Low Testosterone on Hospital Admission with COVID-19 Infection Is Associated with Increased Mortality

Abstract

Objective:

Males, despite equal sex-related susceptibility to COVID-19, appear at a greater risk of poor clinical outcomes and death. This suggests that serum testosterone could be a mediator. The aim of this retrospective study was to evaluate the association between serum total testosterone (TT), other prognostic indicators, and mortality in men with COVID-19.

Methods:

Of the 110 men consecutively admitted to Walsall Manor Hospital (with COVID-19 related symptoms) tested for SARS-CoV-2, 85 were positive and 27 of these men died. Serum TT was compared (rank-sum test) between men negative and positive for SARS-CoV-2, and this was followed by establishing factors associated with mortality in the latter group (rank-sum, logistic, Cox regression analyses). No patient was on testosterone therapy (TTh).

Results:

No significant difference (p = 0.12, rank-sum test) in serum TT between men positive [median TT (IQR) = 3.9 (1.9–7.22) nmol/L, 0 days (median) postadmission] and negative [median TT (IQR) = 5.9 (2.69–10.1) nmol/L, 2 days (median) postadmission] for SARS-CoV-2 was observed. Serum TT was lower (p = 0.0011, rank-sum test) in men with COVID-19 who died [median TT (IQR) = 2.0 (1.5–3.6) nmol/L] compared with survivors [median TT (IQR) = 5.0 (2.6–9.4) nmol/L]. Comorbidities obtained via medication history were not associated with mortality. Mortality (logistic regression) was associated with only age and serum TT (odds ratio: 0.77, 95% confidence intervals [CI]: 0.64–0.91). Survival (Cox regression) was inversely associated with serum TT (continuous variable, hazard ratio [HR]: 0.85) (95% CI: 0.74–0.98), stratified by median, TT ≥3.9 nmol/L (reference, TT <3.9 nmol/L), HR: 0.24 (95% CI: 0.089–0.63).

Conclusions:

Serum TT was inversely associated with mortality in men with COVID-19 and requires measurement at admission and while managing long COVID. Future research should establish whether low serum TT, possibly associated with negative acute phase response, contributes to poorer prognosis and a role for TTh.

Introduction

As the COVID-19 pandemic spreads across the globe, the current death toll (December 24, 2021) in the United Kingdom stands at more than 147,720 individuals.¹ Although overall infection rates are equal for men and women, males appear predisposed to severe infection and death.^2,3 The Sex, Gender and COVID 19 Project, a large-scale global statistical analysis, showed a similar pattern; although males and females are at a near equivalent risk of infection (sex ratio of male: female cases = 1.03:1), male (compared with female) sex is associated with the development of severe disease and deaths.⁴

Peckham et al. conducted a meta-analysis including more than 3 million individual and estimated males to be associated with increased intensive care unit (ICU) admissions (odds ratio [OR] = 2.84; 95% confidence intervals [CI] = 2.06–3.92; p = 1.86 × 10⁻¹⁰ < 0.0001) and mortality (OR = 1.39; 95% CI = 1.31–1.47; p < 0.0001).³

Interestingly, this gender variation was also recognized during the two previous significant outbreaks of coronavirus: the 2003 Severe Acute Respiratory Syndrome and the 2012 Middle East Respiratory Syndrome.^5,6 In particular, during the SARS-CoV-2 outbreak, male fatality rate was 21.9% as compared with 13.2% for females, with twice as many male-to-female deaths in the 0–44-year age range.⁵

As men who contract COVID-19 appear to have worse clinical outcomes compared with women, the possibility of androgen-dependent effects affecting pathogenesis must be considered. We have previously speculated that risk factors could vary in their impact on disease susceptibility and outcomes.⁷ Some studies have suggested that elevated androgens may lead to increased susceptibility to SARS-CoV-2 infection.^8,9 A possible mechanism is that the androgen receptor activation upregulates transcription of the transmembrane protease serine 2 (TMPRSS2).⁸

Evidence for the above association is provided by the observation of androgen deprivation therapies, decreasing TMPRSS2 levels (often overexpressed in prostate cancer) and offering partial protection against COVID-19.⁹ Montopoli et al. estimated SARs-CoV-2 infection in men with prostate cancer to be 76 per 100,000 men when on androgen deprivation therapies, as opposed to 307 per 100,000 men when untreated.⁹ Further, it has been postulated that androgens may be associated with COVID-19 severity as more men with androgenetic alopecia (considered a surrogate of greater testosterone expression) appeared to be hospitalized.¹⁰

