Intersecting Minority Statuses and Tryptophan Degradation among Stimulant-Using, Sexual Minority Men Living with HIV

Abstract

Background:

Disclosure of one’s sexual orientation as a sexual minority (SM) person (i.e., being “out”) may affect HIV-related health outcomes. This longitudinal study examined whether race/ethnicity moderated effects of outness on the plasma kynurenine/tryptophan (KT) ratio, a marker of dysregulated serotonin metabolism due to immune activation that predicts clinical HIV progression.

Methods:

Participants were African American, Hispanic/Latino, and non-Hispanic White, methamphetamine-using SM men living with HIV (N=97) who completed self-report scales of outness and SM stress at baseline for a randomized controlled trial of a positive affect intervention. Linear mixed modeling was used to test whether race/ethnicity and experimental condition moderated the association of baseline outness with the KT ratio at baseline, 6, 12, and 15 months controlling for SM stress, sociodemographics, HIV disease markers, and recent stimulant use.

Results:

The interactions of outness by race/ethnicity and outness by experimental condition on the KT ratio were significant. Greater outness predicted a lower KT ratio over time in non-Hispanic White SM men, but not among SM men of color (MOC). Greater outness predicted a lower KT ratio over time for SM men in the control, but not among those in the intervention arm.

Conclusion:

Being more out may be protective for non-Hispanic White SM men, but not for their SM MOC peers. Outness mattered for participants who did not receive the positive affect intervention. Findings underscore the potentially different contexts and consequences of outness depending on SM men’s race/ethnicity and whether they received a positive affect intervention.

Sexual minority (SM) men are faced with the stress of dealing with stigma (i.e., the co-occurrence of labeling, negative stereotyping, separation, lack or loss of status, and discrimination faced by a group or group members who are held in negative regard by others; Frost, Lehavot, & Meyer, 2015; Link & Phelan, 2001; Meyer, 2003) based on SM status. The Minority Stress Model asserts that SM stress results from psychosocial stressors (e.g., stigma, discrimination) that exceed the burden of stress among non-SM people and affects the health of SM persons (Meyer, 2003). Numerous studies have established that SM stress is related to adverse mental (Pachankis, Cochran, & Mays, 2015) and physical health (Frost et al., 2015) outcomes, including greater human immunodeficiency virus (HIV) viral load (Norcini Pala, Hart, & Steca, 2015) and alterations in leukocyte gene expression indicating immune activation and inflammation (Flentje et al., 2018) in people living with HIV (PLWH). Further, immune activation and dysregulation have emerged as critical mechanisms that underlie mental and physical health outcomes (Hunt, Lee, & Siedner, 2016; Kiecolt-Glaser & Glaser, 2002), and there is some evidence to support a link between SM stress and these immune processes.

Moreover, SM men, particularly men of color (MOC), are disproportionately affected by HIV compared to other groups in the United States (US) (Beyrer et al., 2013; Singh et al., 2014). PLWH experience more chronic inflammation, immune activation, and increased morbidity and mortality than those not living with HIV, despite effective antiretroviral treatments (Hunt et al., 2016; Serrano-Villar et al., 2014). HIV disease processes amplify immune activation, precipitating the release of inflammatory cytokines, such as interferon gamma and interleukin-6. This is partially due to HIV-induced damage to the gastrointestinal tract early in infection, which leads to translocation of microbial products such as lipopolysaccharide to amplify immune activation and inflammation in the periphery (Byakwaga et al., 2014; Klatt, Funderburg, & Brenchley, 2013). This state of chronic immune activation induces the enzyme indoleamine 2,3-dioxygenase-1 (IDO1) that catabolizes tryptophan (Byakwaga et al., 2014; Favre et al., 2010), an essential amino acid precursor for serotonin. HIV may also induce IDO1 independently of proinflammatory processes (Boasso et al., 2007). Tryptophan catabolism by IDO1 can be measured in serum plasma using the ratio of kynurenine to tryptophan, hereafter referred to as the KT ratio(Strasser, Becker, Fuchs, & Gostner, 2017).

