Hypothalamic-Pituitary Axis Function and Adrenal Insufficiency in COVID-19 Patients

COVID-19, which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), identified about 4 years ago when the first case was confirmed in Wuhan, China, has already affected more than half a billion people worldwide and caused more than 6 million deaths [1]. Although SARS-CoV-2 primarily affects the lungs, it is considered a multisystemic disease that involves many organs in the human body [2]. The disease course can be asymptomatic in some patients but may also progress to septicemia and septic shock, resulting in severe organ failure and death [3].

SARS-CoV-2, which enters the body through the respiratory system, causes organ dysfunction via angiotensin-converting enzyme 2 (ACE-2) receptors [4]. The presence of ACE-2 receptors in many endocrine glands such as the pituitary, adrenals, thyroid, pancreas, testes, and ovaries brings endocrine disorders during SARS-CoV-2 [5, 6]. Examples include hypothalamic-pituitary-adrenal (HPA) axis insufficiency, thyroid dysfunction, endocrine pancreas damage, and decreased sex steroids [7, 8]. Apart from organ-specific ACE-2-mediated viral entry and damage, direct viral toxicity and cytokine storm mediated by various cytokines are also known to cause endocrine dysfunction [7]. Additionally, transmembrane protease serine 2 (TMPRSS2) is another cell entry protein for SARS-CoV-2, and it is regulated in response to androgens via transcriptional regulation by the androgen receptor. This may partially explain that men are more likely and severely infected by SARS-CoV-2 than women. In accordance with association between androgens and TMPRSS2, antiandrogens have been considered a potential antiviral therapy in COVID-19 disease; however, clinical studies have given inconclusive results on this topic [9].

In 2003, after the SARS outbreak, cases of pituitary insufficiency were identified, suggesting that SARS-CoV-1 may also affect the pituitary gland [10]. Furthermore, autopsy examinations revealed the presence of this virus in the pituitary of patients with SARS-CoV-1 [11]. SARS-CoV-1 and SARS-CoV-2 are members of the same virus family, sharing 90–99% homology in their proteins. Therefore, SARS-CoV-2 can affect the HPA axis either directly or through an immune-mediated mechanism [12]. On the other hand, the fact that SARS-CoV-2 can cause HPA axis disorders has been clearly demonstrated by recent studies addressing the increased mortality associated with elevated morning basal cortisol and the beneficial use of dexamethasone in the treatment of COVID-19 [13, 14].

Impairment of the HPA axis has been reported during SARS. Leow et al. [10] reported hypocortisolism in 39.3% of SARS survivors as a late complication of the viral infection, of whom 62.5% completely improved within a year. The authors assumed that hypothalamo-pituitary dysfunction was caused by the SARS-CoV-1 infection itself, as the patients had been healthy without any previous glucocorticoid use.

Several case reports and autopsy findings confirmed the detrimental effects of SARS-CoV-2 over the HPA axis [15‒19]. Gonen et al. [20] investigated the frequency of central adrenal insufficiency (AI) in patients with COVID-19 and observed a prevalence of 8.2%. In another study, Urhan et al. [21] reported insufficient cortisol response in 16.2% and 9.3% of the patients when tested with low-dose ACTH stimulation test (LDST) and glucagon stimulation test (GST) at least 3 months following the infection, respectively. The results of large prospective studies are awaited for epidemiologic data regarding HPA dysfunction secondary to COVID-19.