Although neither of the studies mentioned earlier have demonstrated associations between testosterone levels and COVID-19 susceptibility, they appear to have led to a clinical trial evaluating the efficacy of anti-androgen therapy in men diagnosed with COVID-19.¹¹ However, the role of androgens as a susceptibility factor is possibly thrown into doubt by near-equal infection rates in males and females, as previously mentioned.⁴

Interestingly, a recent review by Traish and Morgentaler has shown COVID-19 susceptibility to be inconsistent, as there are more deaths among females in a number of countries.¹² Further, they showed that although TMPRSS2 and angiotensin converting enzyme 2 (ACE2) proteins were upregulated in the prostate this phenomenon was not evident in human and murine lung tissue.¹²

The role of testosterone as a mediator of clinical outcomes is difficult to study due to associations, possibly bidirectional, with various other factors involved in ill health. In a case-control study, Salonia et al. noted lower total testosterone (TT) and luteinizing hormone (LH) and higher estradiol (E2) levels in 286 men with COVID-19 after laboratory testing compared with 281 healthy males.¹³ However, it must be noted that age, ethnic mix, body mass index, comorbidities, and inflammatory markers also differed between the cohorts.

Further, TT levels were inversely associated with ICU admission and mortality, and the analysis was adjusted for clinical and laboratory parameters.¹³ Similar findings were obtained by Rastrelli et al., who studied 31 consecutive men admitted with COVID-19 to a respiratory ICU and showed that lower TT and calculated free testosterone levels were associated with a worse prognosis, including mortality.¹⁴ Dhindsa et al. carried out a prospective cohort study of 152 patients with COVID-19, including 90 men (66 with severe and 24 with milder disease) and measured TT, E2, and insulin-like growth factor-1 (IGF-1) at admission and 3-, 7-, 14-, and 28 days postadmission if the patient remained in hospital.¹⁵

Serum TT was significantly lower in the men with severe infection at baseline, 3 and 7 days (data were not available in the men with milder infection at 14 and 28 days) when compared with those with milder infection. Interestingly, the men with severe infection showed a decrease in serum TT at 3 and 7 days; this was followed by an increase at 14 and 28 days (median TT: admission = 53 ng/dL, 3 days = 19 ng/dL, 7 days = 20 ng/dL, 14 days = 53 ng/dL, 28 days = 102 ng/dL). Although the trend may suggest that the low TT is due to disease severity, more research is required to gain an understanding of the role of TT in men with COVID-19. It was seen that E2, and IGF-1 levels did not vary with disease severity.¹⁵

We wished to evaluate the association between TT, factors routinely measured in patients with COVID-19 at the point of admission, and mortality. Thus, we studied 110 consecutive admissions to Walsall Manor Hospital with symptoms associated with SARS-CoV-2 infection. After the publication by Rastrelli et al., a decision was taken by the Clinical Biochemistry laboratory to carry out TT measurements in the samples sent to the department soon after admission in men suspected of COVID-19.¹⁴ The aim of the study was to evaluate the association between TT and mortality, and whether TT should be included in the biochemical panel that could help predict mortality in men with COVID-19.

Materials and Methods

Walsall Healthcare NHS Trust provides acute hospital and general medical and community health services for the local population of Walsall (United Kingdom) and surrounding areas, circa 270,000. Secondary care services are provided by Walsall Manor Hospital, a university-affiliated hospital, which has 600–650 inpatient beds, taking referrals from more than 60 sites, including primary care health centers and General Practitioner surgeries.

Demographic, length of stay (LOS), and drug histories were obtained from the patient administration system and laboratory data (e.g., TT, LH, full blood count [FBC], C-reactive protein [CRP], and albumin) were collected from the Clinical Biochemistry Department database (Walsall Manor Hospital, Black Country Pathology Services) for 110 consecutive patients hospitalized with a suspected diagnosis of COVID-19, between October 2020 and March 2021. Serum TT was carried out on the first serum sample post-admission according to the protocol drawn up by the laboratory (either requested by the clinician or added by the laboratory staff according to protocol).

