Proactive antithyroid therapy improves survival in hyperthyroidism-induced liver injury: A real-world cohort study

Abstract

BACKGROUND

The initial management of antithyroid drug (ATD)-naive patients presenting with hyperthyroidism and concurrent liver injury represents a critical clinical dilemma. While uncontrolled thyrotoxicosis itself can induce significant hepatocellular damage, clinicians often withhold ATDs due to concerns about exacerbating liver injury through drug-induced hepatotoxicity, leading to delayed definitive therapy and uncertain outcomes.

AIM

To evaluate the efficacy and safety of ATDs in treating hyperthyroidism-induced liver injury (HTI-LI), and to explore its impact on long-term prognosis.

METHODS

In this retrospective and prospective follow-up real-world cohort study, we enrolled 140 hospitalized patients with HTI-LI who were ATD-naïve at diagnosis. Patients were stratified into an ATD treatment group (n = 100) and a non-ATD group (n = 40) based on actual clinical management. Comprehensive longitudinal data on thyroid function, liver biochemistry, coagulation profiles, and clinical outcomes were collected. All-cause mortality was ascertained through active telephone follow-up over a median of 54 months. Multivariate Cox proportional hazards models were used to assess the independent association between ATD use and survival.

RESULTS

ATD therapy led to rapid and significant normalization of thyroid hormones (free triiodothyronine, free thyroxine, thyroid stimulating hormone; all P < 0.05) and was associated with marked improvement in liver injury markers – including alanine aminotransferase, aspartate aminotransferase, total and direct bilirubin, alkaline phosphatase, and γ-glutamyl transferase – as well as restoration of hepatic synthetic function (elevated total protein) and coagulation parameters (reduced prothrombin time and international normalized ratio; all P < 0.05). Critically, ATD use conferred a profound survival benefit: The 3-year mortality rate was 9.4% in the non-ATD group (including liver protection, nutritional support, and symptomatic treatment) vs only 3.2% in the ATD group (P = 0.008). Kaplan-Meier analysis revealed a sustained 97% overall survival in the ATD group over 54 months, compared to a precipitous decline to 72.5% at 36 months. After adjusting for baseline confounders, ATD therapy remained an independent protective factor against death (hazard ratio = 0.24, 95%CI: 0.10-0.60; P = 0.001).

CONCLUSION

These findings demonstrate that, under close monitoring, early initiation of ATD therapy in ATD-naïve patients with HTI-LI is not only safe but also highly effective – simultaneously controlling thyrotoxicosis, promoting hepatic recovery, and substantially improving long-term survival. Our results challenge the prevailing conservative “watchful waiting” paradigm and provide robust evidence to support proactive, individualized ATD use as a cornerstone of integrated management in this high-risk population.

Key Words: Hyperthyroidism; Liver Injury; Liver dysfunction; Antithyroid drug; Graves’ Disease

Core Tip: Hyperthyroidism-related liver injury is common in clinical practice, yet evidence for initial antithyroid drug (ATD) use in ATD-naive patients is insufficient. This study demonstrates that early ATD treatment is safe and effective, can improve thyroid function and liver injury recovery, and may reduce long-term adverse outcomes, providing a reference for clinical decision-making.

INTRODUCTION

Hyperthyroidism is a common endocrine disease affecting approximately 2% of women and 0.2% of men worldwide, with a female-to-male ratio of approximately 10:1[1]. Liver involvement is a common complication that greatly increases the complexity of clinical management[2]. In the context of hyperthyroidism, liver injury is clinically common and the causes are complex and diverse. It may be due to thyrotoxicosis itself, antithyroid drug (ATD) exposure, or other potential liver diseases[3]. Initial management of hyperthyroidism with liver injury constitutes a key treatment challenge. From a pathophysiological point of view, rapid control of thyroid hormone overdose is essential to reduce liver damage caused by thyrotoxicosis. On the contrary, clinicians often hesitate to initiate ATD treatment in the presence of abnormal liver function, which may cause or aggravate drug-induced liver injury (DILI). Such concerns often lead to the adoption of a conservative “close observation” strategy that prioritizes liver protection measures while delaying or avoiding the initiation of ATD treatment. Although this strategy may reduce the risk of DILI, it allows persistent thyrotoxicosis to continue to damage the liver. In view of this high-risk situation, the available treatment methods are still limited to supportive therapy, radioactive iodine (131I) ablation and/or artificial liver support system[4,5]. Thyroidectomy is only suitable for cases with poor tolerance or poor efficacy of ATD[6]. It is worth noting that no studies have retrospectively or prospectively evaluated the safety and efficacy of early initiation of ATD, especially for hyperthyroidism patients with liver injury caused by thyrotoxicosis. In addition, thyroidectomy has the risk of causing permanent hypothyroidism and symptomatic hypocalcemia, which may affect the quality of life and prolong the length of hospital stay[7].

Although this clinical problem is common, high-quality evidence that can provide a basis for management decisions is still extremely scarce. Previous studies have focused on ATD-induced hepatotoxicity or liver enzyme abnormalities that are common in hyperthyroidism, and the investigation of ATD patients with confirmed liver injury is extremely rare[8]. Therefore, the current clinical guidelines do not give clear recommendations for this situation, making treatment decisions largely dependent on the judgment of individual clinicians. The innovation of our work is to systematically evaluate the risk-effectiveness of active initiation of ATD therapy in this high-risk population who have not received ATD therapy. Our results provide important evidence for solving this clinical dilemma and lay the foundation for the establishment of a results-based standardized treatment pathway.

In order to solve this clinical problem, we conducted a cohort study of newly diagnosed patients with hyperthyroidism and liver injury (not treated with ATD). To answer the following questions: (1) To evaluate the efficacy of initial ATD treatment in the recovery of thyroid function; (2) To evaluate the dynamic effects of ATD treatment on key indicators of liver function and coagulation dysfunction; and (3) To evaluate the relationship between ATD treatment and long-term all-cause mortality. We hypothesize that in the case of liver injury mainly caused by thyrotoxicosis, the benefits of timely ATD treatment, that is, rapid control of potential causes, will outweigh the potential risks of drug-related hepatotoxicity. These findings provide solid, evidence-based guidance for this common but challenging clinical scenario, challenge previous conservative treatment concepts, and support the development of standardized, results-oriented initial management strategies.

