key: cord-0772847-m8zgr4t1
authors: Masana, Lluís; Correig, Eudald; Rodríguez-Borjabad, Cèlia; Anoro, Eva; Arroyo, Juan Antonio; Jericó, Carlos; Pedragosa, Angels; Miret, Marcel·la; Näf, Silvia; Pardo, Anna; Perea, Verónica; Pérez-Bernalte, Rosa; Plana, Núria; Ramírez-Montesinos, Rafael; Royuela, Meritxell; Soler, Cristina; Urquizu-Padilla, Maria; Zamora, Alberto; Pedro-Botet, Juan
title: EFFECT OF STATIN THERAPY ON SARS-CoV-2 INFECTION-RELATED MORTALITY IN HOSPITALIZED PATIENTS
date: 2020-11-02
journal: Eur Heart J Cardiovasc Pharmacother
DOI: 10.1093/ehjcvp/pvaa128
sha: 9da46d9d412ea1e8998f06d31c6b777a41478298
doc_id: 772847
cord_uid: m8zgr4t1

AIM: Assessing the effect of statin therapy at hospital admission for COVID-19 on in-hospital mortality. METHODS AND RESULTS: Retrospective observational study. Patients taking statins were 11 years older and had significantly more comorbidities than patients who were not taking statins. A genetic matching (GM) procedure was performed prior to analysis of the mortality risk. A Cox proportional hazards model was used for the cause-specific hazard (CSH) function, and a competing-risks Fine and Gray (FG) model was also used to study the direct effects of statins on risk. Data from reverse transcription-polymerase chain reaction-confirmed 2157 SARS-CoV-2-infected patients (1234 men, 923 women; age: 67 y/o (IQR 54-78)) admitted to the hospital were retrieved from the clinical records in anonymized manner. 353 deaths occurred. 581 patients were taking statins. Univariate test after GM showed a significantly lower mortality rate in patients on statin therapy than the matched non-statin group (19.8% vs. 25.4%, χ(2) with Yates continuity correction: p = 0.027). The mortality rate was even lower in patients (n = 336) who maintained their statin treatments during hospitalization compared to the GM non-statin group (17.4%; p = 0.045). The Cox model applied to the CSH function (HR = 0.58(CI: 0.39-0.89); p = 0.01) and the competing risks FG model (HR = 0.60(CI: 0.39-0.92); p = 0.02) suggest that statins are associated with reduced COVID-19-related mortality. CONCLUSIONS: A lower SARS-CoV-2 infection-related mortality was observed in patients treated with statin therapy prior to hospitalization. Statin therapy should not be discontinued due to the global concern of the pandemic or in patients hospitalized for COVID-19.

The once-in-a-century pandemic caused by the SARS-CoV-2 virus has spread worldwide.

More than 25 million people have been infected less than 1 year from the first reported case in Wuhan, and nearly one million people have died. Many questions about the pathophysiology of the SARS-CoV-2 infection are unanswered. The coronavirus virus enters the cells via binding to the angiotensin-converting enzyme 2 (ACE2) protein, which is located on the cell surface of different tissues. The virus triggers an overwhelming inflammatory and thrombotic response in severe cases after intracellular replication, which leads to severe lung injury (COVID-19) and multiorgan failure 1 . Approximately 15% of patients admitted to the hospital require invasive therapies in intensive care units, and approximately one in five of these patients die 2 .

The reasons why some patients remain asymptomatic and others develop a deadly disease are not known. Several prognostic factors were identified, such as age (maximum lethality in the elderly), sex, race and comorbidities, including hypertension, cardiovascular diseases, obesity and diabetes, are the frequently reported in the more severely ill patients 3 .

The effects of background therapies on prognosis was examined. The relevance of ACE2 in the pathogenesis of the disease highlights the role of drugs targeting the renin-angiotensinaldosterone axis and increasing ACE2 expression, and the withdrawal of these therapies was recommended 4 . However, the clinical data show no effect or a protective effect of these therapies 5 .