Physiological stress caused by conditions such as infection, trauma, surgery, sepsis, and critical illness leads to activation of the HPA axis, decreased cortisol metabolism and cortisol-binding proteins, and consequently increased serum cortisol [22]. Hypercortisolemia triggers many neuroendocrine-immunological adaptive changes in the body, leading to a stress response [23]. However, a number of patients may exhibit relatively low cortisol levels during severe illness. Especially in critical illnesses such as sepsis and septic shock, the adrenal response is insufficient in a range between 15 and 61% of cases, and the increased cortisol demand cannot be provided or peripheral resistance to cortisol may develop [24]. Inadequate cortisol supply cannot be explained by serum cortisol levels solely and the tissue response to cortisol is also very important. It is well known that the effects of cortisol are mediated by glucocorticoid receptor (GCR) expression and affinity in the target cells. Glucocorticoid resistance, which is defined as decreased GCR expression and derangements in post-receptor events, has paramount importance in critically ill patients. In several studies, either decreased GCR expression or reduced affinity of glucocorticoids to cortisol have been shown and determined local cellular/tissue cortisol concentrations which is quite important in critically ill patients [25‒27]. Additionally, by the action of two isoenzymes of 11-β-hydroxysteroid dehydrogenase type 1 and type 2, intracellular glucocorticoids may apparently differ from the blood levels, and these two isoenzymes influence local cortisol concentrations [25‒27].

There is integrity and an adapted response of the HPA axis in the acute phase of COVID-19. The cytokine storm characteristic of COVID-19 and predictor of disease severity occurs after the first week [28]. Thus, in the later stages of infection, the cortisol response is altered, and AI dominates the picture. The main causes of AI are critical illness-related corticosteroid insufficiency (CIRCI) or functional hypopituitarism secondary to an inflammatory stress response [29]. CIRCI describes the impairment of HPA axis during critical illness, and in most of the cases, it is associated with dysregulation of the HPA axis, tissue corticosteroid resistance, and altered cortisol metabolism. In clinical practice, refractory shock unresponsive to fluid resuscitation and vasopressors might suggest the presence of CIRCI, and it has been suggested that patients with probable CIRCI should immediately be treated with glucocorticoids [30]. Moreover, current evidence indicates that there are also some changes resulting in mineralocorticoid dysfunction during critical illness, and hyperreninemic hypoaldosteronism has been demonstrated. In these patients, co-administration of fludrocortisone with glucocorticoids has been suggested in CIRCI [31]. In the literature, there are some studies addressing the hypercortisolemic response to COVID-19-associated acute stress during early period, while others report development of AI, especially in critically ill patients [13, 20, 21, 32‒35].

ACE-2 and TMPRSS2 proteins are required for SARS-CoV-2 to enter the cell, and they are widely expressed in adrenal glands [36]. However, unlike SARS-CoV-1, SARS-CoV-2 has not been isolated from adrenal glands, yet pathologic findings have been associated with the infection [37]. Autopsy reports evidenced damage to the adrenals during COVID-19. Zinserling et al. [19] reported infiltration of the adrenal cortex by CD3+ and CD8+ cells and groups of proliferating cells characterized by large clear nuclei similar to those observed in the lungs which were considered to be caused by direct action of SARS-CoV-2. Another autopsy report revealed acute fibrinoid necrosis of the arterioles in adrenal parenchyma and periadrenal adipose tissue that was associated with vessel necrosis and apoptosis [18]. Santana et al. [17] observed ischemic necrosis, cortical lipid degeneration, hemorrhage, and non-specific focal adrenalitis in 43% of deceased patients with severe COVID-19 but could not distinguish if these changes were due to the virus per se or the effects of the end-stage systemic disease.

A number of case reports revealed that COVID-19 may result in primary AI during or following the disorder. Radiologic findings revealed bilateral adrenal hemorrhage or non-hemorrhagic adrenal infarction in some cases [38, 39]. Bilateral adrenal hemorrhage has been reported previously during the course of other infectious diseases caused by bacteria such as Neisseria meningitidis, staphylococci, and gram-negative bacilli [40]. The underlying mechanism is not fully elucidated yet, but adrenal vascular anatomy with a rich arterial supply and relatively restricted venous drainage was held responsible, and therapeutic anti-coagulation might have contributed in some patients [38]. Primary AI due to non-hemorrhagic adrenal infarction was associated with prothrombotic state of COVID-19 [39]. Anti-phospholipid syndrome was co-occurent in some cases, and bilateral adrenal hemorrhage and infarction were assumed to be the consequences of the two prothrombotic diseases [41]. Autoimmune primary AI was observed in another case report following COVID-19, and the authors discussed that SARS-CoV-2 infection contributed to the progression of the autoimmune disease [42].

