The Association of Thyroid Hormone Changes with Inflammatory Status and Prognosis in COVID-19

Background COVID-19 infection may have multiorgan effects in addition to effects on the lungs and immune system. Recently, studies have found thyroid function abnormalities in COVID-19 cases which were interpreted as euthyroid sick syndrome (ESS) or destructive thyroiditis. Therefore, in this study, we aimed to evaluate the thyroid function status and thyroid autoimmunity in COVID-19 patients. Material and Method. 205 patients were included. The medical history and laboratory parameters at admission were collected from medical records. Serum thyroid-stimulating hormone (TSH), free thyroxine (FT4), free triiodothyronine (FT3), thyroid peroxidase antibody, and thyroglobulin antibody were measured, and patients were classified according to thyroid function status. Results 34.1% of the patients were euthyroid. Length of hospitalization (p < 0.001), rate of oxygen demand (p < 0.001), and intensive care unit (ICU) admission (p=0.022) were lower, and none of the euthyroid patients died. 108 (52.6%) patients were classified to have ESS, 57 were classified as mild, and 51 were moderate. The inflammatory parameters were higher in patients with moderate ESS. In cluster analysis, a high-risk group with a lower median FT3 value (median = 2.34 ng/L; IQR = 0.86), a higher median FT4 value (median = 1.04 ng/dL; IQR = 0.33), and a lower median TSH value (median = 0.62 mIU/L; IQR = 0.59) included 8 of 9 died patients and 25 of the 31 patients that were admitted to ICU. Discussion. Length of hospitalization, oxygen demand, ICU admission, and mortality were lower in euthyroid patients. Moreover, none of the euthyroid patients died. In conclusion, evaluation of thyroid function tests during COVID-19 infection may give information about the prognosis of disease.


Material and Method
e study was approved by the Local Ethics Committee of Marmara University School of Medicine (11.05.2020/ethics no.: 535) and Turkish Ministry of Health.
Two hundred and five patients with reverse-transcription polymerase chain reaction-(RT-PCR-) confirmed COVID-19 who were admitted to Marmara University Education and Research (E&R) Hospital between April and October 2020 were enrolled. None of them have had previous thyroid disease, used any thyroid medications or glucocorticoids, or had pregnancy. All patients were evaluated for thyroid-stimulating hormone (TSH), free triiodothyronine (FT3), free thyroxine (FT4), thyroglobulin antibody (TGAb), and thyroid peroxidase antibody (TPOAb). eir medical history, symptoms at admission, medications, length of hospitalization, thorax computerized tomography (CT) findings, oxygenation, and vital signs were recorded.
e results for complete blood counts, alanine aminotransferase (ALT), creatinine, high sensitive C-reactive protein (hs-CRP), lactate dehydrogenase (LDH), ferritin, d-dimer, and procalcitonin values were collected from the laboratory information system. Blood samples that were taken within 48 hours of admission were centrifuged and serum aliquots were stored at −20°C. Serum TSH, FT3, FT4, TGAb, TPOAb, follicle-stimulating hormone (FSH), luteinizing hormone (LH), testosterone, and estradiol levels were measured from these samples. Written consent has been obtained from each patient after full explanation of the purpose and nature of all procedures used. e patients were classified into six categories: euthyroid, subclinical hypothyroidism, ESS, subclinical hyperthyroidism, overt hyperthyroidism, and central hypothyroidism by two investigators in a double-blinded manner as described in references [14][15][16][17] (Table 2). Seventy patients were diagnosed as euthyroid according to the reference range for TSH (0.4-4 mU/L) [14]. Four patients were diagnosed as subclinical hypothyroidism if TSH is 4-10 mU/L and FT3 and FT4 levels were in the normal reference range [14]. ESS was diagnosed if the patient had low/normal TSH, low FT3, and normal/low/high FT4. Patients with ESS were subdivided as mild, moderate, and severe according to their thyroid function tests. Low FT3, normal TSH, and FT4 were classified as ESS associated with mild disease; low FT3, normal/low TSH, and normal/low/high FT4 were classified as ESS associated with moderate disease. Low TSH, FT3, and FT4 were classified as ESS associated with severe disease [15].  Ferritin levels were measured with a two-site immunoenzymatic assay in Access Analyzer (Beckman Coulter, USA). D-dimer parameter was quantitated with an immunoturbidimetric assay in 3.2% sodium citrated venous plasma (STA Compact, Diagnostica Stago, France). hs-CRP levels were measured nephelometrically (BN Prospec, Dade Behring, Germany). TSH, FT3, FT4, TPOAb, and TGAb parameters were measured by paramagnetic particle, chemiluminescent immunoassays in serum samples (DxI800, Beckman Coulter, USA). e reference range for TSH was 0.34-5.60 mU/L, for FT3 was 2.6-4.37 ng/L (0.061-0.103 pmol/L), and for FT4 was 0.61-1.12 ng/dl (0.144-0.26 pmol/L). TGAb was 0-115 IU/ml, and TPOAb was 0-34 IU/ml. FSH and LH levels were also measured by paramagnetic particle, chemiluminescent immunoassays in serum samples (DxI800, Beckman Coulter, USA). Estradiol and testosterone levels were determined by electrochemiluminescence immunoassay (Modular Analytics E170, Roche Diagnostics, Germany).

