Critical Illness-related Corticosteroid Insu � ciency is Common in Covid-19 and Treatment Provide Survival Bene �


 PurposeCovid-19 is a severe lethal disease characterized with pneumonia and acute respiratory distress syndrome. We aimed to analyze the prevalence of critical illness-related corticosteroid insufficiency (CIRCI) in Covid-19 patients treated in intensive care unit and whether treatment of it would create any survival benefit. MethodsA total of 119 Covid-19 patients in intensive care unit were enrolled into the study. All patients received treatment for Covid-19 according to national guideline including hydroxychloroquine, favipiravir, antibiotics including azithromycin and oseltamivir in some patients. Basal cortisol below 10 µg/dl were accepted as CIRCI and above 34 µg/dl as intact HPA axis. Patients between them were introduced 1 µg cosynthropin stimulation test and delta cortisol below 9 µg/dl were also accepted as CIRCI. All patients who got diagnosis received hydrocortisone 300 mg/day until clinical condition stabilized. Mortality rate was recorded, then. ResultsCIRCI was detected in 50.4% of cases and mean plasma ACTH level was 14.8±6.0 pg/ml. Patients below age 65 had a statistically higher risk (63.4 % vs. 40.2%, p=0.012). Presence of comorbidity did not increase the risk of CIRCI (47.3% vs. 62.5%, p=0.18). Also, severity of pulmonary involvement and intubation did not increase the risk of CIRCI. Total mortality rate was 49.5% and not different in patients with and without CIRCI (45.0% vs. 54.5%, p=0.31) indicating benefit of corticosteroid treatment. ConclusionCIRCI is common in Covid-19 patients treated in intensive care unit and treatment of CIRCI provide survival benefit.

Critical illness-related corticosteroid insu ciency (CIRCI) describes the impairment of the hypothalamicpituitary-adrenal (HPA) axis during critical illness. Inadequate corticosteroid activity is the result of dysregulated systemic in ammation [6]. CIRCI can occur in a variety of acute conditions, such as sepsis and septic shock, ARDS, severe community acquired pneumonia, and other shock states [6,7]. Treatment with various doses of corticosteroid provide survival bene t in sepsis or septic shock [8].
Patients with adrenal insu ciency are at increased risk of infection due to their suppressed immunity which could facilitate the worsening of a SARS-CoV-2 infection into severe acute respiratory distress syndrome [9]. Patients with already known adrenal failure are advised to double the dose of original regimen in mild COVID-19. But in more severe or critical care patients, treatment should include higher doses like in adrenal crisis. Given that suggestions patients who had been diagnosed adrenal failure in critical care should be treated promptly to decrease morbidity and mortality related to that [6][7][8].
Data from SARS-CoV in the past indicate suppression of HPA axis. Study was done on survivors of SARS-Cov infection. Patients were evaluated 3 months after recovery of the disease and HPA axis suppression was detected in 40% of cases, majority of them resolved within a year [10]. In normal physiological response, immune reaction to an infection induces HPA axis and increased secretion of cortisol results in mild immunosuppression to control cytokine toxicity. However, in critical illnesses HPA activation was mostly blunted, leading to corticosteroid insu ciency [11]. Initial immune response to virus in COVID-19 induces an uncontrolled cytokine storm characterized by hyperin ammation and immunosuppression.
Resultant reaction characterized by vasculitis, thromboembolism, and hypotensive shock resembles adrenal insu ciency.
Up to now, there is no study on assessment of adrenal function in COVID-19. The aim of this study was to determine the rate of adrenal failure in critical care patients and the effect of treatment on survival.

