FT3 to FT4 Ratio is Associated with Non-Alcoholic Steatohepatitis and Significant Fibrosis in Euthyroid Subjects with NAFLD

Background & Aims: Studies on the relationship between thyroid function and non-alcoholic fatty liver disease (NAFLD) among euthyroid subjects had shown inconsistent results. Objective of the present study was to exploring the independent relationship between thyroid function parameters and nonalcoholic steatohepatitis (NASH), significant fibrosis (SF) respectively after adjusting other well-identified risk factors. Method: This study enrolled 307 patients with biopsy-proven NAFLD. Thyroid dysfunction defined as serum thyroid-stimulating hormone > 4.5 mIU/l or < 0.5 mIU/l and/ or free thyroxine > 14.41 pmol/l or < 7.86 pmol/l. Results: Stepwise regression analysis showed that the fT3/fT4 ratio, an optimal thyroid function parameter, was associated with NASH and SF in euthyroid subjects with NAFLD. After multivariable analysis, the fT3/fT4 ratio (per 0.1 change) showed significant correlation with NASH (OR 2.03(1.33, 3.11), P = 0.001) and SF (OR 2.11(1.28, 3.46), P = 0.003). When ratio was stratified by quartiles (Q1-Q4) as a categorical variable, the results still significant (Q4 versus Q1 (OR for NASH 4.29(1.68, 10.91), P = 0.002; for SF 4.63(1.49, 14.43), P = 0.008, all P for linear trend < 0.05). The prevalence of NASH and SF rose significantly with increasing in quartiles. Furthermore, ratio was positively correlated with the grade of steatosis, lobular inflammation, hepatocellular ballooning and liver fibrosis stage (all P < 0.05). Subgroup analysis showed that hypertension was a possible effect modification. Conclusion: The findings of the present study confirmed an association between fT3/fT4 ratio and NAFLD in euthyroid subjects in a dose-dependent manner, particularly in non-hypertension adults. There is no trial registration number.

liver biopsy-proven NASH and advanced fibrosis 12-14 . A reasonable explanation is that the thyroid hormones is associated with body fat distribution, metabolic syndrome, and insulin resistance 15, 16 . Recently, some studies have shown that variation in thyroid hormones in the reference range may also have negative effects on health, same to subclinical and clinical hypothyroidism 17 . Meanwhile, the relationship between normal thyroid function and NAFLD has also gained attention. However, there is no consensus. In the previous findings, based on different sample sizes and diagnosis methods of NAFLD, they showed that low-normal thyroid function (higher plasma TSH level [2.5 to 4.5 mIU/L] with a normal thyroid hormone level) 18 , low serum free thyroxine level(fT4) 19 , high serum free triiodothyronine level(fT3) 20, 21 , and high fT3/fT4 ratio 22 were risk factors for NAFLD respectively.
For this reason, we performed a study in euthyroid subjects with NAFLD aimed at exploring the independent relationship between thyroid function parameters (i.e., triiodothyronine (T3), thyroxine (T4), fT3, fT4, fT3/fT4 ratio and thyroid-stimulating hormone (TSH) levels) and biopsy-proven NASH, significant fibrosis respectively after adjusting other well-identified risk factors (e.g. metabolic risk factors and insulin resistance).