Depletion of tryptophan, as indicated by a higher KT ratio, reduces available serotonin (Byakwaga et al., 2014; Strasser et al., 2017). Additionally, increased concentration of catabolites of tryptophan, including kynurenine, inhibits T-cell proliferation (Byakwaga et al., 2014) and T-cell activation (Mándi & Vécsei, 2012) and induces T-cell apoptosis (Terness et al., 2002). Even after accounting for CD4+ T-cell count and viral load, a higher KT ratio increases the risk of neuropsychiatric disorders, such as depressive disorders, possibly due to depletion of serotonin (Bipath, Levay, & Viljoen, 2015; Serafini et al., 2017) and predicts faster clinical HIV progression (Byakwaga et al., 2014; Strasser et al., 2017). These associations with the KT ratio persist in the face of controlled HIV (Hunt et al., 2016), which underscores the importance of understanding stress-responsive pathways relevant to the KT ratio. Among PLWH, these effects are further exacerbated by stimulant use, which is a potent driver of immune dysregulation in the setting of treated HIV (Carrico, Cherenack, et al., 2018; Carrico et al., 2014).

Given adverse effects of SM stress on health, being out about one’s SM status may be a protective or a risk factor for SM persons depending on their context. For example, an association of greater perceived discrimination based on sexual orientation with higher salivary interleukin-6 was observed only among those who engaged in less concealment of, or were more out about, their sexual orientation (Doyle & Molix, 2016). Further, Intersectionality Theory indicates that individuals and groups with multiple stigmatized identities, such as African American SM men, deal with stigma and stress related to the intersection of their multiple identities that are not captured by solely examining issues of each social identity in isolation (Bowleg, 2013; Crenshaw, 2005). For example, African American SM men have been shown to experience significant stigma based on sexual orientation in their cultural communities and marked stigma based on race in predominantly White gay communities (Bowleg, 2013; McConnell, Janulis, Phillips, Truong, & Birkett, 2018; Raymond & McFarland, 2009). This holds for other minority groups, particularly darker-skinned Hispanic/Latino SM men (Ibañez, Van Oss Marin, Flores, Millett, & Diaz, 2012). Thus, outness may be more protective for some racial or ethnic groups than for others (Fekete et al., 2009).

The present study tested whether being African American or Hispanic/Latino moderates the longitudinal association between outness at baseline and the KT ratio over 15 months among SM men living with HIV who use methamphetamine. Outness was hypothesized to lead to a lower KT ratio, a protective association, for non-Hispanic White SM men relative to African American and Hispanic/Latino SM men while adjusting for SM stress, age, education level, income, antiretroviral medication regimen characteristics, detectable viral load, and recent stimulant use. Because participants were taking part in a larger randomized controlled trial (see Methods section), the present analyses tested for two- and three-way interaction effects of outness by both race or ethnicity and experimental condition on the KT ratio.

Method

Participants

Participants were SM men living with HIV who had biologically confirmed recent methamphetamine use based on urine and hair toxicology screening. Men were originally recruited as part of a randomized controlled trial conducted from 2013 to 2017 in the San Francisco Bay Area (Carrico et al., 2019). The trial tested the efficacy of a five-session positive affect intervention to improve and extend the effectiveness of community-based contingency management for stimulant abstinence (Affect Regulation Treatment to Enhance Methamphetamine Intervention Success [i.e., ARTEMIS]), which achieved more durable reductions in HIV viral load (Carrico et al., 2019). Participants were randomized to either the positive affect intervention or an attention-control condition.

Follow-up rates were acceptable; at six, 12, and 15 months, follow-up rates were 88%, 80%, and 71%, respectively. Peripheral venous blood samples were collected at baseline and at six, 12, and 15 months to measure biological markers resulting in up to four KT measurements over 15 months. Participants completed baseline self-report measures of SM stress and outness. The final analyses for the present longitudinal study included 97 African American, Hispanic/Latino, or non-Hispanic White men.

The Institutional Review Board of the University of California at San Francisco approved all procedures, and reliance agreements were made with the University of Miami and Northwestern University. Additional details regarding the randomized controlled trial and primary findings are reported elsewhere (Carrico et al., 2019).

Measures

Time-invariant covariates.

The time-invariant covariates were collected at baseline only as follows:

Demographics.

Participants reported their age in years, race or ethnicity, education level, and annual income level (see Table 1).

Table 1.

