Thyroid Parameters and Kidney Disorder in Type 2 Diabetes: Results from the METAL Study

Objective Diabetic kidney disease is one of the most common microvascular complications of diabetes mellitus. We aimed to analyze the association of thyroid parameters with kidney disorders, especially in euthyroid participants. Methods The data were obtained from a cross-sectional study, the METAL study. Thyroid parameters, including thyroid-stimulating hormone (TSH), free triiodothyronine (FT3), free thyroxine (FT4), triiodothyronine (T3), thyroxin (T4), thyroid peroxidase antibody (TPOAb), and thyroglobulin antibody (TgAb), of 4136 participants with type 2 diabetes were measured. Two structure parameters of thyroid homeostasis, including the sum activity of step-up deiodinases (SPINA-GD) and thyroid secretory capacity (SPINA-GT), and two pituitary thyrotropic function indices, including Jostel's TSH index (TSHI) and the thyrotroph thyroid hormone resistance index (TTSI), were also calculated. Kidney disorders were described according to the presence of reduced estimated glomerular filtration rate (eGFR) and/or higher urinary albumin to creatinine ratio (UACR). Results The prevalence of kidney disorders increased with decreasing FT3 or T3 and increasing FT4 or T4 quartile levels (all P < 0.05). After full adjustment, linear regression showed that UACR levels were negatively associated with FT3 and T3 (P < 0.001). In addition, eGFR was positively associated with FT3 and T3 and was negatively associated with TSH and FT4 levels and TgAb positivity (all P < 0.05). By using binary logistic regression, higher TSH and FT4 and lower FT3 and T3 were associated with kidney disorders (all P < 0.05). Similar results were seen in sensitivity analyses, which were performed in 3035 euthyroid diabetic participants; however, TSH was no longer related to them. The area under the receiver operating characteristic curve (AUROC) of lower FT3 for existing kidney disorder was greater than that for any other thyroid hormones (all P < 0.001). The cutoff value of FT3 for reduced eGFR was 4.39 pmol/L. Regarding thyroid homeostasis parameters, SPINA-GD was negatively associated with three statuses of kidney disorders, and TSHI and TTSI were positively associated with reduced eGFR (all P < 0.05). Conclusions Among patients with type 2 diabetes, elevated TSH and FT4 (or T4), lower FT3 (or T3), TgAb positivity, lower SPINA-GD, and higher TSHI and TTSI were associated with kidney disorders. The lower FT3, even within the normal range (<4.38 pmol/L), may be the factor most related to reduced eGFR compared with other thyroid hormones in diabetic patients.


Introduction
Diabetes mellitus (DM) and its complications have become highly prevalent and have gained increasing attention [1], especially in developing countries. The estimated prevalence of diabetes among a representative sample of Chinese adults was 11.6%, which indicates the importance of diabetes as a public health problem in China [2]. Chronic kidney disease (CKD) is defined by the sustained presence of reduced kidney function or damage, often resulting from diabetes mellitus (DM) and hypertension [3]. Approximately 13% of individuals in the general US population have CKD, and the inci-dence is increasing globally [4][5][6]. Diabetic kidney disease (DKD) is one of the most common microvascular complications of diabetes mellitus and the leading cause of end-stage renal disease (ESRD) worldwide [7]. Approximately 20%-40% of patients with diabetes progress to DKD, and 40% also progress to ESRD [8]. Reduced estimated glomerular filtration rate (eGFR) and higher urinary albumin to creatinine ratio (UACR) levels are two main markers of diabetic kidney status, quantifying renal function and serving as a proxy of renal damage severity, respectively [3,9].
The thyroid gland is one of the most important organs in the human body. It regulates the majority of the body's physiological actions [10]. Thyroid hormone has an impact on renal tubular function and the renin-angiotensin system and is associated with hemodynamic and cardiovascular alterations that interfere with renal blood flow [11]. Conversely, the kidney is not only an organ for the metabolism and elimination of TH but also a target organ of some of the actions of the iodothyronines [12]. Acute kidney injury and chronic kidney disease are accompanied by notable effects on the hypothalamus-pituitary-thyroid axis. Hormonal derangements at the level of the hypothalamicpituitary axis are often seen with the worsening of kidney function, and recent evidence points towards the implication of such hormonal disorders in the genesis of CKD [13]. Thyroid dysfunction causes remarkable changes in glomerular and tubular functions and in electrolyte and water homeostasis [12].
Although several studies have suggested the association of thyroid disorders and CKD or DKD with conflicting results [7,[14][15][16], studies focusing on the possible relationship between thyroid status and eGFR and UACR among a relative number of DM participants, especially in euthyroid DM patients, have been limited. The present cross-sectional study investigated whether thyroid parameter concentrations, including thyroid-stimulating hormone (TSH), free triiodothyronine (FT 3 ), free thyroxine (FT 4 ), T 3 , T 4 , thyroid peroxidase antibody (TPOAb), and thyroglobulin antibody (TgAb), and four parameters of thyroid homeostasis, including the sum activity of step-up deiodinases (SPINA-G D ), thyroid secretory capacity (SPINA-G T ), Jostel's TSH index (TSHI), and thyrotroph thyroid hormone resistance index (TTSI), were related to eGFR, UACR, and the prevalence of kidney disorders, described as reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR in type 2 diabetes.

