Testosterone Deficiency Is a Risk Factor for Severe COVID-19

Background Male sex is related to increased COVID-19 severity and fatality although confirmed infections are similarly distributed between men and women. The aim of this retrospective analysis was to investigate the impact of sex hormones on disease progression and immune activation in men with COVID-19. Patients and Methods We studied for effects of sex hormones on disease severity and immune activation in 377 patients (230 men, 147 women) with PCR-confirmed SARS-CoV-2 infections hospitalized at the Innsbruck University Hospital between February and December 2020. Results Men had more severe COVID-19 with concomitant higher immune system activation upon hospital admission when compared to women. Men with a severe course of infection had lower serum total testosterone (tT) levels whereas luteinizing hormone (LH) and estradiol (E2) levels were within the normal range. tT deficiency was associated with elevated CRP (rs = - 0.567, p < 0.001), IL-6 levels (rs = - 0.563, p < 0.001), lower cholesterol levels (rs = 0.407, p < 0.001) and an increased morbidity and mortality. Men with tT levels < 100 ng/dL had a more than eighteen-fold higher in-hospital mortality risk (OR 18.243 [95%CI 2.301 – 144.639], p = 0.006) compared to men with tT levels > 230 ng/dL. Moreover, while morbidity and mortality showed a positive correlation with E2 levels at admission, we detected a negative correlation with the tT/E2 ratio upon hospital admission. Conclusion Hospitalized men with COVID-19 present with rather low testosterone levels linked to more advanced immune activation, severe clinical manifestations translating into an increased risk for ICU admission or death. The underlying mechanisms remain elusive but may include infection driven hypogonadism as well as inflammation mediated cholesterol reduction causing gonadotropin suppression and impaired androgen formation. Finally, in elderly late onset hypogonadism might also contribute to lower testosterone levels.


INTRODUCTION
Coronavirus disease 2019 (COVID- 19) is still influencing the daily life of people all over the world. Most patients infected with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have developed mild symptoms, while up to 20% need hospitalization due to severe disease course characterized by shortness of breath and hypoxia (1). Several comorbidities including hypertension, obesity, diabetes, cardiovascular disease, and chronic pulmonary disease as well as age were shown to be related to more severe COVID-19 courses (2)(3)(4)(5). Additionally, male sex is related to more severe COVID-19 manifestation, even though there are no sex differences in the absolute number of confirmed COVID-19 cases (6). Currently, up to 60% of hospitalized patients and even up to 82% of patients treated in the intensive care unit (ICU) are men (3,4,6). Accordingly, the case fatality rate (CFR) is 1.7 times higher in men compared to women (6)(7)(8). Inflammatory biomarkers were shown to be associated with COVID-19 morbidity and mortality in men and women: interleukin 6 (IL-6) as main inducer of Creactive protein (CRP) in the liver (9), neopterin reflecting macrophage activation and thus T-helper cell type I (Th1) immune response (10) as well as other acute phase proteins including procalcitonin (PCT) and ferritin (11).
Several reviews have elucidated potential mechanisms underlying these sex differences in COVID-19 patients and will be only shortly summarized hereafter (6,8,12). Biological sex affects immune responses to invading pathogens through hormonal regulation, gene expression and environmental factors (8). Generally, women have a stronger innate and adaptive immune response than men resulting in faster pathogen clearance (13). Many genes located on the sex chromosomes regulate innate immune function by encoding for pattern recognition receptors, cytokine receptors or transcriptional factors (14). On the other hand, androgen response elements and estrogen (E 2 ) response elements are found in promoters of several innate immunity genes thereby affecting their expression (15,16). These mechanisms are suggested to contribute to faster containment and clearance of SARS-CoV-2 in female patients (8). Moreover, sex-associated differences in expression and activity of the virus entry receptor angiotensin-converting 2 (ACE2) and its co-receptor transmembrane protease serine subtype 2 (TMPRSS2) are suggested to contribute to a higher mortality in men (17,18). The ACE2 gene is located on the X chromosome and estrogens were shown to upregulate its expression (19), while TMPRSS2 was shown to be regulated by androgen receptor signalling (20)(21)(22). Preclinical studies of ACE2 tissue expression have shown different results depending on the tissue type (23): ACE2 expression seems to be higher in the lungs, heart or kidney of male (24)(25)(26), while pancreatic ACE2 expression seems to be higher in female (27). However, the relationship of tissue ACE2 expression and circulating ACE2 activity is still not well understood and data from the literature is partial contradictory (23): in human, the ACE2 activity was shown to be higher in healthy men and men with heart failure compared to matched women (28,29) while other studies showed no sex-related differences in serum ACE2 activity (30,31). Interestingly, ACE2 activity does not differ between young and old men but is significantly higher in older compared to younger women (31).
However, whether these mechanisms actually affect disease severity and clinical outcome of COVID-19 patients or explain the sex differences remains unknown. Therefore, we analyzed different sex hormones and their interactions with inflammatory markers to assess the impact on disease severity and outcome in men: the androgen testosterone is the primary male sex hormone playing a key role in the male reproductivity and produced by testicular Leydig cells upon stimulation by luteinizing hormone (LH) released by the pituitary gland and regulated by the gonadotropin-releasing hormone (GnRH) released by the hypothalamus (32). Estradiol (E 2 ) is the primary female sex hormone but also involved in male reproduction and synthesized from androgens by aromatase (33). Sex hormones are transported in the human body unbound in the serum or bound to the sex hormone binding protein (SHBG) (32).