Sample characteristics

	Total Sample (n=97)	ARTEMIS+CM (n = 48)	Attention-Control+CM (n = 49)	p-value
	M (SD)	M (SD)	M (SD)
Age	43.5 (8.7)	43.7 (9.0)	43.3 (8.6)	0.74
Sexual minority stress	12.5 (6.3)	12.3 (6.0)	12.6 (6.7)	1.00
Outness	4.9 (1.9)	4.9 (1.9)	5.0 (1.8)	0.92
Kynurenine/tryptophan ratio	1.6 (0.7)	1.6 (0.1)	1.6 (0.2)	0.07
	n (%)	n (%)	n (%)
Race/ethnicity				0.06
Black/African American	18 (18.6)	9 (18.8)	9 (18.4)
Non-Hispanic White	47 (48.5)	18 (37.5)	29 (59.2)
Hispanic/Latino	32 (33.0)	21 (43.8)	11 (22.5)
Education				0.66†
Less than high school	8 (8.3)	5 (10.4)	3 (6.1)
HS graduate	16 (16.5)	7 (14.6)	9 (18.4)
Some college/trade school	48 (49.5)	21 (43.8)	27 (55.1)
College graduate	15 (15.5)	9 (18.8)	6 (12.2)
Post graduate	10 (10.3)	6 (12.5)	4 (8.2)
Income				0.19†
≤ $4,999	14 (14.6)	7 (14.9)	7 (14.3)
$5,000–$11,999	22 (22.9)	6 (12.8)	16 (32.7)
$12,000–$15,999	26 (27.1)	13 (27.7)	13 (26.5)
$16,000–$24,999	12 (12.5)	6 (12.1)	6 (12.2)
$25,000–$34,999	8 (8.3)	4 (8.5)	4 (8.2)
$35,000–$49,999	7 (7.3)	6 (12.8)	1 (2.0)
≥ $50,000	7 (7.3)	5 (10.6)	2 (4.1)
Protease inhibitor use	32 (33.3)	12 (25.0)	20 (41.7)	0.08
Efavirenz use	5 (5.2)	5 (10.4)	0 (0)	0.06†
ART adherence				0.38
<90% adherence	44 (50.0)	25 (56.8)	19 (43.2)
90% to <100% adherence	32 (36.4)	13 (29.6)	19 (43.2)
100% adherence	12 (13.6)	6 (13.6)	6 (13.6)
Detectable viral load	24 (24.7)	15 (31.3)	9 (18.4)	0.18
Tox+ for stimulants	50 (51.6)	29 (60.4)	21 (42.9)	0.08

Note:

^†indicates analyses requiring exact estimation methods due to quasi-complete or complete separation from zero cells. ARTEMIS=Affect Regulation Treatment to Enhance Methamphetamine Intervention Success. +CM=with community-based contingency management intervention. Values for the time-varying kynurenine/tryptophan ratio were log-transformed.

Experimental condition.

Participants were randomized to either the attention-control arm (coded as 0) or positive affect intervention arm (coded as 1).

SM stress.

The five-item SM stress subscale of the Cultural Assessment of Risk for Suicide (CARS) scale (Chu et al., 2013) measured participants’ experiences of stress due to stigma and discrimination toward sexual-minority persons, such as rejection from others, concealing one’s sexual identity, and internalizing negative attitudes toward oneself as a sexual-minority person. Likert-type response options ranged from 1 (“strongly disagree”) to 6 (“strongly agree”). Higher scores represent more SM stress. Cronbach’s alpha (α) of the present sample was 0.71.

Outness.

The extent to which participants were out as SM persons to other people was measured using the four-item Out to World subscale of the Outness Inventory (Mohr & Fassinger, 2000). Sample items include, “To which degree do your new straight friends know about your sexual orientation?” and “To which degree do your work peers know about your sexual orientation?” Likert-type response options ranged from 1 (“person definitely does not know about your sexual orientation status”) to 7 (“person definitely knows about your sexual orientation status”), with higher scores indicating more outness as a SM person (α =0.91).

Time-varying covariates.

Time-varying covariates were collected at baseline and at six, 12, and 15 months. The measures were as follows:

Medication regimen.