Subjects
Patients aged above 18 years (males and females) were recruited from the intensive care unit at the Bagcilar Training and Research Hospital, Istanbul, Turkey. A total of 119 patients with documented COVID-19 were included in the study.
Patients were admitted to critical care unit mostly for severe pneumonia and related ARDS. Other indications were MODS and myocardial infarction. Indications for admission to critical care were determined by advisory board committee of ministry of health in Turkey [12]. According to that in addition to pulmonary in ltrates on thorax computerized tomography, patients with (a) respiratory rate >30/minute, (b) dyspnea and respiratory failure, (c) patient with oxygen saturation <90% and PaO2 <70mmHg (despite nasal oxygen supply of 5 liters/minutes), (d) Pa02/Fi02 <300, (e) lactate>4mmol/L, (f) bilateral in ltration or multilobar involvement in lung CT, (g) hypotension, (h) skin perfusion disorder, (i) kidney and liver function test disorder, thrombocytopenia and organ dysfunction, (j) immunosuppressive patients, (k) uncontrolled comorbidity, (l) elevated troponin, arrhythmia.
All of the patients needed assisted ventilation either by intubation (n=60) or noninvasive mechanical ventilation (NIMV) (n=59). MODS was described as change in consciousness, respiratory failure, low oxygen saturation, low urine output, increase in creatinine, increase in heart rate, liform pulses, cold extremities, low blood pressure, coagulopathy, thrombocytopenia, acidosis, increase in lactate level and hyperbilirubinemia.
Concomitant diseases were recorded. Subjects with any disease that can affect cortisol secretion such as brain tumors, pituitary diseases, and adrenal diseases were excluded. Patients who were pregnant or have alcohol abuse were also not included. Patients on treatment with drugs that can directly or indirectly cause or affect cortisol metabolism were also excluded. But, medications that patient on treatment for chronic diseases were recorded and included into the study such as antihypertensive medications, antidiabetics etc.
This study was conducted in accordance with the guidelines proposed in the Declaration of Helsinki and was approved by the ethical committee of Bagcilar Training and Research Hospital (2020.05.1.10.042). All subjects or legal representative gave informed consent before entrance into the study.

Study design
All patients were seen in intensive care units. After collection of demographic data, laboratory results used in evaluation and follow-up of a COVID-19 was recorded from patient le e.g. complete blood count, lymphocyte count, C-reactive protein (CRP), brinogen, D-dimer, lactate dehydrogenase (LDH), ferritin, triglyceride, troponin-I, creatine kinase (CK), blood urea, creatinine, aspartate aminotransferase (AST) and alanine aminotransferase (ALT). These laboratory tests were a part of routine assessment and we did not add any other test than those.

Assessment of Adrenal Function
For the diagnosis of adrenal insu ciency in patients in critical care, Society of Critical Care Medicine (SCCM) and European Society of Intensive Care Medicine (ESICM) 2017 recommendations were used [13]. The task force recommended delta cortisol (change in baseline cortisol at 60 min of < 9 μg/dL) after cosyntropin (250 μg) administration and a random plasma cortisol of < 10 μg/dL to be used for diagnosis.
Initially blood drawn from all patients for random cortisol. Patients with random cortisol level below 10 μg/dL were directly accepted as adrenal insu ciency and values above 34 μg/dL were accepted as intact adrenal gland. If random cortisol was between 10-34 μg/dL, patients were directed to cosyntropin stimulation test. But instead of 250 μg which is thought to be supra-physiologic for assessment of these patients, we used 1 μg test. Both tests are equally accurate to identify mortality and time of vasopressor therapy (14). Delta cortisol (change in baseline cortisol at 30 or 60 min of < 9 μg/dL) after cosyntropin (1 μg) administration was accepted as adrenal insu ciency.
Patients who had diagnosed as CIRCI were administered intravenous hydrocortisone at a dose of 100 mg three times daily during ICU stay. If patient recovered from COVID-19 and clinically stabilized hydrocortisone dose was steadily tapered to maintenance dose of 30 mg/day and until reassessment of adrenal function they were suggested to continue hydrocortisone after discharge. Other patients who had initial cortisol level above 34 μg/dL or patients who had > 9 μg/dL increase in cosyntropin test were not given any corticosteroid therapy. All patients also received standard supportive care and COVID-19 speci c medications.

Assessment of Pulmonary Involvement
All patients presenting with dyspnea or symptoms like coughing were directed to computerized tomography of thorax [12]. In order to assess the severity of pulmonary involvement, we strati ed patients into three groups: patients who had involved pulmonary surface area on CT scan less than 25% were accepted as mild, 25-50% moderate, >50% severe. All scans were assessed by same radiologist.

