The relationship between liver histology and thyroid function tests in patients with non-alcoholic fatty liver disease (NAFLD)

Background Data on the role of hypothyroidism in the pathogenesis of non-alcoholic fatty liver disease (NAFLD) and liver fibrosis are conflicting, although selective Thyroid Hormone Receptor (THR)-β agonists have been identified as potential therapy in patients with non-alcoholic steatohepatitis (NASH). Therefore, we investigated the association between hypothyroidism and NAFLD histological features potentially associated with progressive liver disease. Methods Between 2014 and 2016, consecutive patients with histologically proven NAFLD and frozen serum available for thyroid function tests assessment were included. NAFLD was staged according to the NAFLD Activity Score (NAS), and fibrosis according to Kleiner. NASH was defined as NAS ≥4, significant fibrosis as F2-F4 and significant steatosis as S2-S3. Thyroid function tests (TFT; TSH, FT3, FT4, rT3), TPO-Ab and Tg-Ab were also assessed. Results Fifty-two patients were analyzed: median age 54 years, 58% females, LSM 7.8 kPa, 27% diabetics, 14% hypothyroid. At histology, NASH was present in 21 (40%), F2-F4 in 28 (54%) and S2-S3 in 30 (58%) patients. Rates of hypothyroidism were similar independently of the presence of NASH (p = 0.11), significant fibrosis (p = 0.21) or steatosis (p = 0.75). However, hypothyroid patients displayed a higher NAS (p = 0.02) and NASH (p = 0.06) prevalence. At multivariate analysis, TFT were not independently associated with histology. Conclusion Hypothyroidism was highly prevalent in NAFLD patients, and was associated with increased NAFLD activity, but not with fibrosis and steatosis severity. Thus, thyroid dysfunction might play a direct and/or indirect in the pathogenesis of NAFLD and NASH.


Introduction
of chronic liver disease (i.e. viral, autoimmune, drug-induced, vascular and inherited disorders) were excluded. Patients with significant alcohol consumption (20g/day for women and 30g/day for men) were also excluded from the analysis.
Anthropometric and clinical data were collected on the same day of LB, as well as liver stiffness measurement (LSM), through transient elastography (TE). Blood samples were concomitantly collected and served to retrospectively assess thyroid function tests (see below). Thyroid function tests were performed blindly on blood samples collected at the time of liver biopsy, after study approval by our Ethic Committee.
Diabetes and impaired fasting glucose (IFG) were defined according to the standard of care of the American Diabetes Association [4], while dyslipidemia was defined according with the most recent guidelines of the European Society of Cardiology (ESC) and the European Atherosclerosis Society (EAS) [30]. LDL-cholesterol was calculated by the Friedewald Equation.
Data were collected between January and March 2020. Written informed consent was obtained from each patient included in the study. The study protocol was approved by the Institutional Board of our Department (Ethical Committee Milan Area 2) and conforms to the ethical guidelines of the 1975 Declaration of Helsinki.
TPO-Ab and Tg-Ab higher than 34 IU/ml and 115 IU/ml, respectively, were considered positive.
Hypothyroidism was defined as TSH �4.5 μU/l and classified as subclinical or overt hypothyroidism according with the presence of normal or reduced levels of FT4 and FT3, respectively. Euthyroid patients were further divided in two subgroups according to previous suggestion for NAFLD patients [31]: a) "strictly normal" TSH 0.45-2.5 and b) "borderline/ high" TSH >2.5 μU/l.

Statistical analysis
Categorical variables were reported as frequencies (percentages) and continuous variables as median (range). Categorical variables were compared using the χ2 or the Fischer's exact tests; continuous variables were compared using the Student's t test, the Mann-Whitney U-test or the Kruskall-Wallis test, when appropriate. All tests were 2-sided and used a significance level of 0.05. Univariate and multivariate logistic regression analyses were performed to identify factors associated with histological parameters. Quantil regression was used to correct the association between NAS and hypothyroidism for age and BMI. Data handling and analysis were performed with Stat-View package (SAS Institute Inc., Cary, NC).

Thyroid function tests
In the overall cohort, 38 (73%) patients were euthyroid, and 14 (27%) had hypothyroidism. Six out of hypothyroid patients (43%) were taking L-T4 replacement treatment at the time of liver biopsy, because of a previous diagnosis of primary (n = 5) or central (n = 1) hypothyroidism. The remaining eight (57%) patients were affected with newly diagnosed hypothyroidism, including one female with overt hypothyroidism (Table 3).
Patients with hypothyroidism were mostly females (p = 0.002). As expected, they had higher CRP (C-Reactive Protein) (p = 0.009) but lower ferritin (p = 0.07) values (Table 3). At  histology, they displayed a higher prevalence of NASH (p = 0.06) and higher NAS score (p = 0.02). No differences in term of either fibrosis or steatosis severity were noted ( Table 3). The correlation between thyroid function and NAS remained significant after adjusting for age and BMI (coefficient 1.7, 95%CI 0.4-3.3 p = 0.045).
Among euthyroid patients, no significant differences were noted between subjects with TSH <2.5 μU/ml compared with those with TSH between 2.5 and 4.5 μUI/l.

Univariate and multivariate analysis
Clinical variables significantly associated with NASH, significant fibrosis (F2-F4) and significant steatosis (S2-S3) at univariate and multivariate analysis are reported in Table 4. At multivariate analysis, age and significant steatosis were independently associated with NASH, IFG/ diabetes and LSM were independently associated with F2-F3 and low HDL values and NASH with S2-S3 (Table 4).