Participants reported their ART regimen; dichotomous variables were created based on whether they took the following ART medications: protease inhibitors (0=no, 1=yes) and efavirenz (0=no, 1=yes). Protease inhibitors (Carrico, Cherenack, et al., 2018) and efavirenz (Hileman et al., 2015) were included as time-varying covariates because they have been linked to markers of immune activation in PLWH.

Antiretroviral (ART) adherence.

ART adherence was assessed using a visual analog scale (Finitsis, Pellowski, Huedo-Medina, Fox, & Kalichman, 2016; Walsh, Mandalia, & Gazzard, 2002). Two indicator variables were created, one reflecting less than 90% adherence (coded 1) and the other representing less than 100% adherence but at least 90% adherence (coded 1). For both variables 100% adherence was the reference group (coded 0 for each). Less than 100% adherence has been associated with increased inflammatory markers (Castillo‐Mancilla et al., 2019).

HIV viral load.

HIV viral load testing was conducted using the Abbott Real Time HIV-1 assay (Abbott Molecular, Inc.; Des Plaines, IL) to detect plasma HIV RNA. The lower limit of detection was 40 copies/mL. Based on the results, the following dichotomous variable was created: detectable viral load (0=no, 1=yes).

Recent stimulant use.

Toxicology screening of urine samples were conducted on-site using iCup (Redwood Biotech, Inc., Santa Rosa, CA) to confirm recent methamphetamine or cocaine use in the past 72 hours (0=no, 1=yes). The investigators used urine toxicology because they were most interested in recent use. Based on the prior work of this research group (Carrico, Cherenack, et al., 2018; Carrico, Flentje, et al., 2018), the two to three days surrounding the blood draw are highly correlated with self-reported stimulant use in the past 30 days; these draws were anticipated to be most relevant for the measurement of the KT ratio.

Dependent Variable:

KT ratio.

The dependent variable was the KT ratio. It was collected at baseline and six, 12, and 15 months as follows:A reverse-phase high-performance liquid chromatography method was used with fluorescence analysis to determine concentrations of tryptophan at 286 nm and kynurenine at 360 nm. The KT ratio was then obtained and log-transformed (base 10). Additional details regarding this methodology have been published elsewhere (Neurauter et al., 2008).

Data Analysis Plan

A linear mixed model with random intercept was used to test whether a longitudinal effect of outness at baseline on the KT ratio at multiple time points was moderated by whether the participant was a MOC (i.e., either African American or Hispanic/Latino) or randomized to the control versus experimental condition (Outness * Race/Ethnicity; Outness *Experimental Condition; Outness * Race/Ethnicity * Experimental Condition). Thus, the three-way interaction and all constituent two-way interactions were included in the model. These analyses accounted for SM stress, age, education level, income, use of protease inhibitors, use of efavirenz, ART adherence, detectable viral load, and recent stimulant use. Given a significant two- or three-way interaction between outness and race or ethnicity or the experimental condition, the simple slopes for the association between outness and the KT ratio would be tested for MOC and non-Hispanic White men, respectively, or for the control and intervention conditions, respectively, to determine if they were significantly different from zero. Additional, exploratory analyses are presented and discussed in online supplemental materials for this article. These exploratory models included fewer covariates, as more interaction effects were assessed.

In addition to the random-intercept model, a model that included the random intercept plus a random slope for the effect of time on the KT ratio was assessed to determine if the new, more complex random-intercept-and-slope model improved model fit over the initial, more parsimonious random-intercept model. The random slope was based on the association of time with the KT ratio, as we wanted to capture variation in the KT ratio by time point. The Bayesian information criterion (BIC) was used to compare model fit. The BIC may be preferable to other information criteria, as the BIC statistic penalizes models with more parameters in order to minimize the advantage that more complex models have over more parsimonious models (Kline, 2016; Raftery, 1995; Schwarz, 1978). The BIC also accounts for sample size (Kline, 2016; Raftery, 1995; Schwarz, 1978). A smaller BIC value denotes a better relative fit to the data. The grades of evidence corresponding to differences in BIC values between two models are as follows: 0–2 indicates weak evidence of a difference in relative fit, 2–6 reflects positive evidence, 6–10 represents strong evidence, and >10 denotes very strong evidence of a difference between two models (Raftery, 1995). The best performing model was selected.