Study Population
We analyzed 377 patients with polymerase chain reaction (PCR)-proven COVID-19 hospitalized at our department at the Medical University of Innsbruck, Austria, between February and December 2020. Information on past medical history, concomitant medication, clinical characteristics and laboratory parameters were obtained from the local clinical electronic data management. The severity of the disease was categorized according to the score by the WHO Working Group on the Clinical Characterization and Management of COVID-19 infection (34). Events during hospital stay including death, ICU admission and need for mechanical ventilation were recorded for outcome analysis.
The study was conformed to the principles outlined in the Declaration of Helsinki and was approved by the ethics committee of the Innsbruck Medical University and patients had given informed consent (ID of ethical vote: 1167/2020).
We calculated the luteinizing hormone to total testosterone (LH/tT) ratio to specify the hypothalamic-pituitary-gonadal axis (35) and the total testosterone to estradiol (tT/E 2 ) ratio to reflect aromatase activity (36). Reference ranges for total testosterone (tT) levels were based on the guideline of the investigation, treatment and monitoring of functional hypogonadism in males by the European Academy of Andrology (EAA): serum total testosterone (tT) levels were classified to be reduced when tT levels ≤ 230 ng/dL, borderline when tT levels between 231 -350 ng/dL and normal when tT levels > 350 ng/dL (37).

Statistical Analyses
Parameters are reported as n (%) or medians (25th, 75th percentile) since the data were not normally distributed (Shapiro-Wilk test). To test for differences between men and women we used the Mann-Whitney-U test or Pearson chi-square tests. Kruskal-Wallis test was performed to test for significant differences between more than two groups. Analysis of the effect of risk factors on the probability of death or ICU admission during the in-hospital stay was performed with logistic regression analysis (not normally distributed parameters were logarithmized with the natural logarithm). All tests were twotailed and p-values < 0.05 were regarded as statistically significant. Statistical analysis was performed using SPSS Statistics Version 27 (IBM Corporation, Armonk, NY, USA).

