Non-severe immunosuppression might be associated with a lower risk of moderate-severe acute respiratory distress syndrome in COVID-19.

BACKGROUND: Coronavirus disease 2019 (COVID-19) is an emerging infectious disease that has spread rapidly worldwide. The role of immunosuppression among COVID-19 patients has not been elucidated and management may be challenging. OBJECTIVE: To assess differences in severe outcomes of hospitalized patients with COVID-19 according to immune system state. DESIGN: Retrospective single-center observational study with PARTICIPANTS: Conrmed COVID-19 patients. development of event end of follow-up). As a secondary analysis, we performed Kaplan-Meier survival curves of time to moderate or severe ARDS with a log-rank test adjusted by immunosuppression status. A bivariate adjusted Cox proportional-hazards model was used to determine and 95% CIs on the development of the primary endpoint between both groups. All data analyses were conducted using R (R Core Team (2017), Version 3.6.3., R Foundation for Statistical Computing, Statistical hypotheses were tested using p<0.05 as the level of signicance.

from a hypothesized host hyper-in ammatory response and warrants reconsideration of management of IS patients.
The rst published data from China indicated that nearly one-third of patients may subsequently develop acute respiratory distress syndrome (ARDS), which results in a higher likelihood of needing mechanical ventilation (MV) and of death 2 . Older age and the presence of comorbidities have been associated with a higher risk of ARDS, and with progression from ARDS to death 3 .
Patients with immunosuppression were initially thought to be at an increased risk of severe COVID-19. However, preliminary data suggest a three-stage SARS-CoV-2 infection, with two distinct but overlapping subsets 4 : the rst triggered by the virus itself and the second consisting of a deregulated and excessive host immune response, which appears to be the main driver of lung tissue damage 5 . Therefore, we hypothesize that a weaker immune response to the virus might prevent immunosuppressed patients from developing a severe disease.
We aimed was to evaluate differences in the in ammatory response during COVID-19 between immunosuppressed (IS) and non-immunosuppressed (non-IS) patients by assessing the risk of ARDS, the need for mechanical (MV) or non-invasive ventilation (NIV), death and the length of hospitalization.

Study design
This single-center, retrospective, observational study was performed at Hospital Universitario Ramón y Cajal, Madrid, a tertiary hospital in the region with the highest incidence of cases of the pandemic of COVID-19 in Spain 6 . Due to the emergency outbreak, the hospital was completely restructured and nearly all the specialists of the Neurology department were assigned to attend COVID-19 patients. A multidisciplinary group of Internal Medicine, Infectious Diseases, and Neumology experts elaborated on clinical guidelines, which were strictly followed by all the physicians treating COVID-19 patients. All adult inpatients with con rmed or highly suspected SARS-CoV-2 infection from March 18, 2020 to April 4, 2020 attended by a selected group of multidisciplinary physicians were consecutively enrolled (Figure 1). For the analysis, only those with a laboratory-con rmed infection and a minimum of baseline and outcome variables available (detailed below) were included. According to the WHO guidance 7 , con rmation of SARS-CoV-2 was de ned as a positive result of real-time reverse transcriptase-polymerase chain reaction (RT-PCR) assay of nasal and pharyngeal swabs. The RT-PCR assay was performed by expert microbiologists from the same center. The patients were followed-up with until April 9, 2020. This study was approved by the institutional ethics board of Hospital Universitario Ramón y Cajal. Due to the nature of the retrospective chart review, the need for informed consent from individual patients was waived.

Data collection
The electronic medical records of the participants were reviewed by a trained team of physicians from Hospital Universitario Ramón y Cajal during the epidemic period, after selection based on the eligibility criteria. Baseline characteristics included were: demographics, medical history, comorbidities, laboratory examinations, radiological ndings, and clinical and respiratory variables. Additionally, we included such variables as bacterial co-infection and treatments administered. Immunosuppressed patients were de ned as those with an inherited or acquired immunode ciency or taking any drug with an immunomodulatory or immunosuppressive effect (see Supplement Material for a complete description of all IS patients). Corticosteroids were considered immunomodulators at prednisone-equivalent dosages >10 to <40 mg/day, but immunosuppressants at ≥40 mg/day.
The primary endpoint was the development of moderate or severe acute respiratory distress syndrome (ARDS), de ned according to the Berlin De nition 8 . Secondary endpoints were percentages of death among those with a nal outcome, need for mechanical (MV) or non-invasive ventilation (NIV), a composite of need of MV/NIV or death, length of hospitalization to April 9, 2020, and time to moderatesevere ARDS between the cohorts.