Linear mixed models captured the within-group variability in the KT ratio at multiple time points while controlling for potential drivers of inflammation, such as a detectable HIV viral load or recent stimulant use, that may also vary at multiple time points for some in this population of SM men who are living with HIV that use methamphetamine. Additionally, levels of kynurenine and tryptophan vary over time (Dehhaghi, Kazemi Shariat Panahi, & Guillemin, 2019). Even single instances of stress have been found to have effects that lasted longer than expected on biological markers of immune dysregulation (Schubert et al., 2012), and longer-lasting effects of potentially chronic stressors, such as outness or lack thereof, can be accounted for in individual SM men. Thus, despite the presence of time-invariant covariates of interest, a longitudinal approach allowed the inclusion of more than one time point to help improve the precision of measurements for the KT ratio and associated biological markers that affect immunologic balance.

The continuous independent and dependent variables were standardized to facilitate interpretation by using standard-deviation units. The resulting regression coefficients were interpreted in standard-deviation units. The regression coefficients (B) and their standard errors (SE), 95% confidence intervals (95% CI), and p-values were reported. For descriptive statistics, means (M) and standard deviations (SD) were reported for continuous variables; percentages were reported for categorical variables. Non-significant interactions were not reported in the text. Missing data were handled using full information maximum likelihood (i.e., FIML). Primary analyses were performed in Mplus version 8.3 (Muthén & Muthén, 2017).

Results

Preliminary Analyses

Descriptive statistics for the sample appear in Table 1. Outness did not vary by race or ethnicity (F=0.74, p=0.480) or by treatment condition (t=−0.49, p=0.626).

Primary Analyses

The random-intercept model (BIC = 5188.670) fit the data better than the random-intercept-and-slope model (BIC = 5199.417). Results for the random-intercept model of the effect of time and the interactive effects of outness by race/ethnicity and experimental condition at baseline on the KT ratio over 15 months while adjusting for SM stress, use of protease inhibitors or efavirenz, detectable viral load, and recent stimulant use are shown in Table 2. The effect of time on the KT ratio was not significant (B = 0.004, SE = 0.009, p = 0.626, 95% CI: −0.013, 0.021). We reran the model with time as a categorical variable expressed with three indicator variables (i.e., at six, 12, and 15 months), each with baseline (i.e., zero months) as the reference group. However, the results were similar.

Table 2.

A Linear Mixed Model Estimated via Maximum Likelihood of Predictors of Dysregulated Neurotransmitter Synthesis, as Indicated by the Kynurenine/Tryptophan Ratio, in a Sample of People Living with HIV who Use Methamphetamine (N = 97) over 15 Months

Variable	B	SE	p	95% CI
Between Level (Outcome: Kynurenine/Tryptophan Ratio)
Age	0.144	0.076	0.060	−0.006, 0.293
Education level	0.023	0.073	0.756	−0.120, 0.165
Annual income	0.015	0.042	0.714	−0.066, 0.097
SM stress	−0.247	0.090	0.006	−0.423, −0.071
Outness	−0.612	0.102	<0.001	−0.812, −0.412
Racial/ethnic-minority status	−0.754	0.206	<0.001	−1.158, −0.351
Intervention cond.	−0.226	0.221	0.307	−0.658, 0.207
Outness * Race/ethnicity	0.522	0.161	0.001	0.206, 0.838
Outness * Intervention cond.	0.450	0.195	0.021	0.069, 0.832
Race/ethnicity * Intervention cond.	0.301	0.295	0.307	−0.277, 0.878
Outness * Race/eth. * Intervention cond.	−0.209	0.265	0.430	−0.728, 0.310
Within Level (Outcome: Kynurenine/Tryptophan Ratio)
Time	0.004	0.009	0.626	−0.013, 0.021
Protease inhibitor	−0.039	0.138	0.775	−0.310, 0.231
Efavirenz	−1.097	0.308	<0.001	−1.701, −0.492
<90% ART adherence (ref=100%)	−0.140	0.139	0.314	−0.412, 0.132
90% to <100% ART adherence (ref=100%)	−0.144	0.132	0.273	−0.402, 0.114
Detectable viral load	0.443	0.112	<0.001	0.213, 0.653
Recent stimulant use	−0.151	0.096	0.115	−0.338, 0.037

Note. SM stress = sexual minority stress. Racial/ethnic-minority status = being African American or Hispanic/Latino (reference group = being non-Hispanic White). Intervention cond. = being randomized to the intervention condition of the study (reference group = control condition). ART = antiretroviral treatment. All variables were analyzed simultaneously in the same model. We also reran the model with ART adherence (a) at <80% adherence instead of <90% and, alternatively, (b) a continuous variable instead of two indicator variables, but adherence remained a nonsignificant independent variable. Values for the time-varying dependent variable, the kynurenine/tryptophan ratio, were log-transformed.