Baseline Sex Hormones and Lipids in Men With COVID-19
Sex hormones were available from 267 patients (155 men, 112 women) in the first three days after hospital admission. Most hospitalized men with available sex hormones upon initial hospital admission (n = 155) presented with reduced total testosterone (tT) levels ≤ 230 ng/dL (n = 107, 69.0 %), while 22 men (14.2 %) presented with borderline tT levels between 231 -350 ng/dL and 24 men (15.5 %) with normal tT levels > 350 ng/dL. When differentiating men according to their age, men over the age of 60 had a significantly higher prevalence of reduced tT levels (81.5% vs. 52.5 %) and a lower prevalence of borderline tT levels (9.8 % vs. 21.3 %) when compared to men under the age of 60 (p < 0.001). Irrespective of the median age of 67 years, only 15 men out of 155 (9.7 %) were on pharmacological therapy for their benign prostate enlargement with a 5-alpha-reductase inhibitor, generally accepted to not influence the tT concentrations; indeed, these men showed median tT levels of 191 ng/dL (65 -419 ng/dL) and median free testosterone (fT) levels of 3.94 ng/L (2.05 -5.65 ng/L) as compared to men without 5-alpha reductase inhibitor therapy with tT levels of 148 ng/dL (70 -262 ng/dL, p = 0.494) and fT levels of 4.19 ng/L (2.49 -5.89 ng/L, p = 0.510). Interestingly, although tT levels were rather low in the majority of hospitalized men, levels of the gonadotropins luteinizing hormone (LH) and follicle-stimulating hormone (FSH) as well as the steroid hormone estradiol (E 2 ) were within the upper normal range ( Table 1). However, when differentiating men again according to their age, men over the age of 60 presented with slightly elevated LH levels (7.6 U/L [4.8 -12.0]) which were also significantly higher compared to men under the age of 60 (4.8 [3.5 -6.1], p < 0.001). Conversely, tT levels were significantly lower in men over the age of 60 compared to men under the age of 60 (130 ng/dL [50 -198] vs. 219 ng/dL [120 -359], p < 0.001), Cardiovascular disease, arterial hypertension, diabetes mellitus and hypercholesterinemia were frequently encountered in men within our study. Interestingly, serum low-density lipoprotein (LDL) and high-density lipoprotein (HDL) were quite low even in men without vs. with lipidlowering therapy (LDL: 81 mg/dL vs. 57 mg/dL; HDL: 32 mg/ dL vs. 30 mg/dL). Low serum cholesterol, LDL and HDL levels were associated with older age, higher immune activation, WHO-score and temperature and lower SpO 2 with concomitant higher O 2 requirements ( Table 2). Moreover, low tT levels as well as a lower tT/E 2 and higher LH/tT ratio were associated with lower cholesterol, LDL and HDL levels ( Table 3).

Sex Hormones, Disease Severity and Immune Activation
In men (n = 155) lower tT and fT levels upon hospital admission were correlated with older age, higher WHO score and temperature, as well as lower SpO 2 with concomitant higher O 2 requirement and longer hospital stay (Table 3). Also, lower dehydroepiandrosterone (DHEA) levels were significantly correlated with an older age, higher WHO score and lower SpO 2 , again with higher O 2 requirement and longer hospital stay. SHBG was positively correlated with age and hospitalization duration and negatively with BMI and O 2 requirement. Androstenedione (ASD), E 2 , LH and FSH were not related to disease severity. Finally, a higher LH/tT and a lower tT/E 2 ratio were associated with an older age, higher WHO score and temperature as well as lower SpO 2 with accordingly higher O 2 requirements and longer hospital stay ( Table 3).
When analyzing relations of sex hormones with markers of immune activation, tT and fT levels negatively correlated with CRP, IL-6, IL-10, neopterin, PCT, fibrinogen and ferritin levels as well as with leukocyte counts and positively correlated with absolute lymphocyte numbers. Also, other sex hormones correlated widely with different biomarkers of immune activation depicted in Table 3. Interestingly, a higher LH/tT and lower tT/E 2 ratio was associated with higher CRP, IL-6, IL-10, neopterin, PCT, fibrinogen and ferritin levels as well as lower absolute lymphocyte counts. (Table 3)

Testosterone and Estradiol Levels Predict the Outcome in Men With COVID-19
Univariate logistic regression analysis showed that lower tT, fT and DHEA levels, but higher ASD and E 2 levels, as well as a lower tT/E 2 and higher LH/tT ratio, were associated with an increased risk to die  Figure 1). Actually, 19 out of 20 men who died had reduced tT levels ≤ 230 ng/ dL; yet one man who died had a tT level of 244 ng/dL. Moreover, men with a tT/E 2 ratio < 3.30 (n = 52) had a more than 22-fold higher mortality risk when compared to men with a tT/ E 2 ratio > 7. 30 Table 4).
Since men over the age of 80 were less probably transferred to the ICU because of their age, we only included men below the age of 80 (n = 127) in the following analyses. Increased risk for ICU admission during hospital stay was found in men with lower baseline tT or fT levels as well as in those with a lower tT/E 2 ratio   and higher LH/tT ratio. These findings were independent of age, BMI, LDL and the time of symptom onset in multivariate Cox regression analysis (

