Systematic Review of Sex Differences in the Relationship Between Hormones and Depression in HIV

Abstract

Background

Major depressive disorder is the most common neuropsychiatric comorbidity of human immunodeficiency virus (HIV), and women are more frequently affected in the general population and among those with HIV. The rate of depression in HIV is three times higher than the general population. Differences in biomarkers in neuroendocrine and inflammatory pathways are one possible explanation for the increased prevalence of depression in individuals with HIV, especially biological women. Therefore, we aimed to perform a systematic review identifying differences in neuroendocrine factors leading to depression in men versus women with HIV.

Methods

A comprehensive search of 8 databases was performed, followed by title and abstract screening and later full-text screening by two independent researchers. A risk of bias assessment was completed.

Results

Twenty-six full-text articles were included in the review. Significant correlations between depression and neuroendocrine marker levels were found for cortisol (both sexes), testosterone (only in men), oxytocin (only tested in women), and estradiol (only in women). No significant correlation between depression and hormone level was found for prolactin, dehydroepiandrosterone (DHEAS), or sex hormone binding globulin (SHBG). Nearly all studies included only men or women and did not directly compare neuroendocrine markers between the two sexes. One study found that the correlation between cortisol levels and depression scores was stronger in women than men.

Conclusion

Neuroendocrine systems are highly active in the brain and important in the development and persistence of mental illness. Given that HIV can, directly and indirectly, impact hormone signaling, it is likely contributing to the high rate of depression in individuals with HIV. However, few studies explore neuroactive hormones in depression and HIV, nor how this connection may differ between the sexes. More high-quality research is needed in this area to explore the link further and inform possible avenues of treatment.

1. Introduction

Major depressive disorder (MDD) is the most common neuropsychiatric comorbidity of individuals with human immunodeficiency virus (HIV), with the prevalence of depression in HIV-infected people being 2 to 4 times higher than in the general population.¹ Depression in people with HIV also may be treatment-resistant, as a recent Cochrane review showed inconclusive evidence that the standard treatment regimens were effective in this population.² This comorbidity is associated with low quality of life and a wide range of adverse clinical outcomes, including poor antiretroviral adherence, lower CD4 counts, and higher viral loads.¹

It is well established that women experience higher levels of major depression in comparison to their male counterparts.³ This pattern continues in people with HIV, with multiple studies citing a higher rate of depression in women with HIV than men with HIV. For example, in one study of 3,863 people with HIV in Africa, 32% of women with HIV reported depressive symptoms versus 22% of men with HIV.⁴ Women have also been shown to have depressive symptoms of more severe intensity than men.⁵ Despite these differences, there are few studies exploring why depression is more common and more severe in HIV-infected women than men.

One explanation for this discrepancy lies in the biological differences between men and women. Studies have explored possible biological causes for the higher prevalence of depression in individuals with HIV, including increased inflammatory cytokines, neurobiological changes in the central nervous system (CNS), and neuroendocrine pathway disruptions.^6-8 Hormonal pathways are of particular significance due to the previously established relationship between depression and neuroendocrine pathways.^9,10 Some studies have begun to explore this relationship in people with HIV. In one recent study of 65 people with HIV, salivary afternoon basal cortisol levels were examined by remitted depression determined using the SCID. Higher cortisol levels were found in women with remitted MDD than women with no MDD; however, this association was not observed in men with HIV.¹¹

Although the prevalence of depression among people with HIV has been noted, and research has begun to consider sex as a biological variable in this context, there are outstanding questions regarding the role of hormones on different rates of depression in men and women with HIV. This systematic review will explore the biological differences between men and women with HIV and depression by comparing levels of neuroendocrine markers such as cortisol, testosterone, and estrogen. These discoveries will aid in the understanding of the pathophysiology behind depression and HIV and potentially illuminate new therapeutic targets.