Statistical Analysis.
e comparison of the continuous variables among independent groups was performed with Mann-Whitney U and Kruskal-Wallis tests. Consequent measurements were analyzed with the Wilcoxon test. e cross tables of categorical variables were analyzed with chisquare and Fisher's exact tests. e correlation between numerical variables was tested with Spearman's correlation test.
e uni-and multivariate binary logistic regression analyses were performed and odds ratios were reported. L2 (Euclidean) cluster analysis was performed with random start points, where k � 2 clusters were created. p < 0.05 was considered statistically significant. All analyses were executed by using Stata 15.1 software (Stata Corp, Texas 77845 USA).

Comparison of Laboratory Parameters and Outcomes of Euthyroid and Noneuthyroid Patients.
Euthyroid patients were younger (p < 0.001) and mostly female (p � 0.002). e symptoms at admission were similar in both groups. None of the euthyroid patients had died. Additionally, the length of hospitalization was shorter (p < 0.001), and the rate of oxygen demand (p < 0.001), ICU admission (p � 0.022), and mortality (p � 0.029) were lower in the euthyroid group.

Comparison of Laboratory Parameters and Outcomes of
Patients with ESS. One hundred and eight patients were categorized as ESS: 57 were mild and 51 were moderate. ey had higher levels of neutrophil count, LDH, hs-CRP, ferritin, d-dimer, and procalcitonin and lower levels of lymphocyte percent when compared to euthyroid patients. Moreover, subgroup analysis showed that age, neutrophil, and lymphocyte percent, LDH, CRP, ferritin, d-dimer, and procalcitonin levels were significantly different in moderate ESS in comparison with euthyroid and mild ESS cases ( Table 3). As clinical outcomes, the length of hospitalization was longer in the moderate ESS group in comparison with both the euthyroid and mild ESS groups (p < 0.001). More patients in the moderate ESS group needed oxygen (p < 0.001).
e mortality of the patients with ESS was significantly higher than that of the euthyroid patients (p � 0.043). e ICU demand of moderate ESS patients was significantly higher than that of the euthyroid and mild ESS groups (p � 0.001). ree of them had previous thyroid function tests and they were euthyroid. One has needed ICU admission, and nobody had died.
irteen patients were categorized as ESS and hyperthyroid (high FT4, low TSH, and FT3 levels). When all hyperthyroid patients were considered (n � 30, 14.6%), they were older (p � 0.007) and have had higher d-dimer levels (p � 0.002). Four patients (1.95%) had subclinical hypothyroidism. ey were antibody negative and 3 of them were normal before COVID-19. ey have not needed ICU. ree patients (1.46%) were categorized as central hypothyroidism with low FSH and LH levels. Among them, one patient had died.

Characteristics and Laboratory Parameters of Patients Admitted to ICU and Who Had Died.
irty-one patients were admitted to ICU. ey were older in age (p < 0.001) and their length of hospitalization was longer (p < 0.001). D-dimer, procalcitonin, hs-CRP, LDH, ferritin, and neutrophil counts were significantly higher in the ICU group together with low lymphocyte percent (Table 4). Nine patients had died. ey were also older in age (p � 0.001), and their length of hospitalization was longer (p � 0.006). eir neutrophil count, procalcitonin, and d-dimer levels were significantly higher (Table 5).
FT3 and TSH levels were significantly lower in the ICU group (p < 0.001 and p � 0.005, respectively). FT4 levels were higher, but this difference was not significant (p � 0.12). Also, in patients who died, FT3 (p � 0.0025) and TSH levels were significantly lower (p � 0.02).

Predictive Factors of Mortality and ICU Admission.
Univariate logistic regression analysis showed that age, lymphocyte percent, hs-CRP concentration, procalcitonin, and FT3 levels had a significant relation with mortality (Table 6). Age, length of hospitalization, respiratory rate, basal neutrophil count, ferritin, CRP, LDH, D-dimer, TSH, FT3, and FT4 had a significant relation with ICU admission (Table 6).

