INTRODUCTION

Severe acute respiratory syndrome-Coronavirus-2 (SARS-CoV-2), the etiologic agent of Coronavirus disease-2019 (COVID-19), might cause endocrine dysfunction on a wide range, including pituitary, thyroid, adrenal, gonadal, and pancreatic abnormalities¹. Among these, perhaps thyroid hormone dysfunction, in the form of euthyroid sick syndrome (ESS), is the most common abnormality.

It has been reported that COVID-19 might cause thyroid dysfunction due to direct or indirect effects of the SARS-CoV-2 virus on the thyroid gland. This might be in the form of the ESS owing to severe systemic disease or may take the form of acute inflammation due to autoimmune reaction². Several case reports demonstrated the development of subacute thyroiditis after COVID-19^3-5. COVID-19 was also speculated as a trigger for autoimmunity and resultant Graves’ disease in some patients, as well⁶. Several studies evaluated ESS and its severity on the prognosis of patients with COVID-19. These studies showed an increased prevalence of impaired thyroid function and an association between ESS and COVID-19 severity and level of inflammatory parameters^7-9.

Critical illness places great physical stress on the organism, and systemic availability of cortisol is a coping strategy with this stress. Several randomized studies produced conflicting results with respect to the impact of so-called relative adrenal failure-later renamed critical illness-related corticosteroid insufficiency-on clinical outcomes in patients with sepsis¹⁰. Subsequent research revealed the cause of increased systemic availability of cortisol, which was not due to increased adrenal secretion but rather owing to decreased degradation in the liver and kidney and less binding to cortisol-binding proteins in plasma¹¹. Because of increased hospital and intensive care unit (ICU) admission rates of COVID-19 patients and dire outcomes for some, several investigators also evaluated the role of baseline levels of adrenocorticotropic hormone (ACTH) and cortisol on clinical outcomes and mortality in COVID-19 patients. Kanczkowski et al.¹², in their autopsy study, found that though not causing overt adrenal insufficiency, adrenal glands are a frequent target of SARS-CoV-2 and ensuing inflammation in deceased COVID-19 patients. A few case reports described cases of new-onset primary adrenal insufficiency following COVID-19^13-15. Moreover, Mao et al.¹⁶ found the level of serum cortisol significantly lower in ICU patients with COVID-19 compared to ICU patients without COVID-19. A very recent meta-analysis investigated the effect of serum cortisol levels on the prognosis of COVID-19 patients. The authors found significantly higher levels of serum cortisol in patients with severe COVID-19 compared to patients with moderate and mild COVID-19 patients¹⁷. Several treatment options were tried in COVID-19 patients with the hope of reducing mortality. Corticosteroids were one of the most promising among these potential treatment candidates. A meta-analysis by Boppana et al.¹⁸ included 6 randomized controlled trials on 7 thousand patients with COVID-19 who were treated with corticosteroids and sought to answer whether corticosteroids were beneficial in reducing mortality. The analysis revealed that the use of systemic corticosteroids was associated with a reduction in all-cause mortality in patients with COVID-19 who required oxygen therapy or mechanical ventilation.

To the best of our knowledge, no study in the literature evaluated the combined power of thyroid and adrenal axes in the prediction of adverse clinical outcomes in patients with COVID-19. Thus, we aimed to evaluate the predictive ability of thyroid function along with serum cortisol and ACTH levels on clinical outcomes in COVID-19 patients who required ICU admission.

