No evidence for the immunocompetence handicap hypothesis in male humans

The observations that testosterone might be immunosuppressive, form the basis for the immunocompetence handicap hypothesis (ICHH). According to ICHH only high-quality individuals can maintain high levels of testosterone and afford the physiological cost of hormone-derived immunosuppression. The animal and human studies that attempted to support the ICHH by precisely defined impairment of immunity associated with high testosterone levels are inconclusive. Furthermore, human studies have used only selected immune functions and varying testosterone fractions. This is the first study examining the relationship between multiple innate and adaptive immunity and serum levels of free testosterone, total testosterone, DHT and DHEA in ninety-seven healthy men. Free testosterone and marginally DHT levels were positively correlated with the strength of the influenza post-vaccination response. Total testosterone and DHEA showed no immunomodulatory properties. Our findings did not support ICHH assumptions about immunosuppressive function of androgens. In the affluent society studied here, men with higher levels of free testosterone could afford to invest more in adaptive immunity. Since the hormone-immune relationship is complex and may depend on multiple factors, including access to food resources, androgens should be treated as immunomodulators rather than implicit immunosuppressants.

The correlations between immune parameters and controlled factors or androgens. Absolute T lymphocyte count (R = −0.24; p = 0.02) and the strength of the influenza post-vaccination response (R = −0.26; p = 0.01) were negatively associated with participants' age. BMI and body fat were not related to the analyzed immune functions. Among the hormonal factors, fT was positively related to the strength of the influenza post-vaccination response (R = 0.34; p < 0.001), while DHT was positively associated with ROS production (R = 0.22; p = 0.03), IgG serum levels (R = 0.217; p = 0.03), and the strength of the influenza post-vaccination response (R = 0.25; p = 0.01). Both total testosterone and DHEA were not correlated with any of the immune parameters measured ( Table 2).

The relationship between innate immune functions and androgens.
Because some immune functions depend on age and BMI, we checked the relationship between hormone levels and immune function when age and BMI were controlled. Multiple regression analysis models were constructed separately for each immune factor as a dependent variable and each hormone as an independent variable. Among innate immune parameters, the phagocytic uptake was marginally negatively correlated with fT (p = 0.06), while DHT was marginally positively correlated with ROS production (p = 0.07) ( Table 3). However, the supplemental analyses showed that the relationship between fT and phagocytic uptake or DHT and ROS production became nonsignificant when removing statistical outliers for the respective immune parameters from the models. Total testosterone and DHEA were not associated with any of the immune factors tested ( Table 3).
The relationship between adaptive immune functions and androgens. Among the adaptive immune parameters, only the strength of the influenza post-vaccination response was positively predicted by fT (p = 0.03) and marginally significantly (p = 0.06) by DHT. Other immune functions were not correlated with any of the studied hormones (Table 4).
According to the hypothesis that interaction between testosterone and immune function may be mediated by stress hormones 20,40 we repeated the analyses presented in Table 3 and Table 4 using cortisol as another predictor. The analyses show that none of the androgen-immune interactions changed after the adjustment for cortisol levels. The immune response to influenza vaccination can be also analyzed according to the immunological standards, i.e. using a dichotomous scale (negative or positive immune response). A positive response is reflected by at least a four-fold increase in the antibody titer after vaccination 41 . To determine whether the group with a positive immune response (seroconverted) differed in any androgen levels from the group with negative immune response (lack of seroconversion), we compared androgen levels using the Mann-Whitney U test. The seroconverted group (N = 73) appeared to have on average higher fT levels than the group without seroconversion (N = 24) (Z = 2.32; p < 0.02). The serum concentrations of other androgens did not differ between these two groups.

Discussion
Immune system efficacy is a very important element of human physiology, and is perceived as one of the best measures of an organism's biological condition. Any defect in the immune function leads to inadequate protection and poses a significant risk to survival. According to the ICHH, morphological traits that are sexually dimorphic or related to sexual ornamentation may signal biological quality, and therefore could also reflect good immune quality. To test the hypothesis that androgen-dependent male ornamentation is also an honest signal of biological quality, we analyzed various parameters of immunity in relationship to different androgen levels in healthy adult males. In the well-nourished Western society studied here, we found very limited evidence for a relationship between some androgens and adaptive immunity. Our results do not confirm the hypothesis that androgens exert immunosuppressive effect. Below we discuss our results in details, categorized by the immune topic.
