EBioMedicine 65 (2021) 103246 Contents lists available at ScienceDirect EBioMedicine journal homepage: www.elsevier.com/locate/ebiom Research paper Shorter androgen receptor polyQ alleles protect against life-threatening COVID-19 disease in European males Margherita Baldassarria,b,1, Nicola Picchiottic,d,1, Francesca Favaa,b,e, Chiara Fallerinia,b, Elisa Benettib, Sergio Dagaa,b, Floriana Valentinoa,b, Gabriella Doddatoa,b, Simone Furinib, Annarita Gilibertia,b, Rossella Titae, Sara Amitranoe, Mirella Bruttinia,b,e, Susanna Crocia,b, Ilaria Melonia,b, Anna Maria Pintoe, Nicola Iusoa,b, Chiara Gabbif, Francesca Sciarrag, Mary Anna Vennerig, Marco Goric,h, Maurizio Sanaricoi, Francis P. Crawleyj, Uberto Pagottok, Flaminia Fanellik, Marco Mezzullok, Elena Dominguez-Garridol, Laura Planas-Serram,n,o, Agatha Schlu€ term,n,o, Roger Colobranp, Pere Soler-Palacinq, Pablo Lapunzinan,r, Jair Tenorion,r, Aurora Pujolm,n,s, Maria Grazia Castagnat, Marco Marcelliu, Andrea M. Isidorig, Alessandra Renieria,b,e,*, Elisa Frullantia,b,2, Francesca Maria,b,e,2, Spanish Covid HGE, GEN- COVID Multicenter Study a Medical Genetics, University of Siena, Italy b Med Biotech Hub and Competence Center, Department of Medical Biotechnologies, University of Siena, Italy c University of Siena, DIISM-SAILAB, Siena, Italy d Department of Mathematics, University of Pavia, Pavia, Italy e Genetica Medica, Azienda Ospedaliero-Universitaria Senese, Italy f Independent Medical Scientist, Milan, Italy g Department of Experimental Medicine, Sapienza University of Rome, Rome, Italy h Universite Co ^te d’Azur, Inria, CNRS, I3S, Maasai i Independent Data Scientist, Milan, Italy j Good Clinical Practice Alliance-Europe (GCPA) and Strategic Initiative for Developing Capacity in Ethical Review-Europe (SIDCER), Leuven, Belgium k Unit of Endocrinology and Prevention and Care of Diabetes, Center for Applied, Biomedical Research, Department of Medical and Surgical Sciences, University of Bologna, S. Orsola-Malpighi Hospital, Bologna, Italy l Molecular Diagnostic Unit, Fundacio n Rioja Salud, Logron ~ o, La Rioja, Spain m Neurometabolic Diseases Laboratory, Bellvitge Biomedical Research Institute (IDIBELL), Barcelona, Spain n CIBERER, Centro de Investigacio ndez Almagro, 3, 28029 Madrid, Spain n Biomedica en Red de Enfermedades Raras, ISCIII, Melchor Ferna o Spanish Covid HGE p Immunology Division, Genetics Department, Hospital Universitari Vall d'Hebron, Vall d'Hebron Research Institute, Vall d'Hebron Barcelona Hospital Campus, Uni- versitat Auto noma de Barcelona (UAB), Barcelona, Catalonia, Spain, EU q Pediatric Infectious Diseases and Immunodeficiencies Unit, Hospital Universitari Vall d'Hebron, Vall d'Hebron Research Institute, Barcelona, Catalonia, Spain r Institute of Medical and Molecular Genetics (INGEMM)-IdiPAZ, Hospital Universitario La Paz-UAM Paseo de La Castellana, 261, 28046 Madrid, Spain s Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain t Department of Medical, Surgical and Neurological Sciences, University of Siena, Italy u Department of Medicine, Baylor College of Medicine, Houston TX, USA A R T I C L E I N F O A B S T R A C T Article History: Background: While SARS-CoV-2 similarly infects men and women, COVID-19 outcome is less favorable in Received 1 December 2020 men. Variability in COVID-19 severity may be explained by differences in the host genome. Revised 24 January 2021 Methods: We compared poly-amino acids variability from WES data in severely affected COVID-19 patients Accepted 2 February 2021 versus SARS-CoV-2 PCR-positive oligo-asymptomatic subjects. Available online xxx Findings: Shorter polyQ alleles (22) in the androgen receptor (AR) conferred protection against severe out- come in COVID-19 in the first tested cohort (both males and females) of 638 Italian subjects. The association Keywords: between long polyQ alleles (23) and severe clinical outcome (p = 0.024) was also validated in an indepen- Androgen receptor gene dent cohort of Spanish men <60 years of age (p = 0.014). Testosterone was higher in subjects with AR long- * Corresponding author. E-mail address: alessandra.renieri@unisi.it (A. Renieri). 1 Co-first authors 2 Co-last authors https://doi.org/10.1016/j.ebiom.2021.103246 2352-3964/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 Testosterone polyQ, possibly indicating receptor resistance (p = 0.042 Mann-Whitney U test). Inappropriately low serum COVID-19 testosterone level among carriers of the long-polyQ alleles (p = 0.0004 Mann-Whitney U test) predicted the LASSO logistic regression need for intensive care in COVID-19 infected men. In agreement with the known anti-inflammatory action of WES testosterone, patients with long-polyQ and age 60 years had increased levels of CRP (p = 0.018, not account- Viral infection and host genome ing for multiple testing). Interpretation: We identify the first genetic polymorphism that appears to predispose some men to develop more severe disease. Failure of the endocrine feedback to overcome AR signaling defects by increasing testos- terone levels during the infection leads to the polyQ tract becoming dominant to serum testosterone levels for the clinical outcome. These results may contribute to designing reliable clinical and public health meas- ures and provide a rationale to test testosterone as adjuvant therapy in men with COVID-19 expressing long AR polyQ repeats. Funding: MIUR project “Dipartimenti di Eccellenza 2018-2020” to Department of Medical Biotechnologies University of Siena, Italy (Italian D.L. n.18 March 17, 2020) and “Bando Ricerca COVID-19 Toscana” project to Azienda Ospedaliero-Universitaria Senese. Private donors for COVID-19 research and charity funds from Intesa San Paolo. © 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 1. Introduction Research in context Alongside the mode of transmission, viral load, comorbidities, and Evidence before this study demographic factors (such as age and sex), the host genetic back- We searched on Medline, EMBASE, and Pubmed for articles ground appears to play an important role in COVID-19 severity and published from January 2020 to August 2020 using various progression [1 8]. We hypothesized that common polymorphisms combinations of the search terms “sex-difference”, “gender” may contribute to COVID-19 severity, including poly-amino acids AND SARS-Cov-2, or COVID. Epidemiological studies indicate repeat polymorphisms, such as the polyQ tract of the Androgen that men and women are similarly infected by COVID-19, but Receptor (AR). AR contains in its N-terminus domain a polymorphic the outcome is less favorable in men, independently of age. polyQ tract, ranging between 9 and 36 repeated CAG units in the nor- Several studies also showed that patients with hypogonadism mal population [9]. In vitro and in vivo studies have demonstrated tend to be more severely affected. A prompt intervention that the transactivation potential of AR is inversely correlated