2. Methods

We conducted a comprehensive search of 8 databases: PubMed, Medline, Embase and PsycINFO via Ovid, Web of Science Core Collection, Global Index Medicus, ClinicalTrials.gov, and Cochrane Library via Wiley. Per best practices,¹² the search combined natural language searching and controlled vocabulary to reflect the concepts of HIV, hormones, and depression. A complete search strategy is available in Appendix A. The search was conducted in September 2020. We placed no limitations on the language of publication or study design. Title and abstract screening were completed using Rayyan, a web-based systematic review screening tool.¹³ Two independent reviewers reviewed titles and abstracts, and any discrepancies were resolved through consensus or a third party where necessary. Full-text screening was also completed by two independent reviewers, and reasons for exclusion were recorded and reported in Figure 1.¹⁴

Figure 1.

PRISMA Flow Diagram

Inclusion and Exclusion Criteria

Inclusion criteria required HIV, Depression, and measurement of a hormone. Participants needed to be over 18 years old. Depression had to be measured using a validated method that was reported. Both randomized and non-randomized studies were included.

Exclusion criteria included articles that focused on pediatric populations and patients at risk or affected by but not personally diagnosed with human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS). Articles that included both men and women but did not report data for these populations separately were excluded. Single case reports and papers that did not report original data, such as opinion pieces, were excluded.

The data extraction form was developed by one researcher and piloted prior to further refinement. Quality assessment was completed using the Cochrane Collaboration's RoB 2 tool for randomized studies and National Heart, Lung, and Blood Institute (NHLBI) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies for non-randomized studies. Data extraction and quality assessment were completed by an individual researcher, and a second researcher independently confirmed the findings. Quality assessment was guided by official implementation documents developed by Cochrane and NHLBI.^15-17 following the previously established workflow, any discrepancies in findings or opinion between the researchers were resolved through consensus or by a third party. The protocol for the project was registered in PROSPERO (CRD42020157839).

3. Results

3.1. Flow of Studies

Our literature search produced 4,307 articles related to our search terms. Of these, 2,203 remained after duplicates were removed. The abstracts of these articles were screened by at least two reviewers independently, leaving 200 full-text articles to review for eligibility. In addition, 174 full-text articles were excluded for reasons including outcome, population, publication type, and unavailability of the article, leaving 26 full-text articles. Of these full-text articles, 7 included non-hormonal biomarkers, and we decided to include these in a separate paper. Thus, 19 papers remained in the scope of this article. Table 1 describes the characteristics of the included studies and their participants, while Table 2 describes the relevant outcomes of each included study.

Table 1.

