Subclinical hypothyroidism and the risk of chronic kidney disease in T2D subjects: A case-control and dose-response analysis.

Evidence indicated a positive association between subclinical hypothyroidism (SCH) and cardiovascular diseases. But the relationship between SCH and chronic kidney diseases (CKD) remains unclear. A case-control study was performed to ascertain this relationship followed by a meta-analysis. In this hospital-based, case-control study, we recruited 3270 type 2 diabetic patients with euthyroidism and 545 type 2 diabetic patients with SCH. All English studies were searched upon the relationship between SCH and CKD up to October 2016. Meta-analysis was performed using STATA 13.0 software. Our case-control study indicated an association between SCH and CKD in patients with type 2 diabetes [OR (95% CI): 1.22 (1.09-1.36)]. Five observational studies reporting risk of CKD in SCH individuals were enrolled. A significant relationship between SCH and CKD was shown [pooled OR 1.80, (95% CI) 1.38-2.35]. Among normal TSH range, individuals with TSH ≥3.0 μIU/ml had a significantly higher rate of CKD (Fisher exact test, P = 0.027). Dose-response linear increase of CKD events was explored [pooled OR 1.09 (95% CI): 1.03-1.16 per1 mIU/L increase of TSH]. The present evidence suggests that SCH is probably a significant risk factor of CKD in T2D. Linear trend is shown between TSH elevation and CKD in T2D. This relationship between serum TSH and renal impairment in type 2 diabetic patients needs further studies to investigate.

Introduction

Subclinical hypothyroidism (SCH), defined as elevated serum TSH but normal free T4 (FT4) level, is a common endocrine disease. The prevalence of SCH increased with age, about 4 to 10% in adult population [ and 12.7% in diabetic individuals.[

Up to now, the association between SCH and the risk of clinical cardiovascular events, such as coronary heart disease,[ heart failure,[ cognitive [ and stroke,[ has been established in specific subgroups with higher TSH levels. CKD was also associated with an increased risk of cardiovascular outcomes, such as heart failure and death.[ Our previous study showed that the estimated glomerular filtration rate (eGFR) is different between SCH and euthyroid individuals.[

However, association between subclinical hypothyroidism with the risk of CKD is a matter of debate. It is also unknown whether the dose-response relationship between level of TSH and risk of CKD exists. For this reason, we performed this case-control study, and then, our data were combined in a meta-analysis with other eligible observational studies to investigate the relationship between SCH and CKDs. With the results of this study, we hope to clarify the relationship between SCH and CKD in diabetic individuals.

Materials and methods

In this hospital-based, case-control research, a total of 5530 patients with type 2 diabetes (aged 56.3 ± 12.6 years, with a mean duration of known diabetes for 7.7 ± 5.6 years) were investigated during January 2011 and December 2015 in Beijing, China. Patients with SCH were allocated in case group and patients with normal TSH were allocated in control group. Patients with pituitary tumor, thyroid carcinoma, head or neck cancer, pregnancy, severe heart, liver or brain disease were not eligible to be included in this study.[ Obese individuals were excluded from this study to decrease the potential effect of body mass index (BMI).[

Individuals taking TH replacement therapy were excluded. Metformin, one of anti-diabetic drugs, can interfere with TSH measurement.[ The usage of metformin was comparable between cases and controls in our study.

In all patients, serum-free triiodothyronine (FT3), serum-free thyroxine (FT4), thyroid stimulating hormone (TSH), and antithyroid peroxidase antibody (anti-TPO) concentrations were measured. The reference range of normal TSH levels was 0.55 to 4.78 mIU/L. And FT4 was 11.5 to 22.7 pmol/L, FT3 was 11.5 to 22.7 pmol/L. SCH was defined as 19.9 mIU/L>TSH > 4.78 mIU/L with normal FT4. Euthyroidism was defined as normal levels of all three parameters: FT3, FT4, and TSH.

According to individual matching principle, for each type 2 diabetic patient with SCH, 6 type 2 diabetic patients with euthyroidism were selected as controls, matching for gender, age.

Ethics statement

The study was conducted with the approval from the Ethics Committee of Beijing Tongren Hospital, Capital Medical University, and adhered to the tenets of the Declaration of Helsinki. Additionally, the written informed consent was obtained from each participant.