Study Design and
Participants. The data of this study were obtained from a cross-sectional study, the METAL study (Environmental Pollutant Exposure and Metabolic Diseases in Shanghai, http://www.chictr.org.cn, ChiCTR1800017573) [17,18], which was designed to investigate the association between exposure to heavy metals and diabetic complications in Chinese diabetic adults. We enrolled participants from seven communities in the Huangpu and Pudong districts in Shanghai. In 2018, we obtained the list of diabetic patients who were Chinese citizens ≥ 18 years old and had lived in their current area for ≥6 months from the registration plat-form in each community healthcare center and then randomly selected 50% of them (n = 4937) to receive the examination by using SPSS Statistics, Version 22 (IBM Corporation, Armonk, NY, USA). We excluded participants who were missing laboratory results (n = 8) or questionnaire data (n = 116), those who were missing thyroid parameters (n = 6) or UACR results (n = 234), those having a history of thyroid surgery or thyroid disorder treatment (n = 265), or those having a history of glucocorticoid or amiodarone treatment (n = 172) in this study. Finally, 4136 participants were involved in this study. Then, we excluded subjects with abnormal TSH, FT 4 , or T 4 levels and those with FT 3 or T 3 levels below the lower limits of their reference ranges for sensitivity analysis ( Figure 1). All participants provided written informed consent before data collection. The study protocol was approved by the Ethics Committee of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the 1975 Declaration of Helsinki, as revised in 2008.

Laboratory Measurements.
Serum samples for laboratorial assays were obtained by venipuncture after an 8-hour fast. Blood samples were stored at -20°C when collected and shipped by air in dry ice to one central laboratory, which was certified by the College of American Pathologists (CAP), within 2-4 hours of collection.
Creatinine (Cr) and uric acid (UA) were measured by Beckman Coulter AU680 (Brea, USA). Using morning fasting spot urine samples, the concentrations of albumin and CR were measured with a Beckman Coulter AU680 (Brea, USA) using a turbidimetric immunoassay and an enzymatic method, respectively. Reduced eGFR was defined as eGFR < 60 mL/min/1:73m 2 according to the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) for "Asian origin" [22] and regardless of whether the UACR was high. Higher UACR was defined as UACR ≥ 30 mg/g, suggested by an ADA statement [23] and regardless of the value of eGFR. Higher UACR and/or reduced eGFR was defined as the positivity of higher UACR and/or reduced eGFR.

Clinical Parameters.
A questionnaire about sociodemographic characteristics, medical history, family history, and lifestyle factors was adopted during the interview. The same group of trained and experienced personnel in the SPECT-China study (Survey on Prevalence in East China for Metabolic Diseases and Risk Factors, ChiCTR-ECS-14005052, http://www.chictr.org.cn) [24,25] conducted the interviews and clinical examinations, including measurements of weight, height, and blood pressure, according to a standard protocol. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Current smoking was defined as having smoked at least 100 cigarettes in one's lifetime and currently smoking cigarettes [2]. Hypertension was assessed by systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or self-reported previous diagnosis of hypertension by physicians or medication use for hypertension. Dyslipidemia was defined as TC ≥ 6:22 mmol/L, TG ≥ 2:26 mmol/L, LDL-C ≥ 4:14 mmol/L, HDL-C < 1:04 mmol/L, self-reported previous diagnosis of hyperlipidemia by physicians, or medication use for dyslipidemia, according to the modified National Cholesterol Education Program-Adult Treatment Panel III.