DISCUSSION
In the present study we found that low serum total and free testosterone levels upon hospital admission are associated with disease severity indicated by lower oxygen saturation, higher WHO score and increased risk for ICU admission or death during the hospital stay in men with SARS-CoV-2 infection. These data are in line with recent published small cohort studies suggesting that low testosterone levels predict clinical adverse outcome (38,39). Our results strongly suggest that SARS-CoV-2 infected men with the need for hospitalization present with distinct low testosterone levels that are further not compensated by hypothalamic-hypophyseal feedbacks especially in younger men, since LH levels were within the normal range. Testosterone itself was shown to be associated with a reduced cytokine response following cellular immune activation in men but not in women (40). This might be due to already rather low testosterone levels in women who are therefore not further strongly affected by inflammation (41). In addition, androgen receptor expression in male immune cells is more distinctive than in female ones (18,42). Thus testosterone is suggested to be partly responsible for the lower prevalence of autoimmune disease but also for the higher incidence of cancer in men compared to women (43). Interestingly, it was also shown that androgens can reduce lung pathology in influenzainfected mice (44) suggesting that testosterone prevents inappropriate overwhelming immune activation following infection (45). In the case of low testosterone levels, such suppressive mechanism may be reduced, which might result in an imbalance between immunosuppressive and pro-inflammatory regulating mechanism in men (46). It's generally assumed that this can lead to enhanced pro-inflammatory cytokine response following a SARS-CoV-2 infection, also known as a cytokine storm. Therefore, a reduced innate immune response toward viral infections in men compared to women results in a higher and longer persisting viral load (47,48) with a subsequently more pronounced cellular immune response (49). Such situation is mainly present in men with low serum levels of the immunosuppressive acting testosterone. Accordingly, serum testosterone levels were negatively correlated with all investigated inflammatory biomarkers including IL-6, IL-10, neopterin (Th1 immune response), PCT, CRP, fibrinogen and ferritin which all were associated with a poorer clinical course and higher risk for ICU admission or death in SARS-CoV-2 infected patients (10,48,50,51). Vice versa, it was shown in previous studies, that the testosterone production in Leydig cells is also downregulated by immune activation itself (52). The findings of our study, that rather low than high testosterone levels predict mortality in men, contrast with recent findings showing that androgen stimulates the expression of the ACE2 coreceptor TMPRSS2 (53), which further provides access of SARS-CoV-2 to cells (54), suggesting that higher testosterone levels with consequently higher receptor density provides more docking sites for SARS-CoV-2. However, high mortality in COVID-19 is suggested to be caused by hyperinflammation and not actually by higher viral loads (55).
Thus, results of our study suggest that low levels of testosterone might be even more immunosuppressive than high testosterone levels with consequently higher SARS-CoV-2 receptor expression. However, other risk factors for severe COVID-19 disease also interact with hormonal status. Particularly, lower serum testosterone levels are associated with older age, higher BMI and lower cholesterol levels. Several studies have shown that viral and bacterial infections cause decreased cholesterol levels (56) and that the alterations in lipid levels correlate with the severity of the underlying infection (57,58). Since cholesterol is the precursor of testosterone, its deficiency also causes testosterone deficiency (59). Actually, a highly significant positive correlation of testosterone with lipoproteins was also found in our cohort. Interestingly, statin therapy, which was frequently encountered in our patients, was found to cause lower total testosterone levels in men with diabetes mellitus (60). However, testosterone levels did not significantly differ between men with or without lipid lowering therapy and also the predictive value of testosterone was independent of concomitant use of lipid lowering drugs.
Testosterone levels are typically decreased in older men (61) due to several factors, including a decline in ability of Leydig cells to produce adequate testosterone in response to LH stimulation (62), which might also partly contribute to the higher COVID-19 related morbidity and mortality within this vulnerable group. Actually, the prevalence of hypogonadism in geriatric hospitalized men over the age of 65 was found to be 53.3 % (63). This is supported by the finding that higher LH/tT ratio was also related to morbidity and mortality. LH levels were within the normal range suggesting the presence of normogonadotropic hypogonadism (64). The absence of compensatory LH secretion in case of low testosterone levels (65) most probably might be caused by altered GnRH and consecutive LH suppression by cytokines as found in patients with immune activation (66,67). Actually, men were hospitalized in median one week after symptom onset providing a long period for suppression of gonadotropins. The finding that older men over the age of 60 had again lower tT levels compared to young men with concurrent slightly elevated (but still low) LH levels suggests that in older men also the occurrence of late onset hypogonadism might contribute to this rather low tT levels. This is further supported by the positive correlation of the LH/tT ratio with age. Unfortunately, a differentiation of whether these men had age-related Leydig cell dysfunction before COVID-19 or whether low testosterone levels are primarily due to inflammation-related Leydig cell dysfunction or impaired testicular steroid-biosynthesis, cannot be provided by our results.
Testosterone expression can also be suppressed by cortisol which is produced in the adrenal cortex upon stimulation by the adrenocorticotropic hormone (ACTH) secreted by the pituitary gland upon stress-related corticotropin-releasing hormone (CRH) stimulation (68,69). Activation of the hypothalamic-pituitaryadrenal axis (HPA) in patients with COVID-19 reflected by elevated total serum cortisol levels was shown to be associated with an increased mortality (70) and might also contribute to low testosterone levels in these patients. However, testosterone was also shown to suppress CRH-stimulated cortisol production in men (71) which is why loss of this suppressive mechanism might promote stress-induced HPA activation. Unfortunately, we do not have detected serum hormonal levels of the hypothalamic-pituitaryadrenal axis. Interestingly, hydroxysteroid dehydrogenases, which catalyse steroid biosynthesis (e.g. DHEA to androstenedione, androstenedione to testosterone), is also suggested to be suppressed by immune activation (40).
On the other hand, obesity, linked to disease severity and outcome in SARS-CoV-2 infected patients, is considered to be stringently linked to testosterone deficiency (72). Testosterone can be aromatized to estradiol by the aromatase which is found in abundance in the visceral fatty tissue but also in male gonads (typically in Leydig cells (73), placenta, brain, muscle, bone and vascular tissues (74), and stimulated by cytokines such as IL-6 (75) or Tumor necrosis factor alpha (TNF-a) (76). This is supported by the highly significant negative correlation of inflammatory markers with the tT/E 2 ratio. Interestingly, estradiol levels were within the normal range thus representing a disturbed relation of testosterone to estradiol primarily caused by quite low testosterone levels rather than increased aromatase activity. Finally, the median BMI was normal. This would also explain the contrary findings in airway epithelial cells in which estradiol was demonstrated to downregulate ACE2 expression (77). Also, recent study results suggest that estrogen receptor signaling is actually protective in mice infected with SARS-CoV-2 (78) by suppressing ACE2 expression as well as pro-inflammatory pathways (79). Th1 immune activation might have stronger impacts on ACE2 expression than estradiol receptor signaling (80). Moreover, lower levels of cardiovascular-and lung-protective ACE2 might also aggravate existing co-morbidities (81). However, higher estradiol level (but within normal range) as well as a lower tT/E 2 ratio were associated with mortality and morbidity suggesting that this might primarily reflect inflammation-induced aromatase activity, which only slightly increase estradiol levels in men with distinctive testosterone deficiency.
Finally, low DHEA levels upon hospital admission were (similarly to testosterone) also related to higher morbidity and outcome. In vivo experiments showed that DHEA increases macrophage function and promotes a shift of Th1/Th2 balance toward Th1 immunity (82) thus enhancing immunity against viral infection (83). At the same time, DHEA suppresses the expression of various pro-inflammatory cytokines thus preventing overwhelming immune activation (83), suggesting that its deficiency enables overwhelming inflammation with reduced immunity.