When controlling for age and BMI, none of the measured innate immune parameters were related to androgen concentrations. Our results are in contrast to the in vitro study by Marin et al. 42 , which documented an inhibitory effect of testosterone on neutrophil function. The experimental studies on mice with knockout of androgen receptors show, however, that these mice have functional defects of neutrophils including the lower production of proinflammatory cytokines, but retain the normal capacity of phagocytosis and ROS production 43 . We obtained the similar result in men i.e. no effect of androgens on neutrophils functions.
Our study corroborates the correlational study of Prall and Muhlenbien 22 on complement activity and saliva DHEA concentrations. Although the data documenting a potential down-regulatory role of androgens in innate immunity are very limited, the evidence from correlational studies indicates that innate immune factors are independent of serum androgen concentrations.
We found that when controlling for age and BMI, androgens were not associated with the total number of T and B lymphocytes, i.e. major cells involved in adaptive cell-mediated and humoral immunity, respectively. A functional test of lymphocytes showed that androgen concentrations were not related either to lymphocyte proliferation after mitogenic stimulation or to total IgA and IgG immunoglobulin levels. Interestingly, the strength of the immune response to the influenza vaccine was positively predicted by the level of fT and marginally significantly by DHT. This indicates that hormones can affect rather the function of lymphocytes (in response to a stimulus) than their total number in blood.
The majority of human and animal in vitro studies which focused on quantitative analyses of lymphocytes have shown a negative impact of testosterone on lymphocyte proliferation [44][45][46] or antibody production 47 . A limited experimental human study conducted on transsexual Caucasian women found no effect of testosterone administration on IgG levels 48 . The results of correlational studies, however, are inconclusive. Whereas some reported results similar to ours, i.e. observing no association between immunoglobulins and testosterone levels 49 , others found a positive relationship 50 . It is worth noting that in contrast to the study on Filipino men 51 , our study, as well as those of van Anders 49 and Giltay et al. 48 (the latter only on women) were conducted in well-nourished, affluent societies. These differences may suggest the influence of ecological factors on immune-testosterone interaction. In a correlational study linking lymphocyte function with androgens other than testosterone, Prall and Muehlenbien 22 found that saliva DHEA level was negatively correlated with lymphocyte proliferative response.
Because of the small sample size (only 20 men) and no controlling for participant age, adiposity, or health status, the conclusions of Prall and Muehlenbien should be, however, treated with caution. In general, our results on the baseline immune parameters and androgens are compatible with the meta-analysis findings on many species by Foo et al. 51 . They showed that in contrast to experimental studies, there was no significant overall correlation between testosterone and immune function in correlational studies.  There are also inconclusive data for the relationship between androgen concentrations and the ability to produce specific antibodies post-vaccination. Our results show that men with higher levels of fT produce more specific anti-influenza antibodies in response vaccination. This finding is similar to those obtained by Rantala et al. 20 , who found a positive relationship between total testosterone levels and immune response to hepatitis B vaccination. The opposite finding for fT and response to influenza vaccination was reported by Furman et al. 21 . Their men sample, however, was small (N = 34), and the median age of participants was 63 years. Age is an important factor in immune response to vaccination it is because the response tends to decrease with age. In Furman et al. 21 , only 10 out of 37 men (27%) responded positively to the influenza vaccine by showing seroconversion. In our study 75.3% of the vaccinated men were seroconverted, what is similar to the rate obtained in healthy young people in another study (68%-84%) 52 . Furthermore, Furman et al. 21 did not control for adiposity, which may influence the immune response to the influenza vaccine 53 .
Due to the small sample of men who were also vaccinated against tetanus in our study, we did not present the response analyses for this antigen. It is worth noting, however, that although nonsignificant (p = 0.11), the relationship between fT and the response to tetanus vaccination was in the same direction (R = 0.27) as that of the influenza vaccine.