Patients and study design
No informed consent was required because all the data were anonymized. The protocol was in accordance with the Helsinki Declaration and was approved by the ethics committee of the First Affiliated Hospital of Wenzhou Medical University. From our liver biopsy database at The First Affiliated Hospital of Wenzhou Medical University, the clinical records of 1371 subjects during January 2016 and July 2019 were reviewed. Strict exclusion criteria are designed and implemented: (1) ≤ 18-year-old, (2) excessive alcohol consumption (> 140 g/week for men and 70 g/week for women) evaluated by a questionnaire, (3) the steatosis (≤ 5% of liver cells) at histology and history of viral hepatitis, autoimmune hepatitis, or other forms of chronic liver disease, (4) history of malignancy, (5) history of thyroid disease, including clinical hyperthyroidism and hypothyroidism, thyroidectomy, radiofrequency ablation of thyroid gland, (6) with thyroid dysfunction defined as serum TSH > 4.5 mIU/l or < 0.5 mIU/l and/ or fT4 > 14.41 pmol/l or < 7.86 pmol/l, (7) insufficient clinical date. In the end, 307 strictly screened subjects were enrolled in this study. Data was collected from the time of liver biopsy. All variables in this study were objective results stored in the hospital computer system. The presence of diabetes mellitus (fasting blood glucose > = 7.0 mmol/L or treatment with antidiabetic drugs) was recorded. Hypertension was defined as systolic blood pressure ≥ 140 mmHg/diastolic blood pressure ≥ 90 mmHg and/or the current use of anti-hypertensive medication. Smokers were defined as those who had smoked at least one cigarette per day during the previous year.

Liver biopsy
Liver biopsy was performed by senior operators using 16-gauge Hepafix needles under ultrasonography positioning. The liver specimens were fixed in 10% formalin, and was scored by experienced hepatologists who were blinded to the clinical data, treatment allocation, and imaging findings. A scoring system published by Kleiner et al 24 was used. The histological NAFLD Active Score (NAS) is defined as the unweighted sum of the scores for steatosis (0-3), lobular inflammation (0-3), and hepatocellular ballooning (0-2); scores therefore ranged from 0 to 8. Subjects with scores of 5 or greater were diagnosed as NASH. Fibrosis was staged as follows: stage 0 = no fibrosis; stage 1 = perisinusoidal or periportal fibrosis with 3 different patterns: 1a = mild, zone 3, perisinusoidal; 1b = moderate, zone 3, perisinusoidal fibrosis, and 1c = portal/periportal fibrosis; stage 2 = perisinusoidal and portal/periportal fibrosis; stage 3 = bridging fibrosis; stage 4 = cirrhosis. In this study, we pooled the subtype 1a, 1b, 1c of fibrosis into a single F1 score. Significant fibrosis (SF) was defined as stage 2 or greater (≥ 2).

Statistical Analysis
Continuous data were presented as median (1st quartile, 3rd quartile), and categorical variables were expressed in frequency or as a percentage. First, the univariate analysis (student t-test, Mann-Whitney U test, chi-square test) were used to infer the difference between the 2 groups. It was noted that the fT3/fT4 ratio was too small; therefore, we expanded it 10 times and labelled per 0.1 change (henceforth fT3/fT4 ratio (per 0.1 change)). Second, significant variables from the univariate analysis (P < 0.05) were then subjected to stepwise logistic regression analysis (Probability for Stepwise enter: 0.05, remove: 0.1) to evaluate the risk factors for NASH, significant fibrosis respectively. Third, according to the recommendation of the STROBE statement 25 , we simultaneously showed the results from unadjusted, minimally adjusted analyses and those from fully adjusted analyses. Fourth, Spearman's correlation analysis was performed to assess the relationship between thyroid function parameter and histological features. Fifth, subgroup analyses were performed using stratified linear regression models. The modifications and interactions of subgroups were inspected by likelihood ration tests. All of the analyses were performed with the statistical software packages R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empowerstats.com, X&Y Solutions, Inc., Boston, MA). P values less than 0.05 (two-sided) were considered statistically significant.