Characteristics of included studies

Study	Study Design (Country)	Population	Neuroendocr ine Marker (s) (sample type, time of collection, fasting vs non-fasting)	N	CD4 (mean (sd) cells/ mm3)	HIV Viral Load (copies/ml) ††	Population on ART
Antoni 2000	RCT (USA)	Symptomatic HIV+ gay and bisexual men	Cortisol (urine, 24 hour output)	69	418 (237) cells/ml	NR	39.1% (27/69)
Grinspoon 2000	RCT (USA)	HIV+ men	Testosterone (serum), estradiol (serum),SHBG (hematocrit and serum)	Total: 61	NR	NR	NR
				Hypogonad al: 51	175 (30)	191151 (36483)
				Eugonadal: 10	205 (91)	111920 (37766)
Antoni 2005	RCT (USA)	Symptomatic HIV+ gay and bisexual men	Cortisol (urine, 24 hour output)	Total: 25	NR	NR	NR
				Interventio n: 16	528 (273.9)	1202 (1787)	25%
				Control: 9	420.7 (165.5)	2088 (2336)	44%
Kertzner 1993	Longitudinal (USA)	Gay and bisexual men	Cortisol (urine, 24 hour output)	Total: 121*	NR	NR	NR
				HIV+: 80	422 (209)
				HIV−: 41	825 (266)
Wagner 1998a	Longitudinal (USA)	HIV+ men with sexual dysfunction, CD4<400, serum testosterone<450	Testosterone (serum)	54	109	NR	NR
Wagner 1998b	Longitudinal (USA)	HIV+ men with diminished libido and comorbidity of interest**	Testosterone (serum)	23	150	NR	65%
Blick 2013	Longitudinal (USA)	Men with hypogonadism	Testosterone (serum)	849	NR	NR	NR
Seay 2014	Longitudinal (USA)	HIV+women	Oxytocin (plasma)	70	478.1 cells/uL	NR	90.5%
Antoni 1991	Cross-sectional (USA)	Gay men	Cortisol (plasma, 7:30 AM-10:30 AM, fasting)	71	713 [median]	NR	NR
Gorman 1991	Cross-sectional (USA)	Gay men	Cortisol (urine, 24 hour output)	Total: 187	NR	NR	NR
				HIV+: 112	401.8 (226.4)
				HIV-: 75	828.2 (274.9)
Gorman 1992	Cross-sectional (USA)	Gay and bisexual men	Prolactin (serum, majority collected in AM)	Total: 199	NR	NR	NR
				HIV+: 120	409 (222.6)
				HIV−: 79	843.1 (263.7)
Goggin 1998	Cross-sectional (USA)	HIV+ women	DHEAS (serum, 11:00 AM-5 PM)	54	158 (149.9)	NR	NR
Ferrando 1999	Cross-sectional (USA)	HIV+ men	DHEAS, Testosterone (serum, 11:00 AM-4:00 PM)	Total: 169	301 (190, 496) [median (IQR)]	NR	NR
				HIV+ CD4 > 500: 18	681 (547, 748) [median (IQR)]
				HIV+ CD4 200-500:46	296 (230, 361) [median (IQR)]
				AIDS CD4 <200: 105	91 (21, 165) [median (IQR)]
Carrico 2006	Cross-sectional (USA)	HIV+ persons	Cortisol (urine, 24 hour output)	Total: 264	435 (276)	17441 (62991)	NR
				Men: 130	422 (239)	13098 (30481)
				Women: 134	482 (306)	21717 (83385)
Wisniewski 2006	Cross-sectional (USA)	18-50-years in high risk categories†	Cortisol (serum, 8:00 AM-10:00 AM)	209	NR	NR	NR
Lari 2012	Cross-sectional (Iran)	HIV+ men	Prolactin (does not specify), free testosterone (serum, early morning, fasting)	237	NR	NR	NR
Sunchatawi rul 2012	Cross-sectional (Thailand)	HIV+ men	Total testosterone, SHBG (serum, 8:00 AM-11:00 AM)	491	320	<50 copies/ml: 87.3% Undetectable; peak plasma RNA level 157	93.5%
Bekhbat 2018	Cross-sectional study (USA)	Women	Estradiol (does not specify)	Total: 147	NR	NR	NR
				HIV− without depression: 37	1151.5 (915, 1509) [median (IQR)]	NR	N/A
				HIV− with depression: 34	1066.5 (831, 1349) [media (IQR)]	NR	N/A
				HIV+ without depression: 38	436 (309, 606) [median (IQR)]	11500 (3850, 32000) [median (IQR)]	65.8%
				HIV+ with depression: 38	499.5 (313, 631) [median (IQR)]	5500 (340, 33000) [median (IQR)]	50.0%
Laan 2019	Cross-sectional (Netherlands)	Women with or without testosterone insufficiency (TI)	Prolactin, testosterone, SHBG, FH, FSH, 17b-estradiol (plasma)	Total: 49	NR	Undetectable in 78%	78%
				With TI: 18	653 (376)	Undetectable in 88%	89%
				Without TI: 31	572 (272)	Undetectable in 71%	72%

ART: Antiretroviral Therapy; DHEAS: Dehydroepiandrosterone sulfate; FSH: Follicle-stimulating hormone; LH: Luteinizing hormone; SHBG: Sex hormone binding globulin

^*A subset of the initially enrolled 184 patients (109 HIV positive, 75 HIV negative individuals)

^**Comorbidities of interest: low mood, low energy, low appetite/weight loss

^†Risk categories included visiting HIV clinics, methadone maintenance clinics, and homeless shelters

^††Viral load was reported as % Undetectable with cutoff as available. If only numeric viral load was presented that was what was reported.

