Inpatient Administration of Alpha-1-Adrenergic Receptor Blocking Agents Reduces Mortality in Male COVID-19 Patients

Apha-1-adrenergic receptor antagonists (α1-blockers) can suppress pro-inflammatory cytokines, thereby potentially improving outcomes among patients with COVID-19. Accordingly, we evaluated the association between α1-blocker exposure (before or during hospitalization) and COVID-19 in-hospital mortality. We identified 2,627 men aged 45 or older who were admitted to Mount Sinai hospitals with COVID-19 between February 24 and May 31, 2020, in New York. Men exposed to α1-blockers (N = 436) were older (median age 73 vs. 64 years, P < 0.001) and more likely to have comorbidities than unexposed men (N = 2,191). Overall, 777 (29.6%) patients died in hospital, and 1,850 (70.4%) were discharged. Notably, we found that α1-blocker exposure was independently associated with improved in-hospital mortality in a multivariable logistic analysis (OR 0.699; 95% CI, 0.498-0.982; P = 0.039) after adjusting for patient demographics, comorbidities, and baseline vitals and labs. The protective effect of α1-blockers was stronger among patients with documented inpatient exposure to α1-blockers (OR 0.624; 95% CI 0.431-0.903; P = 0.012). Finally, age-stratified analyses suggested variable benefit from inpatient α1-blocker across age groups: Age 45-65 OR 0.483, 95% CI 0.216-1.081 (P = 0.077); Age 55-75 OR 0.535, 95% CI 0.323-0.885 (P = 0.015); Age 65-89 OR 0.727, 95% CI 0.484-1.092 (P = 0.124). Taken together, clinical trials to assess the therapeutic value of α1-blockers for COVID-19 complications are warranted.


INTRODUCTION
Severe coronavirus disease 2019 (COVID -19) has been linked to dysregulated immune responses, including an excessive inflammatory response marked by high levels of proinflammatory cytokines such as interleukin-6 (IL-6) (1,2). Immunosuppressive drugs such as glucocorticoids have become a standard component of treatment for severe  (3), and several trials of anti-cytokine or anti-inflammatory agents are underway or have reported promising results (4)(5)(6)(7)(8)(9)(10). Despite these advances, there remains a need for safe, effective, and widely available therapeutic options.
Adrenergic signaling has been linked to hyperinflammation in models of bacterial sepsis and cytokine release syndrome. In preclinical experiments, a positive feedback loop of adrenergic signaling was identified wherein macrophages responded to catecholamines by producing more catecholamines and inflammatory cytokines; this adrenergic loop could be interrupted by blocking α 1 -adrenergic receptors with prazosin (11). In a retrospective clinical study of patients with acute respiratory distress and pneumonia, exposure to α 1 -adrenergic receptor antagonists (α 1 -blockers) was associated with a significant reduction in risk of mechanical ventilation or death (12). Similarly, a recent retrospective analysis of 25,130 patients with COVID-19 across the United States Veterans Health Administration hospital system showed that outpatient exposure to any α 1 -blocker was associated with decreased in-hospital mortality compared to matched controls not on any α 1 -blocker at the time of hospital admission (13).
These observations have led to the hypothesis that α 1blockers in routine clinical use (e.g., prazosin, doxazosin, tamsulosin, etc.) may be repurposed for COVID-19 treatment (14). We conducted this real-world evidence study based on electronic medical record (EMR) data to determine whether exposure to α 1 -blockers is independently associated with mortality among patients hospitalized with COVID-19.

Patient Characteristics and Outcomes
We gathered and processed data from five hospitals within the Mount Sinai Health System to construct three cohorts of male patients aged 45 years or older: (a) no α 1 -blocker exposure (N = 2,191), (b) an α 1 -blocker-exposed group (N = 436), and (c) a documented inpatient α 1 -blocker-exposed group (N = 343) ( Figure 1A). The most common α 1blocker was Tamsulosin followed by Doxazosin ( Figure 1B). See Materials and Methods for details regarding data processing and cohort generation.
We further assessed the impact of α 1 -blockers on mortality for patients with documented administration of α 1 -blockers while admitted to the hospital (N = 343) compared to unexposed patients. We observed that inpatient α 1 -blocker use significantly reduced the risk of in-hospital mortality overall (OR 0.624; 95% CI 0.431-0.903; P = 0.012) ( Table 2).