Symptoms such as chronic fatigue, nausea, joint pains, headache, and brain fog persisting more than 4 weeks after the initial infection in the absence of other diseases are described as long COVID, and most of these signs and symptoms mimic AI. It is well known that COVID-19 might predispose patients to CIRCI, and it may persist although patients recover from the SARS-CoV-2 infection. This extended period of AI may result from persistence of HPA axis dysfunction or an autoimmune response against key factors involved in adrenocortical cell function. Moreover, high titers of anti-ACTH antibodies were demonstrated in long COVID-19 patients indicating a different pathophysiological mechanism in HPA axis dysfunction [43]. Additionally, a number of patients require glucocorticoid therapy over an extended period of time, and secondary AI should also be considered [44].

Therefore, clinicians should keep in mind primary AI as a possible differential diagnosis when dealing with critically ill patients diagnosed with COVID-19. The clinical manifestations may be attributed to COVID-19 infection, and undiagnosed primary AI may have lethal consequences.

The pituitary gland has a rich blood supply and is highly sensitive to hypoxia, ischemia, and hypovolemia [45, 46]. Pathological conditions such as encephalitis, meningitis, ischemic stroke, trauma, and sepsis are known to cause hypothalamo-pituitary dysfunction through vascular damage, autoimmune and inflammatory changes [45, 46]. Similarly, it can be assumed that SARS viruses exert their effects on the hypothalamus and pituitary gland either by directly invading and damaging these organs or through a systemic inflammatory response mediated by cytokines and inflammatory mediators.

Furthermore, the presence of central hypothyroidism and hypocortisolemia has been demonstrated in SARS outbreak survivors, and autopsies of SARS-CoV-1-infected patients showed the presence of this virus in the pituitary [10, 11]. Therefore, SARS-CoV-2 may cause hypocortisolemia, possibly through direct viral action or through autoimmune mechanisms in the pituitary [12]. On the other hand, mimicry of host ACTH by viral antigens may lead to binding of viral antibodies to host ACTH, causing secondary hypocortisolemia [47].

SARS-CoV-2, a virus from the same family as SARS-CoV-1, is assumed to cause AI by these similar mechanisms. For example, COVID-19 behaves like a systemic vasculitis, causing microvascular damage, endothelial dysfunction, and thromboembolic hypercoagulability [48], leading to ischemic hemodynamic changes in the hypothalamic-pituitary region through direct viral effect [49]. On the other hand, as an example of immune-mediated effects of the virus, in a previous study, we demonstrated a significantly increased risk of hypopituitarism after bacterial and/or viral meningitis and proposed that pituitary dysfunction may be associated with the presence of anti-pituitary antibodies (APA) and anti-hypothalamic antibodies (AHA) [45]. In another recent study, we reported the presence of APA and AHA in patients with COVID-19 who had AI [20].

In Table 1, for the first time in the literature, in a short report, Tan et al. [13] described that patients with COVID-19 had higher cortisol levels than patients without COVID-19 and that hypercortisolemia was associated with the rate of mortality. In addition, the authors did not report AI during the acute phase of SARS-CoV-2 infection. They attributed this to the collection of samples in the first 48 h [13]. Cortisol, a marker of disease severity, is likely to be higher in a more severe disease state [50], but the disease severity scores were not evaluated in the study. In this preliminary study, as a single cortisol measurement was performed within the first 48 h after hospitalization, regardless of the COVID-19 severity, it may not reflect the true predictive potential of cortisol.

There are limited data regarding COVID-19 and adrenal androgens. Recently, Sezer et al. [51] investigated adrenal hormones in 153 patients diagnosed with COVID-19 with various disease severities by using highly sensitive method liquid chromatography-tandem mass spectrometric method (LC-MS/MS). The authors found that dehydroepiandrosterone and dehydroepiandrosterone-sulfate levels were higher in moderately severe patients, while they decreased as the severity of COVID-19 increased, possibly due to suppression of HPA axis.