MATERIALS AND METHODS

Ethical approval

This study was a real-world cohort study, supplemented by active telephone follow-up to confirm final outcomes. The study protocol was approved by the Ethics Committee of the Second Xiangya Hospital, Central South University (No. LYEC2026-0107), and informed consent was waived for patients in the retrospective medical record review component, while verbal informed consent was obtained from all participants during active telephone follow-up.

Study population

We screened hospitalized patients diagnosed with hyperthyroidism with liver injury in the Department of Infectious Diseases from January 1, 2009 to October 31, 2024. This process was completed through the hospital’s electronic medical record system. The diagnosis of hyperthyroidism is based on typical clinical manifestations and serum thyroid function test results [decreased thyroid stimulating hormone (TSH), increased free thyroxine (FT4) and/or free triiodothyronine (FT3)]. Acute liver injury refers to a clinical disease characterized by a sharp deterioration of liver function, and in patients without liver disease or with liver disease but in a relatively stable state, related clinical symptoms occur within 2 weeks after direct or indirect exposure to various liver injury risk factors. According to the definition of the American Association for the Study of Liver Disease, drug-induced liver injury (DILI) refers to liver damage caused by the ingestion of prescription drugs, over-the-counter drugs, herbal products or dietary supplements. Hyperthyroid liver injury is defined as liver dysfunction caused by hyperthyroidism itself, not by ATDs or other liver disease causes[9].

Inclusion criteria: (1) Age ≥ 18 years; (2) Met the diagnostic criteria for hyperthyroidism; (3) No prior use of any ATDs at initial diagnosis (i.e., ATD-naïve); and (4) Presence of liver injury at initial diagnosis, defined as alanine aminotransferase (ALT) and/or aspartate aminotransferase (AST) > 3 times the upper limit of normal, or total bilirubin (TBIL) > 2 times upper limit of normal.

Exclusion criteria: (1) Presence of other clear causes of liver injury at admission (e.g., recent use of herbal medicines or other hepatotoxic drugs); (2) Incomplete medical records; and (3) Missing key baseline clinical data (thyroid function, liver function) that precluded evaluation.

Study design

This study is a single-center, real-world-based cohort study, which was completed in the Second Xiangya Hospital, Central South University. The subjects were hospitalized patients who were diagnosed with hyperthyroidism liver injury and did not receive ATDs treatment from January 2009 to October 2024. According to the actual clinical management, the patients were divided into ATD treatment group and non-ATD treatment group. Baseline characteristics, thyroid function, liver biochemical indicators, coagulation function and long-term prognosis information were collected through electronic medical records, supplemented by active telephone follow-up. The primary outcome was all-cause mortality, and secondary outcomes included improvements in thyroid function and liver injury indicators. This study was approved by the Institutional Ethics Committee and implemented in accordance with the Declaration of Helsinki.

Data collection

We retrospectively collected the following data from electronic medical records: (1) Demographic and baseline characteristics: Age, gender, hyperthyroidism (such as Graves ' disease), comorbidities, etc.; (2) Thyroid function indicators: TSH, FT4, FT3, total triiodothyronine, total thyroxine, adrenergic receptor antibody (thyrotropin receptor antibody), thyroglobulin antibody (double globulin antibody), thyroid peroxidase antibody before and after treatment; (3) Liver injury related indicators: ALT, AST, alkaline phosphatase, γ-glutamyl transferase, TBIL, direct bilirubin, albumin, total protein and international normalized ratio before and after treatment; (4) Other indicators: White blood cells, red blood cell (RBC), platelet before and after treatment; (5) The existence of complications; hospitalization time; disease burden (hospitalization expenses); (6) Treatment and grouping: According to the actual treatment, the patients were divided into “ATD treatment group” and “non-ATD treatment group” (the latter included patients who received only β-blockers or simple supportive care). The type of ATD [methimazole (MMI) or propylthiouracil (PTU)], initial dose and time to start treatment were recorded; and (7) Measurement of results: The main result was all-cause death. Secondary outcomes included: (1) Improvement rate of thyroid function after ATD treatment (defined as FT4 normalization or reduction > 50%); and (2) The normalization or improvement rate of liver function (ALT, TBIL) and coagulation function [prothrombin time (PT)] in the same period.

Follow-up and outcome confirmation

Patient survival status was first confirmed through the hospital’s electronic medical record system. For patients without a clear death record in the system, uniformly trained researchers conducted active telephone follow-up for 54 months to confirm their survival status (alive or deceased). All death information was verified whenever possible through statements from family members.

Statistical analysis

Continuous variables are expressed as mean ± SD or median (interquartile range), and categorical variables are expressed as n (%). The comparison between groups was performed by t-test, Mann-Whitney U test or χ² test, depending on the distribution of data. It is worth noting that none of the 140 patients with hyperthyroidism-induced liver injury (HTI-LI) received ATD treatment before admission. In addition, all subsequent analyses (including survival analysis and Cox regression analysis) were performed only for these 140 patients with HTI LI, while 191 patients with ATDI LI were only used for baseline description and background presentation and were not included in any outcome analysis. For a small number of missing values in baseline variables (missing rate < 5%), we used a full case analysis method, that is, patients with complete data included in the multivariate Cox model only for key variables (age, baseline ALT, baseline FT4, and ATD treatment status). No interpolation processing was performed.

In order to assess the independent impact of ATD treatment on mortality risk, we used a multivariate Cox proportional hazards regression model. The follow-up time was calculated from the date of diagnosis of liver injury until the date of death or the end of the follow-up period. ATD treatment (with/without) was included as the main independent variable in the model, and potential confounding factors (such as age, gender, baseline FT4 level, baseline ALT level, and presence or absence of cardiovascular disease) were adjusted. Considering the small number of events, the multivariate model is limited to ≤ 2 covariates to follow the events-per-variable principle. In addition, propensity score weighting is used as another method to consider multiple confounding factors. The results were expressed as hazard ratio (HR) and 95%CI.

As for the secondary results, we used the paired t-test or Wilcoxon signed rank sum test to compare the dynamic changes of the indicators in the ATD treatment group before and after treatment. All analyses were performed using SPSS 27.0 version software, and two-tailed P values < 0.05 were considered statistically significant.