Comorbidity data (diabetes, ischemic heart disease, chronic obstructive airway disease, hypothyroidism, benign prostatic hypertrophy) were based on the medication(s) that the patients (details were available in 104 men) were on at admission and the electronic discharge summary. None of the men were on testosterone therapy (TTh). The English Indices of Deprivation combine factors of housing, social and economic issues to give a single deprivation score for small areas (known as Lower-Layer Super Output Areas) in England.¹⁶ An overall weighted aggregation index of multiple deprivation (IMD) is generated based on 37 separate indicators, organized across 7 distinct domains of deprivation, and each area is ranked from the least to the most deprived.¹⁶ Ethnicity data were available in 100 (88 white Caucasians, 12 Asians) of the 110 men.

Laboratory measurements

All 110 patients in this study were tested for SARS-CoV-2 infection between October 2020 and March 2021 from nasopharyngeal swabs received at the Department of Microbiology, Walsall Manor Hospital. Detection of SARS-CoV-2 was performed locally by reverse transcription-PCR (negative: 25 men, positive: 85 men) on a commercial assay detecting the ORF-1a/b and E-genes with a reported limit of detection of <300 copies/mL (Roche Cobas; Roche Diagnostics GmbH, Manheim, Germany).¹⁷

All biochemistry analytes were measured during the study period (including TT, LH, CRP, and albumin) via the Roche Cobas 8000 automated immunoassay analyzers (Roche Diagnostics). It was decided to include LH in the laboratory protocol midway through the study in view of the number of low serum TT levels observed in men with COVID-19, during clinical validation by the duty scientist. Recently determined analytical coefficients of variations (CVs) from internal quality controls at three serum concentrations were <3.5% for all biochemistry analytes (except for a TT level at 0.96 nmol/L = 6.9%).

The FBC and differential (including hemoglobin, lymphocytes, platelets, neutrophils) were performed on the Sysmex XN9000 FBC analyzers (Sysmex Corporation, Milton Keynes, United Kingdom) with CVs <5% across the parameters.

Statistical analysis

The TT distribution in the total cohort was skewed, hence nonparametric rank-sum tests were used to check for differences between the groups positive and negative for SARS-CoV-2. The main aim of the study was to determine the factors associated with mortality in men testing positive for SARS-CoV-2 by PCR. Thus, initially separate logistic regression was performed with each of the potential risk factors as independent variables and mortality as the dependent variable.

All the significant factors were then entered into a single model, and this was followed by a repeat analysis with all the significant factors (in the previous analyses) entered into a single analysis. Chi-square tests were conducted to see whether comorbidities derived from the medication history were associated with mortality. The associations mentioned earlier (logistic regression analyses/chi-square test) were reanalyzed via Cox regression survival analyses, with a Kaplan–Meier plot graphically illustrating any associations.

Cox regression was used when the outcome was a number (time to death in our study), whereas logistic regression was a study of factors predicting binary events (dead/alive in our study). Linear regression analysis was used to study the associations between TT and LH in the total group with confirmed COVID-19 and subgroups stratified by mortality.

Results

Table 1 shows that the data were obtained from 110 men admitted with suspected COVID-19, with 25 men testing negative (ICU admission: 4 men, death: 4 men) and 85 men (ICU admission: 18 men, death: 27 men) testing positive for SARS-CoV-2 by PCR respectively (checked at admission). No information was available on the use of high-flow oxygen, continuous positive airway pressure, and invasive mechanical ventilation.

Table 1.

Data in 110 Men Tested for COVID-19 with Serum Total Testosterone Measurements in the Total Group and When Stratified by the Results of the SARS- CoV-2 PCR Tests and Mortality (Men with COVID-19)