Table 2.

Hormone levels by sex and correlation with depression scales

Study	Average Hormone Level		Depression scale (Cut-off)	Correlation between hormones and depression
	Men	Women
Cortisol
Antoni 1991	170 ng/ml	N/A	POMS depression subscale (continuous)	Yes (p =NR)
Antoni 2000	Control: 11 ug/24hrs Intervention: 13 ug/24 hrs	N/A	POMS depression subscale (continuous)	Yes (r = 0.32, p < 0.05)
Antoni 2005	Control: 9 ug/24 hrs Intervention: 14 ug/24 hrs	N/A	POMS depression subscale (continuous)	NR
Carrico 2006	6.1/24h (with square root correction)	N/A	BDI affective subscale (continuous)	No correlation
Gorman 1991	62 ug/24 hours	N/A	HAM-D (continuous)	Yes (r = 25, p = 0.01)
Kertzner 1993	70 ug/24 hours	N/A	HAM-D (continuous)	Yes (r = 0.25, p < 0.01)
Wisniewski 2006	17.5 μg/dL	16.46 μg/dL	CES-D (>23, continuous)	Yes (2% increase in odds of depressive symptoms per unit increase)
DHEAS
Ferrando 1999	158 ug/dL	N/A	BDI (>10 significant depressive symptoms)	No correlation
Goggin 1998	N/A	96.5 ug/dL	HAM-D (continuous)	No correlation
Estradiol
Bekhbat 2018	N/A	No depression: 43.0pg/mL Depression: 37.5pg/mL	CES-D (>16)	NR
Grinspoon 2000	hypogonodal: 12.0 (1.1) pg/m Leugonadal: 14.2 (2.6) pg/mL.	N/A	BDI (>18)	No correlation
Oxytocin
Seay 2014	21.3 pg/ml	N/A	BDI (no cutoff)	Yes (p = 0.05)
Prolactin
Gorman 1992	CDC Stage II: 9.2 (3.2) ng/ml CDC Stage III: 10.2 (3.4) ng/ml CDC Stage IV: 9.9 (3.9) ng/ml	N/A	HAM-D (no cutoff)	No correlation
SHBG
Grinspoon 2000	hypogonodal: 34.0 nmol/L eugonadal: 44.0 nmol/L	N/A	BDI (18)	No correlation
Sunchatawi rul 2012	42.0 nmol/L	N/A	Thai HADS (>10 on the even numbered questions)	No correlation
Free Testosterone
Blick 2013	50 pg/ml	N/A	PHQ-9 (continuous)	NR
Ferrando 1999	87 pg/ml	N/A	BDI (continuous)	No correlation
Grinspoon 2000	hypogonodal: 13 pg/ml eugonadal: 22 pg/ml	N/A	BDI (18)	Yes (p = NR)
Lari 2012	4.8 vs. 6.6 pg/ml	N/A	BDI (18)	Yes (p = NR)
Total Testosterone
Blick 2013	375 ng/dL	N/A	PHQ-9 (continuous)	Yes (p = NR)
Goggin 1998	25.6 dg/sl	N/A	HAM-D (cutoff?)	No correlation
Grinspoon 2000	hypogonodal: 427 ng/d leugonadal: 738 ng/dl	N/A	BDI (18)	Yes (p = NR)
Laan 2018	With insufficiency: 14 ng/dl Without insufficiency: 35ng/dl	N/A	BDI (20)	No correlation
Sunchatawi rul 2012	Median 19 nmol/L	Median 19 nmol/L	Thai HADS (>10 on even numbered questions)	No correlation
Wagner 1998	344 ng/dl	N/A	HAM-D (continuous)	Yes (p = NR)
Wagner 1998	550 ng/dl; week 11:1287 ng/dl	N/A	HAM-D (continuous)	Yes (t = 6.0, p = .000)