Statins are among the more frequently prescribed drugs in the general population. Statins block an early stage of cholesterol synthesis by inhibiting the enzyme hydroxy-methyl-glutaryl CoA reductase. Statins also modulate the production of some downstream intermediates in the reverse transcription-polymerase chain reaction (PCR) and whose infections were acquired in the community were included.

The exposure of interest was the patient's use of statin therapy in the previous year based on the clinical records. The statin therapy was categorized as high intensity (80 mg/day atorvastatin and 20 mg/day rosuvastatin) or low-moderate intensity. The maintenance of therapy during hospitalizations (as included in the medical orders for at least 48 hours) was also recorded.

The present study assessed the effect of background statin therapy on in-hospital SARS-CoV-2 infection-related mortality. The secondary objectives were the effects of statin therapy on surrogate markers of clinical severity.

Continuous variables were tested for normality using the Kolmogorov-Smirnov test. Data are presented as medians and 25th and 75th percentiles for continuous variables with a nonnormal distribution or means and standard deviations (SD) for variables with a normal distribution. Differences between groups were analysed using the non-parametric Mann-Whitney test or Student's parametric t test for continuous variables and the chi-squared test or Fisher's exact test for categorical variables.

Patients on statins were older and had more comorbidities than patients not on statins. To determine the association between statin treatment and mortality, we matched patients from our statin and no-statin groups to balance their baseline characteristics. We used a genetic matching (GM) (supplementary material reference S1 S2 ) procedure because the usual propensity score matching techniques failed to achieve acceptable balance. The GM uses a genetic search algorithm to iteratively determine the weight of each of the covariates to find an optimal balance between matched groups. The matching was 1:1 with replacement and ties (so that one treated unit could be matched to more than one untreated unit after weighting them appropriately) and without callipers. We included all clinical variables prior to hospital admission in the matching procedure. A summary of matching results using standardized differences before and after is provided as supplementary material on-line (Table S1) .

For the matched set, we constructed two types of survival models, a Cox proportional hazards model applied to a cause-specific hazard (CSH) function S3 and a competing-risks Fine and Gray (FG) mode S4 S5 . Models were constructed for a 6-week follow-up.

The CSH considered hospital discharge as a censoring event, i.e., incorporated the assumption that discharged individuals will not die from COVID-19 and are therefore different from simply censored individuals. Although this model under-estimates absolute survival probability, it allowed us to interpret aetiological relationships between covariates (specifically, statin use) and patient outcomes S4 .

The FG model computes the probability of any event (in our case, death or discharge) at time "t" based on the assumption that no other event has occurred. Therefore, it may be used to predict events S6 , and it complements the CSH interpretation. This procedure was performed using a sub-distribution hazard function S5 .

We checked the proportional hazards assumption model and the proportionality of the sub-distribution from the FG model. We stratified all models based on sex because it strongly violated these assumptions. Other smaller violations were ignored. The statins treatment hazard ratio was estimated using the (robust) Huber sandwich estimator.

For the CSH model, we provide survival estimate curves for participants stratified by sex to visualize the effects of statins on death and discharge. These curves were computed by predicting the survival curve for each individual in the dataset then averaging S3 . The same technique was used to plot the cumulative incidence functions of the two events, which illustrates the predictions of the FG model.

Two sub-analyses were also performed using the same techniques within the matched groups. The first approach considered that some patients stopped receiving statins during the 48 hours after hospital admission. These patients were previously included in the "treated" group and now formed a special "withdrawn" group. This analysis removed 26 patients who were censored for any reason prior to 48 h to avoid immortal time bias. The second approach differentiated statin intensity intake as none, moderate or high according to the aforementioned criteria.

Statistical analyses were performed using the R software package version 3.5, and all codes are found at https://github.com/ecorreig/STACOV.

A list of statistical key resources and references is provided in the supplementary material.