	SARS-CoV-2 PCR negative	SARS-CoV-2 PCR positive	SARS-CoV-2 PCR positive	SARS-CoV-2 PCR positive
		(Total group)	Alive	Dead
No. of men	25	85	58 (68.2%)	27 (31.8%)
Ethnicity (white caucasians/asian)	20 (95.2%)/1 (4.8%)	68 (86.1%)/11 (13.9%)	45 (84.9%)/8 (15.1%)	23 (88.5%)/3 (11.5%)
	Median (IQR) Values
Age (years)	68 (56–85)	75 (64–85)	72 (53–82)	77 (73–86)
TT (nmol/L)	5.9 (2.69–10.1)	3.9 (1.9–7.22)	5.0 (2.6–9.4)	2.0 (1.5–3.6)
LH (IU/L)	N/A	9.9 (6.0–14.1), n = 45	9.9 (4.6–16.8), n = 27	9.9 (6.2–12.7), n = 18
Hb (g/L)	129 (108–140)	125 (104–140)	130 (110–143)	122 (99–127)
Lymphocytes ( × 10⁹/L)	0.8 (0.7–1.4)	1.0 (0.7–1.3)	1.1 (0.7–1.3)	0.8 (0.6–1.2)
Neutrophils ( × 10⁹/L)	7.8 (6.5–12.8)	5.2 (3.5–8.4)	5.1 (3.4–8.1)	6.6 (4.2–11.3)
Platelets ( × 10⁹/L)	235 (172–348)	222 (165–310)	229 (169–329)	217 (142–270)
CRP (mg/L)	80 (14–218)	48 (19–108), n = 83	36.5 (13.5–81.5), n = 56	64 (26–122)
Albumin (g/L)	36.5 (32.5–41), n = 24	34 (31–39), n = 73	35 (32–40), n = 51	32.5 (28–38), n = 22
IMD	32.8 (27.3–40.9)	33.3 (16.2–46.5)	33.4 (16.2–42.2)	33.3 (15.4–48.6)
Time between COVID-19 and TT tests (days)	−2 (−5 to 0)	0 (−1 to 0)	0 (−1 to 0)	−1 (−3 to 0)
LOS (days)	14.2 (6.7–21.9)	17.4 (8.7–27.6), n = 84	16.8 (8.0–29.0)	18.7 (13.5–24.6)

Additional data (median and IQR presented in the above table) on serum TT level distribution.

SARS-CoV-2 PCR negative men: Mean: 6.7 nmol/L range: 0.26–15.4 nmol/L, 2 men (8%): serum TT <1.7 nmol/L.

SARS-CoV-2 PCR positive men (85 men): Mean: 5.4 nmol/L, range: 0.086–19.2 nmol/L, 15 men (17.7%): serum TT <1.7 nmol/L.

SARS-CoV-2 PCR positive men alive at final assessment (58 men): Mean: 6.4 nmol/L, range: 0.21–19.2 nmol/L, 7 men (12.1%): serum TT <1.7 nmol/L.

SARS-CoV-2 PCR positive men who had died (27 men): Mean: 3.2 nmol/L, range: 0.086–11.1 nmol/L, 8 men (29.6%): serum TT <1.7 nmol/L.

Comorbidity derived from the medication history and medications (lipid lowering and vitamin D supplements) data in 80 of the 85 men with positive SARS-CoV-2 PCR results are presented next, with chi-square tests used to study differences in the cohort stratified by survival.

Diabetes (survivors: 31.6%, nonsurvivors: 26.1%, p = 0.63), chronic obstructive airway disease/asthma (survivors: 14.0%, nonsurvivors: 26.1%, p = 0.20), hypertension (survivors: 56.1%, nonsurvivors: 78.3%, p = 0.63), ischemic heart disease (survivors: 3.5%, nonsurvivors: 13.0%, p = 0.11), benign prostatic hypertrophy (survivors: 7.0%, nonsurvivors: 8.7%, p = 0.80), lipid-lowering treatment (survivors: 42.1%, nonsurvivors: 65.2%, p = 0.061), and vitamin D supplements (survivors: 22.8%, nonsurvivors: 21.7%, p = 0.92).

Bold indicates statistically significant values.

CRP, C-reactive protein; IMD, index of multiple deprivation; LH, luteinizing hormone; LOS, length of stay.

No significant difference in serum TT or any of the other factors studied was observed between the men diagnosed as having COVID-19 via a positive SARS-CoV-2 PCR result and those with a negative result. The median serum TT was low in both groups (Roche adult male reference ranges: 8.64–29.0 nmol/L aged <50 years, 6.68–25.7 nmol/L aged ≥50 years), with men with COVID-19 having lower levels, although not statistically significant (p = 0.12, rank-sum test).

Additional data (mean, range, number of men with serum TT <castrate level of 1.7 nmol/L) on serum TT are found in the footnote of Table 1. Interestingly, in men with COVID-19 who died, all serum TT values were ≤11.1 nmol/L. Rank-sum tests also showed no significant difference between the SARS-CoV-2 PCR positive and -negative men regarding age (p = 0.46), Hb (p = 0.63), lymphocytes (p = 0.93), platelets (p = 0.39), CRP (p = 0.22), albumin (p = 0.13), and IMD (p = 0.88). Neutrophil (p = 0.0029) levels were significantly lower in men with a positive SARS-CoV-2 PCR result.