BDI: Beck Depression Inventory; CES-D: Center for Epidemiologic Studies Depression Scale; HADS: Hospital Anxiety and Depression Scale; HAM-D: Hamilton Depression Rating Scale; PHQ-9: Patient Health Questionnaire-9; POMS: Profile of Mood States

3.2. Overview of Included Studies

The majority of the studies were conducted in the USA (n=16), with others conducted in Iran (n=1), Thailand (n=1), and the Netherlands (n=1). The studies used a wide variety of screening tools for depression, including the Profile of Mood States (POMS), Patient Health Questionnaire 9 (PHQ-9), Beck Depression Inventory I and II (BDI I/II), Hamilton Depression Rating Scale (HAM-D), Structured Clinical Interview for DSM-IV (SCID), and one used the Thai Hospital Anxiety and Depression Scale (Thai HADS). The number of study participants ranged from 23 - 849 participants. Inclusion criteria varied by study, with some including any people with HIV, many including only biological men or women, and some with other specific criteria (men who have sex with men, men with hypogonadism, etc.). The studies measured various hormones; cortisol (n=7) and testosterone (n=8) were the most frequently measured hormones. In addition, one study measured Prolactin, Testosterone, SHBG, LH, FSH, and 17b-estradiol.¹⁸ The table also includes average CD4 counts, HIV viral load, and % of participants on ART when available in the studies.

3.3. Outcomes

3.3.1. Cortisol

Of the seven papers that measured cortisol, five found some correlation between cortisol levels and measures of depression. ^19-23 One study found no correlation,²⁴ while one did not directly correlate the POMS score with cortisol levels.²⁵ Interestingly, one paper reported a negative correlation between plasma cortisol levels and physiological distress using the POMS depression subscale.²³ The remaining papers reported a positive correlation between depression and plasma cortisol, with r values ranging from 0.25-0.35.^19-22 A study by Wisnieweski found a 2% increase in the odds of experiencing depressive symptoms for every unit (μg/d) increase in cortisol. They also found the correlation stronger in female participants than male participants.²²

3.3.2. Dehydroepiandrosterone Sulfate (DHEAS)

Two studies measured DHEAS.^26,27 Ferrando included only male participants, while Goggin included only female participants. Neither study found any correlation between plasma DHEAS levels and depression score (utilizing the BDI and HAM-D scales, respectively).

3.3.3. Estradiol

Estradiol was numerically lower in women with HIV with CES-D ≤ 15 than in women without both depression and HIV or who had one or the other.²⁸ One study compared estradiol levels in men but did not significantly correlate with depression scores via the BDI (r= 0.15, P=0.24).²⁹

3.3.4. Oxytocin

Only one study measured Oxytocin.³⁰ It only included female participants. They found a U-shaped curve with baseline oxytocin levels and depressive symptoms at ten weeks.

3.3.5. Prolactin

One study by Gormin et al. measured plasma prolactin levels.³¹ They had only male participants, separated into three groups according to the CDC stages of HIV/AIDS infections (utilizing stages II, III, and IV). They did not find any significant correlation between plasma prolactin levels and HAM-D scores in any group.

3.3.6. Sex Hormone Binding Globulin (SHBG)

Two studies measured SHBG, and both included only male participants.^29,32 Neither found a significant correlation between plasma SHBG levels and depression scores.

3.3.7. Total Testosterone (TT)

Seven studies measured total testosterone levels.^{18,27,29,32-35} Most of the studies used the BDI or the HAM-D scales to measure depressive symptoms. Of these, four reported a statistically significant negative correlation between total testosterone levels and depression scores.^29,33-35 Several of these studies were intervention-based and measured depression levels before and after a trial of testosterone supplementation therapy.^34,35

3.3.8. Free Testosterone (FT)

Four studies measured free testosterone, two of which also measured total testosterone and are discussed above.^29,33 Two of these four, both using the BDI scale for depression, found a statistically significant correlation between depression scores and FT levels.^29,36 While Blick did not directly correlate depression and FT; they showed that as FT levels significantly increased, PHQ-9 scores also significantly improved.³³ The remaining study found no correlation between FT and depression.²⁶

3.3.9. Non-hormonal outcomes

While this review focuses on depression and hormone levels as a primary outcome, the included studies sometimes focused on additional outcomes. For example, several studies addressed sexual dysfunction. Others explored nutritional status, exercise, drug use, and neuropsychiatric outcomes. We will not comment on these outcomes here.