Similar to Tan et al.’s [13] study, we reported that the basal cortisol levels were higher in patients with COVID-19 than in controls [20]. In this study, we also detected that ACTH levels were within the normal range in all participants. In another study investigating pituitary functions at least 3 months after acute COVID-19 infection, the ACTH levels were significantly higher in the patient group [21]. This increased ACTH levels may be attributed to the recovery period of “relative AI” seen in critically ill patients. Yoshimura et al. [34] reported a patient with ACTH and GH deficiency developed after recovery from COVID-19. Initially low ACTH and basal cortisol values (1:3 pmol/L [6.1 pg/mL] and 226 nmol/L [8.2 μg/dL], respectively, suggestive of secondary AI) in this patient increased to normal reference limits (3:5 pmol/L [15.8 pg/mL] and 538 nmol/L [19.5 μg/dL]) after 15 months of follow-up [34]. On the other hand, in a study evaluating basal cortisol levels in intensive care unit (ICU) patients, the authors reported lower cortisol levels in patients with COVID-19 [36]. They suggested that the adrenal cortex which is the target organ for the virus in critically ill patients may be the cause of low random cortisol levels [36].

It should be kept in mind that in critically ill patients, inflammatory mediators activate the HPA axis at all levels, aiming to protect the organism via activation of gluconeogenesis, modulation of inflammatory response, and maintenance of intravascular volume. Throughout the critical illness, cortisol levels remain high, but serum ACTH levels tend to decrease, suggesting alternative pathways, mainly activated by inflammatory cytokines, on this regulation. In contrast to healthy individuals, cortisol level in a critically ill patient has been affected by various metabolic alterations such as a reduction in cortisol metabolism, prolongation of cortisol metabolism due to renal dysfunction, changes in CBG and albumin concentrations causing an increase in free cortisol levels, and alterations in GCR functions. These changes are dynamic rather than constant changes which precludes a clear diagnosis of AI in critically ill patients [54].

Furthermore, there are also some other studies predicting AI according to different basal cortisol cut-off values. Etoga et al. [52] found that the rate of AI with basal cortisol ≤414 nmol/L (15 μg/dL) was 86.3%; with basal cortisol ≤138 nmol/L (5 μg/dL), it was 11.2% of patients. However, no dynamic tests were performed for the diagnosis of AI, and disease severity was not considered either. In another study in which ACTH and cortisol levels were randomly measured, the authors suspected AI in 9 (32%) patients with basal cortisol level <298 nmol/L (10.8 μg/dL) [32]. A robust response in cortisol level was not observed in any patient. Moreover, basal cortisol levels were close to the lower end of the reference limits unlike the expectations of a robust HPA axis response during the acute phase of infection. On the other hand, in the same study, ACTH levels were at the lower end of the normal range, consistent with secondary HPA dysfunction [32]. In this study, although a dynamic test was not performed for the diagnosis of AI, thrice repeated basal cortisol measurements may have diagnostic value. In support of this, the study by Gibbison et al. [55] showed that pulsatile release of both ACTH and cortisol was preserved in critical illness. Therefore, a single pointwise measurement of cortisol in critical illness may not truly reflect the HPA axis’ functions. For example, in the presence of intact pulsatility, different results could be obtained if a single sample was taken 1 h earlier or later [55].

The association between COVID-19 disease severity and basal cortisol level has also been studied [32, 35, 51, 53]. Alzahrani et al. [32] demonstrated that although median cortisol levels were similar between the groups of mild, moderate, and severe COVID-19 disease, more patients were detected to have AI in the severe group. Similarly, in another study, patients with moderate-severe disease had a higher rate of AI (38.5 vs. 6.8%) and lower ACTH levels (3.6 pmol/L [16.3 pg/mL] vs. 7.1 pmol/L [32.1 pg/mL]) compared to patients with mild disease [53]. On the contrary, Kumar et al. [35] reported only 3 out of 34 cases with AI in the severe disease group, and basal cortisol levels were higher in patients with severe disease than in other groups. The demonstration of necrosis, hemorrhage, and inflammation indicating direct viral involvement of SARS-CoV-2 in the adrenal glands may explain the lower cortisol levels in individuals with severe disease [17].

