Causal Inference of CNS-regulated Hormones in COVID-19: A Bidirectional Two-sample Mendelian Randomization Study ================================================================================================================ * Yuxuan Sun * Ziyi Ding * Yawei Guo * Jinqiu Yuan * Chengming Zhu * Yihang Pan * Rui Sun ## Abstract **Objectives** We assessed the causal association of three COVID-19 phenotypes with insulin-like growth factor 1 (IGF-1), estrogen, testosterone, dehydroepiandrosterone (DHEA), thyroid-stimulating hormone (TSH), thyrotropin-releasing hormone (TRH), [luteinizing hormone (LH)](https://medlineplus.gov/lab-tests/luteinizing-hormone-lh-levels-test/), and follicle-stimulating hormone (FSH). **Methods** We used a bidirectional two-sample univariate and multivariable Mendelian randomization (MR) analysis to evaluate the direction, specificity, and causality of the association between CNS-regulated hormones and COVID-19 phenotypes. Genetic instruments for CNS-regulated hormones were selected from the largest publicly available genome-wide association studies in the European population. Summary-level data on COVID-19 severity, hospitalization, and susceptibility were obtained from the COVID-19 host genetic initiative. **Results** DHEA was associated with increased risks of very severe respiratory syndrome (OR=4.21, 95% CI: 1.41–12.59), consistent with the results in multivariate MR (OR=3.72, 95% CI: 1.20–11.51), and hospitalization (OR = 2.31, 95% CI: 1.13–4.72) in univariate MR. LH was associated with very severe respiratory syndrome (OR=0.83; 95% CI: 0.71–0.96) in univariate MR. Estrogen was negatively associated with very severe respiratory syndrome (OR=0.09, 95% CI: 0.02–0.51), hospitalization (OR=0.25, 95% CI: 0.08–0.78), and susceptibility (OR=0.50, 95% CI: 0.28–0.89) in multivariate MR. **Conclusions** We found strong evidence for the causal relationship of DHEA, LH, and estrogen with COVID-19 phenotypes. Key words * Mendelian randomization * CNS-regulated hormones * COVID-19 ## Introduction Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has infected more than 452 million individuals and resulted in more than 6 million deaths worldwide as of March 12, 2022 (1). SARS-CoV-2 infection has high variability in susceptibility, hospitalization, and severity, with clinical severity ranging from asymptomatic or mild upper respiratory illness to moderate and severe diseases such as respiratory failure, acute respiratory distress syndrome, and multi-organ failure with fatal outcomes (2). A report on COVID-19 suggests that 80% of infections are mild or asymptomatic, while 20% are moderate to severe (3). At the early infection stage, the innate immune system recognizes SARS-CoV-2 and induces proinflammatory cytokines to remove the viruses (4). However, SARS-CoV-2 disrupts the normal immune response in severely and critically ill patients, leading to an uncontrolled inflammatory response and cytokine storm, resulting in systemic and local damages (4, 5). Understanding the host factors influencing this infectious disease is crucial for elucidating the viral life cycle, identifying risk factors, drug development, and disease prevention. CNS-regulated hormones have received special attention in the context of COVID-19, considering their modulatory effects on cytokines. The gender disparity, with reports that men experience more severe syndromes and have higher mortality than women do, suggests that sex hormones such as estrogen and testosterone may play a role in COVID-19 morbidity and mortality (6), (7). Androgens may be associated with an increased risk of COVID-19 due to their modulatory effects on angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease 2 (TMPRSS2), the two key factors associated with the entrance of SARS-CoV-2 (8). Indeed, anti-androgen therapy in patients with prostate cancer has reportedly protected them against infection with SARS-CoV-2 (9). A recent study also suggested a causal effect of increased testosterone levels with a higher risk of COVID-19 hospitalization and severe disease (10). In contrast, a retrospective study found that older females who received estradiol therapy had a lower fatality risk than older females who did not receive the therapy, indicating a potential immune protective effect of estrogen in females (11). Meanwhile, in a study comparing critically and non-critically ill patients, low insulin growth-like factor-1 (IGF-1) levels showed a potential association with severe forms of COVID-19 (12). Despite the possible effect of CNS-regulated hormones on COVID-19, there appears to be a reverse directional effect of COVID-19 on hormones. A cohort study found that as the severity of COVID-19 increased, the concentration of dehydroepiandrosterone (DHEA) significantly increased (13). However, another observational study of hospitalized patients did not find any difference in DHEA levels between males and females with severe COVID-19 (14). Another study on thyroid function found that thyroid-stimulating hormone (TSH) levels in COVID-19 patients were obviously lower than those in non-COVID-19 patients, which suggested the possible effects of COVID-19 on TSH (15). Although observational studies provide some evidence for the conferral of hormonal effects on COVID-19, the liability of these findings is challenged by confounding factors, reverse causation, and other factors (16). Therefore, the causal impact of hormones on COVID-19 risk has not been ascertained. Mendelian randomization (MR) uses genetic variants following Mendel’s law