3.4. Risk of Bias Assessment

No studies were found to be entirely free of potential risk of bias, as represented in Figures 2, 3 and 4. Within the set of nonrandomized studies, only two studies justified their sample sizes. However, the lack of sample size justification may not be a significant concern in the case of more exploratory research. There was an unclear risk of bias for the majority of studies when considering participation rates. As the studies generally relied on convenience samples, and the discrepancy between those approached and those who agreed to participate is not always clearly articulated, there may be some concern that the study participants may not represent the larger population of interest. Some of the other domains of potential bias, such as the risk of bias associated with the timing of exposure assessment and outcome measurement, are due to the prevalence of cross-sectional studies. Within the three randomized studies, unclear risk of bias was most associated with a lack of details describing the randomization process and adherence to intervention.

Figure 2.

Risk of bias in cross-sectional studies

Figure 3.

Risk of bias in longitudinal studies

Figure 4.

Risk of bias in randomized studies

4. Discussion

Unsurprisingly, cortisol, long linked with stress, was correlated with depression in nearly all studies. DHEAS was measured in men and women in one separate study each and was not significantly associated with depression in either study. Both total and free testosterone were linked to depression in men, although not in women (women tested in only one study). Oxytocin had a complex U-shaped relationship with depression, although only tested in women. Prolactin was only measured in men in one study and was not significant. Finally, we only found two studies that evaluated estradiol. In women, estradiol was numerically lower in those depressed with HIV than controls. In men, estradiol and depression were not significantly correlated.

4.1. Cortisol

Correlations between cortisol and depression in the general population are complex. Most studies show altered cortisol responses in depression.^37-39 Treatment with electroconvulsive therapy for major depression was shown to produce a long-term reduction in cortisol.⁴⁰ Cortisol has been shown to have different effects in men and women, with men having more reactivity and women having a blunted response.⁴¹ HIV is known to alter adrenal function with higher basal cortisol levels and risk for adrenal insufficiency.⁴² HIV medications improve outcomes, but increased inflammation from increased cortisol remains.⁴³ As such, we think our findings are consistent with the literature.

4.2. DHEAS

DHEAS has broad neurologic properties and is known to have both anxiolytic and antidepressant properties.⁴⁴ It acts directly on the gamma-aminobutyric acid (GABA) and N-methyl-d-aspartate (NMDA) pathways, important in mental illness.⁴⁵ A recent meta-analysis showed that DHEAS levels are associated with developing depression, with DHEAS decreasing with age.⁴⁶ Higher levels of DHEAS are protective against depression in the elderly.⁴⁷ HIV is associated with low DHEAS, and low DHEAS is a poor prognostic indicator.²⁶ As such, we would expect DHEAS to be associated with depression in people with HIV, but likely our sample size is too small.

4.3. Testosterone

Testosterone is strongly associated with depression in men.⁴⁸ Testosterone is also strongly associated with depression in women with polycystic ovarian syndrome (PCOS)⁴⁹ In women without PCOS, there is also a correlation between depression and testosterone, although with a complex relationship and potentially J-shaped curve to the testosterone level.⁵⁰ Early studies showed decreased testosterone associated with HIV,⁵¹ but more recent ones with individuals on HIV therapy showed some alterations but less so.⁵² In women, AIDS wasting is associated with androgen deficiency.⁵³ Our study from Laan showed that testosterone was frequently low in women with HIV.¹⁸ Overall, the literature suggests that our findings in men with HIV are to be expected; more data is needed in women with HIV for depression and testosterone.