Discussion
In the present study, we analyzed a well-characterized cohort of consecutive patients with histologically-proven NAFLD, and reported an univariable association between histological  activity and hypothyroidism, whose prevalence was significant (27%), in line with previous observations [6,30] showing similar prevalence of hypothyroidism (either overt or subclinical) in NAFLD patients (15.2%-36.3%). Interestingly, in our cohort most (57%) hypothyroid patients were affected with an undiagnosed primary hypothyroidism, thus suggesting that TSH screening is recommended in patients with NAFLD. In our study, we found that NAFLD patients with hypothyroidism compared with those who were euthyroid had significantly higher NAS values, with an increased, although not statistically significant, prevalence of NASH (89% vs. 53%, p = 0.06). At univariable analysis, this association remained statistically significant also after adjusting for age and BMI. Conversely, we did not find any association between hypothyroidism and the severity of either fibrosis or steatosis, in spite of a slightly increased prevalence of hypothyroidism in patients with F2-F4 and S2-S3 compared with those with lower severity scores. In our study, we found similar prevalence of hypothyroidism (p = 0.11) as well as similar TSH (p = 0.18) values between patients with or without NASH. However, patients with untreated hypothyroidism displayed more severe NAS (p = 0.02) and higher prevalence of NASH (p = 0.06).
Surprisingly, very few data exist regarding the association between hypothyroid and fibrosis, which is the strongest predictor of disease severity also in NAFLD/NASH patients [25]. In our study, fibrosis was not influenced by thyroid function, independently of fibrosis severity. Conversely, Kim et al. in their euthyroid NAFLD cohort reported a higher prevalence of advanced fibrosis (F3-F4) in patients with low (TSH 2.5-4.5 μIU/l) vs. strict-normal thyroid function (TSH 0.4-2.5 μIU/l). Moreover, subclinical hypothyroidism (TSH >4.5) and lownormal thyroid function (TSH 2.5-4.5) were independent predictors of NASH and advanced fibrosis [27]. Additionally, some authors tried to assess the correlation between TSH and fibrosis by using non-invasive tools instead of liver biopsy, whose use is currently limited in clinical practice. In the Rotterdam study, Bano et al. studied nearly 9,500 NAFLD patients and reported that the risk of having liver fibrosis (LSM �8.0 kPa.) positively correlated with TSH levels and negatively with FT4 levels [35]. Conversely, Manka et al. found that low FT3 (but not FT4 or TSH) levels were independently associated with high LSM and NAFLD fibrosis score (NFS) [36]. Due to differences in study design and results, data on the association between hypothyroidism and liver fibrosis deserve further investigations.
Finally, although hypothyroidism may be associated with weight gain and overweight, no conclusive data exist on the association between thyroid dysfunction and steatosis; our study for the first time investigated the relationship between histological steatosis and thyroid function, without finding any association.
Taken together, these results validate the hypothesis that hypothyroidism might be an indirect player in the pathogenesis of NAFLD and NASH, which amplifies several features of the metabolic syndrome, such as endothelial dysfunction, inflammation and insulin resistance.
Interestingly, in our cohort, we did not find any difference between biochemical and histological data of hypothyroid patients on L-T4 or left untreated. In this context, Liu et al found that T4 supplementation improves NAFLD especially in patient with TSH>10 and associated dyslipidemia [37], but only in half of the cases. This may suggest that since the pathogenesis of liver damage in NAFLD is complex and multifactorial, L-T4 alone only partially impacting on this condition.
On the other hand, the availability of liver-selective TH analogs with a higher affinity for the THR-β, the main THR isoform expressed in the liver, might be an interesting treatment also for hypothyroid patients with NAFLD not responsive to L-T4 alone. Among THR-agonists, Resmetirom (MGL-3196), is 28-fold more selective than T3 for THR-β versus THR-α [18], and, consequently, it is not expected to cause any adverse events at heart or bone level. In addition, its effects are mainly intrahepatic, as suggested by the unchanged TSH levels found in treated patients. In healthy subjects it led to reductions in atherogenic lipids, including LDL, Apo-B and triglycerides [38], thus supporting its current advancement in Phase 3 clinical trials.
Our conclusions should be cautiously interpreted as the limited sample size had probably an impact on our analysis and prevented us to gap the conflicting data still existing in literature on this topic. In addition, this study has been conducted in a tertiary-care center, which could have further influenced characteristics of patients included in this study. Finally, the absence of a control group (i.e. patients without NAFLD) as well as the retrospective design of the study reinforces the need for further and prospective analysis, given the heterogeneity of the results of previous studies.
Beyond these limitations, however, are the strengths of our study, which basically rely on the availability of histological examination as well as on the full clinical characterization of our cohort. In fact, in our study we tried not only to assess differences in thyroid function between patients with and without NASH (NAS < vs. �4), but also investigated the association between hypothyroidism and both histological fibrosis and steatosis. In addition to histological data, our study benefits from the extensive availability of complete clinical and biochemical information, as well as from blinded and centralized assessment/reading of both histology and thyroid function tests.
In conclusion, our data suggest that hypothyroidism might be a direct and/or indirect player in the pathogenesis of NASH/NAFLD, though prospective studies are warranted in order to better define the role of hypothyroidism in the pathogenesis of NASH and its associated histological features, as new and promising molecules acting on metabolic pathways regulated by thyroid hormones are on the way.