To further address the potential confounding factors due to non-randomized treatment allocation, we calculated the standardized mean differences between the ATD (+) group and the ATD (-) group for all baseline characteristics (Supplementary Table 1). In addition, we used logistic regression to estimate the propensity score, where ATD acceptance was the outcome variable, and baseline covariates (age, FT3, AST, TBIL, and length of hospital stay) were predictors. The inverse probability treatment weighting method and the weighted Cox proportional hazard model were used to reassess the association between ATD treatment and mortality.

RESULTS

Baseline characteristics of the study cohort

A total of 436 patients with hyperthyroidism and liver injury were initially identified (Figure 1). After excluding 105 patients (33 caused by herbs, 33 caused by other drugs, and 49 incomplete data), a final analysis of 331 patients was performed: (1) 140 patients with HTI-LI were not treated with ATDs at the time of diagnosis; and (2) 191 patients with ATD-induced liver injury. In our cohort, almost all patients treated with ATD were treated with MMI (only 3 patients were treated with PTU). The initial dose was 5-20 mg per day (individualized adjustment based on baseline thyroid hormone levels and body weight), and the maintenance dose was 5-20 mg per day. The median length of hospital stay was 21.5 days (range 7-43 days). We adjusted the dose according to the dynamic monitoring of thyroid function (FT3, FT4, TSH) and liver function (ALT, AST, TBIL), and gradually reduced the dose or discontinued temporarily according to the situation.

Figure 1 Study flow diagram. ATD: Antithyroid drug.

Of the 140 HTI-LI patients who did not receive ATD treatment, 100 (71.4%) received ATD treatment during hospitalization [ATD (+) group], and 40 (28.6%) did not receive ATD treatment [ATD (-) group]. As shown in Table 1, the clinical phenotype of HTI-LI patients at baseline was more severe than that of ATD-induced liver injury patients. Specifically, the incidence of ascites in the HTI-LI group was significantly higher (20.0% vs 2.6%, P < 0.001), and the incidence of hepatic encephalopathy was higher (12.1% vs 2.1%, P < 0.001). In addition, the prognostic scores of patients in the HTI-LI group were also significantly worse, including Child-Pugh classification (P = 0.003), model for end-stage liver disease (MELD) classification (P < 0.001) and MELD-Na classification (P = 0.002). In addition, the median length of stay in HTI-LI was longer (20 days vs 17 days, P = 0.028), and the frequency of hospitalization was also higher (P = 0.002).

Table 1 Comparison of baseline characteristics between hyperthyroidism-induced and antithyroid drug-induced liver injury groups, n (%)/median (interquartile range).

Items	Total (n = 331)	HTI-LI (n = 140)	ATDI-LI (n = 191)	P value
Demographic characteristics
Age (years)	42 (34.5, 52.2)	43 (35.1, 50.9)	42 (34.2, 52.5)	0.975
Male	143 (43.2)	53 (37.9)	90 (47.1)	0.116
BMI (kg/m2)	22.5 (18.2, 25.1)	22.6 (18.3, 25.6)	21.9 (18.1, 23.9)	0.246
Non-drinkers	260 (78.8)	102 (72.9)	158 (83.2)	0.033
Length of hospital stay	17 (8, 26)	20 (11, 32)	17 (10, 27)	0.028
Number of hospitalizations in the past year				0.002
0	260 (78.5)	89 (63.6)	171 (89.5)	0.016
1	45 (13.6)	33 (23.6)	12 (6.3)	0.005
≥ 2	20 (6.0)	15 (10.7)	5 (2.6)	< 0.001
In-hospital medical costs (CNY)	36225 (14870, 42335)	38006 (15242, 43220)	34319 (14182, 40107)	0.589
Clinical presentation
Hepatic injury symptoms	285 (86.1)	123 (87.9)	162 (84.8)	0.550
Hyperthyroid symptoms	264 (79.7)	108 (77.1)	156 (81.6)	0.358
Ascites	33 (9.9)	28 (20)	5 (2.6)	< 0.001
Hepatic encephalopathy	21 (6.3)	17 (12.1)	4 (2.1)	< 0.001
Hyperthyroid heart disease	42 (12.7)	11 (7.8)	31 (16.2)	0.012
Prognostic score
ALBI	-0.52 (-1.84, 0.21)	0.13 (-0.79, 0.27)	-1.35 (-3.13, 0.15)	0.021
Child-Pugh	8.7 (6.3, 11.5)	10.2 (8.9, 12.8)	7.1 (5.8, 9.1)	0.003
MELD	17.4 (12.0, 23.1)	20.6 (15.4, 25.2)	13.2 (10.6, 19.2)	< 0.001
MELD-Na	19.3 (13.1, 24.5)	22.5 (17.4, 27.6)	15.6 (11.4, 20.6)	0.002

The antithyroid drug-induced liver injury group is shown for baseline comparison only and was not included in the primary analyses. BMI: Body mass index; ALBI: Albumin-bilirubin index; MELD: Model for end-stage liver disease; HTI-LI: Hyperthyroidism-induced liver injury; ATDI-LI: Antithyroid drug-induced liver injury.

Importantly, in the HTI-LI group, the ATD (+) and ATD (-) groups maintained a good balance in major demographic and clinical characteristics at baseline (all P values were greater than 0.05; Table 2). This comparability has laid a solid foundation for evaluating the impact of ATD treatment on clinical outcomes.

Table 2 Comparison of baseline characteristics between antithyroid drug-induced liver injury patients with antithyroid drug exposure and those without antithyroid drug exposure, n (%)/median (interquartile range).