We now evaluated the possible predictors of mortality in men diagnosed with COVID-19. Of the 85 men with COVID-19, 58 men survived whereas the remaining 27 men had died (in hospital) before the data collection. Table 1 also shows the time period between the SARS-CoV-2 PCR and the laboratory tests, including TT. Importantly, the time gap between the SARS-CoV-2 PCR result and TT tests was not associated with serum TT (linear regression), p = 0.64 (total cohort), p = 0.98 (men with SARS-CoV-2 PCR negative), and p = 0.29 (men with SARS-CoV-2 PCR positive); hence, the time gap was not considered a confounding variable in further analyses.

Table 1 also shows data of the men with a SARS-CoV-2 PCR positive result stratified by survival. Age (p = 0.018), TT (p = 0.0011), Hb (p = 0.032), and CRP (p = 0.029) were significantly different between the groups, whereas no difference was observed in LH (p = 0.95), lymphocytes (p = 0.11), neutrophils (p = 0.12), platelets (p = 0.38), albumin (p = 0.16), and IMD (p = 0.91) using rank-sum tests. No significant association was apparent between TT and LH in men with a positive SARS-CoV-2 PCR result (total cohort [n = 45]: p = 0.054, survivors [n = 27]: p = 0.089, nonsurvivors [n = 18]: p = 0.089).

No difference was observed in LOS between the men who survived and those who did not (p = 0.47). Comorbidities data obtained from the medication history and medications and the electronic discharge summary (e.g., lipid lowering agents, vitamin D supplements, and thyroxine) are presented in the footnote of Table 1 and are not significantly different between the groups (survivors and nonsurvivors).

This was followed by separate logistic regression models, with mortality as the dependent variable and the factors mentioned earlier as independent variables in the 85 men with a positive SARS-CoV-2 PCR result (Table 2). Age, TT, Hb, and CRP were once again associated with mortality. A single logistic regression model with mortality as the dependent variable and all the earlier mentioned significant factors together with LOS as independent variables was constructed; only age and TT remained associated with mortality (Table 2).

Table 2.

Logistic Regression Analyses with Mortality as the Dependent Variable and Associated Factors in Men with SARS-CoV-2 PCR Positive Results

Logistic regression (dependent variable death)	OR (95% CI)	p
1. Separate models with the factors found next as independent variables
Age (years)	1.05 (1.01–1.09)	0.010
TT (nmol/L)	0.79 (0.67–0.94)	0.006
LOS (days)	0.48 (0.22–1.04)	0.064
Hb (g/L)	0.98 (0.96–1.00)	0.045
Lymphocytes ( × 10⁹/L)	0.73 (0.30–1.77)	0.49
Neutrophils ( × 10⁹/L)	1.08 (0.97–1.20)	0.15
Platelets ( × 10⁹/L)	0.99 (0.99–1.00)	0.43
CRP (mg/L)	1.01 (1.00–1.02)	0.033
Albumin (g/L)	0.94 (0.86–1.03)	0.16
IMD	1.00 (0.97–1.03)	0.89
2. Single model (all significant factors from previous model and LOS)
Age (years)	1.09 (1.03–1.15)	0.002
TT (nmol/L)	0.77 (0.64–0.94)	0.010
LOS (days)	0.97 (0.92–1.02)	0.22
Hb (g/L)	0.98 (0.95–1.01)	0.15
CRP (mg/L)	1.01 (1.00–1.02)	0.12
3. Single model with age and TT as dependent variables
Age (years)	1.06 (1.02–1.11)	0.005
TT (nmol/L)	0.77 (0.64–0.91)	0.003

Bold indicates statistically significant values.

CI, confidence interval; OR, odds ratio.

This was also the case when a similar logistic regression model was constructed, with only age and TT (as they remained associated with mortality in the previous multifactorial model) as the only independent variables (Table 2).

