Lower free triiodothyronine (fT3) levels in cirrhosis are linked to systemic inflammation, higher risk of acute-on-chronic liver failure, and mortality

Background & Aims Advanced chronic liver disease (ACLD) may affect thyroid hormone homeostasis. We aimed to analyze the pituitary–thyroid axis in ACLD and the prognostic value of free triiodothyronine (fT3). Methods Patients with ACLD (liver stiffness measurement [LSM] ≥10 kPa) undergoing hepatic venous pressure gradient (HVPG) measurement between June 2009 and September 2022 and available fT3 levels were included. Clinical stages of ACLD were defined as follows: probable ACLD (pACLD; LSM ≥10 kPa and HVPG ≤5 mmHg), S0 (mild portal hypertension [PH]; HVPG 6–9 mmHg), S1 (clinically significant PH), S2 (clinically significant PH with varices), S3 (past variceal bleeding), S4 (past/current non-bleeding hepatic decompensation), and S5 (further decompensation). Results Among 297 patients with ACLD, 129 were compensated (pACLD, n = 10; S0, n = 33; S1, n = 42; S2, n = 44), whereas 168 were decompensated (S3, n = 12; S4, n = 97; S5, n = 59). Median levels of thyroid-stimulating hormone (TSH) numerically increased with progressive ACLD stage (from 1.2 μIU/ml [pACLD] to 1.5 μIU/ml [S5]; p = 0.152), whereas fT3 decreased (from 3.2 pg/ml [pACLD] to 2.5 pg/ml [S5]; p <0.001). Free thyroxin levels remained unchanged (p = 0.338). TSH (aB 0.45; p = 0.046) and fT3 (aB -0.17; p = 0.048) were independently associated with systemic C-reactive protein levels. Lower fT3 was linked to higher risk of (further) decompensation (adjusted subdistribution hazard ratio [asHR] 0.60; 95% CI 0.37–0.97; p = 0.037), acute-on-chronic liver failure (asHR 0.19; 95% CI 0.08–0.49; p <0.001) and liver-related death (asHR 0.14; 95% CI 0.04–0.51; p = 0.003). Conclusions Increasing TSH and declining fT3 levels are observed with progressive ACLD stages. The association of TSH and fT3 with systemic inflammation suggests a liver disease-associated non-thyroidal illness syndrome. Lower fT3 levels in patients with ACLD indicate increased risk for decompensation, acute-on-chronic liver failure, and liver-related death. Impact and Implications In a large well-characterized cohort of patients with advanced chronic liver disease (ACLD), we found a decline of free triiodothyronine (fT3) throughout the clinical stages of ACLD, paralleled by a numerical increase of thyroid-stimulating hormone (TSH). This suggests a progressive development of a non-thyroidal illness syndrome in association with ACLD severity. Importantly, C-reactive protein independently correlated with TSH and fT3, linking thyroid dysbalance in ACLD to systemic inflammation. Lower fT3 indicated an increased risk for subsequent development of hepatic decompensation, acute-on-chronic liver failure, and liver-related death. Clinical trial number Vienna Cirrhosis Study (VICIS; NCT: NCT03267615).


Introduction
Advanced chronic liver disease (ACLD) causes considerable morbidity and mortality worldwide. 1,2In early asymptomatic stages, ACLD is considered compensated.With the development of portal hypertension (PH)-related complications, including ascites, hepatic encephalopathy (HE), and variceal bleeding, patients progress to decompensated ACLD (dACLD), [3][4][5] which is associated with a considerably worse prognosis. 6,7Acute-onchronic liver failure (ACLF), defined by hepatic and/or extrahepatic organ failure(s), 8,9 may occur in patients with dACLD 6,10 and is closely associated with systemic inflammation. 11here is a complex interplay between chronic liver disease and thyroid function.On the one hand, the liver is essentially involved in thyroid hormone homeostasis, contributing to their (in) activation, transport, and metabolism. 12On the other hand, hypothyroidism is known to impair lipid metabolism and promote hepatic steatosis. 13Accordingly, the risk of non-alcoholic fatty liver disease (NAFLD) increases with lower free thyroxin (fT4) levels. 14oreover, increased thyroid-stimulating hormone (TSH) was associated with NAFLD in a meta-analysis. 15Accordingly, thyroid hormones and thyroid hormone receptor beta agonists are currently evaluated as potential therapies for NAFLD. 16urthermore, a small study including 33 patients with liver cirrhosis reported lower levels of triiodothyronine (T3) and free T3 (fT3), as well as lower T3/thyroxin-binding globulin (TBG) and thyroxin (T4)/TBG ratios. 17A typical altered pattern of thyroid hormone homeostasis with decreased levels of fT3 and similar levels of fT4 was described in cirrhosis, and defined as 'low fT3 syndrome'. 18Moreover, a small retrospective study demonstrated shorter survival among patients with low fT3, 19 linking dysregulated thyroid signaling to impaired clinical outcomes in patients with cirrhosis.
However, abnormalities of the pituitary-thyroid axis across the distinct substages of cirrhosis, as well as the impact of fT3 on different clinical outcomes, have not yet been systematically investigated.Thus, this study aimed (i) to investigate the pituitary-thyroid axis throughout the individual clinical stages of ACLD, 6 (ii) to analyze parameters associated with the dysregulation of thyroid hormones in ACLD, and (iii) to determine the impact of fT3 on clinical outcomes of ACLD.