of independent assortment as unconfounded proxies of an adjustable exposure to examine whether the exposure has causal effects on an outcome, overcoming the limitations of observational studies (17). In this study, we used two-sample MR univariate and multivariable analyses to examine potential causal associations between three types of COVID-19 (very severe respiratory syndrome, hospitalization, and susceptibility) and eight CNS-regulated hormones [IGF-1, estrogen, testosterone, DHEA, TSH, thyrotropin-releasing hormone (TRH), [luteinizing](https://medlineplus.gov/lab-tests/luteinizing-hormone-lh-levels-test/) hormone (LH), and follicle-stimulating hormone (FSH)]. We also performed a bidirectional analysis to detect the possible reverse causal relationship between CNS-regulated hormones and COVID-19. By assessing and establishing the causal role of CNS-regulated hormones, possible hormonal therapies could be applied as therapeutics for COVID-19 to reduce mortality and morbidity in COVID-19 infected patients. ## Materials and Methods ### CNS-regulated hormones Single-nucleotide polymorphisms (SNPs) associated with the levels of the eight selected hormones were selected based on summary-level data from either the Integrative Epidemiology Unit (IEU) open genome-wide association study (GWAS) project (18, 19) or the largest meta GWAS of thyroid-related traits (20). Strict European samples were selected as exposure sets, part of which were from the UK Biobank. The sample sizes of the exposure set varied from 468,343 to 1,000. Genetic instruments for TSH were extracted from a meta-analysis of thyroid hormones in 26,420 participants of European ancestry (21). Estradiol is the most potent form of estrogen in the human body; thus, it was used to represent the function of estrogens (22). In addition, DHEA-sulfate (DHEA-S) was adopted to represent DHEA because it circulates at a far higher concentration than DHEA in the blood (23). All GWAS summary data were adjusted for age, sex, and other study-specific covariates. Detailed information is provided in **Table S1**. ### COVID-19 severity, hospitalization, and susceptibility Summary statistics for the three COVID-19 phenotypes were obtained from release 5 of the COVID-19 Host Genetics Initiative (HGI) GWAS meta-analysis (24). The COVID-19 HGI is a consortium of scientists from over 54 countries working collaboratively to investigate human genetic variation related with very severe respiratory syndrome (cases: 5,582; controls: 709,010), hospitalization (cases: 17,992; controls: 1,810,493), and susceptibility (cases: 87,870; controls: 2,210,804) (**Table S2**). All individuals included in our analysis were of European ancestry. Very severe respiratory syndromes were cases that ended with respiratory support or death from COVID-19. Hospitalizations were cases that involved hospitalized patients. Susceptibility was defined as cases that tested positive, including laboratory-confirmed COVID-19 using reverse transcription-quantitative polymerase chain reaction (RT-qPCR) or serological testing, clinically diagnosed COVID-19 and self-reported COVID-19. In the GWAS meta-analysis, single-variant association analyses of COVID-19 traits were performed after adjusting for age, age 2, sex, age × sex, top principal components for ancestry, and study-specific covariates of each contributing cohort. All studies contributing data to the analyses were approved by the relevant ethics committees. ### Selection of Genetic Instruments for CNS-regulated hormones and COVID-19 traits Genetic instruments were selected from the exposure data with genome-wide significance (*p* < 5 × 10−8). SNPs were excluded if (1) they were not present in the outcome GWAS summary data, (2) SNPs in linkage disequilibrium were within 1-Mb genomic distance (R-squares above 0.01) using the 1000 Genome European reference panel and retained the SNP with the lowest *p*-value, and (3) minor allele frequency was greater than 0.42 for non-inferable palindromic alleles. For exposures with a relatively small number of SNPs, such as DHEA-S and TRH, if an SNP was absent in the outcome dataset, an LD proxy SNP was searched using the 1000 Genome European panel to replace the missing SNP (25). Exposure and outcome GWAS summary statistics were harmonized by aligning the effect alleles. The number of genetic instruments used in bidirectional MR analyses is shown in **Table S1**. The proportion of variance was calculated using the sum of the variance explained by individual SNPs divided by the variance of phenotype (26) (**Table S2**). ### Bidirectional Two-sample MR and Statistical Analysis We conducted bidirectional two-sample MR analyses. CNS-regulated hormones were considered exposures in the forward direction, and COVID-19 severity, hospitalization, and susceptibility were the outcomes. Conversely, COVID-19 severity, hospitalization, and susceptibility were exposures, and CNS-regulated hormones were outcomes in the reverse direction. The procedures for genetic instrument selection were similar in both directions of analysis. We applied three MR methods to estimate causal effects: inversed-variance weighted (IVW) regression/Wald ratio, weighted median (WM), and MR-Egger regression. IVW regression was the primary approach used to derive causal estimates, where at least two exposure SNPs were available for analysis. Under the assumption of no horizontal pleiotropy, the IVW approach could generate an unbiased causal estimate (27). The Wald ratio method was used when only one instrumental SNP was available