Subject characteristics
The current study collective included 307 subjects (24.76%, females; 75.24%, males), 80 (men, 75.47%) and 42 subjects (men, 72.41%) were classified as biopsy-proven NASH and SF, respectively. The baseline characteristics of this cohort were shown in Tables 1 and 2. In both NASH and SF groups, there were higher level of BMI, TC, LDL-c, ALT, AST, type IV collagen, type III procollagen and HOMA-IR with the direction indicating that patients with histological severity reflected a poorer health status with more severe liver damage and active extracellular matrix synthesis and degradation. Contrary to the previously reported studies, there was no difference in prevalence of diabetes mellitus in univariate analysis, but the level of HOMA-IR was significantly greater in subjects with NASH and SF than in subjects without it (4.57 vs.3.12 in NASH; 5.09vs.3.48 in SF, all P < 0.001). One possible explanation was this study's cohorts were younger than in other studies, and insulin resistance predates beta-cell dysfunction 26 . With respect to thyroid function parameters, the NASH group was characterized by a higher level of TSH, T3, fT3 and fT3/fT4 ratio; and the SF group had higher level of T3, fT3 and fT3/fT4 ratio and lower fT4.

Risk factors of NASH, SF respectively
Stepwise logistic regression analysis was performed to evaluate the risk factors for NASH, SF respectively. Variables that were significant in univariate analysis would enter the regression analysis. The results showed that ALT (OR  Table 2). A notable finding was that fT3/fT4 ratio, an optimal thyroid function parameter, was found to be significantly associated with risk factor for NASH and SF.

Correlation between fT3/fT4 ratio and histological features
Univariate linear regression models were used to evaluate the associations between fT3/fT4 ratio and NASH, SF respectively. Meanwhile, we showed the non-adjusted and adjusted models in  14.43) P = 0.008), significant linear trend was observed (all P for trend < 0.05). In addition, the prevalence rate of NASH and SF showed an increasing trend as fT3/fT4 ratio increased (Fig. 1). Spearman's correlation analysis showed that the fT3/fT4 ratio was positively related to more severe histological features of NAFLD presenting the higher grade of steatosis, lobular inflammation, hepatocellular ballooning, and liver fibrosis stage (all P < 0.05) (Fig. 2). Abbreviations: Non-alcoholic steatohepatitis(NASH), significant fibrosis(SF), odds ratio(OR), confidence interval(CI).
Crude model did not adjust for other covariants.
Model Ⅰadjusted for age and sex.

The results of subgroup analyses
As shown in Table 4, the test for interactions was significant for hypertension (P for interaction = 0.011 in NASH, 0.042 in SF), while the test for interactions were not statistically significant for sex, age, smoke, diabetes, HOMA-IR and BMI both in NASH and SF groups (all P values for interactions were larger than 0.05). We observed that there was evidence for an interaction fT3/fT4 ratio and hypertension. The effect sizes of fT3/fT4 ratio on NASH and SF showed significant differences in subjects with or without hypertension. FT3/FT4 ratio (per 0.1 change) was associated with NASH and SF in subjects who were without hypertension (OR for NASH 2.77(1.66, 4.62); OR for SF 2.78(1.54, 5.04)), but had no relationship with NASH and SF in hypertension group. Table 4 Effect size of fT3/fT4 ratio on NASH, SF respectively in subgroups Abbreviations: Non-alcoholic steatohepatitis(NASH), significant fibrosis(SF), odds ratio(OR), confidence interval(CI).