Items	Total (n = 140)	ATD (-) (n = 40)	ATD (+) (n = 100)	P value
Demographic characteristics
Age (years)	43 (35.1, 50.9)	45 (37.3, 52.8)	41 (31.2, 49.2)	0.062
Male	53 (37.9)	17 (42.5)	36 (36.0)	0.601
BMI (kg/m2)	22.6 (18.3, 25.6)	21.6 (18.1, 24.9)	23.5 (19.6, 25.9)	0.082
Non-drinkers	102 (72.9)	28 (70.0)	74 (74.0)	0.786
Length of hospital stay	20 (11, 32)	25 (15, 36)	18 (10, 28)	0.015
Number of hospitalizations in the past year				0.542
0	89 (63.6)	24 (60.0)	65 (65.0)	0.536
1	33 (23.6)	8 (20.0)	25 (25.0)	0.462
≥ 2	15 (10.7)	4 (10.0)	11 (11.0)	0.824
In-hospital medical costs (CNY)	38006 (15242, 43220)	45120 (38217, 48320)	33319 (12286, 39113)	0.021
Clinical presentation
Hepatic injury symptoms	123 (87.9)	34 (85.0)	89 (89.0)	0.428
Hyperthyroid symptoms	108 (77.1)	28 (70.0)	80 (80.0)	0.218
Ascites	28 (20)	7 (17.5)	21 (21.0)	0.310
Hepatic encephalopathy	17 (12.1)	6 (15.0)	11 (11.0)	0.318
Hyperthyroid heart disease	11 (7.8)	3 (7.5)	8 (8.0)	0.612
Prognostic score
ALBI	0.13 (-0.79, 0.27)	0.15 (-0.52, 0.31)	0.10 (-1.03, 0.20)	0.221
Child-Pugh	10.2 (8.9, 12.8)	10.3 (9.0, 13.1)	9.8 (8.1, 11.4)	0.620
MELD	20.6 (15.4, 25.2)	21.8 (16.6, 26.7)	19.4 (12.6, 23.2)	0.102
MELD-Na	22.5 (17.4, 27.6)	24.5 (18.5, 28.2)	20.8 (14.1, 25.6)	0.185

BMI: Body mass index; ALBI: Albumin-Bilirubin Index; MELD: Model for end-stage liver disease; ATD: Antithyroid drug.

Efficacy of ATD therapy on thyroid function and hematological parameters

In order to evaluate the therapeutic effect, we compared the changes of laboratory indexes before and after treatment in ATD (+) group and ATD (-) group (Table 3, Figure 2). At baseline, there was no significant difference in thyroid or coagulation function between the two groups (all P values were greater than 0.05), which confirmed their comparability.

Figure 2 Comparison of clinical parameters between antithyroid drug-treated and non-antithyroid drug groups before and after treatment. A-D: They compare the indicators of blood routine, coagulation function, thyroid function, and liver function before and after treatment in the non-antithyroid drug treatment group, including white blood cell, red blood cell, neutrophil (NEUT), platelet (PLT), hemoglobin, international normalized ratio, prothrombin time (PT), activated partial thromboplastin time (APTT), thyroid-stimulating hormone (TSH), free triiodothyronine (FT3), free thyroxine (FT4), thyrotropin receptor antibody (TRAb), thyroid peroxidase antibody, thyroglobulin antibody, albumin (ALB), alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), direct bilirubin (DBIL), total thyroxine (TT4), and alkaline phosphatase (ALP). Among these, APTT, ALB, ALT, AST, TBIL, TT4, and ALP all showed P < 0.05, indicating statistically significant differences before and after treatment; E-H: They compare the indicators of blood routine, coagulation function, thyroid function, and liver function before and after treatment in the antithyroid drug treatment group, including white blood cell, red blood cell, NEUT, PLT, hemoglobin, international normalized ratio, PT, APTT, TSH, FT3, FT4, TRAb, thyroid peroxidase antibody, thyroglobulin antibody, ALB, ALT, AST, TBIL, and DBIL. Among these, NEUT, TT3, TSH, PT, APTT, FT3, FT4, TRAb, ALB, ALT, AST, TBIL, DBIL, PLT, TT4, ALP, and GGT all showed P < 0.05, indicating statistically significant differences before and after treatment. It presents P values for the differences between the two groups [antithyroid drug (+) vs antithyroid drug (-)] at pre-treatment and post-treatment time points for each laboratory indicator. P values were calculated using an appropriate statistical test (e.g., independent t-test or Mann-Whitney U test). A P value < 0.05 was considered statistically significant. WBC: White blood cell; RBC: Red blood cell; NEUT: Neutrophil; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALB: Albumin; TP: Total protein; TBIL: Total bilirubin; DBIL: Direct bilirubin; ALP: Alkaline phosphatase; GGT: Γ-glutamyl transferase; FT3: Free triiodothyronine; FT4: Free thyroxine; TT3: Total triiodothyronine; TT4: Total thyroxine; TRAb: Thyrotropin receptor antibody; A-TPO: Thyroid peroxidase antibody; TgAb: Thyroglobulin antibody; PT: Prothrombin time; APTT: Activated partial thromboplastin time; INR: International normalized ratio; TSH: Thyroid stimulating hormone; PLT: Platelet; HGB: Hemoglobin; ATD: Antithyroid drug.

Table 3 Comparison of laboratory findings between patients with hyperthyroidism-induced liver injury with antithyroid drug exposure and those without antithyroid drug exposure, n (%)/median (interquartile range).