Evaluation of yroid Function Tests according to Previous
Tests. When previous tests of patients within the last 12 months were evaluated, current median TSH levels were lower than previous levels, but there was not a significant difference (n � 90, p � 0.058). Additionally, current FT3 levels were significantly lower (n � 34, p < 0.001) while for 69 patients, the current FT4 levels were significantly higher (p < 0.001). 32 patients had follow-up TSH levels and there was a statistically significant increase (p � 0.046), 16 patients had FT3 and there was not a significant increase (p � 0.055), and 29 patients had FT4 levels and there was a statistically significant decrease (p � 0.01) according to levels during infection. 10 patients who had euthyroid sick syndrome were euthyroid in control tests. 11 patients who were euthyroid were euthyroid in control tests, 2 patients who were euthyroid were subclinical hypothyroid in follow-up. 2 patients who were hyperthyroid were euthyroid in control tests. From two patients who were subclinical hypothyroid during COVID-19 infection, one was euthyroid and one was subclinical hypothyroid in control tests. One patient who was central hypothyroid during COVID-19 infection was central hypothyroid in the control blood test. When his previous thyroid function status was checked, he was found to be euthyroid.

Discussion
yroid dysfunction rate was 65.8% in this study (21.4% low TSH, 52.6% low FT3, 21.9% high FT4, and 1.95% high TSH). Additionally, we had 108 ESS, 9 subclinical hyperthyroidism, 8 overt thyrotoxicoses, 4 subclinical hypothyroidism, and 3 central hypothyroidism cases. e inflammatory markers, clinical severity score, mortality, and ICU admission were found to be higher in patients with ESS. yrotoxicosis was not associated with increased inflammatory markers, mortality, or ICU admission. yroid dysfunction was present in all nine patients who had died. e severe adult respiratory syndrome (SARS) epidemic in 2002 has had multiorgan effects; even thyroid follicular and parafollicular damages were found in autopsy specimens [19]. ACE-2 is known to be the receptor for coronavirus entry and a recent study demonstrated their presence in thyroid cell cultures [20]. erefore, this might explain the possible direct effect of COVID-19 on the thyroid gland. Muller et al. [5] showed that thyrotoxicosis was mostly evident in COVID-19 + high-intensity ICU (HICU) patients in comparison with patients with COVID-19 + low-intensity ICU (LICU) and COVID-19-HICU. ey have revealed the cause of thyrotoxicosis as an atypical form of subacute thyroiditis in which neck pain was absent due to lymphopenia [5]. Two other studies showed thyrotoxicosis and have interpreted the thyroid dysfunction as thyroiditis primarily. ey suggested that the underlying mechanism for thyrotoxicosis was either cytokine storm or direct effect of SARS-CoV-2 by ACE-2 receptor [6,9].
We found thyrotoxicosis in 17 patients. Moreover, 13 patients had thyrotoxicosis together with ESS as they had suppressed TSH levels with increased FT4 and decreased FT3 levels, a condition which is hard to make a differential diagnosis. Also, in the studies from Italy, Lania et al. and Muller et al. noted that there might be the coexistence of International Journal of Endocrinology ESS and thyrotoxicosis in a group of their patients [5,6]. None of our patients have experienced pain. Only one of these patients was antibody positive, so the possible underlying mechanism might be destructive thyroiditis. Unfortunately, we have not thought to measure thyroglobulin. Severe COVID-19 was related to a consequent decrease in the Treg/ 17 cell ratio, which might result in aggravated inflammatory responses and organ damage [21]. Viral infections are known to activate autoimmunity also by molecular mimicry mechanisms [22]. us, it is obvious that there is a link between the pathogenesis of COVID-19 and autoimmune thyroid disease. In one study, SARS-CoV-2 monoclonal antibodies were applied to different tissues and the thyroid gland was reactive, which has proved that COVID-19 might lead to thyroiditis [23]. However, clinical studies showed contradictory results. In one study, increased TPOAb positivity was detected [24], while increased autoimmunity was not detected in others [9,12]. In the light of this knowledge, we evaluated the TPOAb and TGAb in 205 patients, but we could not observe an increased rate of antibody positivity according to the normal population [25,26]. As autoimmunity may develop in later stages, these patients should be followed up regularly. e most probable mechanism that defines the thyroid dysfunction is ESS. Deiodinases play an important role in the pathogenesis of ESS. In one study, ESS was found to be correlated with disease severity in ICU patients. e possible  International Journal of Endocrinology mechanisms might be the effect of cytokines on HPA axis, thyroid binding proteins, or peripheric metabolism of thyroid hormones [27]. Actually, it is not surprising to think that a severe COVID-19 infection also alters thyroid