Limitations
This was a retrospective explorative analysis of continuous COVID-19 patients hospitalized at our department. Additionally, sex hormones were not available of all men initially included in the study, which might be an unmeasured bias although baseline characteristics of all men and only those with available sex hormones were almost the same ( Table 1). This represents a risk for possible type I and II errors. We further have no information about the hormonal status of the patients before their COVID-19 infection which does not allow any conclusion whether these rather low testosterone levels represent hypogonadal hypogonadism or decreased following SARS-CoV-2 related immune activation.

Conclusion
Our results indicate that men hospitalized due to COVID-19 present with lower testosterone levels showing advanced immune activation and having high risk for an adverse clinical course and a poor prognosis. The origin of testosterone deficiency in these patients might be primarily caused by altered cholesterol biosynthesis in the case of SARS-CoV-2 infection as well as being due to inflammation-induced gonadotrophin suppression. The impact of enzymatic aromatase activation on hypogonadism pathogenesis might be negligible. The latter is supported by the finding that high estradiol levels (but within the normal range) were associated with more severe SARS-CoV-2 infections. Whether these men already had hypogonadotropic hypogonadism before COVID-19 cannot be clarified by our study. Effects of testosterone supplementation on outcome in men with severe SARS-CoV-2 infections should be evaluated in further studies.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by ethics committee of the Innsbruck Medical University. The patients/participants provided their written informed consent to participate in this study.