There are a few possible explanations of higher immune reactivity in more androgenic men (with higher level of fT). Firstly, it is possible that in energetically replete populations (as in the case of our study), men with high genetic quality can afford to invest in more costly adaptive immune arm and at the same time in high androgens production. Such explanation agrees with the hypothesis that only individuals with high immune quality are able to incur the immunosuppressive effect of testosterone. Secondly, the proportional marginal viability cost that is associated with high testosterone level is lower for individuals with higher immune quality. According to Getty et al. 54 the experimental and correlational studies might only reflect absolute cost, whereas it is proportional marginal fitness cost (i.e. in relationship to benefits) that might be crucial for an individual ability to develop sexual signal. Thirdly, because the immune activation in response to infection is related to down-regulation of testosterone levels, higher testosterone in men can be consider as a signal of better physiological condition (including immunity) 55 . Finally, if testosterone is not absolutely immunosuppressive (but rather immunomodulatory), high genetic quality individuals can invest more in maintenance both in high level of viability (reflected by immune efficiency) and in fecundity (active form of androgens).
Our results also indicate that to test the hormonal basis of the immunocompetence hypothesis, the immunological challenge (such as vaccination) may be more informative than baseline immune parameters or functional tests performed on isolated immune cells (such as neutrophil phagocytic uptake or lymphocyte proliferative response). It is because of the nature of immune system network function, when the basal levels of immune parameters occurring in healthy individuals do not necessarily reflect reactivity of immune cells in response to a real infection.
Overall, we did not confirm the ICHH as postulated by Folstad and Karter 2 . There is, however, one problem with this hypothesis. The positive, negative, or lack of relationship between androgens and immune function do not need to undermine the ICHH 54 . In the case of a positive relationship, one can assume that individuals with high testosterone levels can also afford high immunity, and in the case of a negative relationship, that healthy men are trading off between both. The majority of research validating the ICHH has been performed on birds, but results of animal studies are also difficult to interpret because of a lack of "normal range" definition for immunological parameters in healthy individuals. Higher levels of measured leukocytes or immunoglobulins may be interpreted as a higher immune potential (better immunity) or as a result of outgoing infection (weaker immunity) 6 . In the case of individuals living in Western societies, recent data do not support either the ICHH or the positive relationship between the traits perceived as attractive and immune functioning. Foo et al. 56 for instance, found no relationship between components of men's facial attractiveness and selected innate immunity factors. This indicates that testosterone-dependent facial masculinity does not have to signal the immune effectiveness.
In the light of previous results, the main question is why androgens exert different effects on immune cells in vitro and in vivo. Possibilities include molecular mechanisms responsible for local androgen conversion by immune cells, synthesis of specific enzymes, or patterns of sex hormones receptors expressed on immune cells. Firstly, it is possible that in vivo, testosterone can exert indirect effects on immune cells after local conversion by immune cells to estrogenic metabolites and binding to estrogen receptors 57,58 . It is well-known that estrogens up-regulate immune functions, and this may be one reason for the lack of immunosuppressive effect of testosterone observed in our study. Secondly, ARs are expressed by various immune cells and play an essential role in androgen-associated immune regulation (see 15 ). Studies on mice lacking androgen receptors 59 or experimental human studies 60 have shown that the comprehensive impact of hormones on immune response depends not only on hormone concentrations but also on qualitative and quantitative expression profiles of specific immune-cell receptors present in various tissues. This evidence suggests that individual differences in the expression of ARs on various immune cells can explain why the same androgen levels may exert effects of varying magnitude on immunity. Lastly, the effect of AR signaling differs in vitro and in vivo. Yang et al. 61 for instance, showed that AR activation exerted an inhibitory effect on prostatic epithelial cell proliferation in vitro, whereas in vivo, the activation had a stimulatory effect. According to above evidence, to better understand androgen-immune interaction, further investigation of ICHH assumptions should include not only hormone levels but also specific receptor expression patterns and enzyme synthesis by immune cells which convert androgens into other metabolites. The second problem is related to different physiological or ecological factors influencing immune-androgen interaction. Both animal and human studies indicate that immunity development in early life can be crucial. In this respect, McDade et al. 24 suggests three potential elements: "(a) the availability of nutritional resources, (b) the intensity of pathogen exposure and (c) signals of extrinsic mortality risk". The combination of these factors may lead to distinct immunomodulatory effects by androgens. There are reports showing trade-offs between different fitness components when resources are limited 24,62 . Some animal studies showed that food availability can influence the relationship between testosterone levels and immunity 63 and that leptin may prevent the immunosuppressive role of testosterone 64 . Other factors which may affect testosterone-immune interactions are stress 20 and living conditions 65 . Trumble et al. 65 found that cell-mediated response to mitogens was inversely related to testosterone levels in Tsimane men aged 40-89 from an energy-limited subsistence population, which faces a high pathogen burden and is therefore immunologically stressed.