The interaction effect of race/ethnicity by outness on the KT ratio was significant (B = 0.522, SE = 0.161, p = 0.001, 95% CI: 0.206, 0.838). Thus, the magnitude of the effect of outness at baseline on the KT ratio over 15 months depended on whether the SM men in the sample were MOC (i.e., African American or Hispanic/Latino). As shown in Figure 1, further examination of this interaction showed that, on average, outness was associated with a significant decline in KT ratio over 15 months among non-Hispanic White men (reference group; B = −0.612, SE = 0.102, p< 0.001, 95% CI: −0.812, −0.412) but not for MOC (B = −0.090, SE = 0.160, p = 0.575, 95% CI: −0.403, 0.224). Thus, for non-Hispanic White SM men in the sample, there was an average a 0.612 SD decrease, over half a SD, in the KT ratio for each one SD increase in the degree of outness while controlling for covariates. However, there was no average change in the KT ratio for each one SD increase of outness for African American and Hispanic/Latino men in the model.

Figure 1. The Association Between Outness and the Kynurenine/Tryptophan Ratio for Sexual Minority Men of Color and non-Hispanic White Sexual Minority Men (Reference Group).

Note. Race/ethnicity moderates the association of outness at baseline with the kynurenine/tryptophan ratio over 15 months in a linear mixed model. Being more out as a sexual minority person was associated with a significantly lower kynurenine/tryptophan ratio for non-Hispanic White sexual minority men (SMM) but not SMM who are men of color (i.e., African American and Hispanic/Latino).

The interaction effect of outness by experimental condition on the KT ratio was also significant (B = 0.450, SE = 0.195, p = 0.021, 95% CI: 0.069, 0.832); see Figure 2. Specifically, the effect of outness at baseline on the KT ratio over 15 months differed as a function of experimental condition. Further examination of this interaction indicated that, on average, outness was associated with decreases in the KT ratio for men in control condition (reference group; B = −0.612, SE = 0.102, p< 0.001, 95% CI: −0.812, −0.412) but not those in the intervention condition (B = −0.162, SE = 0.199, p = 0.417, 95% CI: −0.552, 0.229). Thus, for SM men in the control condition, there was an average 0.612 SD decrease, over half a SD, in the KT ratio for each one SD increase in the degree of outness while controlling for covariates. However, there was no average change in the KT ratio for each SD increase in outness for SM men in the intervention condition in the model.

Figure 2. The Association Between Outness and the Kynurenine/Tryptophan Ratio for Sexual Minority Men in the Control Condition (Reference Group) and Sexual Minority Men in the Intervention Condition.

Note. Experimental condition moderates the association of outness at baseline with the kynurenine/tryptophan ratio over 15 months in a linear mixed model. Being more out as a sexual minority person was associated with a significantly lower kynurenine/tryptophan ratio for men in the control condition but not in the intervention condition.

Other significant covariates in primary analyses, specifically, SM stress, the use of efavirenz, and having a detectable viral load, are shown in Table 1. On average, having a detectable viral load over 15 months (B = 0.443, SE = 0.112, p< 0.001, 95% CI: 0.213, 0.653) was associated with a higher KT ratio over the same period. In contrast, higher reported levels of SM stress at baseline (B = −0.247, SE = 0.090, p = 0.006, 95% CI: −0.423, −0.071) and use of efavirenz over 15 months (B = −1.097, SE = 0.308, p< 0.001, 95% CI: −1.701, −0.492) were associated with a lower KT ratio over 15 months.