DISCUSSION
Using a racially and ethnically diverse cohort from New York City comprising 2,627 men aged 45 or older hospitalized COVID-19 patients seen between February 24 and May 31, 2020, we found that inpatient use of α 1 -blockers was significantly associated with reduced in-hospital mortality after adjusting for several confounders. In age-stratified analyses, α 1 -blocker exposure appeared more protective in the 55-75 year age group.  Drug repurposing is the process of finding new indications for drugs already in clinical use. The appeal of rapidly validating and deploying an existing drug against a deadly global pandemic is clear, especially if the drug is widely available and affordable. Dexamethasone, now a standard in COVID-19 treatment, is an example of a commonly used drug repurposed for a new indication (3). However, the saga of hydroxychloroquine, which was touted as a cure early in the pandemic but has since proven ineffective, is a cautionary tale (15). The allure of rapid drug repurposing must be balanced against rigorous scientific method. α 1 -blockers, commonly used to treat benign prostatic hyperplasia and hypertension, have become a target for drug repurposing due to preclinical data linking α 1 -adrenergic signaling to pro-inflammatory cytokines which may contribute  to dysregulated immunity and adverse outcomes in COVID-19 (1,2,11). These preclinical findings have been bolstered by recent retrospective clinical analyses linking α 1 -blockers with improved outcomes in hospitalized patients with both COVID-19 and non-COVID-19 respiratory infections (12,13).
In the present study, we found that in-hospital use of α 1 -blockers was independently associated with reduced inhospital mortality after controlling for confounders such as demographics, comorbidities, and clinical factors such as vital signs and lab values. In age-stratified analyses, we observed that this protective effect was more pronounced in the 55-75 year age group. In contrast to the studies by Koenecke et al. and Rose et al., which defined α 1 -blocker exposure based on outpatient prescriptions only, we were able to use inpatient medication administration records to identify patients treated with α 1 -blockers during their COVID-19 hospitalization. The stronger effect seen in the inpatient exposure group than the overall group lends additional support to the hypothesis that α 1 -blockers may have a beneficial effect against COVID-19.
Our results also include tests of association between other common medication classes and COVID-19 outcomes, including beta-blockers, angiotensin-converting enzyme (ACEi) inhibitors, angiotensin II receptor blockers (ARBs), and glucocorticoids. Results pertaining to these other medications should be taken in the context of a selected cohort designed to study α 1 -blocker use and COVID-19 outcomes. That said, it is interesting to note that glucocorticoids were associated with worse COVID-19 outcomes, contrary to the results of a randomized controlled trial (3). This discrepancy may be due to indiscriminate administration of steroids early in the pandemic, as seen in other RWE studies. For example, the use of high-dose steroids was associated with higher odds of death (16). In this study, a high dose was classified by >40 mg daily of methylprednisolone equivalent dosing. For comparison, the equivalent to the RECOVERY trial dosing of 6 mg dexamethasone is 20-30 mg of methylprednisolone (16). Furthermore, corticosteroids were associated with an increased risk of death in patients younger than 60 years without inflammation on admission (17). Thus, the observed effect of steroids in this real-world study may diverge from the effect reported in randomized trials due to factors such as inconsistent dosing, steroid choice, and patient selection early in the pandemic.
Also of interest, α 1 -blocker and beta-blocker exposure were associated with opposite COVID-19 outcomes in our cohort. There is evidence to suggest that β-adrenergic signaling can promote an anti-inflammatory M2 phenotype in macrophages, in contrast to the pro-inflammatory effect of α 1 -adrenergic signaling (11,18). Additional efforts to dissect the interactions between adrenergic signaling and the COVID-19 immune response are warranted. A prior diagnosis of asthma was associated with reduced in-hospital mortality in this analysis. While this observation deserves further scrutiny, it is conceivable that early exposure to inhaled glucocorticoids or β-adrenergic agonists may have contributed to this signal.

Limitations
Our study has several limitations. The cohort did not include women since most α 1 -blockers were prescribed to men, most likely for benign prostatic hyperplasia. Male sex is a recognized   risk factor for adverse COVID-19 outcomes, possibly due to sexspecific differences in immunity (19). Thus, these results may not extrapolate to women. We did not account for different types of α 1 -blockers, which differentially target the three α 1 -adrenergic receptor subtypes.
Importantly, a causal relationship cannot be definitively established between α 1 -blockers and improved COVID-19 outcomes in this retrospective study. Several confounders, such as older age, comorbidities, and hypoxia (an indicator of COVID-19 severity), were more common in the α 1 -blocker group. However, these adverse risk factors would be expected to bias the study result toward the null rather than inflate a protective association. Furthermore, our findings are consistent with prior data (13). Ongoing randomized clinical trials of prazosin (20) and doxazosin (21) against a placebo among hospitalized COVID-19 patients will include women and provide more definitive data on the therapeutic value of α 1 -blockers.
Finally, outpatient medication adherence cannot be evaluated from the EMR. However, inpatient medication administrations are captured by the EMR and provide a definitive record of exposures. Therefore, analyzing in-hospital medication administration is more robust.

CONCLUSIONS
In conclusion, this retrospective study found a protective association between α 1 -blocker exposure and COVID-19 outcomes in a cohort of hospitalized men. These results augment the rationale for studying and repurposing α 1 -blockers as a COVID-19 therapeutic. Thus, we await the results of two ongoing randomized clinical trials (20, 21) to definitively assess the effectiveness of alpha-1-blockers in protecting patients against COVID-19.