Finally, it should be noted that the majority of empirical studies (both experimental and correlational) are mainly designed to measure absolute cost of "signals" (androgens level) which will might be not sufficient to define viability-fertility (i.e. immune-hormones) trade-offs 54 . According to Getty et al. 54 costly signals should be considered as an investment instead of handicaps, and we should avoid the assumptions that individuals who express higher sexual signals must have "wasted" more viability to develop them than lower quality individuals with lower expression of sexual signals. The ICHH prediction is based on an absolute cost associated with the levels of expression sexual signals, whereas it is the marginal fitness cost and benefits (not easily measurable in empirical studies) that might be crucial 54 .
Despite some limitations, the strength of our study is derived from four factors: (a) A relatively large sample of men in their reproductive age, which is important when testing evolutionary hypotheses; (b) a variety of immune biomarkers evaluated; (c) four different androgens studied; (d) controlling for general health status using medical biomarkers.
It is also worth noting that although our study is mainly correlational, it is the first study testing the relationship between different androgens concentration and many other immune functions simultaneously. This means that in contrast to previous studies, our conclusions are about the general immune-androgens in vivo interactions that are based on multiple immune parameters.
In summary, we found that fT and DHT are probably the hormones most relevant for studying immune parameters potentially affected by androgens. These two reactive forms of androgens positively influence adaptive immunity, i.e. the strength of the response to the influenza vaccine, and appear to have no effect on humoral or cell-mediated innate immunity. This finding contrasts with the immunosuppressive function of androgens observed in vivo. The relationship between androgens and immune function is, however, complex, and probably restricted only to particular immune parameters and does not apply to all. Therefore we suggest treating androgens as immunomodulators rather than immunosuppressants, as supported by the important role of ecological factors in moderating the relationship between immunity and androgens. Further investigations regarding the association between serum androgen concentrations, ARs, and various immune functions are required.

Methods
Participants. One hundred and thirty-four men (aged 18.9-36.7 years) who were volunteers and declared no health problems were recruited for the study. Participants who did not complete the study, reported an infection between the two office visits, had above-normal ranges in any blood parameters, had body mass index (BMI) or body fat values above or below the 3 rd standard deviation, and were outliers in sex hormone levels, were excluded from the analyses. The final sample was 97 healthy men (aged 19-36.7 years, mean age 27.2). Thirty-five of these men were also tetanus-vaccinated (aged 20.7-36.7; mean age 28.7) and evaluated for immune response to tetanus toxoid. The lymphocyte T proliferation rate was assessed for 62 participants. Most of the participants lived in Wroclaw (Poland) and the suburbs (85%), had university level education (63%) or be students (24%).

Patient examination scheme.
