Nonalcoholic Fatty Liver Disease Is Not Associated with Thyroid Hormone Levels and Hypothyroidism: A Systematic Review and Meta-Analysis

Nonalcoholic fatty liver disease (NAFLD) is a major health problem. It is one of the most common causes of chronic liver disease with a global prevalence of 25% [1]. The presentation of NAFLD ranges from simple steatosis to nonalcoholic steatohepatitis that could progress to cirrhosis and hepatocellular carcinoma. Its prevalence has increased substantially throughout the world in parallel with the growing epidemic of obesity [2,3]. Moreover, nonalcoholic steatohepatitis-related cirrhosis is expected to surpass other causes of chronic liver disease to become the leading indication for liver transplantation in the next 2 decades [4]. Therefore, identifying risk factors for NAFLD to potentially develop new preventive or treatment strategies is of utmost significance.

Metabolic disorders such as hypertension, hyperlipidemia, diabetes, gallstones, and central obesity are known risk factors for NAFLD [2,5,6]. Thyroid hormones regulate all metabolic pathways, acting on protein, carbohydrates, and lipid metabolism. Reduced thyroid hormone levels are associated with hypometabolism characterized by reduced resting energy expenditure, weight gain, increased cholesterol levels, reduced lipolysis, and reduced gluconeogenesis [7]. Thyroid hormones also have a role in hepatic lipid metabolism and hepatic insulin resistance [8]. Thyroid dysfunction can lead to hyperlipidemia, obesity, and insulin resistance, which are components of metabolic syndrome [9].

Given the hepatic manifestation of metabolic syndrome, several studies have investigated the association between NAFLD and thyroid dysfunction, but have demonstrated inconsistent results. Therefore, to better characterize this association, we conducted a systematic review and meta-analysis of all published observational studies regarding the association of NAFLD and thyroid dysfunction including hypothyroidism and thyroid hormone levels compared to non-NAFLD subjects.

This systematic review and meta-analysis was conducted and reported according to the Meta-Analysis of Observational Studies in Epidemiology Statement [10] and was registered in PROSPERO (No. CRD42016041913).

Two authors (S.U. and V.J.) independently searched published studies indexed in PubMed/MEDLINE and Embase databases from inception to July 2016 using the search strategy that comprised terms for NAFLD and thyroid function without language restriction. References of selected retrieved articles were also manually reviewed.

Our inclusion criteria were (1) published observational studies that addressed the association between NAFLD and thyroid hormone levels (free triiodothyronine [FT₃], free thyroxine [FT₄], or thyroid-stimulating hormone [TSH]) in participants with hypothyroidism, euthyroidism, or hyperthyroidism, or studies that assessed the association between NAFLD and hypothyroidism including either overt hypothyroidism or subclinical hypothyroidism in adult participants; (2) participants without NAFLD were used as a reference group; (3) NAFLD was diagnosed by liver enzyme levels, imaging study, or liver biopsy; and (4) overt and subclinical hypothyroidism were diagnosed by each study's thyroid hormone cutoff levels. To assess the quality of all studies, review articles, case reports, abstracts, and unpublished studies were excluded.

Two authors (S.U. and V.J.) independently reviewed the titles and abstracts of all the citations that were identified. After all the abstracts were reviewed, data comparisons between the 2 investigators were conducted to ensure completeness and reliability. The inclusion criteria were independently applied to all identified studies. Differing decisions were resolved by consensus between the 2 authors.

The quality of each study was independently evaluated by 2 authors (S.U. and V.J.) using the Newcastle-Ottawa Quality Assessment Scale which assessed each study in 3 areas: (1) the selection of the study groups, (2) the comparability of the groups, and (3) the ascertainment of the exposure or outcome of interest for case-control or cohort studies, respectively [11]. Discrepant opinions between authors were resolved by discussion between the authors.

Full-text versions of potentially relevant papers identified in the initial screening were retrieved. If multiple articles from the same study were found, only the article with the most complete data was included. Data concerning author, year of publication, study design, study location, participant characteristics, diagnostic method of NAFLD, and definition of overt and subclinical hypothyroidism were independently extracted by 2 investigators. We contacted the authors of the primary reports to request any unpublished data. If the authors did not reply, we used the available data for our analyses.