for analysis. When the pleiotropy assumption was not satisfied, the WM method would have a lower bias but a higher type I error rate than the IVW method and was therefore used to further verify the inference effects (27). The MR-Egger test was used to assess the horizontal pleiotropy. The intercept of MR-Egger represents the average horizontal pleiotropic effect across the exposure genetic instruments, where the slope of the regression was the casual estimates (28). Multivariable Mendelian randomization (MVMR) analyses were conducted to account for potential confounding factors due to the possible genetic correlation of sex hormones, including estradiol, testosterone, DHEA-S, LH, and FSH (29). Multivariable lasso was used to perform MVMR analyses, as multivariable lasso is mostly a robust multivariate method, accounting for outliers and pleiotropy caused by invalid instruments (30). After combining SNPs of sex hormones, we clumped SNPs for linkage disequilibrium r2 < 0.01 within a 1-Mb genomic distance, excluded overlapping SNPs, and removed SNPs lacking linkage information. The remaining SNPs were used for MVMR analysis. We also conducted MVMR in the reverse direction to account for possible confounding due to the genetic correlation of three types of COVID-19 on CNS-regulated hormones, showing evidence of univariable MR associations with at least one COVID-19 trait. For traits with insufficient SNPs (N ≤ 3) to perform MVMR analyses, genetic instruments were selected with a suggestive genome-wide significant *p-*value threshold (*p* < 1 × 10−5). Sensitivity analyses were performed to verify the robustness of the results. Heterogeneity among the SNPs included in each analysis was examined using the Cochran’s Q test. We used the MR-Egger regression intercept test and MR pleiotropy residual sum and outlier (MR-PRESSO) global test to check for horizontal pleiotropy. The MR-PRESSO method was used to check and correct for outliers (31). *I**2**GX* statistic was used to examine the measurement error in the SNP-exposure association (32). The strength of the selected SNPs was assessed by F-statistic (33) calculated using the total sample size, the number of SNPs used, and the proportion of variance explained. The significance threshold was corrected for multiple testing with each MR analysis using the Bonferroni method (*p*-value < 0.05/number of exposures). In the sensitivity analysis, as the *p*-value evaluated the precision of the effect estimation and given the large number of statistical tests performed, we evaluated each result with a *p*-value < 0.05 individually by considering heterogeneity, pleiotropy, and removal of outliers. All data analyses were performed in R (34) using the *TwoSampleMR* (19), *MendelianRandomization* (35), and *MR-PRESSO* (31) packages. ## Results ### Forward MR Results: Impact of CNS-regulated hormones on COVID-19 #### Univariable MR Analysis using CNS-regulated hormones as exposures As shown in Figure 1, genetically predicted DHEA-S concentrations were nominally associated with very severe respiratory syndrome (OR = 4.21, 95% CI: 1.41 – 12.59, *p* = 0.010). This means that each standard deviation increase in DHEA-S level would dramatically increase the odds of having very severe respiratory syndrome by 4.21-fold. Additionally, DHEA-S was significantly associated with hospitalization (OR = 2.31, 95% CI: 1.13 – 4.72, *p* = 0.021). Along with the insignificant but positive association between DHEA-S and susceptibility (OR = 0.94, 95% CI: 0.67–1.31, p = 0.703), the results showed a possible trend between DHEA-S and COVID-19; DHEA-S had an increased OR and significance level for the three COVID-19 phenotypes (susceptibility, hospitalization, and very severe respiratory syndrome). There was evidence of a negative association between LH level and very severe respiratory syndrome (OR = 0.83, 95% CI: 0.71 – 0.96, *p* = 0.013). In the susceptibility analysis, we did not find significant causal effects of CNS-regulated hormones on COVID-19 phenotypes (**Table S3**). The F statistics of all instrument variables were > 10, with the proportion of variance explained ranging from 0.069% to 17.879%, indicating the absence of weakness in the selected instruments (**Table S2**). ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/12/07/2022.12.07.22283193/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2022/12/07/2022.12.07.22283193/F1) Figure 1. MR analyses the association of CNS-regulated Hormones and COVID-19 phenotypes. **Abbreviations:** CNS, central nervous system; COVID-19, coronavirus disease 2019; DHEA-S, dehydroepiandrosterone sulfate; FSH, follicle-stimulating hormone; IGF-1, insulin-like growth factor 1; LH, luteinizing hormone; MR, Mendelian randomization; TRH, thyrotropin-releasing hormone; TSH, thyroid-stimulating hormone ### Multivariable MR analysis with sex hormones After adjusting for other sex hormones, genetically predicted DHEA-S was still significantly associated with an increased risk of very severe respiratory syndrome at a nominal significance level (OR = 3.72, 95% CI: 1.20 – 11.51, *p* = 0.023) (**Table 1**). Although the association between DHEA-S and hospitalization and susceptibility became insignificant, the trend matched what we found in the univariable analysis; DHEA-S had an increasing effect size and significance level as COVID-19 severity increased. In addition, the effect estimates of LH on very severe respiratory syndrome were invalidated by conditioning the genetic effects of other sex hormones. However, we observed significant