Participants examination was performed in local private clinic. The general study procedure for recruited participants consisted of two visits spaced 4 weeks apart. Both visits were scheduled in the morning hours. The first visit included medical qualification to vaccination, blood sample collection for biochemical, immunological, and hormonal analyses, influenza or both influenza and tetanus vaccine injection, and height, weight, and body fat percentage measurements. Four weeks later, during the second visit, participants submitted to blood sample collection to evaluate the strength of the immune response to vaccination and filled in questionnaires on illnesses after the first visit. All participants were tested for blood morphology and levels of fasting glucose, creatinine, alanine aminotrasferase, aspartate aminotransferase, and C-reactive protein to evaluate general health status. All participants were vaccinated against seasonal influenza using Vaxigrip ® (SanofiPasteur, Immune function measurements. Participants' immune system quality was determined using key immune mechanisms constituting both innate and adaptive immune responses. It is important to note that we used immune functions representing various aspects of immune defense. We assessed the parameters: (1) creating humoral innate immunity (total complement and lysozyme activity); (2) constituting cellular innate immunity (the neutrophil function including phagocytosis and reactive oxygen species production) as well as those (3) involved in cell-mediated immunity (T lymphocyte count and proliferative response after mitogen stimulation), and (4) involved in humoral-mediated immunity (B lymphocyte counts, IgA, IgG and the immune response to influenza or tetanus vaccinations). All measured parameters are included in the test panel for evaluating the immune functioning and were recommended as a screening test for detect primary immunodeficiency 66 .
Innate immune function. Total complement activity was measured in zymosan-activated serum using commercial enzyme immunoassay kit (MicroVue, Quidel ® ). Test procedure was performed according to the manufacturer instruction. Results were expressed in ng/ml. Lysozyme activity of human serum was measured using turbidimetric assay and Micrococcus lysodeikticus cell suspension, and expressed as a difference in absorbance value between control and test sample.
Phagocytic uptake by neutrophils was measured using commercial kits (PHAGOTEST, Glycotope ® ). The mean fluorescent intensity of phagocytosing neutrophils was calculated using WinMDI software.
Reactive oxygen species (ROS) production by isolated neutrophils after PMA stimulation, was measured using luminol-dependent chemiluminescence assay (CL). The results were expressed as area under chemiluminescence curve (AUC) for stimulated neutrophils divided by AUC for controls. The lymphocyte proliferative response after mitogen stimulation was measured using [3H] thymidyne incorporation assay (modified method described in). Results were expressed as stimulation index (SI) calculated as count per minute (CPM) for mitogen-stimulated sample divided by CPM for unstimulated controls.
Total serum immunoglobulin IgA and IgG levels were measured with enzyme-linked immunosorbent assay, using previously calibrated reagents concentration and series of participant's serum samples dilution. Results were expressed in g/L.
The strength of immune response to flu vaccine the titre of specific IgG antibody was measured before and 4 weeks after vaccination 41,52 . Anti-flu antibody titre was measured with the standard ELISA method, using microplates coated with influenza vaccine antigens (Vaxigrip ® , SanofiPasteur) and previously calibrated reagents and participant's serum sample dilution. Immune response to vaccine was expressed as fold-increase in specific antibody titre from pre-to post-vaccination.
The strength of anti-tetanus IgG was measured using commercial kits (DEMEDITEC ® ), and previously calibrated participant's serum sample dilution. Test procedure was performed according to the manufacturer instruction. The results were expressed as fold-increase between pre-to post-vaccination specific antibody concentration.
The immunological methods and test procedure are described in details in the supplementary materials (SI).
Statistical analysis. Normality for variables was assessed with Shapiro-Wilk tests. Despite logarithmic transformation, the majority of variables were still not normally distributed. A series of Spearman correlations using raw data were carried out to test inter-correlation between immune parameters and the relationship between immune functions and participants' age, BMI, percent of body fat, and each hormone. The relationship between immunity and hormones was also tested with different multiple regression models, controlling for known immunomodulatory factors such as participants' age and BMI. Regression models were constructed following natural logarithmic transformation of data (except age) to normalize the residuals distribution. The mean androgen levels between the seroconverted group and group without seroconversion were compared using Mann-Whitney U test. Analyses were carried out using Statistica 12 (StatSoft Inc.). Human Subjects" formulated by World Medical Association in 2013 (https://www.wma.net/policies-post/ wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/). Data availability. The authors declare that the data supporting the findings of this study are available in supplementary information files.