Items	Total (n = 140)	ATD (-) (n = 40)	ATD (+) (n = 100)	P value
Laboratory findings pre-treatment
Blood routine
WBC (× 109/L)	6.21 (5.42, 8.17)	6.45 (5.61, 8.21)	6.03 (5.38, 8.02)	0.570
RBC (× 1012/L)	3.83 (2.94, 4.32)	3.68 (2.92, 4.21)	3.95 (2.98, 4.54)	0.161
PLT (× 109/L)	142.3 (106.5, 204.8)	164.8 (114.4, 210.0)	136.0 (92.3, 190.4)	0.172
HGB (g/L)	114.3 (93.0, 130.9)	111.9 (92.3, 126.6)	118.1 (95.1, 132.2)	0.166
NEUT (× 109/L)	4.91 (2.19, 7.56)	4.72 (2.12, 7.33)	5.15 (2.23, 7.68)	0.590
Hepatic function
ALT (μ/L)	284.9 (128.6, 315.5)	284.4 (128.3, 310.0)	286.1 (130.4, 318.4)	0.983
AST (μ/L)	231.7 (105.3, 283.6)	241.5 (109.2, 294.0)	224.6 (102.6, 281.3)	0.779
TBIL (μmol/L)	272.3 (86.1, 405.4)	280.9 (85.6, 451.1)	269.9 (97.8, 373.9)	0.810
DBIL (μmol/L)	196.5 (55.8, 336.2)	206.2 (50.6, 368.6)	193.2 (68.1, 260.8)	0.705
TP (g/L)	61.92 (48.8, 76.1)	62.47 (49.3, 76.3)	61.54 (48.4, 75.9)	0.630
ALB (g/L)	28.22 (21.8, 34.6)	27.81 (21.6, 34.3)	28.85 (22.7, 34.9)	0.246
GGT (U/L)	83.28 (68.36, 105.5)	74.11 (61.42, 96.32)	98.34 (71.38, 118.0)	0.1268
ALP (U/L)	162.4 (113.6, 214.4)	159.5 (109.2, 208.3)	173.0 (119.2, 226.4)	0.4333
Coagulation function
PT (seconds)	18.87 (14.92, 24.16)	18.67 (14.23, 23.62)	19.05 (15.33, 25.13)	0.781
APTT (seconds)	48.87 (38.02, 61.24)	49.91 (38.36, 62.42)	48.42 (37.92, 60.67)	0.631
INR	1.62 (1.22, 2.15)	1.61 (1.21, 2.14)	1.63 (1.23, 2.16)	0.906
Thyroid function
FT3 (pmol/L)	10.20 (5.12, 13.62)	8.71 (3.24, 13.58)	11.63 (6.50, 14.13)	0.077
FT4 (pmol/L)	19.41 (8.12, 24.55)	17.02 (3.76, 24.00)	21.72 (10.76, 25.00)	0.067
TT3 (nmol/L)	3.50 (2.47, 4.36)	3.22 (2.30, 4.20)	3.77 (2.89, 4.56)	0.114
TT4 (nmol/L)	127.2 (96.5, 150.3)	122.8 (83.5, 146.0)	139.7 (110.0, 156.9)	0.091
TSH (μIU/mL)	0.24 (0.01, 0.42)	0.32 (0.01, 0.51)	0.17 (0.01, 0.34)	0.198
TRAb (IU/mL)	15.08 (8.16, 17.00)	15.22 (7.80, 18.20)	14.99 (9.10, 16.80)	0.904
TgAb (μ/mL)	210.6 (29.1, 370.3)	222.1 (34.1, 349.0)	202.6 (27.9, 400.0)	0.748
A-TPO (μ/mL)	501.2 (39.3, 913.6)	533.7 (39.5, 990.0)	477.9 (39.2, 880.0)	0.5634
Laboratory findings post-treatment
Blood routine
WBC (× 109/L)	5.62 (3.67, 7.03)	5.85 (3.38, 7.76)	5.40 (4.01, 6.64)	0.446
RBC (× 1012/L)	3.93 (3.10, 4.57)	3.57 (2.71, 4.35)	4.25 (3.64, 4.83)	0.003
PLT (× 109/L)	180.0 (93.4, 269.3)	155.7 (79.0, 209.0)	218.2 (127.0, 312.0)	0.021
HGB (g/L)	112.2 (96.0, 128.2)	108.0 (87.0, 121.9)	118.6 (104.0, 133.5)	0.061
NEUT (× 109/L)	3.36 (2.01, 4.33)	3.58 (1.87, 4.53)	3.23 (2.18, 4.02)	0.456
Hepatic function
ALT(μ/L)	48.20 (29.0, 63.4)	51.64 (27.0, 72.3)	45.34 (31.0, 58.3)	0.358
AST (μ/L)	56.18 (31.2, 70.8)	59.73 (33.5, 72.8)	54.25 (29.5, 69.8)	0.558
TBIL (μmol/L)	94.0 (21.8, 121.3)	117.6 (23.0, 139.5)	70.18 (20.0, 107.3)	0.025
DBLI (μmol/L)	71.69 (11.0, 95.2)	90.81 (11.5, 123.6)	53.43 (10.5, 75.2)	0.033
TP (g/L)	66.18 (59.9, 73.4)	64.36 (57.9, 70.2)	69.03 (63.1, 77.0)	0.024
ALB (g/L)	34.28 (31.0, 38.0)	34.09 (30.4, 38.3)	34.69 (32.2, 37.9)	0.566
GGT (U/L)	72.17 (42.6, 83.0)	85.11 (57.8, 90.0)	59.14 (36.6, 66.0)	0.039
ALP (U/L)	106.2 (75.0, 124.5)	120.6 (87.0, 134.0)	91.90 (65.0, 113.5)	0.0005
Coagulation function
PT (seconds)	14.64 (13.1, 15.3)	15.90 (13.6, 17.6)	13.91 (12.5, 15.3)	0.007
APTT (seconds)	41.00 (36.2, 46.7)	43.49 (36.0, 48.3)	40.40 (36.5, 42.2)	0.135
INR	1.21 (0.98, 1.31)	1.34 (1.07, 1.48)	1.12 (0.90, 1.28)	0.017
Thyroid function
FT3 (pmol/L)	6.35 (4.00, 9.21)	7.13 (4.00, 11.00)	5.49 (4.00, 6.46)	0.035
FT4 (pmol/L)	7.12 (2.03, 12.30)	8.90 (1.86, 16.12)	6.09 (2.12, 9.00)	0.033
TT3 (nmol/L)	2.49 (1.81, 3.02)	2.58 (1.89, 3.12)	2.42 (1.79, 2.93)	0.472
TT4 (nmol/L)	64.12 (23.5, 82.1)	71.82 (23.4, 91.0)	55.53 (23.7, 78.0)	0.023
TSH (μIU/mL)	1.02 (0.21, 1.54)	0.52 (0.01, 0.74)	1.53 (0.70, 2.40)	< 0.0001
TRAb (IU/mL)	8.37 (3.50, 10.04)	11.43 (4.71, 15.82)	5.15 (2.53, 7.01)	< 0.0001
TgAb (μ/mL)	140.6 (23.5, 260.3)	183.9 (29.3, 336.0)	109.1 (19.2, 155.3)	0.071
A-TPO (μ/mL)	382.1 (42.5, 712.0)	478.5 (56.0, 768.0)	317.6 (39.5, 678.0)	0.061

WBC: White blood cell; RBC: Red blood cell; PLT: Platelet; NEUT: Neutrophil; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALB: Albumin; TP: Total protein; TBIL: Total bilirubin; DBIL: Direct bilirubin; ALP: Alkaline phosphatase; GGT: Γ-glutamyl transferase; FT3: Free triiodothyronine; FT4: Free thyroxine; TT3: Total triiodothyronine; TT4: Total thyroxine; TRAb: Thyrotropin receptor antibody; A-TPO: Thyroid peroxidase antibody; TgAb: Thyroglobulin antibody; PT: Prothrombin time; APTT: Activated partial thromboplastin time; INR: International normalized ratio; ATD: Antithyroid drug; TSH: Thyroid stimulating hormone; HGB: Hemoglobin.