Study design
This study investigated consecutive patients with suspected ACLD (i.e.liver stiffness measurement [LSM] > − 10 kPa) undergoing HVPG measurement at the Hepatic Hemodynamic Lab of the Vienna General Hospital between June 2009 and September 2022 following a standardized operating procedure. 20We exclusively included clinically stable patients in whom data on TSH and fT3 were available.
Patients with a history of liver transplantation (LT), hepatocellular carcinoma (HCC) or other malignancy, vascular liver disease, portal vein thrombosis (PVT), transjugular intrahepatic portosystemic shunt (TIPS), congestive heart disease, intake of thyroid hormones, or active infection at the time of HVPG measurement were excluded.Moreover, we excluded patients with insufficient available clinical or laboratory data.
Central venous blood withdrawals for assessment of pituitary-thyroid axis parameters and HVPG measurements were conducted under fasting conditions.Clinical events occurring during follow-up (FU) including the development of decompensation events (clinically apparent ascites, variceal bleeding, HE, and ACLF), as well as acute kidney injury, HCC, LT, and (liver-related) death, were documented.

Patient cohorts
Patients were stratified for ACLD severity according to clinical stages modified from D'Amico et al. 6 for baseline characterization of the pituitary-thyroid axis in patients with ACLD.Compensated ACLD (cACLD) was defined by the absence of any previous decompensating event defined and was further subdivided into four distinct stages: probable ACLD (pACLD; LSM > − 10 kPa and HVPG < − 5 mmHg), 21 S0 (mild PH; HVPG 6-9 mmHg), S1 (clinically significant PH [i.e.HVPG > − 10 mmHg] without the presence of varices), and S2 (clinically significant PH with the presence of varices).dACLD was constituted by a history of at least one decompensation event.Patients with dACLD were assigned to one of three substages: S3 (history of acute variceal bleeding), S4 (history of decompensation as a result of non-bleeding decompensating events including ascites and HE), and S5 (further decompensation; i.e. history of two or more different types of decompensating events).

Decompensation events
In cACLD, decompensation was defined by occurrence of clinically evident ascites, HE, or variceal bleeding.In dACLD, further decompensation was defined by either occurrence of a second type of hepatic decompensation or worsening of the first decompensation.Worsening included development of spontaneous bacterial peritonitis (SBP) or refractory ascites in patients with ascites, first hospital (re-) admission caused by HE in patients with previous HE, and variceal rebleeding in patients with a history of variceal bleeding.Finally, ACLF following the EASL/EF-CLIF definition 10,22 was regarded as a fourth and distinct type of decompensating event, which could potentially occur in patients with cACLD and those with dACLD.

Assessment of HVPG and LSM
A standardized HVPG measurement procedure was followed as previously described. 20In brief, a catheter introducer sheath was placed in the right internal jugular vein following local anesthesia.Subsequently, a specifically designed angled-tip balloon catheter 23 was inserted into a large hepatic vein under fluoroscopic guidance.HVPG was calculated by subtracting free from wedged hepatic vein pressure.A total of three measurements were performed per patient, and the median HVPG of these measurements was used for further analyses.Vibrationcontrolled transient elastography (FibroScan ® ; Echosens, Paris, France) under fasting conditions was used for LSM, as previously described. 24tuitary-thyroid axis parameters and routine laboratory parameters All laboratory tests were conducted at the ISO-certified Department of Laboratory Medicine of the Vienna General Hospital.Blood withdrawals were conducted in a standardized manner after HVPG measurement under fasting conditions and after resting in the supine position for at least 30 min.Serum levels of TSH (normal range 0.27-4.2lIU/ml; Elecsys TSH, Roche Diagnostics, Mannheim, Germany), as well as fT4 (normal range 0.76-1.66ng/dl; Elecsys fT4 III, Roche Diagnostics, Mannheim, Germany) and fT3 (normal range 2.15-4.12pg/ml; Elecsys fT3 III, Roche Diagnostics) were analyzed by electrochemoluminescence immunoassay (Roche Diagnostics).Standard laboratory methods were used for the assessment of routine laboratory parameters.Low fT4 and low fT3 were defined as fT4 <lower limit of normal (LLN) and fT3 <LLN, respectively.Elevated TSH was defined as TSH >upper limit of normal (ULN).