Data from 2157 (1234 men and 923 women) SARS-CoV-2-infected patients who were admitted to the hospital were analysed. The median age was 67 years (IQR 54-78 years). A total of 581 patients (38.7% women) were on statin therapy (ST) at admission, and 30% were on high intensity statin therapy. Statin therapy was withdrawn within the first 48 hours of admission in 245 patients (42,2%). The statins were maintained unchanged in 336 patients (57,8%). To assess the effect of withdrawing or maintaining statin therapy we exclude 22 patients censored within the first 48 hours for any reason, 9 in the GM-non-statin group (GM-NST) and 13 in the statin group (ST). More males than females were on statins, and the median age of the statin group was 11 years older than the non-statin (NST) group. The NST group had significantly fewer comorbidities. Demographic and clinical data are shown in Table 1 . Out of the total 353 patients, 16.3% died: 115 (19.8%) in the ST group and 238 (15%) in the NST group (p=0.04) ( Table 2) . Table 1 also reports the demographics and clinical data of the GM groups. The percentage of deaths in the comparable GM-NST group was 25.7%, which was significantly higher than the ST group (19.8%) (p= 0.027) ( Figure 1A and Table 2 ). Although no significant differences were observed in statin intensity, the percentage of deaths was even lower in patients who maintained their statin treatments during hospitalization compared to the GM-NST group (17.4%; p=0.045) ( Figure 1B ). Table 2 also shows several biomarker values, COVID-19-specific therapies and main clinical outcomes sorted by statin therapy groups. Although statistically significant differences were observed in few items, observed rates of inflammation markers and severe clinical outcomes, such as ARDS, acute renal failure or the need for tracheal intubation, were lower in patients on statins, even if not statistically significant. ( Table 2 ).

The effect of statin treatment on overall mortality was also shown using a CSH model stratified by sex (Supplementary material Figure S1 ). A significant difference in the mortality rate was observed between groups (HR = 0.58 with (0.39-0.89) 95% CI; p = 0.01).

The competing risk Fine and Gray analysis ( Figure 2) showed that statins were associated with a significantly lower probability of mortality (sub-distribution HR = 0.60 with (0.39-0.92) 95% CI; p = 0.02) and showed a trend towards a higher probability of achieving hospital discharge. The CHS and FG methods suggest a stronger effect in females, but this result should be confirmed in a larger study.

We used the same methods within the GM groups to analyse differences between patients with and without statin treatment discontinuation (supplementary material Table S2 ). No significant differences were observed, in unadjusted comparisons, between the continuing and discontinuing groups, but the continued statin treatment was associated with lower risk of mortality compared to the NST group. (n=327, HR 0.60 with (0.39-0.92) 95% CI; p = 0.02). No differences were observed between high and moderate statin use.

The present study suggests that statin therapy prior to hospitalization with SARS-CoV-2 is associated with lower mortality compared to matched patients not on statin therapy.

Prehospitalization statin therapy was significantly associated with a lower in-hospital mortality rate in patients with COVID-19. The mortality rate was significantly lower in patients in whom statin therapy was maintained compared to the NST group, but there were non-balanced confounding factors when compared to discontinued, because no further matching was applied to these analyses,. Patients on statins showed a less severe pulmonary effect on X ray examination and better oxygen parameters (PaFi). Although not statistically significant, the results showed lower severe clinical outcomes, such as ARDS, respiratory and renal failure and the need for tracheal intubation in the ST group. Therefore, the first message is that background statin therapy should not be withdrawn based on COVID-19 concerns. Although the withdrawal of statins during hospitalization, primarily in severe cases requiring invasive treatments in the ICU, will not substantially alter its cardioprotective effects, the discontinuation of these treatments in the general population due to concerns related to the SARV-CoV-2 pandemic may lead to a prolonged suspension of statin therapy and a potential increased cardiovascular risk 20 . Our results are consistent with findings from a recent study in a Chinese population that showed that in-hospital statin therapy was associated with a lower severity of COVID-19 17 .