MATERIALS AND METHODS

Patients and Setting

This is a retrospective study that aims to evaluate the prognostic effects of admission serum cortisol level and thyroid function tests on the prognosis of patients admitted to a COVID-19 intensive care unit. The study was carried out in a teaching and research hospital in Karaman, Turkey. All patients who were admitted to the ICU due to COVID-19 were screened for eligibility for the study between March 11, 2020, and December 31, 2020. All patients had either a positive SARS-CoV-2 polymerase chain reaction (PCR) test from nasal swabs and/or a compatible chest computed tomography (CT) along with COVID-19 contact history. Exclusion criteria included the use of thyroid hormone therapy, the presence of hypothalamic-pituitary gland dysfunction history, having the history of radiotherapy or chemotherapy in the last 6 months, the presence of thyroid surgery history and incomplete evaluation of thyroid gland function examination at admission (İndividuals with a history of thyroid disease were included in the study only if they lacked a history of active medication use or surgical intervention). Of the 95 patients who were excluded, the following reasons were ascertained: Being transfer from another service (n=5), having hypothyroidism (n=25), hyperthyroidism (n=7), change of thyroid gland revealed by chest CT scan (n=10), and nodular goiter (n=9), undergoing thyroidectomy (n=6) and radioiodine therapy (n=2), having multihormonal pituitary insufficiency (n=1), and lack of thyroid function test (n=30).

Approval was granted by the Ethics Committee of Karamanoğlu Mehmetbey University (date: 07/12/2020, no: 02-2020/04). This study was performed in line with the principles of the Declaration of Helsinki.

Data Collection

Age, sex, and comorbid medical conditions were recorded for each patient. Laboratory evaluations at ICU admission included complete blood count and differential, serum urea, creatinine, uric acid, lactate dehydrogenase, C-reactive protein (CRP), procalcitonin, and ferritin (daily COVID-19 sampling as part of hospital protocol). Serum levels of cortisol, ACTH, free triiodothyronine (fT3), free thyroxine (fT4), and thyroid-stimulating hormone (TSH) levels were measured both on admission and thereafter two times during hospitalization (weekly COVID-19 sampling as part of hospital protocol). Initial laboratory examinations were taken before patients received any treatment. Serum TSH and thyroid hormones, ACTH, and cortisol measurements were performed with the chemiluminescence method via an autoanalyzer (Advia Centaur XP, Siemens). All study laboratory measurements were performed again during the course of the hospitalization (second and third measurements in addition to admission (first measurement) values).

The ESS was described as having a decreased serum fT3 with normal or low TSH levels⁷. The normal reference values for fT3 at our laboratory were 2.7-4.3 pg/mL. Hypocortisolism was defined as having a serum morning cortisol value below 10 µg/dL¹⁹.

The patient outcomes (discharge or exitus) were obtained from patient charts. The length of ICU stay was determined for each study participant. Patients who were administered pulse prednisone as a part of the COVID-19 treatment regimen and who were receiving thyroid hormone or drugs containing steroids were also recorded. Pulse steroid was given to 46.0% of the patients as mentioned in the national COVID-19 treatment guidelines. All patients received the same glucocorticoid agent. The patients were treated with pulse steroid treatment with an intermediate-acting glucocorticoid in the form of 250 mg methylprednisolone for the first 3 days and 1 mg/kg methylprednisolone for the next 7 days. Data regarding intubation and mechanical ventilation and positive inotropic use were also gathered.

Statistical Analysis

The Statistical Package for the Social Sciences 25.0 software package (IBM, Armonk, NY, USA) was used to analyze the data of the study. A probability value of p<0.05 was considered statistically significant, and two-tailed p-values were used for all statistics.

To determine whether the continuous variables were normally distributed, the Shapiro-Wilks test, histogram, and Q-Q plot were used for continuous variables. Continuous variables were expressed as mean±standard deviation or median (minimum-maximum) depending on the distribution of the variable. Categorical variables were reported as numbers and percentages. To compare the categorical variables between the groups, the chi-square test and Fisher’s exact test were used. The Mann-Whitney U test and the independent t-test were employed to compare two-group comparisons for numeric variables. The correlation was displayed with the Spearman correlation and the point biserial correlation test. Repeated measured ANOVA or the Friedman test with post-hoc Durbin-Conover test were used to analyze repeated measurements of the variables.