Discussion

The present study is one of the few to examine the longitudinal associations of outness at baseline with the KT ratio over time in light of the moderating roles of race or ethnicity and a positive affect intervention while accounting for time, SM stress, demographic characteristics, ART regimen characteristics, ART adherence, detectable viral load, and recent stimulant use. Our results are among the first to demonstrate that the burden of stigma at the intersection of being SM and also MOC is accompanied by negative immunologic consequences of outness. As hypothesized, there was a significant prospective association between outness and an established indicator of tryptophan depletion related to immune activation and dysregulation, the KT ratio, that was moderated by race and ethnicity. Specifically, greater outness was associated with a lower KT ratio over 15 months, a beneficial effect, for non-Hispanic White SM men, on average. However, on average, the association between outness and the KT ratio was not significant for SM MOC. Although outness may be a protective factor for non-Hispanic White men when taking SM stress into account, it is also fruitful to consider the intersection of racial, ethnic, and SM identities. There may be a culturally specific experience of stigma and SM stress for SM MOC that is not captured in general measures and current conceptualizations of SM stress. Additionally, studies may wish to model specific experiences of discrimination based on race or ethnicity and sexual orientation and their effects on both minority stress and immune dysregulation, as experiences of discrimination have been shown to have immunological effects (Cuevas et al., 2020).

Prior literature suggests that outness is linked to adverse health outcomes in the context of a high level of SM stress (Doyle & Molix, 2016), and SM MOC face significantly more stigma based on SM status than non-Hispanic White SM in the US (Moradi et al., 2010). Specifically, studies have reported higher levels of stigma based on SM status in populations of color (Ramirez-Valles, 2007; Vincent, Peterson, & Parrott, 2009; Whitley, Childs, & Collins, 2011) as well as greater SM stress in response to stigma based on sexual orientation among SM MOC (Ghabrial, 2017; Moradi et al., 2010) in the US. Despite the greater SM stress that they experience, SM MOC with social ties to predominantly non-Hispanic White gay communities may also feel under pressure to come out as a show of authenticity or gay pride (Ghabrial, 2017; Moradi et al., 2010). Given that peer support is a significant protective factor for SM MOC (Storholm et al., 2019; Vincent et al., 2017), some SM MOC may be out without having sufficient social support to navigate this important transition. For some SM MOC, greater outness may negatively impact their access to social networks in their racial or ethnic communities. The findings support future study with a more explicitly intersectional approach.

There was a significant interaction such that outness was associated with a lower KT ratio for the attention-control condition, suggesting a protective association, but there was no association in the intervention condition. One possible explanation is that any benefit of outness made more of a difference in the KT ratio for SM men who did not receive the positive affect intervention. In fact, the intervention was successful in improving positive affect and decreasing stimulant use as well as achieving durable and clinically meaningful reductions in viral load (Carrico et al., 2019). Interventions that influence affective states have been found to have beneficial effects on immune functioning (e.g., cognitive-behavioral stress management; Antoni et al., 2016; Carrico et al., 2005). Positive affect may serve as an important buffer against adverse outcomes related to immune functioning. Potentially bidirectional associations between positive affect and the KT ratio could also be studied.

Contrary to hypotheses, SM stress was negatively associated with the KT ratio, which is indicative of a protective association. However, extant studies indicate that SM stress is linked to adverse mental and physical health outcomes for SM persons (Flentje, Heck, Brennan, & Meyer, 2019; Pachankis et al., 2015). One potential explanation for the paradoxical finding is that, although the measure of SM stress used in the present study shows adequate psychometric properties (Chu et al., 2013), other measures need to be developed to better account for the experience of SM stress for diverse samples of HIV-positive men who are also struggling with substance use. Another possibility is that a one-time measure of SM stress is not adequate to capture its effects within a longitudinal study.

Alternatively, people living with substance use disorders tend to score higher than those without these disorders on alexithymia (de Haan, van der Palen, Wijdeveld, Buitelaar, & De Jong, 2014), which is a psychological condition characterized by difficulty in identifying and describing emotions and discriminating between emotions and physical sensations (de Haan et al., 2014). People with alexithymia may show confusion, externalization, and vagueness when directly asked about their affective state (McIntosh, Ironson, Antoni, Fletcher, & Schneiderman, 2016). Generally, alexithymia is associated with more psychological distress, immune dysfunction, non-adherence, and HIV disease progression in persons living with HIV (McIntosh et al., 2016; Temoshok et al., 2008). Thus, those with difficulty understanding their emotions might tend to experience more distress but also lack sufficient insight and awareness to report this distress. SM persons living with HIV who use methamphetamine may be relatively high in alexithymia and, consequently, less able to correctly identify and report their experiences of SM stress. Such participants may score lower on SM stress but have elevated KT ratios. Alexithymia is a more extreme, clinical example, however; lack of insight or awareness may be evident without clinically significant symptoms. Thus, greater insight and awareness regarding experiences of SM stress may be a protective factor, which could explain the negative association between SM stress and the KT ratio. Interventions that increase awareness of SM stress among SM persons (Flentje, 2019) may be helpful. Additional clinical research is needed to test these hypotheses.