to directed toward the most fragile subjects with SARS-Cov-2 repeat length, and Q-tract size can significantly influence androgen- infection is currently the only strategy to reduce mortality. Glu- dependent physiological functions [9 12]. cocorticoid treatment is a cost-effective measure to improve Several lines of evidence lead to the concept that androgens are the outcome of severe cases. Clinical algorithms have been pro- relevant to both SARS-CoV-2 infection and COVID-19 disease presen- posed, but little is known on the ability of genetic profiling to tation; however, they seem to have a Janus bifacial way of action predict outcome and disclose novel therapeutic strategies. [13,14]. On one side, androgens promote the transcription of the TMPRSS2 gene that encodes a serine protease known to prime the Added-value of this study spike (S) protein of coronaviruses, facilitating viral entry into the cells [15]. On the other hand, hypogonadism is known to correlate with In a cohort of 1178 men and women with COVID-19, we used a severe COVID-19 [16] and other chronic conditions, partly due to the supervised Machine Learning approach on a synthetic repre- loss of attenuation of the inflammatory immune response exerted by sentation of genetic variability due to poly-amino acid repeats. testosterone (T) [17 19]. Comparing the genotype of patients with extreme manifesta- tions (severe vs. asymptomatic), we found an association 2. Methods between the poly-glutamine repeat number of the androgen receptor (AR) gene, serum testosterone concentrations, and 2.1. Patients COVID-19 outcome in male patients. Failure of the endocrine feedback to overcome AR signaling defects by increasing testos- We performed a nested case-control study (NCC). Cases and con- terone levels during the infection leads to the fact that polyQ  trols were drawn from the Italian GEN-COVID cohort of 1178 subjects 23 becomes dominant to testosterone levels for the clinical infected with SARS-CoV-2 diagnosed by RT-PCR on nasopharyngeal outcome. swab [2]. Demographic characteristics of patients enrolled in the cohort are summarized in Table 1 according to their clinical status. In Implications of all the available evidence the current NCC study, cases were selected according to the following We identify the first genetic polymorphism predisposing some inclusion criteria: i. CPAP/biPAP ventilation (230 subjects); ii. endo- men to develop a more severe disease irrespectively of age. tracheal intubation (108 subjects). As controls, 300 subjects were Based on this, we suggest that sizing the AR poly-glutamine selected using the sole criterion of not requiring hospitalization. repeat has important implications in the diagnostic pipeline of Exclusion criteria for both cases and controls were i. SARS-CoV-2 patients affected by life-threatening COVID-19 infection. Most infection not confirmed by PCR; ii. non-caucasian ethnicity. Demo- importantly, our studies open to the potential of using testos- graphic characteristics of the subjects in the NCC study are summa- terone as adjuvant therapy for patients with severe COVID-19 rized in Table 1. A similar Spanish cohort, composed of male COVID- having defective androgen signaling, defined by this study as 19 patients (117 cases and 41 controls) was used to validate the 23 PolyQ repeats, and inappropriately low levels of circulating results in another representative European population highly androgens. impacted by COVID-19. All subjects were white European. The Span- ish Covid HGE cohort is under IRB approval PR127/20 from Bellvitge University Hospital, Barcelona, Spain. M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 3 Table 1 Demographics characteristics of the Italian GEN-COVID Cohort and NCC study. Intubation CPAP/BiPAP Oxygen Therapy Hospitalized w/o Oligo^-asymptomatics w/o Ventilation respiratory support hospitalization GEN-COVID Number of Sybjects 108 230 352 188 300 Male/Female 80/28 157/73 208/144 104/84 116/184 Age males (years) 61,52§11,43 62,75§13,48 63,41§14,53 55,99§15,44 47,40§13,23 Age females (years) 63,71§13,96 66,23§15,25 68,40§14,74 52,88§16,39 48,61§11,06 Cases Controls NCC study Number of Subjects 338 300 Male/Female 237/101 116/184 Age males (years) 62,34§12,84 47,40§13,23 Age females (years) 65,53§14,94 48,61§11,06 ^ Oligosymtpomatic: individuals with minor symptoms of COVID-19 (mild fever, cough, sore throat, etc.) 2.2. Ethics 2.6. Total T measurement The GEN-COVID study was approved by the University Hospital of Blood samples were collected after an overnight fast, immediately Siena Ethics Review Board (Protocol n. 16917, dated March 16, 2020). centrifuged at 4 °C and stored at -20 °C until assayed. Serum and This observational study has been inserted in www.clinicaltrial.org plasma total T (TT), SHBG levels in plasma and serum LH were mea- (NCT04549831). The Spanish Covid HGE cohort is under IRB approval sured following standard procedures. PR127/20 from Bellvitge University Hospital, Barcelona Spain. Writ- Serum TT was measured using the Access testosterone assay ten informed consent was obtained from all individuals who contrib- (Beckman Coulter Inc., Fullerton, CA, USA) with a minimum detection uted samples and data. limit of 0.35 nmol/L. Reference range for this assay was 6.07 27.1 nmol/L and liquid chromatography - tandem mass spec- 2.3. Analysis of triplets size in the AR locus trometry (LC-MS/MS) according to a previously validated method provided with reference values between 9.8 28.4 nmol/L [22]. To establish allele sizes of the polymorphic triplet in the AR locus, Thawed plasma underwent 15 min incubation at 56 °C for virus inac- we used the HUMARA assay with minor modifications [20]. Specifi- tivation, and TT measured in 100 ml of plasma, with sensitivity limit cally, we performed a fluorescent PCR followed by capillary electro- being 0.270 nmol/L, imprecision ranging 9.8 to 0.7% and accuracy phoresis on an ABI3130 sequencer. Allele size was established using 90.6 to 101.5% at concentration levels between 1.12 and 39.2 nmol/L. the Genescan Analysis software. A stability test under viral inactivation conditions was performed in 6 samples, revealing a T mean (min-max) % loss of 9.7% (4.6 16.7%). SHBG levels were measured in plasma samples using Quantikine 2.4. Binary representation of WES data ELISA Kit (DSHB G0B, R&D Systems, Minneapolis, MN, USA) according to the manufacturers' instructions. Serum LH was measured using Variants calling was performed according to the GATK4 best prac- “Access LH assay“ a chemiluminescenSert, two-step enzyme immu- tice guidelines, using BWA for mapping, and ANNOVAR for annotat- noassay (Beckman Coulter Inc., Fullerton, CA, USA). Sensitivity for the ing. WES data were represented in a binary mode on a gene-by-gene LH determination is 0.2 mIU/mL. Reference range in adult males for basis. Poly-amino acids triplet repeats were represented in a binary this assay is 1.2 8.6 mIU/mL. mode: long and short repeats in respect to the reference sequence on the genome. A total of 40 genes with 43 triplet repeat regions were 2.7. Statistical analysis taken from UniProtKB (Supplementary Table S1). In the boolean representation of poly-amino acids triplet repeats, for each of these Since serum and plasma T values were not normally distributed, 40 genes two features were defined, Dij and Iij, with Dij being equal the statistical analyses were performed using non-parametric tests. to 1 if gene i in sample j has a repeated region shorter than the refer- When appropriate, transformation was used for skewed data in ence, 0 otherwise, and Iij being equal to 1 if gene i in sample j has a regression models. We used the Mann-Whitney U test to compare T repeated region longer than the reference, 0 otherwise. levels in males with AR long-polyQ (23) versus males with short- polyQ repeat (22). Logistic regression analysis was performed to 2.5. LASSO logistic regression test the contribution of age, T, and the number of polyglutamine rep- etitions on COVID-19 outcome. The only prespecified interaction We adopted the LASSO logistic regression that provides a feature tested was the T by polyQ (categorical). Box-Tidwell procedure was selection method within the classification tasks able to enforce both used to assess linearity and the Hosmer and Lemeshow to assess the sparsity and the interpretability of the results. The weights of the goodness of fit test. Multicollinearity was assessed by variance infla- logistic regression algorithm can be interpreted as the importance of tion factor, and dealt with by dropping the offending variables from the subset of the most relevant features for the task [21]. the analysis on the basis of clinical grounds. The input features of the LASSO logistic regression are the poly-amino acids triplet repeats as well as gender, comorbidity (1 if there is at least 2.8. Role of funders one comorbidity) and age, the latter as a continuous variable normalized between 0 and 1. Comorbidities were defined as the presence of one or The work was financially supported by MIUR project “Diparti- more clinical conditions (i.e. cardiac, endocrine, neurological, neoplastic menti di Eccellenza 2018-2020” to Department of Medical Biotech- diseases) at the time of infection. During the fitting procedure, the class nologies University of Siena, Italy (Italian D.L. n.18 March 17, 2020) slight unbalancing is tackled by penalizing the misclassification of the and by “Bando Ricerca COVID-19 Toscana” project to Azienda Ospe- minority class with a multiplicative factor inversely proportional to the daliero-Universitaria Senese. It was also funded by private donors for class frequencies. The data pre-processing was coded in Python, whereas COVID-19 research and charity funds from Intesa San Paolo “Fondo di for the logistic regression model we used the scikit-learn module with Beneficenza n. b/2020/0119”. The sponsors of the study had no role the liblinear coordinate descent optimization algorithm. in study design, data collection, data analysis, data interpretation, or 4 M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 writing of the manuscript. The authors collected the data, and had full we selected those with amino acid repeats, namely 40 genes, and access to all of the data in the study. They also had the final decision represented them as a boolean variable. Logistic regression with and responsibility to submit the study results for publication. LASSO regularization analysis identified AR as the only protective gene (Fig. 1, panel a). The 10-fold cross-validation provides good per- 3. Results formances in terms of accuracy (77%), precision (81%), sensitivity (77%), specificity (78%) and Area Under the Curve (AUC) score (86%) 3.1. Testing the role of common poly-amino acid repeat polymorphisms (Fig. 1, panel b). The performances of the logistic regression without in COVID-19 outcome LASSO regularization for the selected set of features (age, gender, comorbidity and AR gene) are 79% accuracy, 81% precision, 81% sensi- In order to test the role of common poly-amino acid repeat poly- tivity, 78% specificity, 88% roc-auc. The model shows a slight decrease morphisms in determining COVID-19 clinical severity, we performed of almost all the performance measures when the AR gene is a NCC, selecting the extreme phenotypic ends of our entire GEN- removed from the set (accuracy -1.2%, precision -1.3%, sensitivity COVID cohort (Table 1 and Fig. 1). Among 18,439 annotated genes, -1.4%, specificity -1.2%, roc-auc +0.3%). Finally, the logistic regression Fig. 1. LASSO logistic regression. The bar of the LASSO logistic regression beta coefficients represents the importance of each feature for the classification task (Fig. 1) (Panel a). The positive beta coefficients of the LASSO (upward bars) reflect a susceptible behaviour of the features to the target COVID-19 disease, whereas the negative coefficients (downward bars) a protective action. The calculated odd ratio of AR short repeats (22) is 0.79 i.e. protective. Therefore, the odd ratio of long repeats (23) is 1/0.79 = 1.27 i.e. severity. Panel b: Table reporting the averages and the standard deviations of accuracy, precision, sensitivity, specificity, and ROC-AUC scores for the 10-folds of the cross-validation. M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 5 on the male cohort with the AR gene alone provides results quite Table 3 higher than the random guess (accuracy 58%, precision 71%, sensitiv- Validation in Spanish cohort ity 64%, specificity 55%, roc-auc 55%). Spanish validation (x2) Males global 22 23 Marginal Row Totals 3.2. Validation of polyQ polymorphism by sizing the PolyQ repeat of the Cases 51 (43,6%) 66 (56,4%) 117 (74,1%) AR gene Controls 27 (65,9%)* 14 (34,1%) 41 (25,9%) Marginal Column Totals 78 (49,4%) 80 (50,6%) 158 (Grand Total) In order to validate the results on AR obtained by LASSO logistic * p-value (cases vs controls)=0.014 (Significant at p<0.05) regression, we sized the number of triplets in the male subset (351 subjects) using the gold standard technique that uses a fluorescent PCR reaction followed by the use of GeneScan Analysis softwareÒ in short repeats), and anemia (11% in long repeats vs. 2% in short (Applied Biosystems) [20]. We identified a 98% concordance between repeats), two powerful sings of low androgenicity, and severe erectile the results of the two techniques in measuring the polyQ repeats. dysfunction (22% in long repeats vs. 9% in short repeats) were more Based on the AR polyQ length, male patients were subdivided into frequently reported in subjects with longer repeats, but not osteope- two categories, those having a number of PolyQ repeats less than or nia/osteoporosis (6% in long repeats vs. 7% in short repeats) (Supple- equal to 22 repeats, and those having a number of PolyQ repeats mentary Table S4). These results indicate a trend toward clinical greater