We performed univariate and multivariate logistic regression analyses to determine the independent associates of mortality. According to the univariate logistic regression, among those with a p value less than 0.050, only the variables that had clinical significance were included in the multivariate logistic regression analysis test. We did not include CRP in the regression model because its confidence intervals crossed 1.000.

RESULTS

Baseline Clinicodemographic Features and Laboratory Values

In total, 124 patients (58.1% males) with COVID-19 who required ICU admission were included in the study. The mean age was 70.7±11.3 years. The most prevalent chronic medical comorbidity was hypertension (63.7%), followed by type 2 diabetes mellitus. The majority of the patients (75%) had a positive PCR test for SARS-CoV-2. During the course of the hospitalization, 74.2% of the patients required tracheal intubation and mechanical ventilatory support. Approximately half of all patients required the administration of positive inotropic agents to support blood pressure. Pulse steroid was given to 46.0% of the patients as part of the national COVID-19 treatment guidelines. Table 1 summarizes the baseline clinicodemographic features of all study patients. Admission and repeat laboratory values for hemogram, routine biochemistry, free thyroxine, free triiodothyronine, TSH, ACTH, and cortisol values are shown in Table 2. All controls were created one week apart. Controls were performed on 124 patients upon admission to intensive care (1^st), 103 patients after 1 week (2^nd), and 57 patients after 2 weeks (3^rd).

Sixty-seven patients (54%) were not given pulse steroids during their hospitalization. In relation to patients who were administered pulse steroid treatment, clinical outcomes (length of ICU stay, rates of mechanical ventilation and inotropic drug use, and mortality) were comparable in both groups.

Thyroid and Adrenal Function

During the course of the ICU stay, both fT3 and fT4 showed a gradual decrease from their corresponding admission values. And these trends were statistically significant. Serum TSH did not significantly change during ICU stay. Although repeat values of serum cortisol and ACTH values increased in all patients compared to admission values, these increases were not statistically significant (Table 2).

When admission laboratory values were evaluated in the whole cohort of patients, 89.5% of the patients had ESS. This rate was maintained during the course of the hospitalization [ESS (2^nd)=93.2%, ESS (3^rd)=93.0%] (Table 3).

We also assessed ESS frequency among deceased and survivors who were not given pulse steroid treatment. The ESS frequency was similar at all measurement points during the hospitalization between survivors and deceased patients (Table 3).

Although admission values were comparable, repeat values of serum TSH were significantly lower in patients who were administered pulse steroid compared to patients who were not. On the other hand, fT3 and fT4 measurements were not different. All ACTH measurements, including admission value, were consistently and significantly lower in patients who were administered pulse steroids compared to patients who were not. Similarly, baseline and repeat serum cortisol values were significantly lower in patients who were treated with pulse steroids (Table 4).

Hypocortisolism (serum cortisol value below 10 μg/dL) was present in 25 (20.1%) patients. Inotropic medications were given to 14 (56%) patients with hypocortisolism. The coexistence of ESS and hypocortisolism was evident in 24 (19.4%) patients. The frequency of hypocortisolism in survivors and decedents was 21.4% and 19.5%, respectively. Mortality rate was not different between patients with and without hypocortisolism. In addition, serum fT3 levels were also comparable between patients with and without hypocortisolism. In a cohort of 25 patients with hypocortisolism, 22 received pulse steroid therapy, while 3 did not. Among the 22 patients who received pulse steroid therapy, 15 (68.2%) experienced an adverse outcome (exitus). Of the 3 patients who did not receive pulse steroid therapy, 2 (66.7%) had an adverse outcome. Statistical analysis using the Fisher’s Exact test indicated no significant difference in mortality between these two groups (p=0.704).

Patient Survival and Associated Factors

Two-third of all patients died during ICU stay. When compared, survivors and decedents had comparable age and comorbid disease distribution. As expectedly, mechanical ventilation requirement and inotropic drug administration were significantly more frequent among the deceased compared to survivors. The mean admission serum fT3 and T4 levels were comparable between the survivor and deceased groups. However, repeat values taken during the course of the hospitalization were significantly lower in deceased patients compared to survivors (Table 5). On the other hand, serum TSH values at all time points were lower among deceased compared to survivors. However, only the second repeat TSH value difference reached statistical significance. There was no significant difference in serum basal cortisol and ACTH values between decedents and survivors except for second repeat serum cortisol values. Figure 1 shows changes of serum fT3 level throughout ICU stay in survivor and decedent groups.