Whereas the present study examined aspects of intersectional identities in terms of risk, models that consider intersectional identities can also be studied in terms of resilience. For example, qualitative interviews with SM and gender minority persons of color found that, although intersectionality theory has facilitated the incisive study of joint oppressions and their effects, positive intersectionality emerged as a key theme (Ghabrial, 2017). Specifically, persons who are marginalized at the intersection of multiple identities, such as those based on race or ethnicity and SM status, can embrace their identities and find strength in community with others of similar intersectional identities (Ghabrial, 2017). Positive intersectionality may help to explain why SM persons of color do not necessarily have poorer self-esteem or mental health than non-Hispanic White SM persons.

Although the present study has many strengths, there are notable limitations. For example, participants were SM men who were living with HIV and had recently used methamphetamine, which is a potent driver of immune dysregulation (Carrico et al., 2008). As such, caution must be used in any attempts to generalize these findings to other SM persons, including SM women. Additionally, the associations between factors such as outness and underlying immune processes may be more robust in the PLWH represented in the present sample than in those who are not living with HIV (Flentje et al., 2018). Also, both outness and SM stress were measured once at baseline; these measures may be better measured longitudinally, even if they reflect chronic circumstances for a given participant. In addition, race and ethnicity were measured as proxies for cultural experiences and phenomena that vary or manifest differently between and within groups, such as social conservatism and stigma based on SM status in the context of the US. Further, the study would have been strengthened with measures that directly assessed multiple dimensions of the experiences of SM persons of color, such as the degree to which their racial or ethnic identities are central to their view of their social identities (i.e., centrality) (e.g., Sellers, Rowley, Chavous, Shelton, & Smith, 1997). Also, the study did not directly measure intersectional experiences or test intersectional processes, such as interactions between racial- or ethnic-minority and SM experiences of discrimination or stress. Further, the study occurred in Northern California, which may be more inclusive of sexual and gender minorities than other US areas. Thus, the present findings may reflect conservative estimates, as effects may have greater magnitude in less inclusive locales.

Moreover, this study did not capture how variability in outness to specific groups (e.g., co-workers, new heterosexual friends) affect the associations in the models we tested. For example, results may differ for SM men on the very or moderately high end of outness to some groups (e.g., friends) but very or moderately low end of outness to other groups (e.g., co-workers). Race/ethnicity may further complicate such differences, and more research is needed. Further, although we included multiple measurements of time-varying variables to improve the precision of our analyses, there was no effect of time or the random slope for time. Larger samples may better capture such time effects, including possible differences between White SM men and respective SM MOC in the variability of the KT ratio over time. These are all important directions for future research examining intersectionality and health in SM populations.

In summary, this longitudinal study is among the rare studies to test for differences in the association between outness, which is a factor related to the experience of SM persons and stigma based on SM status, and a marker of immune dysregulation through the lens of intersectionality. The findings suggest that, when studying biological mechanisms of processes relevant to SM stress, such as outness, researchers may benefit from considering the central role of intersectional identities. The potential benefits of interventions that enhance positive affect for reducing SM stress and improving immune function also merit further study.

Public Health Significance Statement:

Based on a marker of dysregulated serotonin metabolism due to immune activation that also predicts clinical HIV progression, non-Hispanic White sexual minority men may experience greater benefit from being more out than sexual minority men of color. There may be different cultural contexts and, thus, different immunologic consequences of outness for sexual minority men of color. Additionally, compared to sexual minority men who received a positive affect intervention, men who did not receive the intervention may have benefited from being more out to others; thus, any protective effects of outness could matter more if sexual minority men are not receiving some form of a supportive or coping intervention.