Treatment of Comorbidities
If patients were stable in concomitant diseases, we did not change their therapy unless clinical condition created a contraindication for the use of medications such renal dysfunction or hepatic failure. For the treatment of diabetes, oral antidiabetics were stopped, and insulin infusion was introduced to control glycaemia. Antihypertensive treatment was continued unless contraindicated or patient becomes hypotensive. All other medications were continued.

Assessment of mortality
In-hospital mortality was recorded with the exact reason.

Assays
All laboratory assays were studied at the central laboratory of Bagcilar Training and Research Hospital. Serum LDH, AST, ALT, Urea, creatinine, CRP, CK, triglyceride was studied with photometric method in Beckman Coulter AU4800, Fibrinogen, D-dimer, was studied in Succeeder sf-8200. Complete blood count was studied with absorption photometry and ow cytometry in Sysmex XN-900.

Statistical analysis
Statistical analyses were performed using SPSS software version 25. Descriptive analyses were presented using means and standard deviations for continuous data and frequencies and percentages for categorical data. Since the variables were normally distributed, independent samples t-test was used to compare the groups. The Chi-Square and Fisher's exact test, where appropriate, were used to compare the percentages in the groups. For the multivariate analysis, the possible risk factors identi ed statistically signi cant with the univariate analyses were further entered into lineer regression analysis to determine independent predictors of adrenal insu ciency. A 5% type-I error level was used to infer a statistical signi cance.

Diagnosis of CIRCI
Critical illness-related corticosteroid insu ciency was detected in 60 (50.4%) patients in intensive care unit with Covid-19 infection. Twenty-ve of them were diagnosed with basal cortisol below 10 μg/dl and 35 of them with delta cortisol below 9 μg/dl at 60th minute of 1 μg cosynthropin test. There were differences in some parameters between either group whether diagnosed with basal cortisol or cosynthropin test (Table 1).

Demographic characteristics
Demographic characteristics of study population were presented in Table 2. Although men (63.8%) dominated study population, the rate of diagnosis was similar in both sex (47.3% vs. 55.8%, p=0.37).
Age is accepted as an important risk factor for Covid-19, especially patients above age of 65 have a higher risk. On the other hand, in this study population younger patients below age 65 had a statistically higher risk of CIRCI (63.4 % vs. 40.2%, p=0.012). We also looked at young patients and 8 patients were below age of 45 and all of them had CIRCI.
Ninety-ve patients (79.8%) had at least one comorbidity. But comorbidity also did not increase the risk of CIRCI, even patients without comorbidity had a relatively higher rate of CIRCI (47.3% vs. 62.5%, p=0.18).

Frequency of Comorbidities and Effect on Diagnosis
The most common comorbidities were hypertension (51.2%), diabetes mellitus (39.4%), coronary artery disease (33.6%), chronic renal failure (17.6) and chronic obstructive pulmonary disease (15.1%). Other comorbidities and rate of diagnosis of CIRCI were presented in Table 2.
The Effect of Medications on CIRCI Eighty-one (68.0%) of patients were on chronic treatment with some medications while hospitalized. None of the patients were on treatment that in uences cortisol metabolism. Whether the patient on treatment with any medication or not did not differ the possibility of CIRCI (46.9% vs 57.8%, p=0.26). Medication list and their effect on CIRCI rate was presented in Table 2.

Laboratory tests
Laboratory results were presented in Table 3. As expected, cortisol levels either basal or in ACTH stimulation test were lower in patients with CIRCI (p<0.0001). All other in ammatory markers related to SARS Covid-19 infection were not different between groups. Although disease severity was described as an increased in ammatory markers and pulmonary involvement, there were no correlation between severity of pulmonary involvement and in ammatory markers such as CRP, brinogen, leukocytosis etc.

Mortality
Total mortality rate was 49.5%. There was no difference in patients with and without CIRCI (45.0% vs. 54.5%, p=0.31). Main reason for mortality was respiratory failure in 84.7% of patients, multiorgan failure in 13.5% and myocardial infarction in 1.7% of cases. Mortality risk was correlated with basal (p=0.02, r=0.209) and mean cortisol level (p=0.01, r=0.224). Among laboratory tests only CRP level was correlated with mortality (p=0.003, r=0,274). Severity of pulmonary involvement (p<0.0001, r=0.542) and intubation (p<0.0001, r=0.501) were correlated with mortality. In linear regression analysis, only basal cortisol level had effect on mortality (p=0.02 R2=0.044). Comorbidities did not have any effect on mortality.