Statistical analysis
Continuous variables were described using means and standard deviations or medians and quartiles depending on whether or not data were normally distributed. The Shapiro-Wilk test was used to test the normality of data distributions. Categorical variables were described using absolute and relative frequencies. The Fisher test was used to analyze categorical variables and the Mann-Whitney U test was used for quantitative variables. Logistic regression models were conducted to study the association between primary and secondary endpoints and immunosuppression; in the case of the length of hospitalization, we used linear regression instead. We performed both unadjusted (crude) and multivariate (adjusted) models, taking into account potential confounding factors (sex, age, and time from onset of symptoms to event or end of follow-up). As a secondary analysis, we performed Kaplan-Meier survival curves of time to moderate or severe ARDS with a log-rank test adjusted by immunosuppression status. A bivariate adjusted Cox proportional-hazards model was used to determine HR and 95% CIs on the development of the primary endpoint between both groups. All data analyses were conducted using R (R Core Team (2017), Version 3.6.3., R Foundation for Statistical Computing, Vienna, Austria). Statistical hypotheses were tested using p<0.05 as the level of signi cance.

Results
From March 18, 2020 to April 4, 2020, 157 patients with highly suspected or con rmed COVID-19 were admitted and attended to by the treating physicians and were therefore selected for this study. After excluding those who did not ful ll the eligibility criteria (17 patients due to a negative test result for SARS-CoV-2 and 2 patients due to insu cient data), 138 patients were included in the analysis. From the nal cohort, 19.6% (n=27) of patients were IS and 80.4% (n=111) were non-IS. Among IS patients, 63% (n=17) were diagnosed with an autoimmune disease (IS-AD) and 37% (n=10) were diagnosed with other diseases (IS-NAD) (Figure 1). Additional information about IS patients is provided in the Supplement Material. Table outlines the demographic, clinical and laboratory baseline characteristics of all patients and both groups. Overall, 68.8% (95 of 138 patients) were male and the median (Q1, Q3) age was 68 (54 -78) years, distributed similarly among both cohorts. No statistically signi cant differences were observed between IS and non-IS in terms of comorbidities, excepting for subsidiary diseases of immunosuppression such as AD (59.3% vs. 2.7%, p<0.001) and organ transplant (18.5% vs. 0%, p<0.001).
Comparison of complications during hospitalization