than or equal to 23 repeats, being 23 repeats the reference hypogonadism for those with longer repeats. Conversely, in the sequence on genome browsers and the reported cut-off value [23- entire male dataset, 6 cases of prostate cancer were found annotated 24]. We found that PolyQ repeats below 22 are enriched in the in the past-medical history, all in the 22 glutamines group, suggest- asymptomatic cohort of males. The difference was statistically signifi- ing an increased prostate sensitivity to androgens in this group. No cant in the group of males younger than 60 years of age in which difference was found in the prevalence of BPH or 5-alpha-reductase genetic factors are expected to have a major impact (p-value 0.024 inhibitors use. by x2 test) (Table 2; Supplementary Table S2). As the reduced signal transduction of AR might be partially com- pensated by higher T levels, we tested whether the decreased AR 3.3. Validation of polyQ polymorphism in the Spanish Cohort negative feedback was sufficient to overcome larger polyQ repeats size (Fig. 2). Logistic regression was performed to investigate the We then sized the polyQ repeat in an independent cohort consist- joint effect of T level and polyglutamine receptor length on the likeli- ing of 158 <60 years old Spanish males without known comorbidities hood that subjects require intensive care during COVID infections, (117 cases and 41 controls). The association with shorter repeats ( adjusting for age in the model. The logistic regression model was 22) and protection was confirmed (p-value 0.014 by x2 test) (Table 3). highly significant (x2 (3) = 18,881, p < 0.0001), with the model explaining 7.5% (Nagelkerke /R2) of the variance in COVID-19 out- come (Supplementary Table S5). To test whether the association 3.4. Males with longer polyQ have receptor resistance between T and the outcome changes when the polyQ is short ( 22) or long (23), an interaction term was included in the model. A sig- To functionally link the length of the PolyQ repeats to AR function- nificant interaction was found (p-value 0.018), suggesting impaired ality, we measured TT in 183 men using LCMS/MS (Supplementary feedback as a predictor of the worst outcome, namely intubation or Table S2). TT was higher in patients carrying 23 vs 22 glutamines CPAP/BiPAP versus hospitalization not requiring respiratory assis- (13.45 vs 11.23 nmol/L, p-value 0.042), reflecting reduced negative tance. To provide an intuitive graphical representation, we plotted feedback from the less active receptors present in patients carrying a the ratio between TT serum concentrations and polyQ number vs. PolyQ repeat of 23. This difference was evident also comparing the clinical outcome (Supplementary Figure 1). Results show a TT value and polyQ repeats in the case and the control group (Fig. 2). decreased mean ratio, a sign of an inappropriate rise of TT for increas- ing polyglutamine repeats, and association with a worse outcome 3.5. Unbalanced T-AR axis in males with longer polyQ repeats (p = 0.0004). The hormonal status of the entire male cohort revealed lower TT 3.6. Inflammatory phenotype in males with longer polyQ repeats and calculated free T levels and higher SHBG levels with increasing age (Supplementary Table S3). Finally, we tested the relationship between the AR polyQ repeat To evaluate whether the AR receptor reduced activity resulted in size and 5 laboratory markers of immunity/inflammation, including signs and symptoms of hypogonadism, subjects were interviewed, CRP, Fibrinogen, IL6, CD4 and NK count. We found that older (60) post-infection, using a modified version of the AndrostestÒ [25]. males with AR polyQ tract 23 have a higher (55.92 versus 48.21 mg/ Interviews were available for 61 subjects (43 short and 18 long) rep- dl) mean value of CRP (p-value 0.018, not accounting for multiple resentative of the extremes genotypes (19 and 25 repeats) of the testing) and lower mean value of Fibrinogen and a trend of higher IL6 cohort. An Androtest score 8 was found in 38% of men with longer (Table 4). repeats as compared to 16% of those with 19 glutamines (likelihood ratio, p = 0.046). Similarly, cryptorchidism (11% in long repeats vs. 2% 4. Discussion Table 2 We employed machine learning methodologies to identify a set of PolyQ alleles correlation with COVID-19 outcome - males with age <60. genes involved in the severity of COVID-19. In the presence of very Males <60 high dimensionality, as for instance in a WES study, it is crucial to <22 >23 Marginal Row Totals select the most predictive genes representing patterns of variation (mutations or variants) in subjects with different classes of response Cases 52 (59,1%) 36 (40,9%) 88 (48,1%) Controls 71 (74,7%)* 24 (25,3%) 95 (51,9%) (i.e., disease state: from asymptomatic to severe cases). This problem Marginal Column Totals 123 (67,2%) 60 (32,8%) 183 (Grand Total) is even more complex in diseases where multiple genes are involved * p-value (cases vs controls) =0.024 in determining the severity and clinical variability of the pathology. Here, we wanted to represent poly-amino acids repeat 6 M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 Fig. 2. Relationship between Total Testosterone and polyQ repeats in the case and the control group. Box-plot showing values of Total Testosterone (TT), expressed in nmol/L, in subjects with shorter (22) and longer (23) polyQ repeats in AR gene grouped between controls (left panel) and cases (right panel). The TT median value, represented by the black horizontal line, is higher in patients with 23 polyQ repeats in the case group, (**p-value = 0.023; Mann-Whitney U test). No statistically significant difference was present in the control group (p-value = 0.088; Mann-Whitney U test). polymorphisms that are typically missed in classical GWAS analysis, COVID-19, some patients are unable to compensate for the reduced which concentrates on bi-allelic polymorphisms. AR activity with higher T levels [27]. The result is a status of reduced We used a machine learning approach and logistic regression with androgenicity even in the presence of apparently normal T values a LASSO regularization to test if using such a simplified representa- [27]. tion could lead to a reliable prediction of extreme clinical outcomes As T is known to have an immunomodulatory activity attenuating (asymptomatic versus severely affected). This approach enabled us to inflammatory immune responses [26 32], we hypothesized that a predict such clinical outcomes with 77% sensitivity. long PolyQ repeat would lead to a pro-inflammatory status heralded AR contains a highly variable polyglutamine repeat (poly-Q) by increased proinflammatory markers [19,33] by conferring located in the N-terminal domain of the protein, spanning from 9 to decreased AR transcriptional activity. Conversely, men with a more 36 glutamine residues in the normal population [5]. AR polyQ length active receptor (short PolyQ tract) would