Statistical analysis
The number (n) and proportion (%) of patients with the parameter of interest were reported for categorical variables.Continuous data were presented using median and IQR.For comparison of continuous non-normally distributed variables between two groups, the Mann-Whitney U test was implemented.The Kruskal-Wallis test was performed for comparing continuous variables in three or more groups.Pearson's Chi-square test or Fisher's exact test, as appropriate, was used for group comparisons of categorical variables.Spearman's rho (q) was used to assess correlations.Linear regression analysis was conducted to investigate factors associated with TSH and fT3.Clinical outcomes at half-year intervals up to 2 years of FU were assessed using cumulative incidence calculation.Gray's test, as previously described, 25 was computed for cumulative incidence comparison.Fine and Gray competing risk regression models using the R package cmprsk 25,26 were calculated to evaluate whether TSH, fT4, and fT3 levels were associated with the risk of clinical events of interest.Apart from TSH, fT4, and fT3, well-established risk factors for worse outcomes in ACLD (i.e.age, Child-Turcotte-Pugh [CTP] score, serum creatinine, sodium, HVPG as a marker for PH, and C-reactive protein [CRP] as a parameter of systemic inflammation) as well as sex were evaluated by univariate and multivariate analysis.Etiological cure, HCC, LT, and non-liverrelated death were considered as competing risks for both cumulative incidence comparisons and competing risk regression models.In both the linear regression models and the competing risk regression models, parameters that were associated with the parameter of interest in univariate analysis (p <0.100) were included in the multivariate model.Moreover, TSH, fT3, or fT4 as the main target parameters, as well as BMI, CTP score, and HVPG as established parameters of pathophysiological and prognostic relevance, were included in the multivariate models.IBM SPSS 27.0 statistic software (IBM, Armonk, NY, USA), R 4.2.1 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria), and GraphPad Prism 8 (GraphPad Software, La Jolla, CA, USA) were used for statistical analyses.A two-sided p value of <0.05 was considered statistically significant.

Ethics
The study was approved by the ethics committee of the Medical University of Vienna (No. 1493/2016 and No. 1262/2017) and was performed in accordance with the current version of the Helsinki Declaration.All patients included between 2018 and 2022 are part of the prospective Vienna Cirrhosis Study (VICIS; NCT: NCT03267615) and gave their written informed consent before study inclusion.The ethics committee waved the need for informed consent for the retrospective part of the study.

Patient characteristics
In total, 297 patients with ACLD with a median age of 56.0 years and male predominance (67.0%) were included in the study.The patient flow chart is depicted in Fig. 1.Baseline characteristics of the included patients are provided in Table 1.The main etiologies were alcohol-related liver disease (53.9%), viral hepatitis (16.5%), and metabolic dysfunction-associated steatohepatitis (MASH; 11.8%).Overall, the study included 56.6% of patients with dACLD most frequently caused by ascites (85.7% of patients with dACLD).The median HVPG was 17 mmHg, and the median model for end-stage liver disease (MELD) was 11 points.Pituitary-thyroid axis parameters in patients with different clinical stages of ACLD With progressive ACLD severity, patients exhibited numerically higher levels of TSH (from 1.2 lIU/ml [pACLD] to 1.5 lIU/ml [S5]; p = 0.152; Fig. 2 and Table S1).Still, TSH mostly remained within the normal range with only overall nine patients (3.0%; no difference between distinct clinical stages; see Table S1 S1), significantly increased with progressive clinical ACLD stage.
Interestingly, as detailed in Table S2, TSH and fT3 did not differ between distinct BMI strata in both patients with cACLD and patients with dACLD.

