reported. This inconsistent literature was critically reviewed by Meyer-Bahlburg. ^152,153 He concluded that findings were overall inconsistent among sexual minority men, ¹⁵² although about one-third of sexual minority women manifested elevated androgen levels while otherwise showing no endocrine abnormalities. ¹⁵³ Methodological differences between studies rendered comparisons and final conclusions impossible. Taken together, this body of research did not support a neurohormonal hypothesis of sexual orientation ¹⁵⁴ despite more recent claims to the contrary. ¹⁵⁵ In parallel, research among intersex that are born with ambiguous genitalia often show nonheterosexual proclivities in adulthood due in part of differential sex hormone sensitivities. ^156,157 A critical factor ignored in these studies was whether sexual and gender minority stress could explain inconsistent findings.

Meyer-Bahlburg acknowledged that some early researchers of sex hormones and sexual orientation were rather cautious in their conclusions. ¹⁵² In particular, the inconsistencies in observed HPG axis patterns might not be the primary cause of sexual minority orientation, but rather a secondary consequence related to unmeasured factors such as psychosocial stress. ^153,158 To exemplify this, critics considered a longitudinal study among military men undergoing stressful training that revealed plasma testosterone levels were lowest during the earlier novice phase compared to the later senior phase. ¹⁵⁹ Unfortunately, this emerging knowledge from stress science was not applied to understanding the mixed HPG axis literature on sexual orientation and gender identity.

Sexual Orientation and GC Functioning

Despite speculation that psychosocial stress might be involved in study inconsistencies, ¹⁶⁰ these studies of HPG axis functioning never took sexual or gender minority stress into account. Work by our group aimed to address this lacuna in a comprehensive psychoneuroendocrine study.

We provided the first evidence that sexual orientation modulates GC reactivity in a gender-based fashion. ¹⁶¹ That is, lesbian/bisexual women hyperreacted similar to heterosexual men, while gay/bisexual men hyporeacted similar to heterosexual women. In terms of cardiovascular parameters, we found that gay/bisexual men showed higher heart rate throughout in comparison to heterosexual men. ¹⁶² Studying sexual minorities represents a means to further understand the role of gender-based factors in stress reactivity paradigms that are otherwise believed to differ solely according to biological sex.

In addition to highlighting how one’s sexual orientation modulates HPA axis reactivity, we also identified a sex hormone covariation: cortisol systemic output was positively associated with estrogen and progesterone concentrations among lesbian/bisexual women but not among heterosexual women. This sex hormone covariation effect was controlled for in our main analyses. This being said, our curiosity led us to assess sex hormone variations further. Serendipitously, 2 additional saliva samples were collected 15 minutes before and 15 minutes after the psychosocial stressor separate from the 10 samples destined for cortisol. In analyses that adjusted for key covariates, we found that lesbian/bisexual women had higher overall testosterone and progesterone concentrations as a group than heterosexual women, while men showed no differences in HPG axis functioning by sexual orientation. ¹⁶⁰

That lesbian/bisexual women would manifest higher overall levels of testosterone is in accord with some existing reports, ^163,164 but also in disaccord with studies showing no differences. ^165,166 As discussed earlier, a longstanding proposal of this mixed literature has been that elevated androgens among sexual minority women might be explained by environmental stress. ¹⁵³ Consistently, lesbian/bisexual women had higher stress reactive cortisol 40 minutes after the stressor than heterosexual women. ¹⁶¹ Moreover, lesbian/bisexual women had higher progesterone concentrations than heterosexual women, something that has not been previously shown and is not easily explained. There is in fact very little literature on progesterone concentrations, which is somewhat surprising since it is a precursor of GCs and all sex hormones. Given that GCs, testosterone, and progesterone share common molecular precursors like pregnanolone, it is highly probable that coactivation of HPA axis and HPG axis is at play here.