be protected because they correlates with receptor functionality, with shorter polymorphic glu- can tame the inflammatory response and increase survival regardless tamine repeats typically associated with higher and longer PolyQ of serum T levels. We found that -CRP-, one of the main inflammatory tracts with lower receptor activity [5]. AR is expressed in both males markers, was higher in subjects with a long AR PolyQ tract. This and females, but the bioavailability of its ligands T and dihydroT observation not only is in line with the known anti-inflammatory (DHT) differs significantly, being much higher in males. As previous function of T, but also reinforces the functional importance of the AR studies linked male hypogonadism to a poorer outcome in COVID-19 PolyQ tract and its association with COVID-19 clinical outcome. Fur- patients we decided to focus on male patients and demonstrated that thermore, this observation suggests that CRP is hierarchically more shorter polymorphic glutamine repeats (22) confer protection relevant than serum T level, which can be inappropriately normal against life-threatening COVID-19 in a subpopulation of individuals and mask a status of low androgenicity in men with a long PolyQ with age <60 years. repeat. We also confirmed the association between polyQ size and recep- The allele distribution of the PolyQ repeat length varies among tor activity. Specifically, we showed that longer polyQ size (23) is different populations, with the shortest in Africans, medium in Cau- associated with higher serum T levels, suggestive of impaired nega- casians, and longest in Asians [34]. Interestingly, WHO data on mor- tive feedback (p=0.004 at Mann-Whitney U test) at the level of the tality rates during the first pandemic wave indicated a higher fatality hypothalamus and pituitary gland. While this is compensated in rate in China and Italy (https://covid19.who.int/) [35] with respect to healthy subjects [26], during non-gonadal illnesses (NGI) such as African. Hence, AR polyQ length variability could represent an M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 7 Table 4 Correlation between polyQ repeats in AR gene and laboratory values CRP M60y cases CRP M<60y cases Triplets Mean Count Triplets Mean Count 22 48,21 78 22 54,5 43 23 55,92 38 23 26,41 29 p-value = 0.018 (Significant at p<0,05) p-value = 0.2 Fibrinogen M60y cases Fibrinogen M<60y cases Triplets Mean Count Triplets Mean Count 22 401,33 57 22 316,93 22 23 320,34 27 23 356,91 19 p-value = 0.093 p-value = 0.53 IL6 times the upper limit of normal M60y cases IL6 times the upper limit of normal M<60y cases Triplets Mean Count Triplets Mean Count 22 54,56 40 22 40,43 17 23 75,78 16 23 31,8 14 p-value = 0,249 p-value = 0,81 CD4 Lymphocytes M60y cases CD4 Lymphocytes M<60 cases Triplets Mean Count Triplets Mean Count 22 264,06 32 22 503,68 16 23 357,52 21 23 396,13 15 p-value = 0.22 p-value = 0.45 NK Cells M60y cases NK Cells M<60y cases Triplets Mean Count Triplets Mean Count 22 70,71 28 22 147,3 13 23 102,25 16 23 107,14 14 p-value = 0.179 p-value = 0.098 explanation for the observed differences in death rate. Moreover, organizing a clinical trial where patients selected based on their Africans seem to be more prone to infection [36]. This observation serum T concentration and polyQ repeat size are randomized to could be due to a more active AR receptor, leading to a higher expres- receive T vs. placebo. Such study could introduce the concept that a sion of TMPRSS2, a protease essential for SARS-CoV-2 spread [15]. simple genetic test measuring the AR polyQ repeat can be used in Different studies have shown an association between hypogonad- male patients to screen for those who are more likely to benefit from ism or long polyQ repeats and severe COVID-19 [16,37] and other T therapy. chronic obstructive pulmonary diseases [17,18]. Our results are in Variants of another X-linked gene, TLR7, have been associated line with these initial observations and provide a possible mechanism with severe COVID-19 outcomes in young men [6]. In the 2 reported explaining these associations. The present study brings these obser- families, the rare TLR7 mutations segregated as a highly penetrant vations to the next level, revealing that is the overall androgenic monogenic X-linked recessive trait. While variants in TLR7 gene are effect -resulting from the interaction of polyQ polymorphism and cir- expected to account for a small number of severely affected cases, culating T levels- that predicts the need for intensive care. In infected our findings involve a much larger number of subjects, as long polyQ men, we observed impaired feedback no longer sufficient to compen- alleles are relatively common (27%) [40]. Overall, X-linked genetic sate for the reduced AR transcriptional activity, leading to the conclu- variants keep coming up as important for defining severe COVID-19 sion that polyQ tract length is hierarchically more important than cases in males. serum T levels. This concept helps to solve some inconsistencies, In conclusion, we present a method that can predict if subjects including the early reports of a slightly better outcome in prostate infected by SARS-CoV-2 are at risk for life-threatening complications. cancer patients -who tend to have smaller polyQ repeats, as in our This approach has 77% accuracy, 81% precision, 77% sensitivity, and cohort - when compared to other cancers. Interestingly, previous 78% specificity. Furthermore, we present evidence suggesting that a studies failed to link polyQ with mortality, in healthy subjects [26] or more active AR has the potential to confer protection against COVID- individuals with chronic diseases such as diabetes mellitus [38]. 19 severity. If confirmed, these observations should be followed by Thus, the observed association between low androgenicity and out- properly conducted clinical trials exploring if T replacement may come seems related to the hyperinflammatory state present in severe decrease morbidity and mortality in patients affected by the most COVID-19. severe forms of the disease. Finally, as shown by regression analysis, An improvement in peak oxygen saturation in men receiving T ORs ranges between 1.26 and 1.45, therefore the risk of carrying a replacement therapy has been demonstrated in a randomized con- longer AR is much smaller than other already known strong predic- trolled trial [39] and could be one of the mechanisms responsible for tors such as age and sex, but still is highly significant, relatively com- the observed protective effect of AR’s with shorter polyQ tract in mon, and among the very few known genetic predictors of COVID-19 COVID-19 patients. The observations reported in this study prompt outcome. 