To graphically illustrate the difference between survival in the men stratified by admission TT levels, we plotted a Kaplan–Meier survival curve (Fig. 1) based on a Cox regression analysis (Table 3). Initially, age and TT were included as continuous independent variables in the Cox regression (Table 3—Model 1) and interestingly age was not associated with survival. When TT was stratified by the median TT of 3.9 nmol/L (men with COVID-19) and age (Table 3—Model 2), we found that TT <3.9 nmol/L (reference ≥3.9 nmol/L) was associated with lower mortality, whereas stratified age was not associated with mortality.

FIG. 1.

Kaplan–Meier plot of survival by time stratified by median TT at admission of 3.9 nmol/L. TT, total testosterone.

Table 3.

Cox Regression Survival Analysis with the Men with SARS-CoV-2 PCR Positive Results Stratified by Median Age (75 Years) and Median Serum Total Testosterone (3.9 nmol/L) and a Kaplan–Meier Plot Based on Model 3 of the Cox Regression

Cox survival analysis	HR (95% CI)	p
Model 1: Age and serum TT as continuous variables
Age (years)	1.02 (0.99–1.06)	0.230
TT (nmol/L)	0.85 (0.74–0.98)	0.021
Model 2: Age and serum TT stratified by median values
Age ≥75 years	1.18 (0.52–2.66)	0.690
Age <75 years	Reference
TT ≥3.9 nmol/L	0.24 (0.089–0.63)	0.004
TT <3.9 nmol/L	Reference
Model 3: Serum TT stratified by median values
TT ≥3.9 nmol/L	0.24 (0.089–0.63)	0.004
TT <3.9 nmol/L	Reference

Bold indicates statistically significant values.

HR, hazard ratio.

Table 3—Model 3 included only stratified TT as the independent variable; once again, TT <3.9 nmol/L (reference ≥3.9 nmol/L) was associated with lower mortality. The Kaplan–Meier survival curve (Fig. 1) was based on the Cox regression seen in Model 3 (Table 3) and very clearly demonstrates the poorer prognosis in men with serum TT at admission <3.9 nmol/L. Thus, all the analyses showed that lower TT levels were associated with mortality.

Discussion

The aim of this study was to evaluate the role of serum TT at admission as a predictor of mortality from COVID-19. Serum TT was inversely associated with mortality in all the analyses carried out; p = 0.0011 (rank-sum), p = 0.006 (unadjusted logistic regression), and p = 0.021 (unadjusted Cox regression). Hence, it could be considered a significant predictor of mortality in men diagnosed with COVID-19. Logistic regression showed that age and serum TT at admission were the only two factors that were associated with mortality (R² = 0.21).

Interestingly lymphocytes, neutrophils, CRP, albumin, and IMD scores did not predict mortality. Serum TT was the only survival factor (age was not significantly associated) on the Cox regression survival analysis, with a hazard ratio (HR) suggesting an inverse relationship (0.85). Thus, although age at the point of admission was predictive of mortality (logistic regression), it was not associated with time from admission to death (Cox regression). Interestingly, serum TT was inversely associated with death and time from admission to death.

Men with a serum TT <3.9 nmol/L (median) were at a significantly higher risk of mortality than their counterparts with a higher TT value. Unfortunately, we do not have serum TT levels before the diagnosis of COVID-19 to study an association between change in serum TT and mortality. This would have also provided information as to whether the low serum TT was related to a negative acute phase response.

In this study, serum TT was low in all the groups and this could be due to a negative acute phase reaction (secondary to infection). The lower values seen in the men who died could reflect severity of the underlying infection. Previously, in a case-control study, Salonia et al. showed that serum TT together with LH and other inflammatory markers were associated with disease severity and death.¹³ Other case series too have shown lower serum TT to be associated with adverse COVID-19 related outcomes.^14,15 Data from Dhindsa et al. showed low serum TT at days 3 and 7 postdiagnosis, but gradual recovery after 14 and 28 days.¹⁵ This pattern suggests that the reduction of serum TT may be related to COVID-19. Interestingly, no association between serum TT and LH levels was observed in our study.

ACE2, the key receptor-binding domain for SARS-CoV-2, is expressed in the testes (spermatogonia, Leydig and Sertoli cells), in addition to other organs such as lungs, heart, kidneys, intestines, and liver.¹⁸ Thus, the low serum TT that we have observed in men with COVID-19, with lower levels in the men who died could be related to direct damage of the testes by SARS-CoV-2 as well as due to a local inflammatory response.¹⁹ Further, inhibition of pituitary function could also be due to cytokines mediating inflammation.²⁰

Lanser et al. followed up 377 people (230 men, 147 women) admitted with acute Covid-19 infection and found that a TT on admission of <3.4 nmol/L was associated with increased CPK and IL-6, most significantly an 18-fold increase in mortality.²¹ Thus, the low serum testosterone levels associated with SARS-CoV-2 infection are possibly a combination of primary and secondary hypogonadism.