4.4. Oxytocin

Oxytocin is closely aligned with perinatal depression, which was excluded in this work, with higher oxytocin being protective.⁵⁴ However, oxytocin is known to have antidepressant effects in animal models and post-mortem studies of individuals who had been depressed.⁵⁵ Sanwald et al. evaluated epigenetic regulation of the gene coding for oxytocin; they found significant differences between men and women.⁵⁶ Specifically in depression, oxytocin coding is less activated in women than men. However, oxytocin was not correlated with depression severity overall.⁵⁶ In addition, oxytocin immunoreactivity was found to be decreased in hypothalamic neurons in people with HIV.⁷ Thus, we would expect differences between men and women in depression and oxytocin’s effects. We also would expect complex relationships with oxytocin which our data show, although more testing should be done, especially in men.

4.5. Prolactin

Like oxytocin, prolactin has been studied in depression among peripartum women, although the link between depression and prolactin is somewhat equivocal.⁵⁷ In mice, prolactin levels in the hypothalamus are increased due to stress, which is thought to regulate depression.⁵⁷ Electroconvulsive therapy, an effective treatment for depression, increases prolactin in people with depression.⁵⁸ In men with HIV, bioactive prolactin was elevated, while the response of immunoreactive and bioactive prolactin suggested a decrease in dopaminergic tone.⁵⁹ In a population of Nigerian women, they found that prolactin levels did not vary significantly between those with HIV and controls at coordinated phases of their menstrual cycle.⁶⁰ We did not see an association between depression and prolactin, but we only found one study in men.

4.6. Estradiol

One study of estradiol in men found men under 60 years old with depression had higher estradiol levels.⁶¹ Estradiol has been shown to regulate the expression of monoamine oxidases, the serotonin transporter, and the serotonin-1A receptor in women.⁶² Given the known role of serotonin and monoamine oxidases in depression, estradiol likely has profound impacts on depression in women. In men, estradiol has not been associated with depression.⁶³ In addition, there is data showing that estradiol can regulate infections, including HIV, and that estradiol may be protective against HIV, although it depends on the phase of the infection.⁶⁴ Overall, estradiol needs more study in depression and HIV but is more relevant for women, with our data being consistent with that finding.

4.7. Progesterone

Progesterone and its metabolite allopregnanolone are important in both men and women and are synthesized in the nervous system, the adrenal gland, and, in women, the ovaries.⁶⁵ Allopregnanolone acts directly on the GABA receptor, and progesterone is known to have a key role in neuroprotection. Low progesterone has been associated with depression and anxiety.⁶⁶ Similar to estradiol, there is data showing that progesterone can regulate infections, including HIV, and that progesterone may protect against HIV early in the disease course. However, later in the disease course, women progress to AIDS faster than men.⁶⁴ We did not find papers with progesterone in our search for hormones and depression in HIV, which the literature suggests is an area that should be further studied.

4.8. Limitations

This systematic review has several important limitations. First, despite the large body of literature screened, relatively few studies met inclusion criteria. Second, these studies were heterogeneous, generally including different measures of depression and considering different biomarkers. Third, few studies included both male and female participants, limiting our ability to draw conclusions regarding the role of sex in this context. Finally, methodologically, most of the studies featured incomplete reporting. Lack of details regarding selection, randomization, timing, and adherence creates difficulties when assessing research and applying findings. However, it is possible that reporting standards changed since the time of the articles’ initial publications, as several studies were published in the early 1990s. Overall, the lack of statistical conclusions and consistency between the studies left us unable to perform a meta-analysis of this data. The prevalence of depression in people with HIV and its importance in treatment outcomes is a large and important gap in existing research. More meaningful work should be done in this area to better understand the pathophysiology of depression in people with HIV and how gender may affect this comorbidity. We have intentionally avoided describing these results as evidence bioreductionism, and acknowledge that depression in people with HIV is a complex disease process that undoubtedly includes psychological and social influences such as stigma, living with a chronic illness, and other social determinants of health experienced by people with HIV. However, we maintain that neuroendocrine markers and other biological factors are worth studying as one aspect of the pathophysiology.