The inverse variance method was used to estimate the pooled odds ratios (OR) and corresponding 95% CI, and data were pooled using a random-effects model given the high likelihood of between-study variance. For studies with several multivariable-adjusted estimates, we extracted those reflecting the greatest degree of control for potential confounders. All continuous data are summarized as the mean difference (MD) along with 95% CI. Sensitivity analysis was performed to evaluate the robustness of results, in which pooled estimates were computed omitting 1 study in each turn. The heterogeneity of effect size estimates across these studies was quantified using the Q statistic, its p value, and I² (p < 0.10 was considered significant). An I² value of 0-25% indicates insignificant heterogeneity, 26-50% low heterogeneity, 51-75% moderate heterogeneity, and 76-100% high heterogeneity [12]. Publication bias was assessed using a funnel plot and the Egger regression test [13]. Data analysis was performed using Comprehensive Meta-Analysis 3.3 software from Biostat Inc. (Englewood, NJ, USA).

The initial search yielded 485 articles; 460 articles were excluded based on the title and abstract review. A total of 25 articles underwent full-length review. Eleven articles were excluded (4 articles had no comparison group, 3 articles did not report a subject of interest, and 4 articles did not report an outcome of interest). Fourteen observational studies (9 cross-sectional studies [14,15,16,17,18,19,20,21,22], 4 case-control studies [23,24,25,26], and 1 cohort study [27]) involving 7,191 NAFLD patients and 30,003 controls were included in the meta-analysis. Table 1 describes the detailed characteristics and quality assessment of the included studies.

A total of 5 studies [14,15,16,20,27] involving 26,454 participants, 6 studies [14,15,16,20,23,27] involving 27,070 studies, and 9 studies [14,15,16,17,18,20,25,26,27] involving 32,347 participants were included in our meta-analyses for assessing the association between NAFLD and subclinical hypothyroidism, overt hypothyroidism, and overall hypothyroidism, respectively. NAFLD was not associated with subclinical hypothyroidism (pooled OR = 0.63, 95% CI: 0.18-2.20, p = 0.47, I² = 97%, p_{heterogeneity} <0.01), overt hypothyroidism (pooled OR = 1.37, 95% CI: 0.78-2.41, p = 0.27, I² = 81%, p_{heterogeneity} = 0.08), or overall hypothyroidism (pooled OR = 1.21, 95% CI: 0.91-1.61, p = 0.18, I² = 72%, p_{heterogeneity} = 0.02) compared with non-NAFLD controls (Fig. 1).

In order to assess the stability of the results of our meta-analyses, we performed a 1-study removed sensitivity analysis. For subclinical hypothyroidism, removal of Lee et al. [27] moved the overall effect to favor NAFLD with an OR of 1.31 (95% CI: 1.12-1.53), suggesting that Lee et al. was partly the reason for the high between-study heterogeneity. For overt hypothyroidism, removal of Lee et al. moved the overall effect to favor NAFLD with an OR of 1.76 (1.28-2.41), suggesting that Lee et al. also could contribute to the presence of heterogeneity. For overall hypothyroidism, statistically similar results were obtained after sequentially excluding each study in all meta-analyses, suggesting the results were robust.

To investigate potential publication bias, we examined the contour-enhanced funnel plot of the included studies. The vertical axis represents study size (standard error by log OR), while the horizontal axis represents effect size (log OR). When assessing the studies for OR of overt hypothyroidism, publication bias is present as there are more studies that favor a positive log OR (positive results), but the p value for the Egger regression test is not significant (p = 0.12). When assessing the studies for OR of subclinical hypothyroidism and overall hypothyroidism, the plots exclude bias since there are symmetrical distributions of studies on both sides of the mean. The Egger tests were not significant (p = 0.49 and 0.61, respectively).

A total of 6 studies [15,16,17,19,21,22] involving 6,294 participants, 7 studies [15,16,17,19,21,22,27] involving 24,838 studies, and 9 studies [15,16,17,19,20,21,22,24,27] involving 26,147 participants were included in our meta-analyses for assessing the association between NAFLD and FT₃, FT₄, and TSH, respectively. Patients who had NAFLD did not show a significant difference concerning FT₃ (pooled MD = 0.04, 95% CI: -0.11 to 0.19, p = 0.61, I² = 96%, p_{heterogeneity} <0.01), FT₄ (pooled MD = -0.27, 95% CI: -0.70 to 0.16, p = 0.22, I² = 98%, p_{heterogeneity} <0.01), and TSH (pooled MD = 0.11, 95% CI: -0.03 to 0.24, p = 0.12, I² = 88%, p_{heterogeneity} = 0.03) compared with non-NAFLD controls (Fig. 2).