The low serum TT seen in men infected by SARS-CoV-2 and the poor prognosis associated with low serum TT makes it imperative that serum TT is regularly measured in patients with long COVID. It is interesting that there is some overlap of the symptoms (fatigue, postexertional malaise, and cognitive dysfunction) experienced by men suffering from long COVID and adult-onset TD.^22–26 Future research on long COVID should include a study of the associations between serum TT and various disease-related outcomes at various time points.

Further, it would be interesting to study a cohort who have been on TTh before being infected by SARS-CoV-2. The association between serum TT (together with free/bioavailable testosterone, LH, serum estradiol, and dihydrotestosterone) in these men receiving exogenous testosterone, disease susceptibility, and disease outcomes should be conducted at various time points while suffering from COVID-19 and long COVID. This could throw light on the role of serum TT and the possible use of TTh in men with COVID-19.

There are strengths and weaknesses to our study. The number of men were small, but it is comparable to other series published thus far. We would have liked to have serum TT before SARS-CoV-2 infection to relate baseline TT with mortality. We would have also wished for the serum TT measurements to be at the point of admission, as opposed to the first sample post-admission. Further, it would have been interesting to see whether men with low serum TT before COVID-19 had a worse prognosis.

Unfortunately, serum cortisol and sex hormone binding globulin (SHBG) measurements were not available; SHBG levels would have enabled us to have included calculated free/bioavailable testosterone as another independent variable. We would also have liked LH measurement in the entire cohort. Further, estimation of all the factors used as independent variables at various time points during the admission would have allowed us to see whether changes in these parameters during admission could be associated with clinical outcomes.

A larger cohort of COVID-19 negative men (only 25 men in our study compared with 85 COVID-19 men) would also have enabled us to determine with greater certainty (p = 0.12, rank-sum) whether serum TT in men with COVID-19 had lower hormone values. Further, although we found the serum TT cut-off of 3.9 nmol/L to be strongly predictive of mortality and may be of clinical use at this point as shown in Figure 1, validation from future research is required.

Conclusions

In a cohort of 85 consecutive men admitted to a secondary care hospital with a positive test for SARS-CoV-2, we established that serum TT levels soon after admission were a predictor of mortality in men with COVID-19, which adds new information. Our data add to the results of other studies, suggesting a possible role of serum TT in COVID-19 prognosis. Importantly, when serum TT was included in regression models, all other prognostic measurements (apart from age in the logistic regression) lost significance, suggesting that serum TT should be included as part of a prognostic panel in men admitted with a positive SARS-CoV-2 result.

We also suggest that serum TT be measured and associations with clinical outcome be studied in men with long COVID at various time points. A study of COVID-19 in men on TTh should also be carried out. All these may provide some information as to whether TTh could be considered for a clinical trial in men with low serum TT developing COVID-19 and long COVID.

Footnotes

Authors' Contributions

M.L.: design of study, data analysis, and preparation of article. S.R.: design of study, data analysis, and preparation of article. A.H.: design of study, data analysis, and preparation of article. A.P.: design of study, data analysis, and preparation of article. M.K.: design of study, data analysis, and preparation of article. G.H.: design of study, data analysis, and preparation of article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Abbreviations Used

References

Coronavirus Disease Statistics. Available at: https://coronavirus.data.gov.uk/details/deaths (accessed October 5, 2021).

Márquez

, Trowbridge

, Kuchel

, Banchereau

, Ucar

. The lethal sex gap: COVID-19. Immun Ageing. 2020; 17(1):13.

Peckham

, de Gruijter

, Raine

, et al. Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission. Nat Commun. 2020; 11(1):6317.

Global Health 5050. Sex, gender and COVID-19. Available at: https://globalhealth5050.org/covid19 (accessed October 5, 2021).

Karlberg

, Chong

DSY

, Lai

WYY

. Do men have a higher case fatality rate of severe acute respiratory syndrome than women do?. Am J Epidemiol. 2004; 159(3):229–231.