Other Correlation Analysis
There was a positive correlation between age and severity of pulmonary in ltration (p=0.04, r=0.188) and presence of any comorbidity (p=0.0001, r=0.321).

Discussion
The results of this study showed that adrenal insu ciency is common among patients with COVID-19 treated in critical care units. Compared to other studies, the prevalence was similar to patients with severe sepsis reported by Annane et al. [15]. Treatment with corticosteroids decreased mortality in these patients.
CIRCI was rst introduced in 2008 by a task force convened by the Society of Critical Care Medicine to describe the impairment of the HPA axis during critical illness [16]. Severity of the critical illness causes inadequate cellular corticosteroid activity resulting in dysregulated systemic in ammation. Three major pathophysiologic events are considered to constitute CIRCI: dysregulation of the HPA axis, altered cortisol metabolism, and tissue resistance to glucocorticoids [7,17]. Plasma clearance of cortisol is markedly reduced during critical illness, due to suppressed expression and activity of the primary cortisolmetabolizing enzymes in the liver and kidney. Furthermore, despite the elevated cortisol levels during critical illness, tissue resistance to glucocorticoids is believed to occur because of insu cient glucocorticoid receptor alpha-mediated anti-in ammatory activity.
The clinical spectrum of COVID-19 varies from asymptomatic form to clinical conditions characterized by respiratory failure that requires mechanical ventilation to sepsis, septic shock, and MODS. Direct injury to the lung tissue from a viral infection-mediated local in ammatory response is one of the proposed mechanisms behind the pulmonary manifestations of COVID- 19 [18]. Cytokine storm syndrome (CSS) is an accentuated immune response to viral infection. CSS is characterized by unremitting fever and multiorgan involvement, including ARDS and acute cardiac and renal injury.
People with COVID-19 and ARDS have classical serum biomarkers of CRS, including elevated CRP, LDH, D-dimer, and ferritin [19]. This clinical history is similar to other infections resulting in ARDS and respiratory failure. For this reason, adrenal gland dysfunction is possible in patients with COVID-19. This study is an example for this relationship.
Proposed mechanisms for the development CIRCI are ful lled in COVID-19. This study, for the rst time, demonstrated the possibility of compromised adrenal function in severe COVID-19 treated in ICU. CIRCI is commonly seen in patients admitted to intensive care units for different reasons. CIRCI occurred in 54.3% of the patients with multiple injuries [20] and 0-48% in severe community-acquired pneumonia [21]. Mortality among patients with CIRCI was higher than normal adrenal function especially patients with abnormal response to cosyntropin had a higher mortality risk [20]. Similar nding was also observed in our study. Patients who had basal cortisol level less than 10 mcg/dl had a mortality rate 28.0% while patients who had abnormal response to cosynthropin (< 9 mcg/dl increase) had a mortality rate of 57.1%. This shows that baseline cortisol is not enough to assess adrenal function and response to cosyntropin may have a predictive role in assessing mortality.
In contrast to low cortisol levels in CIRCI group, ACTH level was inappropriately low or below 10 pg/ml indicating suppression at hypothalamic or pituitary level. Cortisol insu ciency during critical disease may result in structural damage in any point of the HPA axis from hypothalamus to adrenal gland itself. In physiological states, acute phase of stress causes and increase in cortisol level and either due to in ammatory reaction or negative feedback suppression ACTH levels were found relatively low. If patients survive the acute phase of disease but do not recover from the disease, they will enter chronic phase of critical disease, which is characterized by atrophy of the adrenal gland due to inadequate stimulation, by the pituitary with a resultant secondary adrenal failure [22].
According to WHO data on 31st May 2020 con rmed cases of COVID-19 throughout the world was 5.934.936 with 397.166 deaths and mortality rate of 6.6%. Main reasons for mortality were hypoxic respiratory failure, MODS and septic shock. Independent from each other every etiology has a potential to induce CIRCI [17,23]. It may be speculated here that CIRCI might play role in at least some of those deaths. In our study, mortality rate was comparable in patients with and without adrenal dysfunction. This was probably related to the corticosteroid replacement after diagnosis of adrenal insu ciency. This shows that detection of adrenal insu ciency early after admission to intensive care unit and proper treatment may be lifesaving.