Assessment of CKD

Estimated GFR (eGFR) is the preferred measurement of kidney function in the current study. It is estimated using the Modification of Diet in Renal Disease Study (MDRD) equation: eGFR (mL/min/1.73 m2) = 186 × (serum creatinine)−1.154 × (Age)−0.203 × (0.742 if female) (conventional units).[ The moderate CKD was defined as eGFR of < 60 ml/min/1.73 m2.

Meta-analysis

PubMed, EMBASE, and the Cochrane Library were searched for studies to January 2016, using the terms “subclinical hypothyroidism,” “chronic kidney disease,” “chronic kidney dysfunction,” “chronic kidney insufficiency,” “chronic kidney failure,” “chronic renal failure,” “chronic renal dysfunction,” “chronic renal insufficiency,” “end-stage renal disease,” or “end-stage kidney disease,” and “risk factors.” Terms were explored when possible within each database. To avoid missing any relevant studies in the search, reference lists of key articles were also searched for relevant articles that could have been missed. English language restriction was applied.

This research was designed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement[ and recommendations from the Cochrane Collaboration and Meta-analysis of Observational Studies in Epidemiology (MOSE),[ using a methodology extensively described in our previous paper.[

References were included if the following criteria were met: cross-sectional study involving individuals 18 years or older. Evaluation of TSH and GFR. Reporting of multivariate adjusted odds risks (ORs) and 95% confidence intervals (CIs) for CKD associated with SCH versus reference, or reported ORs and 95% CIs for low, middle, and high-range TSH versus reference, respectively.

Studies were excluded if they met the following criteria: studies without specific sample origins. Studies with a sample size less than 50. Insufficient data in the research.

Statistics

Data were analyzed by an unpaired Student t test, Fisher exact test, X2 test, and multiple logistic regressions. We assessed the within- and between-study variation or heterogeneity by testing Cochran's Q-statistic.[ Heterogeneity was quantified with the I2 metric, which is independent of the number of studies in the systematic review.[ The pooled OR was estimated using fixed effects (FE, Mantel and Haenszel) and random effects (RE, DerSimonian and Laird) models. When there is heterogeneity between studies, the pooled OR was estimated using the random effects model.

Three studies provided ORs for each TSH level category that was eligible for dose-response analysis. An overall OR was obtained for the highest TSH level category compared with the lowest TSH level category based on 3 studies.[ Once the dose-relationship between the TSH level and the CKD risk was established, a restricted cubic spline regression model was used to reveal a fixed-effects potential nonlinearity model at 4 points (at the fifth, 35th, 65th, and 95th percentiles), via generalized least-square regression considering the relationship between each calculated OR for each 1 mIU/L TSH increment. If the P value for the “goodness-of-fit”/heterogeneity of the previous model was <0.05, a random effects potential nonlinearity model was evaluated. If the test for the nonlinear association was not significant, the simple generalized least-squares trend model without the restricted cubic spline model was used to test the linear hypothesis.[ Begg and Egger tests were performed to assess the publication bias. The Egger regression asymmetry test and the regression asymmetry plot tend to suggest the presence of publication bias more frequently than the Begg approach.[ Statistical manipulations were undertaken using program STATA (version 13.0, StataCorp LP, TX). Based on the previously reported odds ratio (1.3), sample size of our case-control study had 70% power to replicate the previous finding (alpha has been set at 0.05).[ Power analysis for case-control samples was carried out by Power and Sample Size Calculation (version 3.1.2, 2014).

Results

As a result, 545 cases and 3270 controls were selected for investigation. No significant differences were found in age, duration of known diabetes, BMI, systolic blood pressure, or biochemical parameters, except diastolic blood pressure, estimated glomerular filtration rate (eGFR) and LDL cholesterol between SCH and euthyroid groups (seen in Table 1). After adjustment for age, duration of diabetes, glycosylated hemoglobin (A1C), BMI, blood pressure, and LDL cholesterol, SCH was found to be associated with CKD [odds ratio (95% CI): 1.22 (1.09–1.36)].

Table 1

Clinical data of euthyroid and SCH.

In 3270 individuals with normal TSH, 125 of 2720 individuals (4.59%) with a TSH level between 0.55 and 3.0 μIU/mL had CKD, while 55 of 550 individuals (10.0%) with a TSH level between 3.0 and 4.78 μIU/mL had CKD. Individuals with TSH ≥3.0 μIU/mL had a significantly higher rate of CKD (Fisher exact test, P = 0.027) in euthyroid group. Further analysis showed that 85 of 545 individuals in SCH group had CKD (15.60%). Prevalence of CKD in SCH group was significantly higher than that in control group (Fisher exact test, P < 0.001).