FU and clinical outcomes
The median FU period was 419.0 days.Overall, 25.3% of patients (n = 75) experienced at least one (further) decompensation event.Details on the rates of different decompensation events are given in Table 3.Ten patients (3.4%) developed HCC during FU.TIPS implantation was conducted in 5.7% (n = 17) of patients, and 19 patients (6.4%) underwent LT.Furthermore, 11.1% (n = 33) of patients died with 20 of these deaths (60.6%) being attributed to liver disease.

Discussion
In this study, we demonstrated that the pituitary-thyroid signaling axis becomes gradually impaired with progressing stages of ACLD and that low fT3 levels are of critical prognostic value in a large well-characterized cohort of patients with ACLD.Specifically, lower fT3 levels across the distinct clinical ACLD substages were paralleled by increases in TSH, whereas fT4 remains mostly unchanged.Importantly, fT3 and TSH were independently linked to CRP, suggesting a key role of systemic inflammation in the dysregulation of thyroid hormone signaling in ACLD.Furthermore, lower fT3 levels correlated with parameters of liver dysfunction, PH, endothelial dysfunction, and hyperdynamic circulation, indicating a potential role of altered thyroid signaling in ACLD progression.Finally, we demonstrated that fT3as both a continuous variable and dichotomized  variableconfers important prognostic value, as it was an independent predictor of ACLF development and liver-related mortality.
It has long been known that thyroid hormone dysregulation occurs in liver cirrhosis 17 with abnormalities of thyroid hormones described in 13-61% of patients. 12Typically, patients with ACLD exhibit decreased total T3, fT3, and fT4, as well as increased reverse T3 compared with healthy individuals, while usually maintaining normal levels of TSH. 12,27,28Accordingly, in our study the majority of patients with ACLD showed euthyroid function.Furthermore, previous studies have shown that lower fT3 is associated with hepatic dysfunction (i.e. higher Child-Pugh scores), 12,27 whereas there were no differences in the levels of fT4 between patients with compensated and decompensated cirrhosis. 27,28his is in line with our data, demonstrating that fT3 significantly declines with progressive clinical stage of ACLD, along with increasing TSH levels with progressive clinical ACLD stage.Interestingly, the prevalence of elevated TSH was low (3.0% of patients) and did not differ between different stages of ACLD, and median levels of fT4 were virtually identical in different substages of ACLD, indicating that patients with dACLD, despite exhibiting progressively lower fT3 seem to mostly remain in a euthyroid state.
These thyroid hormone alterations correspond to a nonthyroidal illness syndrome or euthyroid sick syndrome, which can be described to occur in patients with severe or chronic illness. 29,30Different underlying pathogenetic mechanisms have been proposed, including a downregulation of type 1 deiodinase, hampering the conversion of T4 to T3, reduced hepatic clearance of reverse T3, or changes in the concentration of thyroid hormone binding proteins in patients with ACLD. 12Notably, we demonstrate a strong linear correlation between lower fT3 and higher CRP, which strongly suggests a critical contribution of systemic inflammation to lowered fT3 levels with progressive ACLD severity.
2][33] It has been demonstrated that chronic inflammation leads to altered transport and metabolism of thyroid hormones. 29,34Moreover, systemic inflammation has also    Etiological cure, hepatocellular carcinoma, liver transplantation, and non-liver-related death were considered as competing risks.Cumulative incidences were compared using Gray's test.ACLF, acute-on-chronic liver failure; fT3, free triiodothyronine.
been linked to a suppression of the pituitary-adrenal axis in ACLD, 35 and indeed, CRP was the only parameter independently linked to TSH in our study population.Thus, chronic inflammation in patients with advanced stages of ACLD may contribute to the dysregulation of thyroid hormone homeostasis both by impairment of the pituitary-thyroid axis and by influencing hormone transport and metabolism.Although a detrimental effect of low fT3 (i.e.fT3 <LLN) on survival in patients with cirrhosis has been previously described, 19 our study adds to this finding by demonstrating that lower fT3 levelseven when remaining in the normal rangewhen considering them as continuous, and not only as a binary parameter, are strongly linked to development of ACLF and liverrelated death.On univariate analyses, lower fT3 levels also predicted (further) decompensation in patients with ACLD, whereas TSH and fT4 did not show any correlations with clinical outcomes.Mechanistically, fT3 showed a correlation to parameters of hepatic dysfunction (i.e.MELD and albumin), PH (i.e.HVPG), endothelial dysfunction (i.e.VWF 36,37 ), and hyperdynamic circulation (i.e.proBNP 7 ), suggesting a role of fT3 in ACLD progression.At this point, further mechanistic studies are required to decipher the impact of fT3 hormones on ACLD progression and particularly on immune dysfunction (cirrhosis-associated immune dysfunction).
Our study also has limitations.Firstly, we did not assess levels of thyroglobulin or reverse T3 and can thus not comment on their role in altered thyroid hormone signaling in ACLD.Thyroglobulin and reverse T3 were previously reported to be elevated in liver cirrhosis. 12,38Secondly, the retrospective design and the long study period represent another limitation.Importantly, we started to systematically assess fT4 and fT3 in April 2018 in patients who were prospectively included and followed up in the VICIS study, which includes 96.4% (n = 287/297) of patients in this study.By assessing thyroid hormones in consecutive patients undergoing HVPG measurement, we can largely exclude selection bias.Thirdly, we could not measure intrahepatic deiodinase expression/activity and systemic T3/T4-binding globulin levels to assess potential loco-hepatic hypothyroidism and dysregulation of the thyroid hormone metabolism in ACLD, respectively.Moreover, we did not assess anti-thyroglobulin or anti-thyroid peroxidase antibodies.However, no patient with autoimmune thyroid disease or intake of antithyroid medication was included in our study.Furthermore, only 11.8% of patients included in this study had MASH as primary liver disease etiology.Further research is required to explore the specific impact of thyroid hormone imbalance in patients with metabolic dysfunction-associated steatotic liver disease and MASH.Finally, we did not assess nutritional status or sarcopenia, which may influence fT3 levels.
In conclusion, our large observational study showed a significant increase in TSH and decline of fT3 levels with progressive clinical stage of ACLD.Mechanistically, lower fT3 seems to be linked to liver dysfunction, PH, endothelial dysfunction, and hyperdynamic circulation.Interestingly, both lower fT3 and higher TSH were associated with biomarkers of systemic inflammation.
Finally, and clinically most relevant, lower fT3 levels were independent predictors of hepatic decompensation, ACLF, and liver-related mortality.Based on our study results, regular monitoring of thyroid hormonesparticularly fT3in patients with ACLD is warranted.Further studies are required to determine whether specific therapeutic interventions correcting thyroid hormone imbalances may benefit patients with ACLD.