Finally, it is important to keep in mind that several issues related to basal cortisol measurement may affect the reliability of the results. More than 90% of cortisol circulating in the blood is bound to proteins, and changes in the levels of binding proteins can alter the measured total cortisol without affecting the free fractions. In a previous study, about half of critically ill patients had low total cortisol levels despite lack of AI [56]. Although measurement of free fractions is not routinely used, this should not be overlooked. On the other hand, in the literature, cortisol level was generally measured by electrochemiluminescence immunoassay (ECLIA) [13, 20, 21]. These non-extraction methods are susceptible to interference and have poor sensitivity and specificity. In some specific situations, a LC-MS/MS would be preferable to obtain more reliable results [51]. However, it is not widely available.

To date, only a few studies have investigated AI in COVID-19 patients using a dynamic test rather than basal cortisol measurement [20, 21, 34, 35]. Previously, Urhan et al. [21] investigated pituitary functions at least 3 months after acute COVID-19 infection. In that study, peak cortisol response higher than 345 nmol/L (12.5 μg/dL) was considered sufficient for LDST. This cut-off value for LDST was compatible with an adequate cortisol response to standard-dose ACTH test [57]. As a result, they demonstrated that the peak cortisol values in LDST were insufficient in 7 (16.2%) out of the 43 patients [21].

In another study, Gonen et al. [20] investigated neuroendocrine changes in patients with COVID-19, particularly the presence of AI by using LDST. The minimum peak cortisol level of healthy controls after LDST was accepted as a cut-off level for the assessment of AI in patients with COVID-19. AI was detected in 4 (8.2%) patients and disappeared in 2 patients who survived after about 6 months during follow-up [20]. Similarly, Clark et al. [58] found that adrenal function was preserved 3 months after admission with COVID-19. In contrast, in a case report, an insufficient response to LDST in a patient with COVID-19 persisted with AI 3 months after discharge [34].

Diagnosis of AI is a matter of debate in critically ill patients, and Annane et al. [29] did not recommend the use of LDST due to insufficient evidence. They suggested basal cortisol levels less than 276 nmol/L (10 μg/dL) as a diagnostic threshold for CIRCI. Since the use of LDST in critically ill patients is controversial, we excluded ICU patients in our two previous studies. In the context of COVID-19, in non-ICU patients, CIRCI is not considered a common problem, but more data are needed.

To the best of our knowledge, so far, only one study has used the high-dose (250 μg) ACTH stimulation test (HDST) to detect AI in COVID-19 patients [35]. In the study by Kumar et al. [35], AI was assessed by 1 h after administration of intramuscular 250 μg cosyntropin. They used <497 nmol/L (18 μg/dL) after HDST as a cut-off value for the diagnosis of AI and found AI in 64.7% of patients. The CIRCI was present in 18.3% of patients defined by a delta cortisol value of <248 nmol/L (9 μg/dL) during HDST.

Previously, it was reported that both LDST and HDST may be used with appropriate cut-off cortisol values for the diagnosis of AI [57]. Therefore, testing with high doses may not be necessary; moreover, subclinical AI may be masked with supraphysiologic doses of 250 μg ACTH.

As far as we know, the GST was only analyzed by our group to identify AI in COVID-19 patients [21]. In this study, a cortisol response lower than 295 nmol/L (10.7 μg/dL) was considered an inadequate response to GST, and four (9.3%) out of the 43 patients showed an inadequate cortisol response to GST. On the other hand, in a case report by Yoshimura et al. [34], ACTH and cortisol responses to insulin tolerance test (ITT) were markedly blunted 3 months after discharge, and it was consistent with secondary AI. In addition, 12 months after discharge, both ACTH and cortisol responses to ITT were still inadequate but had partially improved since the last test which may suggest a reversible dysfunction in the HPA axis [34].