hormone metabolism and causes ESS. ESS related to COVID-19 was seen in 16-48% of cases in different reports [3,4,[7][8][9][10][11][12]. However, these studies have several limitations, such as small sample size [3,7,10,11,13], drug interferences [3,6], and limited test panel [3,5,6,8]. Only two studies had evaluated thyroid autoimmunity [9,12], and prior thyroid function tests were assessed only in another [8]. ESS was diagnosed in 16.9% of COVID-19 (n � 71) patients with high fatality rate by Zhang et al. [10], and they have correlated thyroid dysfunction to increased neutrophil count, CRP, LDH, and CK, and low lymphocyte count. But this study was retrospective with a relatively small sample size and thyroid autoantibodies were not measured. Campi et al. [12] have declared the ESS rate as 40% of COVID-19 (n � 144) patients with higher serum cortisol, CRP, and IL-6 levels. Association of ESS with increased inflammatory markers was also confirmed by Zou et al. [4] who has found FT3 and CRP as predictive factors for disease severity. Gao et al. [7] have evaluated thyroid function tests of 100 severely ill COVID-19 patients and have found reduced FT3 levels associated with all-cause mortality. Shwarz et al. [11] also found low FT3 levels associated with higher mortality and ICU admission. Our ESS rate was 52.6% in 205 patients, making it the second largest study after the study by Lania et al. [6]. We found ESS to be associated with increased inflammatory parameters and WHO severity score, ICU admission, and mortality rates. In contrast to other studies, we have classified ESS patients as mild and moderate. Both inflammatory parameters and disease severity were found to be significantly high in moderate ESS in comparison with the mild ESS and euthyroid patients. In moderate ESS, TSH is normal or low, FT3 is low, and FT4 is high. So it is predicted that low FT3 is not enough for the severity of COVID-19.
In univariate and multivariate regression analysis, low FT3 was found to predict mortality. Also, we created a different scenario with low T3, low TSH, and high T4 by cluster analysis. Accordingly, the patients with a lower median FT3 (p < 0.001), higher median FT4 (p � 0.032), and a lower median TSH values (p < 0.001) had significantly higher risk for mortality (n � 8; 7.48% vs n � 1; 1.11%; p � 0.039). We also compared these two risk clusters for ICU admission rate, and high risk cluster included 25 of 31 patients admitted to ICU (23.3% vs. 6.67%, p � 0.001). We believe that risk clusters established for our study according to thyroid functions are valuable for the prediction of ICU admission risk and mortality. COVID-19-induced thyroid dysfunction might be due to a primary thyroid injury (thyrotoxicosis; atypical thyroiditis), a secondary injury at hypothalamic or pituitary level, or both of them [28]. In this study, three patients were assigned as secondary hypothyroidism confirmed by low FSH and LH levels and low sex hormones. One of these patients had died.
When the strength of our study is considered, it is one of the largest patient groups in the literature with all thyroid function tests, thyroid antibodies, previous thyroid, and pituitary function tests which were evaluated individually in a double-blinded manner by two investigators together with clinical outcomes. But our study also has several limitations. A control group was not enrolled, and follow-up thyroid function tests and antibodies were planned, but patients have refused to come to the hospital because of the pandemic.
ere are a lot of conflicts about COVID-19 and thyroid. What we want to say is that if you are a euthyroid, you will be lucky. In this study, the length of hospitalization, the rate of oxygen demand, and ICU admission rate were lower in the euthyroid patients. Moreover, none of the euthyroid patients died. Furthermore, the worst scenario might be to fall into high-risk group. Hence, the prognosis of patients who are in high-risk cluster with low FT3 (median � 2.34 ng/L; IQR � 0.86), a high median FT4 value (median � 1.04 ng/dL; IQR � 0.33), and a low median TSH value (median � 0.62 mIU/L; IQR � 0.59) are poor and mortality is increased.
We believe that COVID-19 will have effects on the thyroid gland, especially in respect to autoimmunity, and the pituitary gland. Future studies with follow-up measurements should be planned.

Data Availability
Data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
Ceyda Dincer Yazan, Onur Elbasan, Tugce Apaydin, Saida Dashdamirova, and Hulya Gozu conceptualized the study and collected and recorded the demographic, clinical, and laboratory data. e clinical findings of the patients were reported by Uluhan Sili. Tayfun Yigit, Onder Sirikci, and Goncagul Haklar studied and reported the thyroid function tests of the patients. e PCR results of the patients were reported by Aysegul Karahasan Yagci. Can Ilgin analyzed the data. Ceyda Dincer Yazan drafted the manuscript. Hulya Gozu and Goncagul Haklar revised all versions of the manuscript. All authors read and approved the final manuscript.