8 M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 33 Declaration of Competing Interest Department of Preventive Medicine, Azienda USL Toscana Sud Est, Italy 34 The authors declare no competing interests. Territorial Scientific Technician Department, Azienda USL Toscana Sud Est, Italy 35 Additional information Clinical Chemical Analysis Laboratory, San Donato Hospital, Are- zzo, Italy 36 GEN-COVID Multicenter Study (https://sites.google.com/dbm. Chirurgia Vascolare, Ospedale Maggiore di Crema, Italy 37 unisi.it/gen-covid) Department of Health Sciences, Clinic of Infectious Diseases, Francesca Montagnani2,22, Laura Di Sarno1,2, Andrea Tommasi1,2,5, ASST Santi Paolo e Carlo, University of Milan, Italy Maria Palmieri1,2, Massimiliano Fabbiani22, Barbara Rossetti22, Gia- 38 Division of Infectious Diseases and Immunology, Fondazione como Zanelli2,22, Fausta Sestini20, Laura Bergantini23, Miriana D’Ales- IRCCS Policlinico San Matteo, Pavia, Italy sandro23, Paolo Cameli23, David Bennet23, Federico Anedda24, Simona 39 Department of Internal Medicine and Therapeutics, University Marcantonio24, Sabino Scolletta24, Federico Franchi24, Maria Anto- of Pavia, Italy nietta Mazzei25, Susanna Guerrini25, Edoardo Conticini26, Luca Can- 40 Department of Anesthesia and Intensive Care, University of tarini26, Bruno Frediani26, Danilo Tacconi27, Chiara Spertilli27, Marco Modena and Reggio Emilia, Modena, Italy Feri28, Alice Donati28, Raffaele Scala29, Luca Guidelli29, Genni Spargi30, 41 Department of Medical and Surgical Sciences for Children and Marta Corridi30, Cesira Nencioni31, Leonardo Croci31, Gian Piero Cal- Adults, University of Modena and Reggio Emilia, Modena, Italy darelli32, Maurizio Spagnesi33, Paolo Piacentini33, Maria Bandini33, 42 HIV/AIDS Department, National Institute for Infectious Diseases, Elena Desanctis33, Silvia Cappelli33, Anna Canaccini34, Agnese Ver- IRCCS, Lazzaro Spallanzani, Rome, Italy zuri34, Valentina Anemoli34, Agostino Ognibene35, Massimo Vaghi36, 43 III Infectious Diseases Unit, ASST-FBF-Sacco, Milan, Italy Antonella D’Arminio Monforte37, Esther Merlini37, Federica Gaia Mir- 44 Department of Biomedical and Clinical Sciences Luigi Sacco, Uni- aglia37, Mario U. Mondelli38,39, Stefania Mantovani38, Serena Ludo- versity of Milan, Milan, Italy visi38,39, Massimo Girardis40, Sophie Venturelli40, Marco Sita40, 45 Infectious Diseases Clinic, Department of Medicine 2, Azienda Andrea Cossarizza41, Andrea Antinori42, Alessandra Vergori42, Ospedaliera di Perugia and University of Perugia, Santa Maria Hospi- Arianna Emiliozzi2,22,42, Stefano Rusconi43,44, Matteo Siano44, Arianna tal, Perugia, Italy 46 Gabrieli44, Agostino Riva43,44, Daniela Francisci45,46, Elisabetta Schiar- Infectious Diseases Clinic, "Santa Maria" Hospital, University of oli45, Pier Giorgio Scotton47, Francesca Andretta47, Sandro Panese48, Perugia, Perugia, Italy Stefano Baratti48 Renzo Scaggiante49, Francesca Gatti49, Saverio Giu- 47 Department of Infectious Diseases, Treviso Hospital, Local seppe Parisi50, Francesco Castelli51, Eugenia Quiros-Roldan51, Melania Health Unit 2 Marca Trevigiana, Treviso, Italy Degli Antoni51, Isabella Zanella52, Matteo Della Monica53, Carmelo 48 Clinical Infectious Diseases, Mestre Hospital, Venezia, Italy. Piscopo53, Mario Capasso54,55,56, Roberta Russo54,55, Immacolata 49 Infectious Diseases Clinic, ULSS1, Belluno, Italy Andolfo54,55, Achille Iolascon54,55, Giuseppe Fiorentino57, Massimo 50 Department of Molecular Medicine, University of Padova, Italy Carella58, Marco Castori58, Giuseppe Merla58, Filippo Aucella59, 51 Department of Infectious and Tropical Diseases, University of Pamela Raggi60, Carmen Marciano60, Rita Perna60, Matteo Bas- Brescia and ASST Spedali Civili Hospital, Brescia, Italy setti61,62, Antonio Di Biagio62, Maurizio Sanguinetti63,64, Luca 52 Department of Molecular and Translational Medicine, University Masucci63,64, Serafina Valente65, Maria Antonietta Mencarelli5, of Brescia, Italy; Clinical Chemistry Laboratory, Cytogenetics and Caterina Lo Rizzo5, Elena Bargagli23, Marco Mandala 66, Alessia Molecular Genetics Section, Diagnostic Department, ASST Spedali Giorli66, Lorenzo Salerni66, Patrizia Zucchi67, Pierpaolo Parravicini67, Civili di Brescia, Italy Elisabetta Menatti68, Tullio Trotta69, Ferdinando Giannattasio69, 53 Medical Genetics and Laboratory of Medical Genetics Unit, A.O. Gabriella Coiro69, Fabio Lena70, Domenico A. Coviello71, Cristina Mus- R.N. "Antonio Cardarelli", Naples, Italy sini72, Giancarlo Bosio73, Enrico Martinelli73, Sandro Mancarella74, 54 Department of Molecular Medicine and Medical Biotechnology, Luisa Tavecchia74, Lia Crotti75,76,77,78,79 University of Naples Federico II, Naples, Italy 22 55 Dept of Specialized and Internal Medicine, Tropical and Infec- CEINGE Biotecnologie Avanzate, Naples, Italy 56 tious Diseases Unit, Azienda Ospedaliera Universitaria Senese, Siena, IRCCS SDN, Naples, Italy 57 Italy Unit of Respiratory Physiopathology, AORN dei Colli, Monaldi 23 Unit of Respiratory Diseases and Lung Transplantation, Depart- Hospital, Naples, Italy 58 ment of Internal and Specialist Medicine, University of Siena, Italy Division of Medical Genetics, Fondazione IRCCS Casa Sollievo 24 Dept of Emergency and Urgency, Medicine, Surgery and Neuro- della Sofferenza Hospital, San Giovanni Rotondo, Italy 59 sciences, Unit of Intensive Care Medicine, Siena University Hospital, Department of Medical Sciences, Fondazione IRCCS Casa Sollievo Italy della Sofferenza Hospital, San Giovanni Rotondo, Italy 25 60 Department of Medical, Surgical and Neuro Sciences and Radio- Clinical Trial Office, Fondazione IRCCS Casa Sollievo della Soffer- logical Sciences, Unit of Diagnostic Imaging, University of Siena, Italy enza Hospital, San Giovanni Rotondo, Italy 26 61 Rheumatology Unit, Department of Medicine, Surgery and Neu- Department of Health Sciences, University of Genova, Genova, rosciences, University of Siena, Policlinico Le Scotte, Italy Italy 27 62 Department of Specialized and Internal Medicine, Infectious Dis- Infectious Diseases Clinic, Policlinico San Martino Hospital, eases Unit, San Donato Hospital Arezzo, Italy IRCCS for Cancer Research Genova, Italy 28 63 Dept of Emergency, Anesthesia Unit, San Donato Hospital, Are- Microbiology, Fondazione Policlinico Universitario Agostino zzo, Italy Gemelli IRCCS, Catholic University of Medicine, Rome, Italy 29 64 Department of Specialized and Internal Medicine, Pneumology Department of Laboratory Sciences and Infectious Diseases, Fon- Unit and UTIP, San Donato Hospital, Arezzo, Italy dazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy 30 65 Department of Emergency, Anesthesia Unit, Misericordia Hospi- Department of Cardiovascular Diseases, University of Siena, tal, Grosseto, Italy Siena, Italy 31 66 Department of Specialized and Internal Medicine, Infectious Dis- Otolaryngology Unit, University of Siena, Italy 67 eases Unit, Misericordia Hospital, Grosseto, Italy Department of Internal Medicine, ASST Valtellina e Alto Lario, 32 Clinical Chemical Analysis Laboratory, Misericordia Hospital, Sondrio, Italy Grosseto, Italy M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 9 68 Study Coordinator Oncologia Medica e Ufficio Flussi Sondrio, We thank the CINECA consortium for providing computational resour- Italy ces and the Network for Italian Genomes (NIG) http://www.nig.cineca.it 69 First Aid Department, Luigi Curto Hospital, Polla, Salerno, Italy for its support. We thank private donors for the support provided to A. 70 Local Health Unit-Pharmaceutical Department of Grosseto, R. (Department of Medical Biotechnologies, University of Siena) for the Toscana Sud Est Local Health Unit, Grosseto, Italy COVID-19 host genetics research project (D.L n.18 of March 17, 2020). 