For the first time, Tan et al. [13] reported an association between mortality and basal cortisol value ≥745 nmol/L (27 μg/dL) in patients with COVID-19. They compared the cortisol levels in the first 3 days of admission with their fate during the following 42 days. The median survival was 36 days in patients with basal cortisol levels <745 nmol/L (27 μg/dL) and 15 days in patients with basal cortisol levels ≥745 nmol/L (27 μg/dL), and it was concluded that doubling of serum cortisol levels was associated with increased mortality by 42%. On the contrary, Ahmadi et al. [33] reported that a one-unit increase in cortisol level was associated with a reduction of 26% in the risk of death.

As expected, patients with high basal cortisol levels had lower median survival times [59]. Similarly, in critically ill patients with severe sepsis or septic shock without COVID-19, the survivors had lower basal cortisol levels than non-survivors (704 nmol/L (25.5 μg/dL) versus 979 nmol/L (35.5 μg/dL) [60]. In another COVID-19 study, authors demonstrated that basal cortisol levels were significantly lower among survivors than non-survivors (370 nmol/L [13.4 μg/dL] versus 690 nmol/L [25 μg/dL]) [35]. These findings may reflect the physiological response of the HPA axis with a more severe disease leading to higher cortisol levels.

Patients with AI are more prone to infections when compared to a healthy population. Inadequate dosage of glucocorticoids and significantly decreased natural killer cell cytotoxicity in this patient group were reported to be among the factors increasing the risk of infection [61, 62]. Moreover, patients with AI (primary or secondary) had higher mortality rates from infectious diseases [63]. Increased incidence of COVID-19 infection and higher hospitalization rates were observed in patients with AI when compared to the global population [62]. The most dreaded complication is adrenal crisis since 6% of adrenal crises resulted in death [64]. There are no data reporting death rates in patients with AI or adrenal crises during COVID-19.

Guidance has been developed for the management of patients with AI during COVID-19 [65]. Asymptomatic patients were recommended to continue with regular daily glucocorticoid regimen, while patients with fever, fatigue, cough, and body pains were recommended to take 20 mg hydrocortisone 4 times daily. In cases of clinical deterioration, the dose of hydrocortisone is recommended to be increased to 50 mg four times daily [65]. It has been recommended to divide doses into equal periods rather than doubling the daily doses as the blood concentration of hydrocortisone is more stable in cases of periodic therapy. Hospitalized patients are recommended to be replaced with major stress doses, 100 mg intravenous hydrocortisone and continuous infusion of 200 mg hydrocortisone/24 h [65].

Glucocorticoids are known to have immune-modulatory effects, and there are studies investigating the effects of steroids on the course of infectious diseases and ARDS [66, 67]. The clinical trials evidenced the essential role of glucocorticoids during the treatment of severe COVID-19 infection. RECOVERY trial demonstrated a decrease in 28-day mortality rates with dexamethasone therapy (6 mg/day up to 10 days) in patients who needed oxygen support, with no beneficial effects in those without need for respiratory support [14]. A prospective meta-analysis confirmed the decrease in 28-day mortality with glucocorticoid use in critically ill patients [68]. These data changed the management of severe COVID-19, although glucocorticoid therapy was not previously recommended [69].

The timing of glucocorticoids is of utmost importance during the treatment of SARS-CoV-2 infection. The glucocorticoid treatment may detrimentally affect the innate immune response and cause an increase in viral replication during the early phase, while it may have life-saving effects when applied during cytokine storm [14, 37]. The generally preferred glucocorticoid treatment protocol during the pandemic was from the RECOVERY trial; however, there is need for further evidence to determine the optimal duration and dose of glucocorticoids for COVID-19.

The short course of corticosteroid treatment was considered relatively safe, but clinicians should keep in mind the possibility of HPA axis suppression with prolonged use of steroid therapy. The prevalence of tertiary AI was 14.7% in patients treated with glucocorticoids due to COVID-19 organizing pneumonia in a study [70]. To the best of our knowledge, this was the only study reporting the prevalence of HPA axis suppression following glucocorticoid treatment in COVID-19 patients, and there is need for further studies.