To further triangulate methods and facilitate interpretation of these complex associations, we ran an analysis assessing intercorrelations among the HPG axis and HPA axis dynamics. ¹⁶⁰ We found that cortisol dynamics (area under the curve formulas) were positively associated with testosterone and progesterone fluctuations while accounting for age, sex, gender roles, and sexual orientation. This supplemental result underlines the need to account for HPG axis functioning in studies of GC dynamics. By the same logic and as previously demonstrated, adjusting for sex hormones is necessary to represent biological sex modulation of GCs and is important in determining gender effects in stress reactivity.

Sexual Orientation and AL

Our group has successfully applied the AL model to the study of LGB health among young to middle-aged adults. ^167,168 Contrary to sexual minority stress theory, however, we were surprised to find that LGB people did not show higher AL indexing 21 biomarkers than heterosexuals. ¹⁶⁷ In fact, gay/bisexual men had lower AL levels ¹⁶⁷ and decreased cortisol reactivity in response to a psychosocial stressor ¹⁶¹ compared to heterosexual men. By contrast, lesbian/bisexual women showed tendencies toward higher cortisol reactivity compared to heterosexual women. ¹⁶¹ In addition, no differences in AL were found between women by sexual orientation. ¹⁶⁷ Interestingly, LGB people who had fully “come out” to their family and friends showed lower levels of morning cortisol than those who had not completely disclosed their sexual identity. ¹⁶⁷

A subanalysis of retrospective coping strategies during sexual identity formation revealed that LGB young adults who enacted less avoidance regarding their sexual orientation showed lower AL. In addition, those who sought social support reported less perceived distress. ¹⁶⁸ Central to integrating one’s self-identity as a sexual minority is stigma management that occurs by developing strategies to help cope with the stressful realities of a marginalized identity. Taken together, individual (eg, coping strategies) and interpersonal (eg, social support) factors may promote resilience. Resilience is here defined as a dynamic process that promotes positive adaptation among individuals exposed to severe adversity, trauma, and stress. ^{169 -172}

To confirm whether our AL findings would generalize in the United States, collaborators and I conducted an analysis of nearly 14 thousand people from the 2001-2010 National Health and Nutrition Examination Survey. ¹⁷³ Consistent with our original findings from Canada, gay men showed lower AL compared to heterosexual men. By contrast, bisexual men showed the highest AL. Bisexual men are a minority within a minority that may not garner the same social supports as gay men. This replication expands our original AL findings that did not distinguish between gay and bisexual men due to restricted power. ¹⁶⁷ Interestingly, again no differences in AL were detected for women by sexual orientation that we speculate may be due to gender inequalities in sociocultural and economic resources. ¹⁷³

Existing AL studies among LGBT people highlight important subgroup differences that can only be demarcated in analyses that disaggregate by sex. It is clear that LGBT subgroups are heterogeneous and cannot be treated as a uniform group. Indeed, we speculate that each LGBT subgroup experiences unique combinations of risk and protective factors that have not been explored in AL studies. For example, subgroup behaviors toward diet, exercise, and body image may render gay men more physically fit, which could explain their lower AL compared to heterosexual men ¹⁷³ ; however, this remains to be empirically tested or applied to an aging population. Among women, AL does not differ by sexual orientation or even by gender roles in separate research by our group ¹⁴¹ .

Critically, stigma has not yet been measured but is assumed to implicitly differ at the group level among LGBT people when compared to heterosexuals. This is a crucial limitation that we will need to be addressed in future research that accounts for more psychosocial factors. Lastly and perhaps most importantly, transgender individuals have yet to be included in published studies of AL. This omission is problematic given the exponentially higher rates of physical and mental health disparities among transgender people compared even to LGB subgroups. ¹⁷⁴ Pioneering fieldwork by Zachary DuBois conducted among transgender men has shown that transgender-specific stressors (eg, transitioning stress, “coming out,” public restrooms) are associated with increased diurnal HPA axis production ¹⁷⁵ as well as deleterious indices of cardiovascular and immune functioning. ¹⁷⁶ More research using stress biomarkers are needed to understand the health needs of transgender and gender diverse people.