71 U.O.C. Laboratorio di Genetica Umana, IRCCS Istituto G. Gaslini, We also thank the COVID-19 Host Genetics Initiative (https://www.cov Genova, Italy. id19hg.org/), MIUR project “Dipartimenti di Eccellenza 2018-2020” to 72 Infectious Diseases Clinics, University of Modena and Reggio the Department of Medical Biotechnologies University of Siena, Italy Emilia, Modena, Italy. and “Bando Ricerca COVID-19 Toscana” project to Azienda Ospedaliero- 73 Department of Respiratory Diseases, Azienda Ospedaliera di Universitaria Senese. We also thank Intesa San Paolo for the 2020 char- Cremona, Cremona, Italy ity fund dedicated to the project N. B/2020/0119 “Identificazione delle 74 U.O.C. Medicina, ASST Nord Milano, Ospedale Bassini, Cinisello basi genetiche determinanti la variabilita clinica della risposta a COVID- Balsamo (MI), Italy 19 nella popolazione italiana”. 75 Istituto Auxologico Italiano, IRCCS, Department of Cardiovascu- lar, Neural and Metabolic Sciences, San Luca Hospital, Milan, Italy. Data availability and data sharing statement 76 Department of Medicine and Surgery, University of Milano- Bicocca, Milan, Italy The samples referenced here are housed in the GEN-COVID 77 Istituto Auxologico Italiano, IRCCS, Center for Cardiac Arrhyth- Patient Registry and the GEN-COVID Biobank and are available for mias of Genetic Origin, Milan, Italy. sharing. The sequencing data are deposited in http://www.nig.cineca. 78 Istituto Auxologico Italiano, IRCCS, Laboratory of Cardiovascular it/, specifically, http://nigdb.cineca.it) and available for consultation. Genetics, Milan, Italy For further information, you may contact the corresponding author, 79 Member of the European Reference Network for Rare, Low Prev- Prof. Alessandra Renieri (e-mail: alessandra.renieri@unisi.it). alence and Complex Diseases of the Heart-ERN GUARD-Heart Spanish COVID HGE Supplementary materials Sergio Aguilera-Albesa80, Sergiu Albu81, Carlos Casasnovas82,13, Valentina Ve lez-Santamaria82,13, Juan Pablo Horcajada83, Judit Vil- Supplementary material associated with this article can be found, lar , Agustí Rodríguez-Palmero84,13,14, Montserrat Ruiz13,14, Luis M 83  s Troya86, Juan Valencia-Ramos87, Marta Gut88 in the online version, at doi:10.1016/j.ebiom.2021.103246. Seijo85, Jesu 80 Navarra Health Service Hospital, Pamplona, Spain 81 References Institut Guttmann Foundation, Badalona, Barcelona, Spain 82 Bellvitge University Hospital, L’Hospitalet de Llobregat, Barce- lona, Spain [1] Benetti E, Giliberti A, Emiliozzi A, et al. Clinical and molecular characterization of 83 COVID-19 hospitalized patients. PLoS One 2020;15(11):e0242534 Published 2020 Hospital del Mar, Parc de Salut Mar, Barcelona, Spain Nov 18. doi: 10.1371/journal.pone.0242534. 84 University Hospital Germans Trias i Pujol, Badalona, Barcelona, [2] Daga S, Fallerini C, Baldassarri M, et al. Employing a systematic approach to bio- Spain banking and analyzing clinical and genetic data for advancing COVID-19 research 85 [published online ahead of print, 2021 Jan 17]. Eur J Hum Genet. 2021;1-15. Clínica Universitaria de Navarra, Madrid, Spain doi:10.1038/s41431-020-00793-7. 86 Infanta Leonor University Hospital, Madrid, Spain [3] Zhang X, Tan Y, Ling Y, et al. Viral and host factors related to the clinical outcome 87 of COVID-19. Nature 2020;583(7816):437–40. doi: 10.1038/s41586-020-2355-0. University Hospital of Burgos, Burgos, Spain 88 [4] Ellinghaus D, Degenhardt F, Bujanda L, et al. Genomewide association study of CNAG-CRG, Centre for Genomic Regulation (CRG), Barcelona severe Covid-19 with respiratory failure [published online ahead of print, 2020 Institute of Science and Technology (BIST), Carrer Baldiri i Reixac 4, Jun 17]. N Engl J Med. 2020 NEJMoa2020283. doi: 10.1056/NEJMoa2020283. 08028, Barcelona, Spain [5] Zhang Q, Bastard P, Liu Z, et al. Inborn errors of type I IFN immunity in patients with life-threatening COVID-19. Science 2020:eabd4570. Epub ahead of print. PMID:32972995. doi: 10.1126/science.abd4570. Contributors [6] van der Made CI, Simons A, Schuurs-Hoeijmakers J, et al. Presence of genetic var- iants among young men with severe COVID-19. JAMA 2020;324(7):1–11 Epub ahead of print. PMID:32706371; PMCID: PMC7382021. doi: 10.1001/ EF, FM, AR designed the study. CF and IM, were in charge of biologi- jama.2020.13719. cal samples’ collection and biobanking. MB, FF were in charge of clinical [7] Bastard P, Rosen LB, Zhang Q, et al. Autoantibodies against type I IFNs in patients data collection. MB, FF, AR, and FM performed analysis/interpretation of with life-threatening COVID-19. Science 2020;370(6515):eabd4585. Epub 2020 clinical data. UP, FF and MM performed T measurement by LC-MS/MS. Sep 24. PMID:32972996. doi: 10.1126/science.abd4585. [8] Pivonello R, Auriemma RS, Pivonello C, et al. Sex disparities in COVID-19 severity EDG, AS, AP and LPS performed the validation of association between and outcome: are men weaker or women stronger? [published online ahead of shorter repeats and protection in a Spanish cohort. MM and AI critically print, 2020 Nov 26]. Neuroendocrinology 2020 10.1159/000513346. doi: reviewed the manuscript and interpreted clinical data/androgen physio- 10.1159/000513346. [9] Callewaert L, Christiaens V, Haelens A, Verrijdt G, Verhoeven G, Claessens F. Impli- pathological processes. SA and MB were in charge of DNA isolations cations of a polyglutamine tract in the function of the human androgen receptor. from peripheral blood samples. FV, GD, AG, RT carried the sequencing Biochem Biophys Res Commun 2003;306(1):46–52. doi: 10.1016/s0006-291x(03) experiments. EB, NP, SF, CG, MG and MS, performed bioinformatics and 00902-1. [10] Simanainen U, Brogley M, Gao YR, et al. Length of the human androgen receptor statistical analyses. EB, NP, SD, CF, and SC prepared Figures and Tables. glutamine tract determines androgen sensitivity in vivo. Mol Cell Endocrinol EB, NP, AMP, FPC, AR, EF and FM wrote the manuscript. All authors have 2011;342:81–6. doi: 10.1016/j.mce.2011.05.011. reviewed and approved the manuscript. [11] Tirabassi G, Cignarelli A, Perrini S, et al. Influence of CAG repeat polymorphism on the targets of testosterone action. Int J Endocrinol 2015;2015:298107. doi: 10.1155/2015/298107. Acknowledgements [12] Lindstro€m S, Ma J, Altshuler D, et al. A large study of androgen receptor germline variants and their relation to sex hormone levels and prostate cancer risk. Results from the national cancer institute breast and prostate cancer cohort consortium. J This study is part of the GEN-COVID Multicenter Study, https://sites. Clin Endocrinol Metab 2010;95(9):E121–7. doi: 10.1210/jc.2009-1911. google.com/dbm.unisi.it/gen-covid, the Italian multicenter study aimed [13] Wambier CG, Goren A. Severe acute respiratory syndrome coronavirus 2 (SARS- at identifying the COVID-19 host genetic bases. Specimens were pro- CoV-2) infection is likely to be androgen mediated. J Am Acad Dermatol 2020;83 vided by the COVID-19 Biobank of Siena, which is part of the Genetic (1):308–9. doi: 10.1016/j.jaad.2020.04.032. [14] Pozzilli P, Lenzi A. Commentary: testosterone, a key hormone in the context of Biobank of Siena, member of BBMRI-IT, of Telethon Network of Genetic COVID-19 pandemic. Metabolism 2020;108:154252. doi: 10.1016/j. Biobanks (project no. GTB18001), of EuroBioBank, and of RD-Connect. metabol.2020.154252. 