Discussion
In this retrospective, single-center, observational study with inpatients with COVID-19 in Madrid, Spain, a better outcome, in terms of lower risk to moderate or severe ARDS, was observed among a cohort of patients with non-severe immunosuppression as compared to non-IS. A trend toward a shorter time to moderate-severe ARDS and a shorter hospitalization were also observed. After stratifying by source of IS, the protective effect seemed to be mainly driven by AD. However. no differences in time to moderatesevere ARDS, need for MV/NIV, or a composite of MV/NIV and death were detected between IS and non-IS despite lower proportions for the rst cohort. A comparison of both groups showed no differences in the death ratio.
Immunosuppression has been widely considered a risk factor for infections, with a higher incidence and a worse outcome, including those caused by a respiratory virus 9 . For example, in uenza infection was observed to be associated with a higher risk of more hospitalizations, a longer length of virus shedding, a more severe disease, and complications requiring intensive care and MV 10 . Additionally, several studies consider immunosuppression to be a risk factor of a more severe disease in MERS-CoV infection [11][12][13][14] . Previous works have already described risk factors predicting a worse outcome in COVID-19 patients, such as older age 3, 15-17 , comorbidities (hypertension, diabetes, or vascular diseases) 3, 15, 17 , -and laboratory ndings, with special attention paid to those indicating hyperin ammation or cytokine storm syndrome 18,19 , like elevated serum D-dimer, ferritin, C-reactive protein, or interleukin (IL)-6 levels 3, 15-17 . Nevertheless, none of these works has speci cally assessed immunosuppression as a risk factor in COVID-19 patients. A population-based study in China evaluated cancer patients with recent surgery or chemotherapy, and found a higher risk of severe events 20 . In addition, recent data about patients with moderate or severe immunosuppression associated to haematologic malignancy 21,22 and solid organ transplant recipients 23, 24 detected higher death ratios compared to general population. In line with these results, a large population-based study with over 17 million subjects developed in UK found a higher mortality among organ transplant, immunosuppression, haematological malignancies, and several autoimmune diseases 25 . However, limitations to interpret these results are that all national COVID- 19 inpatients are compared to about 40% of general population (with a potential selection bias) and no description about treatments and baseline characteristics of these groups is reported.
In our study, a lower proportion of IS patients developing moderate-severe ARDS was observed. We decided to assess this variable as a primary endpoint due to its higher speci city in detecting more in ammatory patients. This contrasts with previous data 21-25 about immunosuppression, but differences might reside in the grade of immunosuppression, being less severe in our cohort. To delve into a possible condition explaining these results, we further differentiated IS patients between those with an AD and those with other diseases (such as cancer or an organ transplant).  27 have been observed. In COVID-19, both innate immune hyperactivation and adaptive immune dysregulation have been hypothesized to be responsible of ARDS 28 . For this reason, several immunomodulatory drugs, such as corticosteroids, intravenous immune globulin, and cytokine inhibitors with a mild effect over immune system have been proposed depending on whether the hyperin ammation stage is suspected 18,19,22 . Preliminary data suggest a potential bene t of methylprednisolone in terms of the risk of death after the development of ARDS 3 , though clinical trials are required before recommending this therapy. All these results support the notion that immunosuppression, at least if non-severe might confer a protective effect in the most severe stages of the disease. We found that the lowest risk of moderate-severe ARDS was detected in IS-AD patients. This contrasts with evidence of an increased risk of infections observed in AD, through the presence of neutralizing autoantibodies against pro-in ammatory cytokines. These autoantibodies may have the ability to interfere with key cytokine such as IL-6, IFN-g, granulocyte/macrophage-colony stimulating factor (GM-CSF), IL-17, and IL-22 resulting in a lower Th1 and Th17 in ammatory response 29 . We have not assessed the risk of infection by SARS-CoV-2 in IS patients. But it could be hypothesized that a complex interaction between AD and a non-severe immunosuppression in COVID-19 could exert a protective effect of severe outcomes related to an innate immune hyperactivation.
No differences in death rate were observed in our cohort. The reasons for these results might be explained by the smaller sample size of patients with a nal outcome, the short follow-up, or unknown confounding factors. For example, during cytokine storm syndrome, ARDS is not the single in ammatory complication observed, as evidence of multi-organ dysfunction, with acute cardiac injury 30 and liver and renal impairment 3, 16, 17, 31 , has been described. This might be attributable to the widespread distribution of angiotensin converting enzyme 2 -the functional receptor for SARS-CoV-2-in multiple organs 32 . In addition, the results may be biased by the fact that IS patients may never get to MV/NIV, being considered too disabled, so they are more likely to die than non-IS patients. Also, several factors related to the development of ARDS that were not associated with death have been described 3 , which indicates that different pathophysiological changes -from hospital admission to the development of ARDS and from the development of ARDS to death-may exist.
Thus, our results might have relevance in terms of establishing recommendations for IS patients. Discontinuation of immunosuppressive drugs can be considered or suggested by either asymptomatic patients or treating physicians concerned about possible a worse course if a SARS-CoV-2 infection develops. But this may have implications with respect to diminishing the underlying disease control, with potential fatal outcomes in either cancer or organ transplant patients or the reactivation 33 or even rebound 34 of disease activity among autoimmune disorders. Taking this data into account, careful individual decision-making about maintaining immunosuppressive drugs must be performed in noninfected or even mild COVID-19 patients. Second, less aggressive anti-in ammatory management among IS patients might be contemplated, lowering the risk of bacterial co-infections. Third, social and preventive care recommendations might be reconsidered for these patients.
This study has several limitations. First, a potential selection bias might have occurred, as only a part of hospitalized patients (those attended by the treating physicians reported) were evaluated and IS might be admitted more easily by its own condition than non-IS. For this reason, patients directly admitted to ICU from the emergency room were not included. Second, this study was conducted at a single center with a limited sample size. Third, the cohort of IS patients is heterogeneous regarding diseases and immunosuppressive drugs, which may limit external validity to all type of IS patients. In addition, small sample size limits subgroup analyses by source of immunosuppression. Fourth, due to the lack of evidence-based treatment protocols, the treating physicians took different management approaches (especially with anti-viral drugs or corticosteroids), which could have altered the development of the outcomes. And nally, the retrospective character of the study and the short follow-up time warrant caution in interpreting the data. Further studies with a prospective design and a larger sample size, including outpatients, might gain a better understanding of the role that immunosuppression plays among COVID-19 patients.
In conclusion, in our cohort non-severe immunosuppression was associated with a lower risk of moderate-severe ARDS, a trend toward a reduced need for MV/NIV, a shorter hospitalization, and a longer time before moderate or severe ARDS occurs. We found the mortality rate was not increased in IC patients. This reinforces that there is a potential protective effect of immunosuppression against a possible hyperin ammation host response observed in SARS-CoV-2 infection and warrants reconsideration of discontinuing systematically immunosuppressive drugs in patients with severe underlying diseases.

Declarations
Demographic, clinical and laboratory characteristics at admission    Time to moderate or severe ARDS. ARDS: acute respiratory distress syndrome.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementalMaterial.docx