After treatment, the ATD (+) group showed statistically significant and significant improvement in thyroid function indicators. The levels of FT3 (P = 0.035), FT4 (P = 0.033) and total thyroxine (P = 0.023) in serum were significantly decreased, while TSH (P < 0.0001) and thyrotropin receptor antibody (P < 0.0001) were significantly increased, indicating that the control effect on thyrotoxic state was significant. At the same time, the coagulation function index was also significantly restored compared with that before treatment, and the RBC count (P = 0.003) and platelet count (P = 0.021) were significantly increased. On the contrary, the ATD (-) group showed only slight and insignificant changes in these parameters, which further indicated that the observed improvement effect was related to the intervention effect of ATD.

Impact of ATD therapy on liver and coagulation function

We next evaluated the effect of ATD therapy on liver and coagulation profiles. At baseline, liver and coagulation indices were comparable between the ATD (+) and ATD (-) groups (all P > 0.05; Table 3, Figure 3).

Figure 3 Matrix of P values for comparisons of clinical parameters between antithyroid drug-treated and non-antithyroid drug groups before and after treatment. A: It compares the differences in white blood cell, red blood cell, neutrophil, international normalized ratio, total triiodothyronine, and thyroid stimulating hormone between the antithyroid drug (ATD) treatment group [ATD (+)] and the non-ATD treatment group [ATD (-)] before and after treatment. Before treatment, there were no statistically significant differences between the two groups for any of these indicators (P > 0.05). After treatment, the differences in red blood cell, international normalized ratio, and thyroid stimulating hormone between the two groups were statistically significant (P < 0.05); B: It compares the differences in prothrombin time, activated partial thromboplastin time, free triiodothyronine, free thyroxine, thyrotropin receptor antibody, and albumin between the ATD treatment group [ATD (+)] and the non-ATD treatment group [ATD (-)] before and after treatment. Before treatment, none of the differences between the two groups were statistically significant for any of the indicators (P > 0.05). After treatment, the differences in prothrombin time, free triiodothyronine, free thyroxine, and thyrotropin receptor antibody between the two groups were statistically significant (P < 0.05); C: It compares the differences in thyroid peroxidase antibody, thyroglobulin antibody, alanine aminotransferase, aspartate aminotransferase, total bilirubin, and direct bilirubin between the ATD treatment group [ATD (+)] and the non-ATD treatment group [ATD (-)] before and after treatment. Before treatment, there were no statistically significant differences between the two groups for any of these indicators (P > 0.05). After treatment, the differences in total bilirubin and direct bilirubin between the two groups were statistically significant (P < 0.05); D: It compares the differences in platelet, hemoglobin, total thyroxine, alkaline phosphatase, γ-glutamyl transferase, and total protein between the ATD treatment group [ATD (+)] and the non-ATD treatment group [ATD (-)] before and after treatment. Before treatment, there were no statistically significant differences between the two groups for any of these indicators (P > 0.05). After treatment, the differences in platelet, total thyroxine, alkaline phosphatase, γ-glutamyl transferase, and total protein between the two groups were statistically significant (P < 0.05). Rows and columns represent laboratory indicators measured at pre-treatment and post-treatment time points in patients stratified by ATD use: (1) ATD (+) indicates patients receiving ATD therapy; and (2) ATD (-) indicates patients not receiving ATD therapy. Each cell contains the P value from a statistical test (e.g., paired t-test, Wilcoxon signed-rank test, or correlation analysis) assessing the relationship or difference between the corresponding row and column parameters. P values less than 0.05 are considered statistically significant. WBC: White blood cell; RBC: Red blood cell; ATD: Antithyroid drug; NEUT: Neutrophil; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALB: Albumin; TP: Total protein; TBIL: Total bilirubin; DBIL: Direct bilirubin; ALP: Alkaline phosphatase; GGT: Γ-glutamyl transferase; FT3: Free triiodothyronine; FT4: Free thyroxine; TT3: Total triiodothyronine; TT4: Total thyroxine; TRAb: Thyrotropin receptor antibody; A-TPO: Thyroid peroxidase antibody; TgAb: Thyroglobulin antibody; PT: Prothrombin time; APTT: Activated partial thromboplastin time; INR: International normalized ratio; TSH: Thyroid stimulating hormone; PLT: Platelet; HGB: Hemoglobin; Pre: Pre-treatment; Post: Post-treatment.

Analysis of the results after treatment showed that the ATD (+) group significantly improved liver injury and synthetic dysfunction. The key indicators of cholestasis and hepatocyte injury, including TBIL (P = 0.025), direct bilirubin (P = 0.033), alkaline phosphatase (P = 0.0005) and γ-glutamyl transferase (P = 0.039), were significantly reduced. At the same time, the liver synthesis function was also improved, which was manifested by a significant increase in total protein (P = 0.024). In addition, coagulation function was also restored, PT was significantly shortened (P = 0.007), and international normalized ratio was decreased (P = 0.017). These results suggest that ATD therapy can not only control hyperthyroidism, but also actively promote liver injury and liver synthesis function recovery.

Identification of factors independently associated with mortality

In order to understand the predictors of long-term survival, we performed multivariate Cox regression analysis (Figure 4). The model shows that there is a significant and independent correlation between the dynamic changes of key laboratory indicators during treatment and the risk of mortality. Specifically, changes in TSH (P = 0.001), PT (P = 0.006), TBIL (P = 0.004), and MELD score (P = 0.001) were important predictors. It is worth noting that the extension of hospital stay has also become an independent risk factor (P = 0.003). Most importantly, ATD treatment was a powerful independent protective factor that reduced the risk of death (P = 0.001).