associations between estradiol and very severe respiratory syndrome (OR = 0.09, 95% CI: 0.02 – 0.51, *p* = 0.006), hospitalization (OR = 0.25, 95% CI: 0.08 – 0.78, *p* = 0.016), and susceptibility (OR = 0.50, 95% CI: 0.28 – 0.89, *p* = 0.019). The different results of LH and estradiol levels between univariate and multivariable analyses indicated that sex hormones interacted with each other during COVID-19. Notably, estradiol consistently increased the effect size and significance level as COVID-19 severity decreased, as in the univariate analysis, which further supported progressive interactions with COVID-19. Other than estradiol and DHEA-S, we did not find any significant associations between CNS-regulated hormones and COVID-19. Meanwhile, in the MVMR analysis, hormones other than testosterone also exhibited a similar increasing OR trend as DHEA-S in the univariate analysis, suggesting possible progressive immune interactive mechanisms during COVID-19. View this table: [Table 1.](http://medrxiv.org/content/early/2022/12/07/2022.12.07.22283193/T1) Table 1. MVMR analysis of CNS-regulated hormone SNPs and COVID-19 severity and susceptibility ### Reverse MR Results: Testing for bidirectionality using COVID-19 traits as exposures We did not find evidence supporting reverse causality between DHEA and all three COVID-19 phenotypes. However, there was evidence of the promoting effects of very severe respiratory syndrome on IGF-1 (OR = 1.01, 95% CI: 1.00 – 1.02, *p* = 0.040), and susceptibility on testosterone (OR = 1.05, 95% CI: 1.03 – 1.06, *p* < 0.001) after outlier correction (**Table S4**). In MVMR analysis, after adjusting for other COVID-19, very severe respiratory syndrome still had promoting effects on testosterone (OR = 1.11, 95% CI: 1.03 – 1.19, *p* = 0.003) and hospitalization was found to have significant decreasing effects on testosterone (OR = 0.89, 95% CI: 0.84 – 0.94, *p* < 0.001) (**Table S5**). ### Sensitivity analysis #### Test for horizontal pleiotropy In univariate analysis with the COVID-19 phenotype as the outcome, the MR-Egger regression intercept test did not find any horizontal pleiotropy in the association between eight CNS-regulated hormones and three COVID-19 phenotypes (all, *p* > 0.05) (**Table S6**). However, the MR-PRESSO global test suggested horizontal pleiotropy in the association of IGF-1 and testosterone with COVID-19 phenotypes (*p*-value < 0.05). In the bidirectional analysis using CNS-regulated hormones as outcomes, there was no evidence of horizontal pleiotropy in any associations between COVID-19 and hormone traits using the MR-Egger regression intercept test (all *p*-values > 0.05). The MR-PRESSO global test suggested potential horizontal pleiotropy effects associated with very severe respiratory syndrome COVID-19 and susceptibility to IGF-1 and the three COVID-19 phenotypes with testosterone (*p* < 0.05). Apart from IGF-1 and testosterone, no horizontal pleiotropy effect was found in either direction for any of the three COVID-19 phenotypes, which indicates the validity of the significant findings in the univariable MR analysis (**Table S7**). ### Test for heterogeneity of instruments As shown in **Table S6**, in the univariate analysis with COVID-19-severity as an outcome, heterogeneity was found in IGF-1 and testosterone estimates. Cochran’s Q test indicated heterogeneity in the association of IGF-1 with very severe respiratory syndrome (*p =* 0.001) and susceptibility (*p* = 0.007). Heterogeneity was only associated with very severe respiratory syndrome (*p* < 0.001). For the analysis using COVID-19 as exposure, heterogeneity was found in the association among very severe respiratory syndrome, susceptibility, and IGF-1 and in the association between very severe respiratory syndrome and testosterone (*p <* 0.001 for all analyses) (**Table S7**). ### Test for measurement error In the analysis using COVID-19 severity as an outcome, only TSH was found to have a low I2GX statistic associated with COVID-19, which indicated a potential bias to MR-Egger estimators (**Table S6**). In the analysis using very severe respiratory syndrome as exposure, only DHEA-S had an I2GX statistic > 0.900, which suggested that MR-Egger estimates of all other hormones could be biased by measurement errors (**Table S7)**. We did not find any measurement errors in the hospitalization analysis based on the I2GX statistics. Like very severe respiratory syndrome, almost all the associations between susceptibility and hormones were biased by measurement errors, except for DHEA-S. ## Discussion ### Main findings Bidirectional two-sample MR analyses using large publicly available genomic datasets to analyze the causal association between the genetic liability of CNS-regulated hormones and COVID-19 susceptibility, hospitalization, and severity. We found strong evidence supporting the causal associations of DHEA and LH with very severe respiratory syndrome COVID-19 and the association of DHEA with hospitalization due to COVID-19. We did not find evidence supporting an association between CNS-regulated hormones and COVID-19 susceptibility. In MVMR analyses, we found evidence of a negative association between estradiol and COVID-19 susceptibility, hospitalization, and severity, adjusted for other sex-related hormones. In the reverse direction, we found that very severe respiratory syndrome was associated with IGF-1 levels, and COVID-19 susceptibility was associated with testosterone levels. Together with previous observational studies, we revealed possible causal associations