Of note, we do not see transgender individuals in our results. Transgender individuals were not excluded; however, they did not come up in our search. It could be because of a lack of biomarker testing or understudy of a marginalized population. Indeed, we know depression is very high among transgender individuals, with a recent Canadian study of transgender women with HIV who had a rate of depression of 45%.⁶⁷ Further work to evaluate hormone therapy in this population is warranted.

Also, we did not see reports of women on hormonal contraceptives. These were, again, not excluded but not found in our search. A review looking at hormonal contraceptives showed improving mental health in women with premenstrual dysphoric disorder.⁶⁸ They report there are not consistent findings of mood disorders with hormonal contraceptives, although there are individual groups with susceptibilities. Further study in women with HIV would be helpful.

Finally, given our narrow inclusion criteria the majority of studies were from the USA. We did not limit countries and would have preferred more representation but were limited by the literature. Specifically, while we found papers from Iran, Thailand, and the Netherlands, we did not find papers from Africa or South America. We suspect the range of countries will expand with more interest in depression in people with HIV.

5. Conclusion

This systematic review of hormones in people with HIV using sex as a biological variable found substantial differences in relevant hormones in their importance for depression by sex. However, these were nearly all studies done in one sex, not comparisons between both. Given the importance of hormones in depression and the impact of HIV on these pathways, better understanding the pathophysiology of the neuropsychiatry in HIV is crucial. Since hormone levels vary significantly by sex yet are important for both biological sexes, further work in individual biological sexes and comparison between groups is needed. Additionally, there was a paucity of data overall to understand this pathophysiology. We call for further research in this area to improve the lives of people with HIV.

Highlights.

• -Significant correlations between HIV-associated depression and hormone levels were found for cortisol (both sexes), testosterone (only in men), oxytocin (only tested in women), and estradiol (only in women). -No significant correlation between depression and hormone level was found for prolactin, dehydroepiandrosterone (DHEAS), or sex hormone binding globulin (SHBG) and no studies measured progesterone.

• -Nearly all studies included only men or women and did not directly compare biomarkers between the two sexes.

• -More research is needed to compare hormone signaling among those with HIV and depression by sex.

Appendix A: Search Strategy for Embase via Ovid

1. exp Human immunodeficiency virus/

2. ("human immunodeficiency virus" or "human immunodeficiency viruses" or "HIV-1" or "HIV-2" or HIV or "LAV-HTLV-III" or "HTLV-III" or "Lymphadenopathy-Associated Virus" or "Lymphadenopathy Associated Virus" or "Lymphadenopathy-Associated Viruses" or "Lymphadenopathy Associated Viruses" or "AIDS virus" or "AIDS viruses" or "Acquired Immune Deficiency Syndrome Virus" or "Acquired Immune Deficiency Syndrome Viruses").tw,kw.

3. (human adj2 ("t cell" or "t-cell") adj2 ("leukemia virus*" or "lymphotropic virus")).tw,kw.

4. or/1-3

5. exp progesterone derivative/

6. exp sex hormone/

7. (progesterone or estrogen* or oestrogen* or testosterone or estradiol or dihydrotestosterone or androgen* or progestrogen* or endocrinopath* or algestone or hydroxyprogesterone* or pregnenedione or epitestosterone or hydroxytestosterone* or methenolone or mehtyltestosterone or estrane* or anaprotin or andractim or dihydroepitestosterone or androstanolone or mesterolone).tw,kw.

8. exp biological factor/

9. (cytokine* or interferon* or interleukin* or chemokine* or intercrine* or chemotactic or "growth differentiation factor* ").tw,kw.

10. or/5-9

11. 4 and 10

12. exp mood disorder/

13. (depressive or depression or depressed or depressions or MDD or mood or (mental adj1 (health or illness* or disorder*)) or psychiatr* or psycholog*).tw,kw.

14. or/12-13

15. 11 and 14

16. exp Adults/

17. exp child/

18. exp infant/

19. exp adolescent/

20. or/17-19

21. 20 not 16

22. 15 not 21

23. exp human/

24. exp animal/

25. 24 not 23

26. 22 not 25

27. ..dedup 26