Data Sources
This retrospective study utilized de-identified electronic medical record (EMR; Epic Systems, Verona, WI) data from five member hospitals within the Mount Sinai Health System (MSHS) in the New York City metropolitan area (MS BI Brooklyn, MS St. Luke's, The Mount Sinai Hospital, MS Queens Hospital, and MS West). De-identified EMR data were obtained via the Mount Sinai Data Warehouse (https://labs.icahn.mssm. edu/msdw/). COVID-19 was diagnosed by real-time reverse transcriptase polymerase chain reaction (RT-PCR)-based clinical tests from nasopharyngeal swab specimens. In total, we identified 8,442 MSHS patients with PCR confirmed diagnosis of COVID-19 from February 24 through May 31, 2020, during the peak of the pandemic in NYC.
We retrieved patient demographics, social history, medication history, and disease comorbidities from the EMR including age, gender, race/ethnicity, smoking status, asthma, chronic obstructive pulmonary disease (COPD), hypertension, obstructive sleep apnea, obesity, diabetes, chronic kidney disease, human immunodeficient virus (HIV) infection, cancer, coronary artery disease, atrial fibrillation, heart failure, chronic viral hepatitis, alcoholic non-alcoholic liver disease, and acute kidney injury (AKI). Patients aged ≥ 89 years were assigned an age of 89 to prevent re-identification. Medications by prescription or hospital administration captured in EMR from January 1, 2019, till May 31, 2020, were included in the medication history. We identified disease comorbidities through their corresponding ICD-10-CM codes before hospital admission and during hospitalization.
We also extracted data from each hospital encounter, including vital signs and laboratory data at the time of presentation, and medications administered during hospitalization. Vital sign and laboratory data extracted included: white blood cell count (WBC), serum creatinine, anion gap, potassium, alanine aminotransferase (ALT), body mass index (BMI), temperature, oxygen saturation, heart rate, respiratory rate, systolic blood pressure (SBP) and diastolic blood pressure (DBP).
This study was approved by the Mount Sinai institutional review board (IRB): IRB-17-01245.

Study Design
This was a retrospective EMR-based study designed to test the independent association of α 1 -blocker exposure with inhospital death among COVID-19 patients. We first identified 6,218 inpatients positive for COVID-19 in one of five hospital systems within the MSHS as of May 31, 2020 ( Figure 1A). The majority (93%) of α-1 -blocker users in this cohort were men aged 45 or older. Therefore, we restricted the analysis cohort to men aged 45 or older (N = 2,627) to limit confounding due to the associations between older age/male sex with both the exposure (α 1 -blocker usage) and the outcome (COVID-19 outcomes).
The primary endpoint was in-hospital mortality. We defined two possible outcomes for each hospitalization: in-hospital death (deceased) or discharged to home or other locations not associated with acute medical care (recovered). The duration of hospitalization was calculated from the beginning of the hospital encounter till death or discharge.
Potential confounders in the analysis included demographic characteristics, comorbidities, baseline labs and vitals, and exposure to medications used to treat hypertension, hyperlipidemia, and inflammation. Confounders were selected a priori based on the literature, clinician input, and data completeness. Detailed medication names included in these categories are shown in Supplementary Table 1. Certain comorbidities, e.g., obesity, are reported as conditions that patients carried prior to hospital admission. We required that at least 85% of patients report a value for a potential confounder for it to be included in the analysis. Some potential confounders that have been associated with severe forms of COVID-19 since the beginning of the pandemic did not meet this threshold, e.g., baseline Ferritin and LDH measures. However, we did include Ferritin values recorded at hospitalization as there was reasonable coverage (75.1%).

Statistical Analyses
Patient characteristics were summarized as median and interquartile range (IQR) for continuous variables or mean and standard deviation (SD). We displayed categorical variables as number and percentage (%). We performed a statistical test of hypothesis for differences using the Kruskal-Wallis test or two sample t-test for continuous variables, and the χ 2 test for categorical variables.
We employed multivariate logistic regression models with potential confounders to estimate the odds ratio and corresponding 95% confidence interval for COVID-19 inhospital mortality (deceased = 1) vs. recovery (recovered = 0) associated with α 1 -blocker use. We adjusted for the following confounders, which were selected a priori: age, hospital stay duration, race, smoking status, BMI, temperature, O2 saturation, heart rate, respiratory rate, hypertension, asthma, COPD, obstructive sleep apnea, obesity, diabetes mellitus, chronic kidney disease, HIV, cancer, coronary artery disease, atrial fibrillation, heart failure, chronic viral hepatitis, liver disease, AKI, ICU stay, WBC, creatinine, anion gap, potassium, and ALT. We used two-tailed test to estimate the probability of event under the null hypothesis which was the in-hospital mortality rate, and we used the binomial distribution for α 1 -blocker with the prescription rate as the probability.
Statistical significance was defined as a two-sided P-value < 0.05, unless otherwise noted.

DATA AVAILABILITY STATEMENT
The data analyzed in this study is subject to the following licenses/restrictions: our data collection, cleaning, and quality control framework makes use of proprietary data structures and libraries, so we are not releasing or licensing this code. We provide implementation details in the methods section to allow for independent replication. Requests to access these datasets should be directed to li.li@sema4.com.