Figure 4 Univariate Cox regression analysis of factors associated with time to death. The plot presents the P values for each variable derived from univariate Cox proportional hazards models examining the association with mortality. Variables include changes in thyroid-stimulating hormone, prothrombin time, albumin, total bilirubin, model for end-stage liver disease score, presence of hyperthyroid symptoms, length of hospital stay, age, and antithyroid drug (ATD) treatment status [ATD (-) and ATD (+)]. P values < 0.05 were considered statistically significant. TSH: Thyroid stimulating hormone; PT: Prothrombin time; ALB: Albumin; TBIL: Total bilirubin; MELD: Model for end-stage liver disease; ATD: Antithyroid drug.

Long-term survival benefit of ATD therapy

We conducted telephone follow-up to confirm the long-term prognosis of all 140 HTI-LI patients (Table 4). The survival advantage brought by ATD treatment is significant and persists over time. At 1 year, the mortality of ATD (+) group was significantly lower than that of ATD (-) group (2.0% vs 10.0%) (P = 0.005). This significant difference was maintained at the two-year follow-up point (3.1% vs 11.1%, P = 0.005) and the three-year follow-up point (3.2% vs 9.4%, P = 0.008). Consistently, the ATD (+) group also had significantly fewer hospitalizations and readmissions during the follow-up period (all P < 0.05).

Table 4 Comparison of follow-up results between patients with hyperthyroidism-induced liver injury with antithyroid drug exposure and those without antithyroid drug exposure, n (%)/median (interquartile range).

Items	Total	ATD (-)	ATD (+)	P value
Follow up for 1 year	n = 140	n = 40	n = 100
Death	6 (4.28)	4 (10.00)	2 (2.00)	0.005
Number of hospitalizations in the past year	1 (0, 2)	2 (1, 3)	1 (0, 2)	0.024
Follow up for 2 years	n = 134	n = 36	n = 98
Death	7 (5.22)	4 (11.11)	3 (3.06)	0.005
Number of hospitalizations in the past year	1 (0, 2)	2 (1, 3)	1 (0, 2)	0.026
Follow up for 3 years	n = 127	n = 32	n = 95
Death	6 (4.72)	3 (9.37)	3 (3.15)	0.008
Number of hospitalizations in the past year	1 (0, 2)	2 (1, 3)	1 (0, 2)	0.031

ATD: Antithyroid drug.

Kaplan-Meier survival analysis (Figure 5) demonstrated this result: The ATD (+) group maintained an overall survival rate of 97% throughout the 54-month follow-up period, while the survival rate of the ATD (-) group decreased sharply to 72.5% at 36 months. The cumulative survival curve showed significant differentiation in the early stage and continued to pull away over time. In view of the limited number of deaths (n = 6), we prefer the parsimonious modeling method. In the unadjusted Cox model, ATD treatment was associated with a 76% reduction in the risk of death (HR = 0.24, 95%CI: 0.10-0.60, P = 0.001). This association remained significant after adjusting for age and gender (HR = 0.26, 95%CI: 0.11-0.63, P = 0.003). Similar results were obtained when the baseline MELD score was adjusted only (HR = 0.29, 95%CI: 0.12-0.70, P = 0.006). In the inverse probability treatment weighting-adjusted analysis, ATD treatment was associated with a 72% reduction in the risk of death (HR = 0.28, 95%CI: 0.11-0.72, P = 0.008), which was consistent with the results of the unadjusted model.

Figure 5 Overall survival comparison between antithyroid drug-treated and non-antithyroid drug groups. The table shows the overall survival probabilities over time (in months) for patients receiving antithyroid drug (ATD) therapy [ATD (+)] and those not receiving ATD therapy [ATD (-)]. ATD: Antithyroid drug.

DISCUSSION

This cohort study provides important insights into the management of patients with newly diagnosed hyperthyroidism complicated with liver injury. Our results show that initiating ATD treatment for this high-risk population can not only effectively control thyrotoxicosis, but also significantly improve liver function and coagulation function, and most importantly, it can also bring significant survival benefits. These results challenge the current strategy of delaying the use of ATD and adopting radioactive iodine therapy[10], and support a paradigm shift towards active management of potential thyrotoxicosis.

The most important finding of this study is the significant survival advantage associated with ATD treatment. Patients treated with ATD achieved a survival rate of 97% throughout the 54-month follow-up period, while the survival rate of patients in the non-ATD group decreased to only 72.5%, and the gap continued to widen within 36 months. The survival curve diverged at an early stage, and the gap continued to widen over time, indicating that the benefits of ATD treatment were both immediate and long-lasting. This finding is in contrast to the previous literature. The latter believes that ATDs may further cause drug-induced liver injury, and in patients with impaired liver function, 131I therapy should be the first choice to achieve timely and effective control of hyperthyroidism[11]. Our results suggest that under close monitoring, ATD therapy may provide a safe and effective alternative to address potential thyrotoxicosis and protect liver function.

Our results show that concerns about the adoption of conservative strategies - that ATD may aggravate the original liver damage – have not been supported by evidence. On the contrary, the liver indexes measured in patients treated with ATD were significantly improved, including ALT, AST, TBIL and PT. This suggests that the hepatotoxic effects of uncontrolled thyrotoxicosis in this population far outweigh the potential risk of ATD-induced liver injury. An analysis of data from a retrospective study in China showed that patients with Graves’ disease-related hyperthyroidism and severe liver dysfunction mainly showed symptoms related to liver disease, and low-dose MMI was safe and effective in these patients[12]. The improvement of coagulation function (especially coagulation time) is of clinical significance, which reflects the recovery of liver synthetic function and the reduction of the risk of bleeding complications. Notably, He et al[13] compared the hepatotoxicity risk of PTU and MMI using quantitative systemic toxicology methods, and found that the simulated DILI incidence of PTU (300 mg/day) was 21.2%, while the predicted incidence of MMI (30 mg/day) was 0%. This indicates that the hepatotoxicity of MMI may not be mediated by conventional mechanisms such as oxidative stress, which provides a new theoretical basis for the priority selection of MMI in clinical practice[13].