Statistical Analysis.
We performed survey analyses with IBM SPSS Statistics. All analyses were two-sided. A P value < 0.05 was taken to indicate a significant difference. Normally and nonnormally distributed continuous variables were expressed as mean ± SD and median (interquartile range), respectively. Categorical variables are presented as percentages. Pearson's χ 2 test was used for dichotomous variables. Associations among thyroid parameters and two kidney parameters (UACR and eGFR) were analyzed using linear regression models with each measure as the outcome. The regression models were adjusted for age, sex, HbA1c, BMI, UA, the duration of diabetes, current smoking, dyslipidemia, hypertension, and the use of metformin. UACR and eGFR were ln-transformed because of their skewed distribution. The results were expressed as B values and 95% confidence intervals (CIs). The relationship of thyroid hormones and parameters of thyroid homeostasis with reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR (all categorical variables) were assessed by logistic regression. The regression models were adjusted for age, sex, HbA1c, BMI, UA, the duration of diabetes, current smoking, dyslipidemia, hypertension, and the use of metformin. The results were expressed as odds ratios (ORs) (95% CIs). Area under the receiver operating characteristic curve (AUROC) analyses were used to compare the predictive powers of thyroid parameters for reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR. Sensitivity analyses were performed in euthyroid participants with type 2 diabetes.

Clinical Characteristics of the Participants.
This study recruited 4136 participants with type 2 diabetes, with a mean age of 67.23 years old (SD 8.63, max 99, min 23); among them, 1983 were men. The clinical characteristics of all these participants are shown in Table 1. The mean duration of diabetes of the participants was 9 years, and the mean level of HbA1c was 7.2%. The prevalence of reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR was 5.58%, 26.0%, and 28.1%, respectively.

Prevalence of Kidney Disorders according to Thyroid
Hormone Quartile Levels. Then, we analyzed the prevalence of reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR according to the quartile levels of thyroid hormones ( Figure 2). In all participants, we observed that the prevalence of reduced eGFR decreased with increasing levels of FT 3 and T 3 quartiles (both P < 0:001). In terms of the prevalence of higher UACR and higher UACR and/or reduced eGFR, we found that they were negatively associated with quartiles of FT 3 or T 3 (both P < 0:001) and were positively associated with quartiles of FT 4 (P = 0:029 and 0.011, respectively). Although there was a significant difference between the prevalence of higher UACR and/or reduced eGFR and TSH levels, it was not linearly correlated but was instead a "U shape"; that is, the lowest and highest quartiles of TSH have a higher prevalence of higher UACR and/or reduced eGFR. We obtained similar results when we grouped the total participants according to TSH reference ranges (<0.27 mIU/L, 0.27~4.20 mIU/L, and >4.2 mIU/L). The prevalence of reduced eGFR was 6.7%, 5.0%, and 8.0% (P = 0:008); the prevalence of higher UACR was 53.3%, 25.0%, and 31.0% (P < 0:001); and the prevalence of higher UACR and/or reduced eGFR was 53.3%, 26.8%, and 34.4% (P < 0:001), respectively.

Association of Thyroid Hormones and Antibodies with the
Prevalence of Kidney Disorders. Given that thyroid status was associated with kidney disease risk factors (UACR and eGFR), we evaluated the adjusted odds ratios (ORs) for the prevalence of kidney disorder status (reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR). Adjusted ORs were calculated after adjusting for age, sex, the duration of diabetes, current smoking, HbA1c, UA, BMI, dyslipidemia, hypertension, and current use of metformin using binary logistic regression models. As shown in Figure 3, higher TSH (OR 1.062; 95% CI 1.004, 1.124; and P = 0:036) and FT 4 (OR 1.090; 95% CI 1.020, 1.164; and P = 0:011) and lower FT 3 (OR 0.276; 95% CI 0.200, 0.380; and P < 0:001) and T 3 (OR 0.119; 95% CI 0.063, 0.226; and P < We then compared the AUROC for reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR between TSH, FT 3 , T 3 , FT 4 , and T 4 . We found that the AUROC for FT 3 was 0.742 for reduced eGFR, 0.552 for higher UACR, and 0.571 for higher UACR and/or reduced eGFR, which was higher than other thyroid parameters (