We used two types of survival models that revealed that statins were associated with a significantly lower mortality (Figure 2 and Figure S1 ), suggesting a stronger beneficial effect in women. Although this point should be addressed in a focused study, a biological interaction between sex and statin effects must not be excluded 21 .

Statins were discontinued in 42% of patients. In our study, 78% of statin withdrawal was coincident to antiretroviral prescription, primarily lopinavir/ritonavir, suggesting a concern about the drug-drug interactions, however, an even lower mortality rate was observed in patients who remained on statins, although a firm conclusion on the effects of statin continuation can not be drawn from our study because confounding factors as a significant different antiretroviral administration (supplementary material Table S2 ).

Despite the use of a robust matching method to balance the main demographic and comorbidity variables, some baseline characteristics were not totally balanced. Patients on statins more frequently received angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor blockers (ARB) and insulin therapy. The impact of these drugs on COVID-19 therapy remains controversial, and the low absolute differential number of patients receiving these treatments do not explain the overall differences observed.

Lipid concentrations are modified during COVID-19 infection. A more severe infection results in lower cholesterol concentrations 22 . Although our study was not focused on the association between lipids and COVID-19, lower total and LDL cholesterol concentrations were observed in the statin group ( Table 1) . The possible effect of statins on the COVID-19 prognosis should be elucidated based on their non-lipid mechanisms, which are referred to as pleiotropic effects 23 . Statins exert effects on inflammation in vitro by interfering with several intracellular proinflammatory signalling cascades 1, 6 . Although, its effect on inflammatory mechanisms at the clinical level remains controversial 24 , our study observed a trend to lower concentrations of inflammatory biomarkers (e.g., white blood cells, ferritin, and C reactive protein). Statins exhibit antioxidant and antithrombotic activities and ameliorate endothelial dysfunction 25 .

Dyslipidaemia was recently associated with the risk of pulmonary thromboembolism in patients with COVID-19, and statins were suggested as a preventive therapy 26 .

In addition, statins may target other SARS-CoV-2-specific mechanisms. As mentioned above, statins upregulated ACE2 expression in rabbits 10 , which could be a negative effect of statins 27 . In contrast, some studies suggest a protective effect of ACE2 expression in ARDS 28 , and statin use was recommended during the previous SARS pandemic (Middle East Respiratory Syndrome) 29 . A direct interference with SARS-CoV-2 replication mechanisms was suggested recently 7 . Based on these findings, a role for statins in treating COVID-19 was proposed 8,30 .

The present study has some limitations. It was a retrospective observational study, and causality could not be extrapolated from our data. The GM analysis reduced the clinical distance between the statin and non-statin groups, but it introduced some uncertainty. Because all hazard and risk analyses performed in the GM groups were designed to compare mortality between the statin and non-statin groups, the effects of other covariates on risk were not extrapolated from our results, and specific analyses are warranted.

In conclusion, background statin therapy exerted a beneficial effect on the in-hospital mortality of SARS-CoV-2-infected patients. The maintenance of statin therapy during admission correlated with an even better prognosis. The potential beneficial effect of statin therapy on 

COVID-19: from epidemiology to treatment

Presenting Characteristics, Comorbidities, and Outcomes among 5700 Patients Hospitalized with COVID-19 in the New York City Area

Baseline Characteristics and Outcomes of 1591 Patients Infected With SARS-CoV-2 Admitted to ICUs of the Lombardy Region

Association of Inpatient Use of

Enzyme Inhibitors and Angiotensin II Receptor Blockers with Mortality among Patients with Hypertension Hospitalized with COVID-19

Association of GM: Genetic [search algorithm based] Matched; COPD: Chronic Obstructive Pulmonary Disease; ACE: Angiotensin-converting enzyme; ARB: Angiotensin receptor blockers; SGLT2: Sodium Glucose Co-Transporter 2; GLP-1R: Glucagon-Like Peptide-1 Receptor; NOACs: Novel Oral Anticoagulants; HDL: High-Density Lipoprotein

* P value between statin and non-statin matched groups