Background: The cellular transmembrane protease serine 2 (TMPRSS2) cleaves the spike protein of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to let viral cell entry. Patients (pts) with prostate carcinoma (PCa) are expected to be at higher risk for coronavirus disease 2019 (COVID-19). However, TMPRSS2 transcription is significantly down-regulated by luteinizing hormone releasing hormone agonists (LHRH-a) and enzalutamide used to treat different PCa stages and this could alter the ability of SARS-CoV-2 to enter target cells. This study aims to evaluate if Single Nucleotide Polymorphisms (SNPs) in the TMPRSS2 gene are associated to a different risk for developing severe COVID-19. Methods: This is the translational part of a large retrospective-prospective real world study designed to estimate whether anti-androgen therapy for PCa is associated to a lower severity of COVID-19. Blood samples were collected from volunteer pts affected with PCa treated at our Institution who developed COVID-19. To be included, pts should receive at least LHRH-a for locally advanced o metastatic hormone sensitive or resistant disease. Genomic DNA of PCa pts was genotyped with the Axiom Human Genotyping SARS-CoV-2 array. Raw genotyping data were analyzed using the Axiom Analysis Suite and genotypes of 219 TMPRSS2 variants were extracted. Identified SNPs were used to compare reference and alternative allele counts between PCa pts and a control cohort of healthy European subjects (n = 2848; DOI:10.1056/NEJMoa2020283) collected in the Milan area (Italy) with chi squared test. Uni- and multivariate logistic regressions were performed to investigate associations between such identified SNPs and severe COVID-19, defined as any between the need for hospitalization or access to an intensive care unit or oxygen therapy. Results: 45 pts were enrolled from November 2020 to May 2022. Median age at COVID-19 detection was 77 years (range 51-94). 10/45 (22%) pts had severe COVID-19, but none died. At COVID-19 diagnosis, 4/45 pts (9%) were receiving chemotherapy for PCa, 13/45 (29%) new generation hormonal agents, 24/45 (53%) LHRH-a, the remaining 4/45 (9%) other therapies. Compared to the control cohort, we identified 52 TMPRSS2 variants associated with the susceptibility to SARS-CoV-2 infection in the cohort of PCa pts (P-value < 0.05). Among them, 3 SNPs were associated with severity in univariable logistic regression models: 2 resulted to be protective (rs9974589, p 0.01, OR 0.26; rs28442123, p 0.04, OR 0.30), and one was associated with increased risk of severe COVID-19 (rs2070788, p 0.01, OR 3.8). A multivariate logistic model (p = 0.0004, AUC = 0.88) was performed considering rs11908828 (p = 0,0109, OR = 32.25), rs28442123 (p = 0.03, OR = 0.11) and rs2070788 (p = 0.02, OR = 6.33). Conclusions: These preliminary findings point to a possible role of TMPRSS2 SNPs in modulating the severity of COVID-19 in PCa pts. Validation in wider series is needed.