Fig. 2 .
Fig.2.Plasma levels of thyroid-stimulating hormone, free thyroxin, free triiodothyronine, and C-reactive protein in patients with different stages of ACLD.Statistical differences between the groups were assessed using Kruskal-Wallis test.Patients were stratified by EASL stages with the addition of patients with pACLD (i.e.HVPG < − 5 mmHg and LSM > − 10 kPa).HVPG, hepatic venous pressure gradient; LSM, liver stiffness measurement; pACLD, probable advanced chronic liver disease.

Fig. 4 .
Fig.4.Cumulative incidence of first/further hepatic decompensation, ACLF, and liver-related death within 2 years of follow-up stratified by fT3 levels.Etiological cure, hepatocellular carcinoma, liver transplantation, and non-liver-related death were considered as competing risks.Cumulative incidences were compared using Gray's test.ACLF, acute-on-chronic liver failure; fT3, free triiodothyronine.

Table 2 .
Levels of parameters of the pituitary-thyroid axis, liver function, and systemic inflammation in patients with cACLD and dACLD.

Table 4 .
Impact of fT3 on the risk of (i) decompensation or further decompensation, (ii) ACLF, and (iii) liver-related death.
(continued on next page)

Table 4
(continued)Univariate and multivariate Fine and Grey competing risk regression models are shown.Etiological cure, hepatocellular carcinoma, liver transplantation, and non-liver-related death were considered as competing risks.P values in bold denote statistical significance.ACLF, acute-on-chronic liver failure; asHR, adjusted sHR; fT3, free triiodothyronine; HVPG, hepatic venous pressure gradient; sHR, subdistribution hazard ratio.

Table - S4. Impact of parameters of hepatic dysfunction, kidney function and systemic inflammation on (i) free triiodothyronine (fT3) and (ii) thyroid stimulating hormone (TSH) levels.
Univariate and multivariate multivariate linear regression models are shown.

Table - S7. Impact of free thyroxin (fT4) on the risk of (i) decompensation/further decompensation, (ii) acute-on-chronic liver failure (ACLF) and (iii) liver-related death.
Univariate and multivariate multivariate competing risk regression models are shown.Etiological cure, hepatocellular carcinoma, liver transplantation and non-liver-related death were considered as competing risks.