After the initiation of ATD treatment, thyroid function was rapidly improved, FT4 and FT3 were significantly decreased, and TSH was significantly increased, confirming that ATD can maintain its efficacy even in the case of liver injury. A systematic review published in 2024 summarized the clinical characteristics of 271 patients with ATD-related liver injury, showing that most patients (79.6%) recovered completely after drug withdrawal, with a median recovery time of 45 days[8]. Importantly, the improvement of blood routine indicators, especially RBC count, suggests an improvement in thyrotoxic anemia and reflects the overall clinical treatment effect. The stability of white blood cell counts throughout the treatment indicated that the risk of ATD-induced agranulocytosis was low in this population. Consistent with a previous study, severe drug-induced hepatotoxicity usually occurs within the first three months after the initiation of ATD treatment; therefore, it is necessary to regularly monitor liver function and timely adjust drug dose during ATD treatment[14]. Recently, Annabi et al[15] reported a case of fulminant liver failure, which occurred within 48 hours of the onset of thiamine, indicating that a small number of patients may have hypersensitivity reactions, which is worthy of clinical vigilance.

We performed a multivariate analysis and found an independent association between several factors and mortality, including changes in thyroid hormone releasing hormone (TSH), coagulation time (PT), total bile acid (TBIL), and MELD score during treatment. These underscore the importance of close monitoring of thyroid function and liver biochemical indicators throughout the treatment phase. After adjusting for potential confounding factors, the strong association between ATD treatment and reduced mortality remained, which further confirmed the independent protective effect of ATD treatment. It is worth noting that the systematic review of Khan et al[8] found that patients treated with PTU usually had higher bilirubin and transaminase levels, and their mortality was significantly higher than that of patients treated with MMI. This suggests that when ATD therapy is needed, patients with liver injury should give priority to the use of MMI.

The clinical significance of this study is extremely important. The current suspension or non-use of ATD drugs due to concerns about ATD-induced hepatotoxicity may put patients at risk of further aggravation of the disease and even death due to uncontrolled thyrotoxicosis. An interdisciplinary expert consensus published in 2025 also pointed out that the diagnosis and management of hyperthyroidism patients with abnormal liver function requires the participation of the Department of Endocrinology, the Department of Hepatology and the Department of Clinical Pharmacy to make decisions[16]. Our results show that for patients with hyperthyroidism complicated with liver injury who have not received ATD treatment before, ATD treatment should be initiated under close monitoring, rather than waiting and delaying treatment. This treatment strategy is in line with pathophysiological principles, because treatment for its etiology is crucial for liver recovery. A case report pointed out that discontinuation of ATDs due to concerns about liver injury may lead to rapid deterioration of the disease[17].

Our findings may be explained by multiple mechanisms. Uncontrolled hyperthyroidism can lead to metabolic hyperfunction, which increases the oxygen demand of the liver and reduces the oxygen supply due to high output heart failure and relatively low blood pressure[18]. This mismatch may lead to centrilobular necrosis and progressive liver injury[19]. By rapidly controlling thyrotoxicosis, ATD can reduce metabolic needs, improve cardiac output, and restore hepatic blood flow, thereby contributing to the recovery of liver function. In addition, the reversal of thyrotoxicosis-related catabolic processes may also promote the regeneration and synthesis of hepatocytes[20]. In recent years, the association between gut microbiota and drug-induced liver injury (DILI) has also attracted attention. A review by Li et al[21] points out that the specific composition and abundance of gut microbiota are closely related to the severity of drug-induced liver injury. This may provide new targets for the prevention and treatment of ATD-related hepatotoxicity in the future.

This study has many advantages. First, we specifically focused on the liver injury caused by hyperthyroidism itself in newly diagnosed patients with ATD, thus solving a well-defined clinical problem. Secondly, we collected longitudinal comprehensive data on thyroid function, liver function indexes and clinical results, so as to analyze the therapeutic effect in detail. Third, the use of active telephone follow-up ensures the complete confirmation of survival results and minimizes follow-up bias. Fourth, we used multivariate Cox regression and sensitivity analysis to control for potential confounding factors.

This study has several limitations. First, although we adjusted the confounding factors of measurement and weighted the propensity score, residual confounding from unmeasured factors (such as subjective risk assessment by clinicians) could not be excluded. However, the consistency of the results of various analytical methods supports the reliability of our findings. Second, the sample size is relatively small, especially in the non-ATD group (40 patients), which may limit the statistical power of some subgroup analyses. Third, non-randomized design means that treatment allocation is based on clinical judgment and may introduce indication bias. However, the comparability of baseline characteristics between the two groups and the robustness of the results of sensitivity analysis have alleviated this concern to some extent. Fourth, longer hospital stays and higher costs may reflect greater disease severity, which may influence treatment decisions and outcomes. Fifth, the research objects are mainly Chinese, and the research results may not be directly extended to other ethnic groups. Sixth, the missing data is mainly limited to a few baseline laboratory parameters (such as ALT, FT4); although the missing rate is low (< 5%), a complete case analysis may still introduce selection bias, and the results should be interpreted carefully. Due to the retrospective data collection, we were unable to obtain detailed causes of death for all patients, which made it impossible to conduct a more detailed analysis to distinguish between liver-related mortality and non-liver-related mortality. Finally, this study conducted a number of statistical tests, which may increase the risk of false positive results. In view of the exploratory nature of our analysis and the limited number of tests, we did not use rigorous Bonferroni correction to avoid overly conservative estimates that would mask clinically relevant signals. Therefore, readers should interpret the uncorrected P value as exploratory evidence rather than confirmatory conclusion; all findings are worthy of verification in an independent cohort.

In the future, more prospective studies with larger sample size and multi-center design are needed to verify our findings and determine the optimal timing and dose strategy of ATD treatment for this population. In addition, it is necessary to further study whether specific patient subgroups (such as patients with other underlying liver diseases) can benefit from ATD treatment and whether their benefits are different. In addition, the comparative study of the effects of different ATD drugs (MMI and PTU) in this context is helpful to guide the selection of drugs. Finally, incorporating our findings into clinical practice guidelines and developing standardized treatment regimens may improve the prognosis of this group.

CONCLUSIONS

In conclusion, this study shows that for patients with hyperthyroidism and liver injury who have never received ATDs treatment, initiating ATD treatment can effectively control thyroid function and significantly improve liver function. Importantly, our study found that the use of ATDs in such patients can bring significant survival benefits. In other words, the benefits of initiating ATD treatment under close monitoring far outweigh the risks. These findings are helpful for clinical decision-making and provide guidance for the development of evidence-based treatment programs for such patient groups.

ACKNOWLEDGEMENTS

We would like to thank all participants for collecting the data of this study.