between CNS-regulated hormones, especially sex hormones, and COVID-19 severity, suggesting that hormonal therapies could be potential treatments for COVID-19 severe patients. ### Hormones and COVID-19 The effects of DHEA on COVID-19 are contradictory. On the one hand, DHEA can exert immunomodulatory and anti-inflammatory functions, which means that DHEA could have potential protective effects. Conversely, as a type of androgen, DHEA is suspected to be associated with an increased risk of COVID-19. Our analysis clearly indicated that DHEA is associated with an increased risk of COVID-19, especially the risk of very severe respiratory syndrome COVID-19. As mentioned before, DHEA, as a type of androgen, has regulatory effects on the ACE-2 receptor and TMPRSS2, promoting the fusion of the SARS-CoV-2 virus into host cells, thereby increasing the susceptibility and severity of COVID-19 (36, 37). Besides its androgenic effect, DHEA also acts as a powerful inhibitor of glucose-6-phosphate dehydrogenase (G6PD) (38). G6PD deficiency has been demonstrated to enhance cell infection with human coronavirus 229E (HCoV 229E) (39). Considering that SARS-CoV-2 is also a human coronavirus, DHEA may greatly increase the risk of COVID-19 due to its inhibition of G6PD. In contrast to androgens, estrogen is the main reason why the female immune system responds more efficiently to pathogens (40). In the analysis, when adjusting for all other sex hormones, estrogen was found to have significant decreasing effects on all types of COVID-19, which supports the possible protective effects of estrogen and indicates the potential for the implementation of hormonal therapy (41). In fact, clinical trials are in progress to test the effect of additional estrogen on COVID-19 patients (42). Moreover, the effects of DHEA and estrogen all displayed a progressively increasing effect size and significance level as the COVID-19 severity increased. Two possible mechanisms may play a role in this pattern. The first is the difference in viral load between severe and mild cases. A study of the viral load during COVID-19 reported that compared with mild cases, patients with severe COVID-19 tend to have higher viral loads (43). Considering the effects of DHEA in enhancing the fusion of viruses into host cells, DHEA might result in a higher risk for people with higher viral loads than those with lower viral loads. Another possible explanation is the mechanism of COVID-19. The key factors affecting severity are an uncontrolled immune response and cytokine storms. Estrogen has powerful anti-inflammatory effects due to its inhibition of proinflammatory cytokines such as IL-1β, IL-6, and others (44). As COVID-19 severity increases, uncontrolled immune responses produce more proinflammatory cytokines; therefore, the anti-inflammatory effect of estrogen becomes more powerful, which leads to increasing effect size and significance level. We did not find any association between testosterone and the three types of COVID-19. A recent MR study on serum testosterone and bioavailable testosterone found that only bioavailable testosterone level was associated with a higher risk of hospitalization due to COVID-19, indicating that different forms of testosterone may have different effects on COVID-19 (10). Besides, we found the promoting effect of COVID-19 susceptibility on testosterone levels. Some studies have observed decreased testosterone levels in severe cases. However, other studies have also proposed that the inflammatory response caused by SARS-CoV-2 viruses may trigger testosterone secretion to initiate an anti-viral reaction considering the anti-inflammatory and immune-modulatory effects of testosterone (45-47). Besides testosterone levels, we also observed that COVID-19 severity was associated with increased levels of IGF-1. However, a phenome-wide association study (48) found that genetically determined severe COVID-19 was associated with lower IGF-1 levels. The contradiction of the results suggests that further studies and an understanding of IGF-1’s mechanisms and pathways are needed. ### Additional findings Our findings also elucidate the possible interactive effects of sex hormones on COVID-19. We found significant effects of LH on very severe respiratory syndrome COVID-19. A recent COVID-19 case-control study involving 89 men with COVID-19 suggested that impaired testosterone in infected males may stimulate LH release to maintain testosterone levels based on negative-feedback regulation(46). Analysis using COVID-19 as exposure indicated that COVID-19 were associated with increased testosterone levels. Combined with the anti-inflammatory function of testosterone, increased LH levels could decrease the risk of severe COVID-19. Another possible reason could be the interaction between LH and estradiol. Univariate analysis did not find a possible causal association between estradiol and COVID-19. In contrast, after adjusting for other hormones in multivariable analysis, estradiol was found to be significantly negatively associated with all three types of COVID-19 with alterations in the effect direction and significance level of LH. LH is the upstream hormone of testosterone and estrogen, which controls the secretion and transformation of these hormones based on the negative-feedback loop. The true effect of LH may be influenced by both estrogen and testosterone levels. ### Strengths and limitations The strengths of our study include the integration of univariate, multivariate, and bidirectional MR analyses using genetic