According to the level of serum TSH, normal serum TSH individuals were divided into 4 groups (0.55–1.49, 1.50–2.49, 2.50–3.49, 3.50–4.79 μIU/mL). Compared with the first group (reference), participants in SCH group had a significantly lower mean of eGFR, multivariable adjusted mean ± SE (−13.96 ± 4.09 mL/min, P = 0.001) (Table 2).

Table 2

Characteristics of included studies in the meta-analysis.

Meanwhile, Pearson correlation analysis suggested that there was a significant inverse correlation between TSH and eGFR levels (r = −0.26; P < 0.01), marginally significant association between FT3 concentrations and eGFR (r = 0.18; P = 0.052), and nonsignificant association between FT4 concentrations and eGFR. Anti-TPO was also nonsignificant between CKD and non-CKD in SCH group.

Meta-analysis between TSH levels and the risk of CKD

A total of 28 relevant studies were screened. After systematic review, only 5 studies plus our research were included in the meta-analysis ([2, 22–25] in Table 2). The studies provided 3154 cases with SCH and 35270 euthyroid individuals. The results of the literature search are shown in Fig. 1. The characteristics of included studies are listed in Table 2. Risk of CKD was significantly raised in SCH group versus that in the euthyroid group (summary OR 1.80, 95% CI 1.38–2.35) (Fig. 2). Participants from 6 studies were divided into 2 groups: diabetic and nondiabetic individuals. Diabetic individuals ([2, 22] and our study) showed a significant association between SCH and CKD [OR, 1.83 (95% CI: 1.03–3.27), P = 0.01]. There was also a significant association between SCH and CKD in nondiabetic individuals (26–28) [OR, 1.78 (95% CI: 1.47–2.14), P < 0.001].

Figure 1

Flowchart demonstrates those studies that were processed for inclusion in our meta-analysis.

Figure 2

Overall OR with 95% CIs for the risk of CKD for each 1 mIU/L increase in the TSH level. The area of each square stands for the weight of each study in the meta-analysis. The diamond shows the overall OR; the horizontal lines indicate the 95% confidence intervals (CIs).

Nonlinear and linear dose-response analyses

The potential nonlinearity of the relationship was evaluated using a restricted cubic spline regression model. The fixed-effects nonlinear model was reasonable (chi2 = 53.77, P < 0 .000) (Fig. 3A), no significant heterogeneity in identified studies was found (goodness of fit chi2 = 8.23, P = 0.144). Further step was explored to evaluate whether dose-response relationship was nonlinearity. Association between a nonlinear trend for TSH elevation with CKD was marginally significant (chi2 = 6.61, P = 0.037). Therefore, a linear dose-response relationship was also analyzed using random-effects analysis, and the results showed a significant positive linear relationship (chi2 = 9.23, P = 0.0024, Fig. 3B). Dose-response linear increment of CKD events was observed (summary OR 1.09 95% CI: 1.03–1.16 per incremental increased in TSH of 1 mIU/L). Significant heterogeneity among these studies was observed (goodness of fit chi2 = 11.44, P = 0.0033).

Figure 3

OR of CKD for each TSH level increment according to a linear or nonlinear dose-response model. A, Nonlinear dose-response model based on restricted cubic splines for TSH concentrations at 4 points: the fifth, 50th, and 95th percentiles. B, Linear dose-response model; P for the linear model <0.001. The lines with long dashes stand for the point-wise 95% confidence intervals for the fitted nonlinear trend (solid line). The lines with short dashes stand for the linear trend.

Publication bias

Both methods indicated symmetry in the funnel plot and the regression asymmetry plot, which gave no evidence of publication bias (Begg test: z = 1.28, P = 0.19. regression asymmetry test: s = 0.80 and P = 0.38).

Discussion

Our study reported the association between SCH and CKD in the hospital-based, case-control research. This association was independent of age, BMI, duration of diabetes, A1C, blood pressure, and LDL cholesterol. Subsequently, dose-response meta-analysis first quantified the relationship between TSH level and risk of CKD.