For ACTH and basal cortisol measurement, there is no consensus on the cut-off plasma cortisol value during acute disease below which AI should be considered. In comparison to healthy individuals, critically ill patients have altered cortisol metabolism, regulation, and carrier protein levels. These confounding factors may lead to misinterpretation of AI in critically ill patients. Thus, values ranging from 276 to 993 nmol/L (10–36 μg/dL) have been proposed as cut-off values for “relative AI” in critically ill patients [29, 71]. Some researchers have proposed 414 nmol/L (15 μg/dL) [72], and others have proposed 497 nmol/L (18 μg/dL) [73]. A cut-off basal cortisol value of less than 414 nmol/L (15 μg/dL) for mild stress or less than 690 nmol/L (25 μg/dL) for severe stress has been suggested, according to the severity of critical illness [74]. So, it is clear that this issue is still controversial. There are different cut-off values referenced in many studies, and some authors have even stated in their study’s methodology that they have chosen the cut-off values arbitrarily [32].

Furthermore, since the pulsatility of ACTH and cortisol might not be impaired in critical illness, repeated measurements are more useful than single cortisol measurements [55]. Also, changes in the levels of cortisol-binding proteins can alter the measured total cortisol without affecting the free fractions [56]. This should be taken into account, especially during critical illnesses. If available, the LC-MS/MS method should be used to obtain more reliable results [51].

For the ACTH stimulation tests, it is controversial to diagnose AI in critically ill patients using ACTH stimulation tests, and various stimulated cortisol levels have been used to define CIRCI [22, 29]. But it has been suggested that it may be more beneficial than random basal cortisol [75]. Additionally, delta cortisol (post-stimulated cortisol minus basal cortisol of <248 nmol/L [9 μg/dL]) was also considered to define CIRCI as recommended by current guideline [29]. On the other hand, given the unreliability of all these assays in patients with CIRCI, most clinicians in routine clinical practice prefer hydrocortisone replacement as a test-therapeutic approach, especially in patients with septic shock refractory to fluid resuscitation [71]. However, this benefit is likely to be related to the treatment of CIRCI per se rather than the anti-inflammatory and similar effects of corticosteroids [13].

In non-critically ill patients, it is a matter of debate whether the 1 μg or 250 μg ACTH stimulation test should be used for the diagnosis of central AI [57, 76, 77]. We prefer the use of LDST in our clinical practice, as it was more concordant with ITT when compared to 250 μg ACTH [78]. However, the role of ACTH stimulation test in critically ill patients is highly debated due to its poor reproducibility and high distribution volume. Moreover, cortisol response to ACTH test may be diminished due to maximally stimulated HPA axis during critical illness. Although the validity and usefulness of LDST have not been evaluated extensively and specifically in acute COVID-19, a recent study has evaluated the value of LDST in 39 patients with CIRCI due to sepsis. The authors concluded that LDST may be useful in sepsis patients presenting as CIRCI [79]. On the other hand, one of the main limitations of 250 μg ACTH stimulation test is the possibility of false-positive results due to supra-physiological dose [20].

COVID-19 may result in AI; thus, routine screening of adrenal functions in these patients should be practiced. Although the tests used in screening are controversial, the preference of LDST, especially as a dynamic test, or easily applicable repeated basal cortisol measurements would be a guide for clinicians for assessment of AI in patients with COVID-19.

The authors declare no conflicts of interest that could be perceived as prejudicing the impartiality of the research reported.

This research did not receive any specific grants from any funding agencies in the public, commercial, or not-for-profit sector.

E.D., A.H., M.S.G., and F.K. participated in planning and conducting the study, including producing and analyzing data and writing the manuscript. E.D., A.H., and Z.K. were responsible for statistical analysis and language modification. Z.K. and K.U. assisted with planning of the study and data interpretation. E.D., A.H., and K.U. contributed to the conception and design of the study and contributed to data production and interpretation. M.S.G. and F.K. conceptualized the study, supervised the study, and contributed to interpretation of data. All authors participated in critical revision of the manuscript and approved the version to be published.

Emre Durcan and Aysa Hacioglu have equally contributed to this manuscript.