In addition, deceased patients had significantly higher serum CRP and procalcitonin values compared to survivors.

We performed univariate and multivariate logistic regression analyses to evaluate independent predictors of mortality. In multivariate analysis, only serum fT3 level appeared as a significant predictor of all-cause mortality (Table 6). Age, comorbid conditions, serum cortisol, ACTH values, and inflammatory markers were not independent associates of mortality.

DISCUSSION

The salient findings of the present study were as follows: (1) ESS was very common among COVID-19 patients admitted to ICU (2). Serum fT3 level was comparable on admission; however, it remained significantly lower during ICU stay in decedents compared to survivors (3). Serum fT3 was the sole predictor of all-cause mortality among several factors, including inflammatory markers, kidney function, and comorbid conditions (4). One-five of our patients had hypocortisolism. However, hypocortisolism rates, serum cortisol, and ACTH levels were comparable between the decedents and survivors (5). Serum ACTH and cortisol levels were significantly different throughout the ICU stay between patients who were given and not given pulse steroids. Values apart from admission might reflect the effects of pulse steroid administration on the hypothalamic adrenal axis (6). Coexistence of ESS and hypocortisolism was evident in 24 (19.4%) patients. However, this cooccurrence did not pose a mortality disadvantage.

A study by Świstek et al.²⁰ found that ESS was present in 38.1% of COVID-19 patients, with higher levels of inflammatory markers and mechanical ventilation need. The study found that ESS increased the risk of death by 3.1 times. ESS was also found to be a prognostic factor even in mild cases. Sparano et al.²¹  found that the frequency of ESS was 57% among hospitalized COVID-19 patients. Serum fT3 level was found to be inversely related to inflammatory markers and an independent associate of mortality.

Although it might be associated with de novo development of Graves’ disease and subacute thyroiditis, ESS remains the most common thyroid pathology in COVID-19 patients with serum fT3 levels significantly lower than non-severe patients²². ESS is an independent predictor of all-cause mortality in these patients²³. Apparently, as the severity of COVID-19 increases, the frequency of ESS also increases²⁴. In our cohort of patients, the frequency of ESS on admission was 89.5%. Different from the studies published to date, we also evaluated thyroid function during the course of the ICU stay. In our patients, the rate of ESS never descended below 90% during ICU stay. However, the frequency of ESS on admission was not statistically significant between the decedents and survivors. The patients who died in ICU had significantly lower fT3 levels compared to survivors. Serum fT4 values were significantly lower in decedents during hospitalization compared to survivors, and serum fT3 was found to be an independent and significant predictor of all-cause mortality.