Our findings demonstrated a relationship between SCH and CKD in T2D individuals. SCH is an asymptomatic stage of hypothyroidism. Thyroid hormone plays an important role in kidney growth and maintenance of many of its functions.[ Hypo-thyroid hormones may decrease cardiac output, increase peripheral vascular resistance, resulting in intrarenal vasoconstriction.[ SCH is also complicated with endothelial dysfunction,[ which leads to small vessel dysfunction.[ Other pathways could include altered iodine metabolism and decreased peripheral sensitivity to hormone.[ Therefore, the pathways mentioned above could explain the association between higher TSH levels and reduced renal function. Conversely, it has also been indicated that decreased kidney function may affect thyroid hormone derangements (e.g., serum TSH bioactivity and TSH clearance) due to metabolic acidosis, iodine retention, selenium deficiencies, and metabolic adaptation to malnutrition.[

Our evidence showed that the prevalence of CKD increased with elevated TSH level, 4.78% in patients with TSH from 0.55 to 3.0 μIU/mL, 9.16% in patients with TSH from 3.0 to 4.78 μIU/mL, and 15.60% in SCH. This relationship between CKD and TSH level was independent of age, sex, duration of diabetes, A1C, BMI, blood pressure, and LDL cholesterol.

Furthermore, a significant inverse association between eGFR and TSH levels was observed throughout the normal and high TSH ranges, which was in accordance with previous studies.[ Moreover, decreasing in renal blood flow associated with low thyroid function within the clinically normal range has been shown, which also supports our findings.[

The glomerular filtration rate in SCH patients is lower than that in euthyroid individuals. It is known that autoimmune thyroid disease can lead to the deposition of immune-complexes in renal glomeruli.[ However, in our research, the association with eGFR was roughly similar to SCH with or without anti-TPO, which is in accordance with the results of Asvold BO.[ This indicates that damage to the kidney caused by autoimmune thyroid disease is not likely an explanation for the association between SCH and CKD in our study.

Our findings add to a growing body of literature demonstrating a relationship between SCH and CKD. The sample of this study could detect the power of 70%. So sample size was enlarged by this evidence-accumulated study. The association between SCH and CKD was evidenced in this meta-analysis. Heterogeneity among studies in this meta-analysis disappeared when subgroup analysis was performed according to participants who were subdivided into diabetic individuals and nondiabetic individuals. Association between SCH and CKD remains significant regardless of the status of diabetes [(OR, 1.83 (95% CI: 1.03–3.27) vs 1.78 (95% CI: 1.47–2.14)], which may explain the heterogeneity of CKD risk associated with SCH in this meta-analysis. In addition, differences in the age and sex structure of the study subjects may modify the association between thyroid hormones and endpoints.[

Subsequently, dose-response analysis indicated that the risk of CKD increased 1.09 times per 1 mIU/L increase of TSH. Additionally, the link between TSH and CKD appears to be a linear relationship. Controversies exist as to the reference range for “normal” TSH. Some clinicians have advocated reducing the TSH upper limit from 4.5 to 5.0 mIU/L to 2.5 to 3.0 mIU/L based on a higher risk of progression to organ damage.[ Our evidence supports this viewpoint. The rate of CKD with TSH levels from 3.0 to 4.78 μIU/mL is significant higher than that with TSH levels from 0.55 to 3.0 μIU/mL (9.17% vs. 4.86%). So optimal cutoff level of TSH would warrant exploration in future studies.

Limitation

First, the cross-sectional nature of our study research is limited to its ability to establish causal relationships between SCH and CKD. It does not necessarily infer that treatment of SCH will reduce the risk of CKD. However, renal function was ameliorated in patients with hypothyroidism after thyroxine treatment.[ While the mechanistic link between these 2 entities remains unknown. Second, in the absence of a consensus, the reference range for “normal” TSH has been the focus of considerable debate. So the link between SCH and CKD requires qualitative analysis. A linear dose-response relationship between TSH and CKD is supplementary to this limitation. Lastly, the possibility of residual confounding and other unmeasured covariates cannot be excluded. Evidence showed that obese was thought to be a protective factor for renal function deterioration.[ Therefore, obese individuals were excluded from this study. Effect of SCH on serum creatinine levels may be due to changes in muscle metabolism independent of its effects on GFR.[ Further exploration needs to be warranted.

Conclusions

The present evidence suggested that SCH could be a significant risk factor for CKD in T2D. Linear trend between TSH elevation and CKD exists in this study. The exact mechanism by which high serum TSH leads to renal impairment in T2D remains unclear, which would prompt future studies.