Except age, all other baseline demographic characteristics including comorbidities, drug use, need for mechanical ventilation in the form of intubation or NIMV and severity of pulmonary in ltration were similar in patients with or without adrenal insu ciency. Patients with adrenal insu ciency were younger like in previous studies [24]. Prognosis in patients with adrenal insu ciency was worse in subgroup of delta cortisol below 9 mcg/dl compared to basal cortisol below 10 mcg/dl both in ours and previous studies [24]. In delta cortisol group, patients were even younger than basal cortisol and normal adrenal function group. These patients also had higher CRP and LDH levels compared to basal cortisol group. These ndings indicate a more severe disease in this subgroup. Even they were treated with corticosteroid; mortality was higher than basal cortisol group but similar to normal adrenal function group. Only patients who had a basal cortisol level below 10 mcg/dl had a bene t over mortality according to our results if they are treated with steroid.
COVID-19 was related to increase in in ammatory (CRP, Fibrinogen, LDH, ferritin, leukocytosis, lymphopenia), thrombotic (D-dimer, thrombocytopenia, hypertriglyceridemia), and myocardial injury (Troponin-I and CK) markers [3]. These are speci c markers to the disease itself and there was no difference in any parameters. But, in subgroups, CRP and LDH level were higher in patients with adrenal insu ciency diagnosed with delta cortisol level compared to basal cortisol level. This nding might indicate a more severe disease in this group. This may explain why the mortality rate was higher in those patients.
Comorbidities were more likely detected in severe patients [25]. In Chinese case series, preexisting comorbid conditions increased mortality by 5.6-10.5% [5]. Also, according to Italian Higher Health Institute (Istituto Superiore di Sanità) report, patients without any comorbidities accounted for only 2.8% of those died from COVID- 19 [26]. On the other hand, although 79.8% of our study population had at least one comorbid disease, mortality was not increased. The presence of disease may be a confounding factor in previous studies since other factors such as disease treatment and control rate before hospitalization, severity or grade of the disease and in-hospital treatment are also important. There were no data about them both in previous studies and ours.
There is no standard treatment of COVID-19. Corticosteroids were widely used during outbreaks of severe acute respiratory syndrome and Middle East respiratory syndrome. But their e cacy is controversial. Although critical patients were more likely to require corticosteroids therapy, its use was associated with higher mortality, longer length of stay, a higher rate of bacterial infection and hypokalemia [27}. For this reason, empirical treatment with corticosteroids may be harmful in COVID-19. Instead, assessment of adrenal function and proper treatment makes more sense.
There are some limitations of this study. First, study includes limited number of patients. But, by being the rst example, sample size is acceptable. Second, we used 1 mcg cosynthropin test instead of 250 mcg. Standard recommendation was 250 mcg cosyntropin test. But comparative studies report equal value of both tests for the evaluation of critical cases. We can estimate that if we used 250 mcg test, the prevalence possibly would be higher. Third, our mortality rate was similar to patients with intact adrenal function, which we attribute it to corticosteroid replacement. We did not have an arm that is not treated for adrenal insu ciency due to ethical reasons. So, we don't know whether mortality will be higher if not treated. Four, although chronic medications that patients were using and the ones that started for the treatment for COVID-19 as far as we know does not in uence adrenal function, separately, we cannot exclude the interaction of polypharmacy.
In conclusion, adrenal insu ciency is very common in COVID-19. Treatment of these patients with corticosteroids help to decrease mortality to the level of normal adrenal function patients. For this reason, all patients admitted to intensive care unit whether the clinical picture resembling adrenal insu ciency or not has to be assessed for adrenal function and if diagnosed treated properly.

Declarations
Con icts of interest/Competing interests All of the authors here declare that there is no con ict of interest that could be perceived as prejudicing the impartiality of the research reported; or declare any nancial or other potential con ict of interest.

Funding
This research did not receive any speci c grant from any funding agency in the public, commercial or notfor-pro t sector.

Availability of data and material
Data about the results obtained are available and ready for evaluation in part of transparency.  Table 3. Laboratory results