As we performed sensitivity analysis, statistically similar results were obtained after sequentially excluding each study at a time in all meta-analyses for FT₃, FT₄, and TSH, suggesting the results were robust. For assessing the studies for the MD of FT₃ and FT₄, publication biases are present as there are more studies that favor negative results (decreased FT₃ and FT₄ in NAFLD compared with controls) but the Egger regression tests showed nonsignificant publication bias (p = 0.73 and p = 0.86, respectively). For assessing the MD of TSH, the plot excludes bias with the nonsignificant Egger test (p = 0.98).

This is the first meta-analysis on the association between NAFLD and thyroid dysfunction involving around 7,000 NAFLD patients and 30,000 controls, and showed that NAFLD patients had no significant differences in thyroid hormone levels including FT₃, FT₄, and TSH compared to non-NAFLD controls. We also found no associations between NAFLD and subclinical, overt, and overall hypothyroidism.

Liangpunsakul and Chalasani [23] published the first study in 2003 demonstrating the significantly higher prevalence of hypothyroidism in a nonalcoholic steatohepatitis group in comparison to a control group in a case-control study involving around 600 subjects. Since then, there have been a number of studies addressing this potential association, but most of them were conducted on a small scale. There are a few large-scale studies that have demonstrated significant results. Ittermann et al. [18] conducted a population-based study in Germany including 3,600 healthy individuals that found only a significant association between NAFLD and FT₄, but not FT₃, TSH, hypothyroidism, or hyperthyroidism. Chung et al. [14] performed a population-based study in Korea involving 4,648 healthy subjects and showed that the prevalence of NAFLD was independently associated with hypothyroidism and the grade of hypothyroidism in a dose-dependent manner. Liu et al. [19] performed another health survey study in China including almost 2,600 participants and showed that FT₃ was independently associated with NAFLD, but FT₄ was only significantly associated with NAFLD in a subgroup of postmenopausal women. However, there have also been a few large trials that found no association. Zhang et al. [26] conducted a population-based study involving more than 1,300 Chinese subjects and showed that TSH is not an independent risk factor for NAFLD. Lee et al. [27] conducted a study involving almost 20,000 healthy Korean subjects and found no association between NAFLD and hypothyroidism in all subtypes (overt and subclinical hypothyroidism). Interestingly, both studies that demonstrated significant and nonsignificant results comparably included variables in multivariate models including age, gender, alcohol, smoking, and metabolic parameters and both were comparably comprised of Asian and Caucasian populations. However, there are variations in the definition of hypothyroidism and the measurement of thyroid hormones.

Although hypothyroidism is associated with components of metabolic syndrome [9] and NAFLD is considered as the hepatic manifestation of metabolic syndrome, and even though there are several plausible mechanisms that could explain this possible link, our meta-analysis, which included all published observational studies, could not find any association between NAFLD and hypothyroidism. Leptin is considered as one of the explanations of this association because it is found to be increased in patients with hypothyroidism [28] and it is also found to be higher in NAFLD as it can promote hepatic insulin resistance and be involved in hepatic fibrogenesis [29,30]. Fibroblast growth factor-21, a member of fibroblast growth factor family, could also provide a role in this association as increased serum fibroblast growth factor-21 levels were observed in patients with hypothyroidism [31]. Its level was also correlated with intrahepatic triglyceride content and associated with increased vulnerability to metabolic stressors that could accelerate the progression of NAFLD [32]. Further large-scale studies are needed to clarify this potential association and should adjust for all potential confounding factors and include comparisons between different ethnicities and gender.

Several limitations of this study should be acknowledged. First, statistical heterogeneity is present in this study. This might be explained by the differences in the populations enrolled, definitions of hypothyroidism, methods used to check thyroid hormone profile, and study designs. Second, this is a meta-analysis of observational studies, which can only demonstrate an association, not causality, between NAFLD and hypothyroidism. Third, most of the included studies used imaging studies to detect hepatic steatosis. Although these imaging techniques are widely accepted as the diagnostic tools of choice for screening NAFLD because they are widely available and easy to perform, they have limited accuracy in detecting mild steatosis compared to liver biopsy, which is considered the gold standard [33].

In conclusion, our meta-analysis demonstrates no significant association between NAFLD and subclinical, overt, or overall hypothyroidism, and we also found no significant difference in thyroid hormone levels between participants with and without NAFLD.

We thank Matthew Roslund for validation of the search.

This article does not contain any studies with human participants or animals performed by any of the authors. No informed consent was obtained.

The authors have nothing to disclose. There are no potential competing interests.

V.J. conceived the study, assessed the quality of the studies, and drafted the manuscript. S.U. searched the literature, assessed the quality of the studies, performed the statistical analysis, and drafted the manuscript. A.S. participated in the statistical analysis and drafted the manuscript. All authors read and approved the final manuscript.