Ahmadzadeh

, Mobaraki

, Mousavi

, Aghazadeh-Attari

, Mirza-Aghazadeh-Attari

, Mohebbi

. The risk factors associated with MERS-CoV patient fatality: A global survey. Diagn Microbiol Infect Dis. 2020; 96(3):114876.

Ramachandran

, König

, Hackett

, Livingston

, Strange

. Managing clinical heterogeneity: An argument for benefit-based action limits. J Med Diagn Ther. 2018; 1:034701.

Hoffmann

, Kleine-Weber

, Schroeder

, et al. SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor. Cell. 2020; 181(2):271–280.

Montopoli

, Zumerle

, Vettor

, et al. Androgen-deprivation therapies for prostate cancer and risk of infection by SARS-CoV-2: A population-based study (N = 4532). Ann Oncol. 2020; 31(8):1040–1045.

10.

Goren

, Vaño-Galván

, Wambier

, et al. A preliminary observation: Male pattern hair loss among hospitalized COVID-19 patients in Spain—A potential clue to the role of androgens in COVID-19 severity. J Cosmet Dermatol. 2020; 19(7):1545–1547.

11.

US National Library of Medicine. ClinicalTrials.gov. Available at: https://clinicaltrials.gov/ct2/show/study/NCT04446429 (accessed October 5, 2021).

12.

Traish

, Morgentaler

. What's testosterone got to do with it? A critical assessment of the contribution of testosterone to gender disparities in COVID-19 infections and deaths. Androgens. 2021; 2(1):18–35.

13.

Salonia

, Pontillo

, Capogrosso

, et al. Severely low testosterone in males with COVID-19: A case-control study. Andrology. 2021; 9(4):1043–1052.

14.

Rastrelli

, Di Stasi

, Inglese

, et al. Low testosterone levels predict clinical adverse outcomes in SARS-CoV-2 pneumonia patients. Andrology. 2021; 9(1):88–98.

15.

Dhindsa

, Zhang

, McPhaul

, et al. Association of circulating sex hormones with inflammation and disease severity in patients with COVID-19. JAMA Netw Open. 2021; 4(5):e2111398.

16.

UK Data Service. GeoConvert Census Support. Available at: https://geoconvert.mimas.ac.uk/application/step1credentials.cfm (accessed October 5, 2021).

17.

Pfefferle

, Reucher

, Nörz

, Lütgehetmann

. Evaluation of a quantitative RT-PCR assay for the detection of the emerging coronavirus SARS-CoV-2 using a high throughput system. Euro Surveill. 2020; 25(9):2000152.

18.

Wenham

, Smith

, Morgan

. COVID-19: The gendered impacts of the outbreak. Lancet. 2020; 395(10227):846–8.

19.

Selvaraj

, Ravichandran

, Krishnan

, Radhakrishnan

, Manickam

, Kandasamy

. Testicular atrophy and hypothalamic pathology in COVID-19: Possibility of the incidence of male infertility and HPG axis abnormalities. Reprod Sci. 2021; 28(10):2735–2742.

20.

Pal

. COVID-19, hypothalamo-pituitary-adrenal axis and clinical implications. Endocrine. 2020; 68(2):251–252.

21.

Lanser

, Burkert

, Thommes

, et al. Testosterone deficiency is a risk factor for severe COVID-19. Front Endocrinol (Lausanne). 2021; 12:694083.

22.

Araujo

, Esche

, Kupelian

, et al. Prevalence of symptomatic androgen deficiency in men. J Clin Endocrinol Metab. 2007; 92:4241–4247.

23.

Hackett

, Kirby

, Edwards

, et al. British Society for sexual medicine guidelines on adult testosterone deficiency, with statements for UK Practice. J Sex Med. 2017; 14:1504–1523.

24.

Nalbandian

, Sehgal

, Gupta

, et al. Post-acute COVID-19 syndrome. Nat Med. 2021; 27(4):601–615.

25.

Sansone

, Mollaioli

, Limoncin

, et al. The Sexual Long COVID (SLC): Erectile dysfunction as a biomarker of systemic complications for COVID-19 long haulers. Sex Med Rev. 2021:S2050–0521(21)00085-8. [Epub ahead of print].

26.

Sansone

, Mollaioli

, Ciocca

, et al. Addressing male sexual and reproductive health in the wake of COVID-19 outbreak. J Endocrinol Invest. 2021; 44(2):223–231.