During the critical illness, cortisol secretion normally increases as a stress response²⁴. However, in some critically ill patients, this increase is not sufficient to meet the needs of the body. When a disorder of the hypothalamic-pituitary-adrenal (HPA) axis is present and not appropriately treated with high doses of hydrocortisone, acute stress may induce a life-threatening adrenal crisis. In the absence of adrenal insufficiency, stress doses of hydrocortisone are also used to lessen the requirement for vasopressors in patients suffering from septic shock. Research on the HPA axis during critical illness have led to the realization that neither of these disorders can be classified as “critical illness-induced corticosteroid insufficiency” or CIRCI. Instead, these results advised adopting the term CIRCI for a syndrome that may develop in individuals who are critically sick for an extended period of time. Patients who rely on organ support for weeks may develop central adrenal insufficiency due to increased systemic glucocorticoid availability, suppressed circulating cortisol-binding proteins, and suppressed hepatic/renal cortisol metabolism. This negative feedback inhibition at the hypothalamus/pituitary is exacerbated by high levels of other glucocorticoid receptor ligands and drugs like opioids. The adrenal cortex may become physically and functionally damaged, leading to inadequate cortisol production and potentially contributing to persistent vasopressor demand and encephalopathy, impeding recovery²⁵. Patients with adrenal crisis usually present with nonspecific symptoms and signs, including severe fatigue, weakness, myalgia, postural dizziness, nausea, vomiting, abdominal pain, and fever²⁶. These symptoms are also common in an acute COVID-19 infection, and it might be difficult to differentiate between the features of acute adrenal crisis and acute COVID-19 infection. Since adrenal insufficiency clinic and COVID-19 clinical findings can be confused with each other, we created a treatment plan for relative adrenal insufficiency by combining the cortisol cut-off value with clinical findings. Given the dynamic nature of the pituitary-adrenal axis, single measurements of serum cortisol cannot fully explain the adrenal reserve. Some studies accepted a serum cortisol level below 10 µg/dL as a hypocortisolism response to critical illness¹⁹. Das et al.²⁴ found that among patients with severe and mild COVID-19, hypocortisolism rates were 38.5% and 6.8%, respectively. In our study, the frequency of hypocortisolism was 20.1%, which was lower than the findings of Das et al.²⁴, because Das et al.²⁴ used 15 µg/dL as the cut-off value in their study, unlike the 10 µg/dL we used in our study. Patients with severe disease had significantly lower serum ACTH values compared to patients with mild COVID-19²⁴. Another case-control study involving noncritical patients hospitalized for COVID-19 revealed that, compared to normal subjects, serum levels of ACTH were significantly lower, and cortisol was higher in COVID-19 patients²⁷. A significant number of our patients received pulse steroid treatment as part of COVID-19 treatment guidelines. However, admission values reflected the steroid-naive status before pulse steroid administration. Hypocortisolism frequencies were comparable between the survivors and decedents. Moreover, neither serum cortisol nor serum ACTH values were independent predictors of all-cause mortality. There was no correlation between serum fT3 and cortisol values, either.

Study Limitations

Some limitations of the current study are worthy of mention. First, our sample size was not large, and this might have led to a failure to unravel subtle differences in thyroid hormone and adrenal hormone levels. Moreover, ESS was almost universal among our patients, and this precludes us from meaningfully comparing patients with and without ESS in terms of study outcomes. Second, pulse steroid administration in a significant portion of patients might affect adrenal axis evaluation during the hospitalization period. It might also have impacted the thyroid axis to some extent. However, several strengths of this study make our results contributive to the current literature. For the first time, we evaluated the combined prognostic impact of thyroid and adrenal axes on the prognosis of critically ill COVID-19 patients. Second, we also evaluated hormone values throughout the hospitalization but not at a single time point, as was the case in the previous studies in the literature.

CONCLUSION

In conclusion, ESS was prevalent in critically ill COVID-19 patients. One-five of all patients also had hypocortisolism. Although the frequencies of ESS and hypocortisolism were similar between survivors and decedents, patients who died during ICU stay had significantly lower levels of serum fT3. In addition, fT3, but not serum cortisol and ACTH levels, was an independent predictor of all-cause mortality.

Ethics

Ethics Committee Approval: Approval was granted by the Ethics Committee of Karamanoğlu Mehmetbey University (date: 07/12/2020, no: 02-2020/04).

Informed Consent: Retrospective study.

Authorship Contributions

Surgical and Medical Practices: M.K., T.E., R.Y., Concept: M.K., S.K., N.K., Design: M.K., S.K., R.Y., Data Collection or Processing: T.E., R.Y., N.K., Analysis or Interpretation: S.K., T.E., N.K., Literature Search: M.K., T.E., R.Y., Writing: M.K., S.K., T.E., R.Y., N.K.

Conflict of Interest: No conflict of interest was declared by the authors.

Financial Disclosure: The authors declared that this study received no financial support.