instrument variables drawn from the largest GWASs to date. We also included a variety of CNS-regulated hormone levels and COVID-19 phenotypes. Furthermore, follow-up sensitivity analyses such as the MR-Egger intercept test, Cochran’s Q test demonstrated the robustness of our findings. The F-statistic of the instrument variables also illustrated the strength of our selected SNPs. Finally, applying multivariable and univariable approaches is an advantage highlighted by the converging findings for COVID-19 phenotypes. We minimized the chances of population stratification between exposure datasets and outcome datasets by using GWAS of the European population for CNS-regulated hormones and by excluding the 23andMe cohort from COVID-19 datasets. Our study has the following limitations. The selected genetic instruments generally explain a small to moderate proportion of the variance in exposure, although well within the typical ranges for complex traits (49). Investigations of genetic variants associated with CNS-regulated hormones are needed. In addition, some exposures had only one or two SNPs, which were not suitable for sensitivity analyses. The use of weak genetic instruments may limit our ability to detect subtle, causal associations. However, the roughly similar pattern between univariate and multivariate analysis, and the high F-statistic, suggested the robustness and reliability of our results. Considering the possible sex effects, we used the UK Biobank sex-stratified data to assess the effects of testosterone, estrogen, IGF-1, and TSH on COVID-19. The results showed an effect direction consistent with that of the univariate two-sample MR analysis (**Table S8**). In conclusion, our study found a possible causal association between CNS-regulated hormones and COVID-19. In contrast to previous observational studies focusing on a single hormone, such as testosterone or estrogen, our study was the first to use MR analysis to investigate the causal association between CNS-regulated hormones and COVID-19 severity, hospitalization, and susceptibility. We found that genetically predicted DHEA and LH had significant causal associations with COVID-19; and that significant effects of estrogen on COVID-19 were observed after adjusting for other hormones, supporting the feasibility of hormonal therapy for COVID-19. Future studies may need to be designed to further understand the mechanisms and pathways underlying the causal relationship between CNS-regulated hormones and COVID-19, with the goal of developing potential therapeutic strategies. ## Data Availability GWAS summary data of seven hormone traits are available in IEU open GWAS project ([https://gwas.mrcieu.ac.uk/](https://gwas.mrcieu.ac.uk/)) with GWAS ID as ukb-d-30770\_irnt for IGF-1, ukb-d-30800_irnt for Estradiol, ukb-d-30850_irnt for Testosterone, met-a-478 for DHEA-S, prot-a-3102 for TRH, prot-a-529 for LH, and prot-c-3032_11_2 for FSH. GWAS summary data of TSH are collected in a GWAS meta-analysis of thyroid-related traits. Summary statistics (release 5) of GWAS meta-analysis for the three COVID-19 phenotypes are available in the COVID-19 Host Genetics Initiative (HGI) at [https://www.covid19hg.org/results/r5/](https://www.covid19hg.org/results/r5/). [https://gwas.mrcieu.ac.uk/](https://gwas.mrcieu.ac.uk/) ## Ethics approval and consent to participate The requirement for informed consent was waived because the study analyzed public databases. ## Availability of data and materials GWAS summary data of seven hormone traits are available in IEU open GWAS project ([https://gwas.mrcieu.ac.uk/](https://gwas.mrcieu.ac.uk/)) with GWAS ID as ukb-d-30770\_irnt for IGF-1, ukb-d-30800_irnt for Estradiol, ukb-d-30850_irnt for Testosterone, met-a-478 for DHEA-S, prot-a-3102 for TRH, prot-a-529 for LH, and prot-c-3032_11_2 for FSH. GWAS summary data of TSH are collected in a GWAS meta-analysis of thyroid-related traits.(20) Summary statistics (release 5) of GWAS meta-analysis for the three COVID-19 phenotypes are available in the COVID-19 Host Genetics Initiative (HGI) at [https://www.covid19hg.org/results/r5/](https://www.covid19hg.org/results/r5/). ## Declaration of interest All authors declare no conflicts of interest. ## Funding This study is supported by Research Start-up Fund of the Seventh Affiliated Hospital, Sun Yat-sen University (Grant No. 392016). ## Acknowledgements We thank the COVID-19 Host Genetics Initiative ([https://www.covid19hg.org/](https://www.covid19hg.org/)) for making their data publicly available. We thank the IEU open GWAS project ([https://gwas.mrcieu.ac.uk/](https://gwas.mrcieu.ac.uk/)) for making the summary statistics of 7 hormones traits publicly available. * Received December 7, 2022. * Revision received December 7, 2022. * Accepted December 7, 2022. * © 2022, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## References 1. 1.COVID W. Dashboard. Geneva: World Health Organization 2020. 2. 2.Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. Jama. 2020;323(13):1239–42. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2020.2648&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 3. 3.Team E. The epidemiological characteristics of an outbreak of 2019 novel coronavirus diseases (COVID-19)—China, 2020. China CDC weekly. 2020;2(8):113. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.46234/ccdcw2020.032&link_type=DOI) 4. 4.Yang L, Liu S, Liu J, Zhang Z, Wan X, Huang B, et al. COVID-19: immunopathogenesis and Immunotherapeutics. Signal Transduct Target Ther. 2020;5(1):128. 5. 