Discussion
In this cross-sectional study performed among 307 euthyroid subjects, the main finding was that the fT3/fT4 ratio was associated with biopsy-proven NASH and SF in subjects with NAFLD, independently of well-known metabolic risk factors. Besides, in the subgroup analyses, an effect modification by hypertension on the association was found. Inconsistent with our conclusion, a study in euthyroid elderly Chinese, which diagnosed NAFLD based on ultrasonography and included thyroid function parameters (TSH, fT3 and fT4), found that serum fT4 level was significantly associated with the risk for NAFLD (odds ratio 0.847) 19 . Another study with a similar approach from China found high levels of serum fT3 was significantly associated with NAFLD among middle-aged euthyroid subjects independently of known metabolic risk factor (odds ratio: 1.253) 20 . Kim et al, based on noninvasive marker to diagnose liver fibrosis, found a significant correlation between low-normal thyroid function / TSH with advanced fibrosis in the US general population 18 . On the one hand different diagnostic methods of NAFLD are the most important reason for inconsistent results, on the other hand the aforementioned studies did not set the variable fT3/fT4 ratio. Liver biopsy is the gold standard for NAFLD diagnosis, the accuracy of using noninvasive markers and ultrasonography for the diagnosis of NALFD may influenced by the misclassification 27 and hyposensitivity 28 . Eline et al 21 and Fatma et al 22 previously reported that higher fT3/fT4 ratio but not TSH is a risk factor for NAFLD independent of metabolic parameters, which is consistent with ours.
The pathophysiological mechanisms between fT3/fT4 ratio and NAFLD are multifactorial and not fully understood. It has been hypothesized that as an indicator of peripheral deiodinase activity, the higher fT3/fT4 ratio, higher conversion rate from fT4 to fT3, is a compensatory response to improve energy consumption 29 . An increase in expression of the genes for type I iodothyronine 5′-deiodinase and activity in adipose tissue of obese humans had been observed 30 . Moreover, a positive correlation between fT3/fT4 ratio and BMI, waist circumference, HOMA-IR in a cohort of euthyroid women have been reported 31 . All of the above could partially explain the potential relationship between fT3/fT4 ratio and NAFLD.
Our study also explored a possible effect modification by hypertension on the association between ratio and NASH, SF respectively. We concluded that the association was significant in subjects without hypertension, but not with hypertension. Gu et al has demonstrated that FT3 and FT4 are positively related to the prevalence of elevated blood pressure in euthyroid adults, which may have caused the phenomenon observed in our study 32 . However, further studies are required to validate and explicate the effect modification noted in our study.
Although our cross-sectional study did not reveal an in-depth mechanism, our finding illustrates the necessity to actively assess serum fT3/fT4 ratio in patients with NAFLD. From a clinical point of view, this present finding suggests the assessment of the fT3/fT4 ratio should be included in the multidisciplinary baseline assessment of patients with NAFLD, and that decreasing the fT3/fT4 ratio may be a promising potential therapy for curing NAFLD or preventing progression. Several studies 33, 34 had demonstrated a beneficial effect of low dose Levothyroxine replacement therapy on NAFLD with a decrease in the prevalence of NAFLD and serum liver enzymes.
To our knowledge, our study is the first and largest analysis to date evaluating the association between thyroid function parameters and biopsy-proven NAFLD in euthyroid population. However, there are also limitations to this study. Firstly, the cross-sectional design of the study was possible to investigate associations but not causalities. Secondly, since the levels of thyroid antibody were not detected, we were unable to explore the possible effects of the impending thyroid autoimmunity on the association between NAFLD and thyroid function. Finally, this study does not reflect institutional and regional diversities because the cohorts of this study were composed of Chinese patients who were recruited at a single hospital.

Conclusion
This study shows that the fT3/fT4 ratio was independently associated with NASH and significant fibrosis in a dose-dependent manner in euthyroid subjects with NAFLD, particularly in non-hypertension adults. These associations persisted after further adjustment for obesity, metabolic risk factors, and insulin resistance. Future studies should elucidate the exact role of fT3/fT4 ratio in the development or progression of NAFLD, and explore the feasibility of adjusting ratio in the prevention and treatment of NAFLD.

Ethics approval and consent to participate
The study was approved by the ethics committee of the First Affiliated Hospital of Wenzhou Medical University(Wenzhou, Zhengjiang, China). No informed consent was required because all the data were anonymized.
HBX and JMW designed this study, HBX, YQG, JZZ, ZQX and YL Collected data. HBX and JMW conducted the statistical analyse. HBX, YQG, JZZ and ZQX interpreted data. HBX and JMW reviewed the result and wrote manuscript. HBX, YQG, ZQX, YL and JZZ revised manuscript. All authors read and approved manuscript.

Figure 2
Correlation analysis between fT3/fT4 ratio and histological features spectrum of NAFLD.