10 M. Baldassarri et al. / EBioMedicine 65 (2021) 103246 [15] Hoffmann M, Kleine-Weber H, Schroeder S, et al. SARS-CoV-2 cell entry depends [28] Gubbels Bupp MR, Jorgensen TN. Androgen-induced immunosuppression. Front on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor. Immunol 2018;9:794. Published 2018 Apr 17. doi: 10.3389/fimmu.2018.00794. Cell 2020;181(2):271–80 .e8. doi: 10.1016/j.cell.2020.02.052. [29] Hoebe K, Janssen E, Beutler B. The interface between innate and adaptive immu- [16] Rastrelli G, Di Stasi V, Inglese F, et al. Low testosterone levels predict clinical nity. Nat Immunol 2004;5(10):971–4. doi: 10.1038/ni1004-971. adverse outcomes in SARS-CoV-2 pneumonia patients [published online ahead of [30] Janeway Jr. CA, Medzhitov R. Innate immune recognition. Annu Rev Immunol print, 2020 May 20]. Andrology 2020 10.1111/andr.12821. doi: 10.1111/ 2002;20:197–216. doi: 10.1146/annurev.immunol.20.083001.084359. andr.12821. [31] Lai JJ, Lai KP, Zeng W, Chuang KH, Altuwaijri S, Chang C. Androgen receptor influ- [17] Van Vliet M, Spruit MA, Verleden G, et al. Hypogonadism, quadriceps weakness, ences on body defense system via modulation of innate and adaptive immune and exercise intolerance in chronic obstructive pulmonary disease. Am J Respir systems: lessons from conditional AR knockout mice. Am J Pathol 2012;181 Crit Care Med 2005;172(9):1105–11. doi: 10.1164/rccm.200501-114OC. (5):1504–12. doi: 10.1016/j.ajpath.2012.07.008. [18] Mohan SS, Knuiman MW, Divitini ML, et al. Higher serum testosterone and dihy- [32] Medzhitov R, Janeway Jr. C. Innate immunity. N Engl J Med 2000;343(5):338–44. drotestosterone, but not oestradiol, are independently associated with favourable doi: 10.1056/NEJM200008033430506. indices of lung function in community-dwelling men. Clin Endocrinol (Oxf) [33] Pierotti S, Lolli F, Lauretta R, et al. Androgen modulation of pro- inflammatory and 2015;83(2):268–76. doi: 10.1111/cen.12738. anti-inflammatory cytokines during preadipocyte differentiation. Horm Mol Biol [19] Mohamad NV, Wong SK, Wan Hasan WN, et al. The relationship between circulat- Clin Investig 2010;4(1):483–8. doi: 10.1515/HMBCI.2010.059. ing testosterone and inflammatory cytokines in men. Aging Male 2019;22 [34] Ackerman CM, Lowe LP, Lee H, et al. Ethnic variation in allele distribution of the (2):129–40. doi: 10.1080/13685538.2018.1482487V. androgen receptor (AR) (CAG)n repeat. J Androl 2012;33(2):210–5. doi: 10.2164/ [20] Allen RC, Zoghbi HY, Moseley AB, et al. Methylation of HpaII and HhaI sites near jandrol.111.013391. the polymorphic CAG repeat in the human androgen-receptor gene correlates [35] Porcheddu R, Serra C, Kelvin D, Kelvin N, Rubino S. Similarity in case fatality rates with X chromosome inactivation. Am J Hum Genet. 1992;51(6):1229–39. (CFR) of COVID-19/SARS-COV-2 in Italy and China. J Infect Dev Ctries 2020;14 [21] Tibshirani R. Regression shrinkage and selection via the lasso. Journal of the Royal (2):125–8 Published 2020 Feb 29. doi: 10.3855/jidc.12600. Statistical Society. Series B (Methodological) 1996;58(21):267–88 .36. [36] de Lusignan S, Dorward J, Correa A, et al. Risk factors for SARS-CoV-2 among [22] Fanelli F, Belluomo I, Di Lallo VD, et al. Serum steroid profiling by isotopic dilu- patients in the oxford royal college of general practitioners research and surveil- tion-liquid chromatography-mass spectrometry: comparison with current immu- lance centre primary care network: a cross-sectional study. Lancet Infect Dis noassays and reference intervals in healthy adults. Steroids 2011 Feb;76(3):244– 2020;20(9):1034–42. doi: 10.1016/S1473- 3099(20)30371-6. 53. doi: 10.1016/j.steroids.2010.11.005. [37] McCoy J, Wambier CG, Herrera S, et al. Androgen receptor genetic variant predicts [23] Mengual L, Oriola J, Ascaso C, Ballesca JL, Oliva R. An increased CAG repeat length COVID-19 disease severity: a prospective longitudinal study of hospitalized in the androgen receptor gene in azoospermic ICSI candidates. J Androl 2003;24 COVID-19 male patients [published online ahead of print, 2020 Sep 25]. J Eur (2):279–84. doi: 10.1002/j.1939-4640.2003.tb02673.x. Acad Dermatol Venereol 2020 10.1111/jdv.16956. doi: 10.1111/jdv.16956. [24] Mo €hlig M, Arafat AM, Osterhoff MA, et al. Androgen receptor CAG repeat length [38] Wong PW, Lee HM, Lau ESH, et al. Interactive effects of testosterone and the polymorphism modifies the impact of testosterone on insulin sensitivity in men. androgen receptor CAG repeat length polymorphism on cardiovascular- renal Eur J Endocrinol 2011 Jun;164(6):1013–8. doi: 10.1530/EJE-10-1022. events and mortality in men with diabetes. Diabetes Metab Res Rev 2019;35(1): [25] Millar AC, Lau ANC, Tomlinson G, Kraguljac A, Simel DL, Detsky AS, Lipscombe LL. e3081. doi: 10.1002/dmrr.3081. Predicting low testosterone in aging men: a systematic review. CMAJ 2016 Sep [39] Caminiti G, Volterrani M, Iellamo F, et al. Effect of long-acting testosterone treat- 20;188(13):E321–30. ment on functional exercise capacity, skeletal muscle performance, insulin resis- [26] Eendebak RJ, Huhtaniemi IT, Pye SR, et al. The androgen receptor gene CAG repeat tance, and baroreflex sensitivity in elderly patients with chronic heart failure a in relation to 4-year changes in androgen-sensitive endpoints in community- double-blind, placebo-controlled, randomized study. J Am Coll Cardiol 2009;54 dwelling older European men. Eur J Endocrinol 2016;175(6):583–93. doi: (10):919–27. doi: 10.1016/j.jacc.2009.04.078. 10.1530/EJE-16-0447. [40] Canale D, Caglieresi C, Moschini C, et al. Androgen receptor polymorphism (CAG [27] Zitzmann M, Nieschlag E. The CAG repeat polymorphism within the androgen repeats) and androgenicity [published correction appears in Clin Endocrinol receptor gene and maleness. Int J Androl 2003;26(2):76–83. doi: 10.1046/j.1365- (Oxf). 2005 Oct; 63(4):482]. Clin Endocrinol (Oxf) 2005;63(3):356–61. doi: 2605.2003.00393.x. 10.1111/j.1365-2265.2005.02354.x.