5.Lopez Bernal J, Andrews N, Gower C, Gallagher E, Simmons R, Thelwall S, et al. Effectiveness of Covid-19 vaccines against the B. 1.617. 2 (Delta) variant. N Engl J Med. 2021:585–94. 6. 6.Jin JM, Bai P, He W, Wu F, Liu XF, Han DM, et al. Gender Differences in Patients With COVID-19: Focus on Severity and Mortality. Front Public Health. 2020;8:152. 7. 7.Papadopoulos V, Li L, Samplaski M. Why does COVID-19 kill more elderly men than women? Is there a role for testosterone? Andrology. 2021;9(1):65–72. 8. 8.Nyce J. Alert to US physicians: DHEA, widely used as an OTC androgen supplement, may exacerbate COVID-19. Endocrine-Related Cancer. 2021;28(2):R47–R53. 9. 9.Bhowmick NA, Oft J, Dorff T, Pal S, Agarwal N, Figlin RA, et al. COVID-19 and androgen-targeted therapy for prostate cancer patients. Endocr Relat Cancer. 2020;27(9):R281–R92. 10. 10.Liu L, Fan X, Guan Q, Yu C. Bioavailable testosterone level is associated with COVID-19 severity in female: A sex-stratified Mendelian randomization study. Journal of Infection. 2022. 11. 11.Seeland U, Coluzzi F, Simmaco M, Mura C, Bourne PE, Heiland M, et al. Evidence for treatment with estradiol for women with SARS-CoV-2 infection. BMC Med. 2020;18(1):369. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 12. 12.Ilias I, Diamantopoulos A, Botoula E, Athanasiou N, Zacharis A, Tsipilis S, et al. Covid-19 and Growth Hormone/Insulin-Like Growth Factor 1: Study in Critically and Non-Critically Ill Patients. Front Endocrinol (Lausanne). 2021;12:644055. 13. 13.Vaez Mahdavi MR, Kaboudanian Ardestani S, Rezaei A, Mohammadi S, Rajabnia Chenary M, Gharegozlou B, et al. COVID-19 patients suffer from DHEA-S sufficiency. Immunoregulation. 2020;3(2):135–44. 14. 14.Beltrame A, Salguero P, Rossi E, Conesa A, Moro L, Bettini LR, et al. Association Between Sex Hormone Levels and Clinical Outcomes in Patients With COVID-19 Admitted to Hospital: An Observational, Retrospective, Cohort Study. Frontiers in immunology. 2022;13. 15. 15.Wang W, Su X, Ding Y, Fan W, Zhou W, Su J, et al. Thyroid Function Abnormalities in COVID-19 Patients. Front Endocrinol (Lausanne). 2020;11:623792. 16. 16.Schuemie MJ, Ryan PB, Hripcsak G, Madigan D, Suchard MA. Improving reproducibility by using high-throughput observational studies with empirical calibration. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences. 2018;376(2128):20170356. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1098/rsta.2017.0356&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30082302&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 17. 17.Smith GD. Mendelian randomization for strengthening causal inference in observational studies: application to gene× environment interactions. Perspectives on Psychological Science. 2010;5(5):527–45. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1745691610383505&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26162196&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000287023100004&link_type=ISI) 18. 18.Elsworth B, Lyon M, Alexander T, Liu Y, Matthews P, Hallett J, et al. The MRC IEU OpenGWAS data infrastructure. BioRxiv. 2020. 19. 19.Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018;7:e34408. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.7554/eLife.34408&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29846171&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 20. 20.Porcu E, Medici M, Pistis G, Volpato CB, Wilson SG, Cappola AR, et al. A meta-analysis of thyroid-related traits reveals novel loci and gender-specific differences in the regulation of thyroid function. PLoS Genet. 2013;9(2):e1003266. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pgen.1003266&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23408906&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 21. 21.Bos MM, Smit RAJ, Trompet S, van Heemst D, Noordam R. Thyroid Signaling, Insulin Resistance, and 2 Diabetes Mellitus: A Mendelian Randomization Study. J Clin Endocrinol Metab. 2017;102(6):1960–70. 22. 22.Salole EG. Estradiol. Analytical profiles of drug substances. 15: Elsevier; 1986. p. 283–318. 23. 23.Prough RA, Clark BJ, Klinge CM. Novel mechanisms for DHEA action. J Mol Endocrinol. 2016;56(3):R139–55. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6Mzoiam1lIjtzOjU6InJlc2lkIjtzOjk6IjU2LzMvUjEzOSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDIyLzEyLzA3LzIwMjIuMTIuMDcuMjIyODMxOTMuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 24. 24.Initiative C-HG. The COVID-19 host genetics initiative, a global initiative to elucidate the role of host genetic factors in susceptibility and severity of the SARS-CoV-2 virus pandemic. European Journal of Human Genetics. 2020;28(6):715. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 25. 25.Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1086/519795&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17701901&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 26. 26.Shim H, Chasman DI, Smith JD, Mora S, Ridker PM, Nickerson DA, et al. A multivariate genome-wide association analysis of 10 LDL subfractions, and their response to statin treatment, in 1868 Caucasians. PLoS one. 2015;10(4):e0120758. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pone.0120758&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25898129&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 27. 27.Burgess S, Foley CN, Allara E, Staley JR, Howson JMM. A robust and efficient method for Mendelian randomization with hundreds of genetic variants. Nat Commun. 2020;11(1):376. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=31953392&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 28. 28.Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International journal of epidemiology. 2015;44(2):512–25. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ije/dyv080&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26050253&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 29. 29.Burgess S, Thompson SG. Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol. 2015;181(4):251–60. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/aje/kwu283&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25632051&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 30. 30.Grant AJ, Burgess S. Pleiotropy robust methods for multivariable Mendelian randomization. Stat Med. 2021;40(26):5813–30. 31. 31.Verbanck M, Chen C-Y, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nature genetics. 2018;50(5):693–8. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0099-7&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29686387&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 32. 32.Bowden J, Del Greco M F, Minelli C, Davey Smith G, Sheehan NA, Thompson JR. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. International journal of epidemiology. 2016;45(6):1961–74. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 33. 33.Pierce BL, Ahsan H, Vanderweele TJ. Power and instrument strength requirements for Mendelian randomization studies using multiple genetic variants. Int J Epidemiol. 2011;40(3):740–52. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ije/dyq151&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20813862&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000293618300025&link_type=ISI) 34. 34.Team RC. R: A language and environment for statistical computing. 2013. 35. 35.Yavorska OO, Burgess S. MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. International journal of epidemiology. 2017;46(6):1734–9. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ije/dyx034&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28398548&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 36. 36.Samuel RM, Majd H, Richter MN, Ghazizadeh Z, Zekavat SM, Navickas A, et al. Androgen signaling regulates SARS-CoV-2 receptor levels and is associated with severe COVID-19 symptoms in men. Cell Stem Cell. 2020;27(6):876-89. e12. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.stem.2020.11.009&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 37. 37.Clinckemalie L, Spans L, Dubois V, Laurent M, Helsen C, Joniau S, et al. Androgen regulation of the TMPRSS2 gene and the effect of a SNP in an androgen response element. Mol Endocrinol. 2013;27(12):2028–40. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1210/me.2013-1098&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24109594&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 38. 38.Gordon G, Mackow MC, Levy HR. On the mechanism of interaction of steroids with human glucose 6-phosphate dehydrogenase. Arch Biochem Biophys. 1995;318(1):25–9. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1006/abbi.1995.1199&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=7726568&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1995QU07700004&link_type=ISI) 39. 39.Vick DJ, editor Glucose-6-phosphate dehydrogenase deficiency and COVID-19 infection. Mayo Clinic Proceedings; 2020: Elsevier. 40. 40.Al-Lami RA, Urban RJ, Volpi E, Algburi AM, Baillargeon J, editors. Sex hormones and novel corona virus infectious disease (COVID-19). Mayo Clinic Proceedings; 2020: Elsevier. 41. 41.Bennink HJC, Foidart J-M, Debruyne FM. Treatment of serious COVID-19 with testosterone suppression and high-dose estrogen therapy. European Urology. 2021. 42. 42.University T. Estradiol and Progesterone in Hospitalized COVID-19 Patients. [https://ClinicalTrials.gov/show/NCT04865029](https://ClinicalTrials.gov/show/NCT04865029); 2021. 43. 43.Liu Y, Yan LM, Wan L, Xiang TX, L. A, Liu JM, et al. Viral dynamics in mild and severe cases of COVID-19. Lancet Infect Dis. 2020;20(6):656–7. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 44. 44.Al-Kuraishy HM, Al-Gareeb AI, Faidah H, Al-Maiahy TJ, Cruz-Martins N, Batiha GE-S. The looming effects of estrogen in Covid-19: a rocky rollout. Frontiers in Nutrition. 2021;8. 45. 45.Rastrelli G, Di Stasi V, Inglese F, Beccaria M, Garuti M, Di Costanzo D, et al. Low testosterone levels predict clinical adverse outcomes in SARS-CoV-2 pneumonia patients. Andrology. 2021;9(1):88–98. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/andr.12821&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) 46. 46.Kadihasanoglu M, Aktas S, Yardimci E, Aral H, Kadioglu A. SARS-CoV-2 pneumonia affects male reproductive hormone levels: a prospective, cohort study. The Journal of Sexual Medicine. 2021;18(2):256–64. 47. 47.Younis JS, Skorecki K, Abassi Z. The Double Edge Sword of Testosterone’s Role in the COVID-19 Pandemic. Front Endocrinol (Lausanne). 2021;12:607179. 48. 48.Papadopoulou A, Musa H, Sivaganesan M, McCoy D, Deloukas P, Marouli E. COVID-19 susceptibility variants associate with blood clots, thrombophlebitis and circulatory diseases. PLoS One. 2021;16(9):e0256988. 49. 49.Brion MJ, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42(5):1497–501. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ije/dyt179&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24159078&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F12%2F07%2F2022.12.07.22283193.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000326726000041&link_type=ISI)