Association of Thyroid Homeostasis Parameters with
Kidney Disorders. To reveal the relationship between thyroid homeostasis parameters and kidney disorders, we further evaluated the adjusted odds ratios (ORs) of SPINA-G D , SPINA-G T , TSHI, and TTSI for the prevalence of kidney disorders. Adjusted ORs were calculated. As shown in Table 4, lower SPINA-G D was significantly associated with these three kidney disorder statuses, and higher TSHI and TTSI were also significantly associated with reduced eGFR (all P < 0:05).
3.6. Sensitivity Analyses. Further, we wanted to explore the relationship between thyroid parameters and reduced kidney function in participants with normal thyroid function and without euthyroid sick syndrome (ESS). Similar results were found for all participants.
Among these euthyroid participants, the prevalence of reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR increased significantly along with decreasing levels of FT 3 and/or T 3 . We also found that the prevalence of higher UACR and higher UACR and/or reduced eGFR increased with the levels of FT 4 and T 4 ( Figure 2). UACR levels were negatively associated with the level of FT 3 , while eGFR was positively associated with the level of FT 3 and T 3 but was negatively associated with the level of FT 4 , T 4 , and TgAb positivity ( Table 2). In logistic analysis, we also found lower FT 3 or T 3 levels, but higher FT 4 or T 4 levels were associated with the increased risks of reduced eGFR and higher UACR and/or reduced eGFR. Higher UACR was only negatively associated with FT 3 (Figure 3). The AUROC of FT 3 was 0.698 for reduced eGFR, 0.548 for higher UACR, and 0.562 for higher UACR and/or reduced eGFR, which was higher than other thyroid parameters (Table 3). Regarding the association of thyroid homeostasis parameters with kidney disorders in euthyroid participants, it was found that only lower SPINA-G D was significantly associated with these three kidney disorder statuses (Table 4).

Discussion
Diabetic nephropathy, as one of the complications of diabetes, is related to the health and longevity of diabetic patients. The gold standard for the diagnosis of DKD is through kidney biopsy, which is difficult to achieve in a large population of samples. Several studies have suggested that morning UACR, as an important risk predictor for diabetic nephropathy and renal events that may provide prognostic information [26,27], is an easy way to obtain the same information as a 24 h urine collection, which has been considered the gold standard [26,28]. Thus, in our study, we focused on two main indictors of CKD and DKD [29], reduced eGFR and higher UACR.
Metformin, as a first-line treatment for type 2 diabetes, was reported to have some effect on the hypothalamicpituitary-thyroid axis [30], and it was also associated with a decrease in the levels of TSH, possibly by enhancing the effects of thyroid hormones in the pituitary and activating adenosine monophosphate-activated protein kinase (AMPK) [31]. Thus, in our study, the current use of metformin was taken into account in all regression analyses.
In our study, among all participants, we found that the lowest and highest quartile levels of TSH had a higher prevalence of reduced eGFR, higher UACR, and higher UACR  Journal of Diabetes Research and/or reduced eGFR. Previous studies reported that both hypothyroidism and hyperthyroidism worsen kidney function directly by affecting renal blood flow, GFR, tubular function, electrolyte homeostasis, electrolyte pump functions, and kidney structure (e.g., decreased glomerular volume and area) and increasing peripheral vascular resistance [14][15][16], leading to higher serum CR concentrations and a lower eGFR. Several researchers also found an association of subclinical hypothyroidism with renal injury, such as macroalbuminuria in DM subjects [32][33][34][35]. Similarly, in our current study, the prevalence of these three kidney disorders was significantly different when all participants were grouped by the reference range of TSH, the distribution of which looked like a "U" shape. Thus, we hypothesized that thyroid dysfunction, including overt and subclinical hypothyroidism or hyperthyroidism, was associated with kidney disorders in DM participants. On the other hand, it is worth pointing out that this association failed to be found when it was analyzed among participants with normal thyroid function in this study. Conflicting evidence has been Odds ratio (95% confidence interval) Figure 3: Associations of thyroid hormones and antibodies with reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR. They were analyzed by using logistic regression. The regression models were adjusted for age, sex, HbA1c, BMI, UA, the duration of diabetes, current smoking, dyslipidemia, hypertension, and current use of metformin.

Journal of Diabetes Research
reported about the association of TSH and kidney status in euthyroid DM subjects [14,36,37]. The HUNT study reported that TSH within the reference range was negatively associated with eGFR [14] in type 2 diabetes, while others [7,38] reached a different conclusion.
In terms of other thyroid hormones, we observed increased prevalence of reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR along with decreased levels of FT 3 and T 3 . After full adjustment, FT 3 and T 3 levels were significantly negatively associated with UACR and were positively associated with eGFR. In logistic analysis, lower FT 3 and T 3 were also found to be associated with increased risks of reduced eGFR, higher UACR, and higher UACR and/or reduced eGFR. Furthermore, though we found that the AUROC of FT 3 for these three patterns of kidney disorders was much higher than that for any other thyroid hormones, only the AUROC of FT 3 and T 3 for reduced eGFR in all participants was larger than 0.7, which may give some significant values. The cutoff values of lower FT 3 and T 3 for reduced eGFR were 4.38 pmol/L and 1.49 nmol/L, respectively. These results were in agreement with some other cohort and cross-sectional studies that suggested that lower T 3 has been consistently found to be the most common disturbance in patients with kidney disorders and that it was an independent predictor of survival in various illness states [12,36,[39][40][41][42]. However, only a few studies have focused on the relationship between normal thyroid hormone levels and renal events in DM participants. In our present study, these results were strongly consistent even though FT 3 and T 3 levels were in the normal reference range. In addition, FT 3 but not FT 4 was reduced in kidney disorders, suggesting that low FT 3 is unlikely to be an in vitro artifact because substances inhibiting binding are expected to affect FT 4 to a greater extent than FT 3 . Thus, reduced FT 3 seems to reflect a true selective T 3 deficiency due to the inhibition of 5′ -deiodinase, which is a catalyzing enzyme for the production of T 3 from circulating T 4 [39,43]. Higher FT 4 and lower T 3 concentrations were reported to be strongly and independently associated with renal events [44]. In this study, we obtained the same results.
It has been suggested that autoimmune thyroid disease may lead to the deposition of immunocomplexes in the renal glomeruli and may be responsible for endothelial dysfunction and subsequent microalbuminuria [45]. In the present  reported that the association with eGFR was roughly similar for hypothyroidism with and without TPOAb positivity, which suggested that hypothyroidismassociated kidney dysfunction seems to be more related to the decline in thyroid hormone levels than to thyroid autoimmunity [46]. Finally, we explored the relationship of thyroid homeostasis parameters, including two structural parameters of thyroid homeostasis (SPINA-G D and SPINA-G T ) and two pituitary thyrotropic function indices (TSHI and TTSI), with kidney status. By extending the classic concept of separate measurements of thyroid hormone parameters, these markers add new qualitative and quantitative dimensions to the evaluation of thyroid homeostasis, offer a more integrated and systemic view, and deliver important insights into the physiology of pituitary-thyroid feedback control [20]. Though it was reported that SPINA-G T was correlated with creatinine clearance [20,21], we found that lower SPINA-G D was significantly associated with three kidney disorder statuses, and higher TSHI and TTSI were significantly associated with reduced eGFR. Therefore, it was suggested that kidney disorders may be related to the maximum stimulated activity of step-up deiodination and pituitary thyrotropic function.
Though our study had some strengths, including novelty, the analysis of community dwelling participants with a relatively large sample size, and strong quality control, there were also some limitations. First, because of the cross-sectional study, we can evaluate association but not causality. We conducted a Mendelian randomization (MR) analysis to examine the causality of the association between TSH and eGFR in 10603 participants recruited from the Survey on Prevalence in East China for Metabolic Diseases and Risk Factors (SPECT-China, ChiCTR-ECS-14005052, http://www.chictr .org.cn) [25,47]. By a genetic approach that limits residual confounding and reverse causation in observational conventional epidemiological studies, TSH and eGFR were not causally associated, which suggested that genetically elevated TSH concentrations may not affect renal function [48]. The mechanism linking kidney function and thyroid homeostasis remains unclear. Thyroid dysfunction may trigger kidney failure, and renal failure might trigger an allostatic response of thyroid function as well, i.e., euthyroid sick syndrome [49] or thyroid allostasis in critical illness, tumors, uremia, and starvation (TACITUS) [50], albeit in slight form. This is known from previous studies and is plausible because the kidney expresses deiodinases that catalyze the conversion of T 4 to T 3 . Further prospective and longitudinal studies are needed. Second, UACR and thyroid function were estimated only once, which could result in misleading classifications. Third, the levels of rT 3 were not measured in this study.

Conclusion
Among type 2 diabetes patients, elevated TSH and FT 4 (or T 4 ), lower FT 3 (or T 3 ), and TgAb positivity were associated with higher UACR and lower eGFR levels. Elevated TSH and FT 4 (or T 4 ), lower FT 3 (or T 3 ) levels, and thyroid homeostasis parameters (SPINA-G D , TSHI, and TTSI) were associated with the prevalence of kidney disorders. The lower FT 3 , even within a normal reference range, was the factor most related to reduced eGFR compared with all other thyroid hormones. Multidisciplinary collaborative care, including nephrologists, endocrinologists, or internal medicine specialists, is needed and appears to positively impact the clinical, humanistic, and economic outcomes of patients with diabetes [51,52].

Data Availability
The data used to support the findings of this study are included within the article.