Risk of Hepatotoxicity in Patients With Gout Treated With Febuxostat or Benzbromarone: A Propensity Score–Matched Cohort Study

Abstract

Objective

The objective of this study was to evaluate and compare the risk of hepatotoxicity associated with the use of febuxostat and benzbromarone in patients with gout.

Methods

New users of febuxostat or benzbromarone with monitoring of liver function at least three times in a year after initiation of the study drugs were identified from an electronic medical record database. Propensity score matching (PSM) was performed between the two groups 1:1 matched for age, sex, and pretreatment alanine aminotransferase (ALT) and aspartate aminotransferase (AST). Kaplan‐Meier analysis was used to estimate the probability of hepatotoxicity (defined as ALT or AST > 3× upper limit of normal). Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox regression. Subgroup analysis was performed based on age, body mass index, and comorbidities.

Results

A total of 2,338 patients with gout were eligible. A total of 37% of patients experienced Common Terminology Criteria for Adverse Events version 5 grades 1 to 3 for AST or ALT abnormality. After PSM, 488 febuxostat users were matched, with 488 participants receiving benzbromarone with a mean follow‐up of 1.20 years. The incidence of hepatotoxicity was 39.6 and 16.8 per 1,000 person‐years for febuxostat users and benzbromarone users, respectively. Febuxostat use was associated with a significantly greater risk of hepatotoxicity than benzbromarone (adjusted HR 2.75, 95% CI 1.28–5.91), especially in patients with elevated transaminases at baseline. Findings did not differ according to prespecified subgroups.

Conclusion

Febuxostat use is associated with a significantly greater risk of mild‐to‐moderate perturbation of liver function compared to benzbromarone in patients with gout.

INTRODUCTION

Although urate‐lowering therapy (ULT) is widely available and effective, gout remains poorly treated, partly because of concerns about drug side effects. Allopurinol is recommended as the preferred first‐line agent, although this can cause a rare hypersensitivity syndrome that is more common in people of Southeast Asian descent (eg, Chinese, Korean, Thai). ¹ Febuxostat has comparable ULT efficacy to allopurinol. ² Benzbromarone is a potent and effective uricosuric agent primarily used in Asia and was withdrawn from some European markets due to the risk of rare, severe, and potentially lethal liver injury. ³ That limited its expanded global approval. However, the benzbromarone drug withdrawal decision has been questioned based on the estimated risk of hepatotoxicity estimated ⁴ , ⁵ to be <1:17,000. Moreover, benzbromarone has high efficacy and safety even for patients with chronic kidney disease, ⁶ as well as a superior efficacy of urate‐lowering in the urate underexcretion subtype that comprises at least ~60% of gout. ⁷

SIGNIFICANCE & INNOVATIONS.

• The incidence of aspartate aminotransferase or alanine aminotransferase abnormality and hepatotoxicity in patients with gout taking benzbromarone or febuxostat was investigated using a propensity score–matched cohort study design.

• In patients with gout, treatment with febuxostat is associated with a significantly greater risk of hepatotoxicity compared to benzbromarone.

• This study indicates that severe hepatotoxicity with benzbromarone is rare and occurs less frequently than with febuxostat.

Approximately one‐quarter of people with gout have metabolic dysfunction–associated steatohepatitis (MASH), which in turn may increase the risk of drug‐induced liver injury. ⁸ Thus, monitoring of serum liver function tests (LFTs) has been recommended during ULT. ⁹ , ¹⁰ Although rare severe hepatotoxicity cases were reported for both febuxostat and benzbromarone, ³ , ¹¹ the incidence of abnormal liver function tests for febuxostat has been reported to be 2% to 13% (average ~3.5%), in which most adverse events are mild to moderate in severity and reversible after discontinuation of the medication, ¹² and for benzbromarone is 0.1% in clinical trials. ¹³

A meta‐analysis included both two randomized control trials and one cohort study identified that benzbromarone had relatively lower alanine aminotransferase (AST) or aspartate aminotransferase (ALT) value than febuxostat at 10 weeks to 12 months, whereas the included studies were relatively short term and often had a small sample size. ¹⁴ In contrast, other cohort studies reported that no significant difference in transferase levels between the two groups. ¹⁵ , ¹⁶ There is currently a lack of comparative data on hepatic safety of relatively long‐term febuxostat or benzbromarone use. Understanding the risk of hepatotoxicity and ULT use may help guide decisions about specific ULT agents. The aim of this real‐world cohort study was to evaluate and compare the risk of hepatotoxicity associated with the use of febuxostat and benzbromarone in patients with gout.

MATERIALS AND METHODS

Study cohort and design

The study cohort was from an electronic medical record database Biobank Information Management System (BIMS) (Haier) at the Affiliated Hospital of Qingdao University. The database included people seen at the Shandong Gout Clinical Medical Center with a diagnosis of gout and demographic characteristics, serum biochemical test value, and treatment regimens recorded at each clinic visit since 2016. Race and ethnicity were assessed through self‐reported methods using a fixed set of standardized categories. Patients who started febuxostat or benzbromarone and had at least three tests of ALT or AST in a year after initiation of the study drugs were included in the analysis. Patients were excluded for the following reasons: age <18 years old, no follow‐up data, allopurinol starters, and baseline ALT or AST above 120 U/L. Febuxostat‐starters were matched 1:1 to benzbromarone‐starters according to age, sex, ALT, and AST value at the baseline time. The flowchart is shown in Figure 1.

Figure 1.

Flow of patients in the analysis. ALT, alanine aminotransferase; AST, aspartate aminotransferase; Ps, propensity score.

Every patient provided written informed consent to import their electronic health records into the BIMS. Patients and/or the public were not involved in the design, conduct, reporting, or dissemination plans of this research. The study was approved by the ethics committee of the Affiliated Hospital of Qingdao University. The data used in this study are available upon agreement from the scientific committee.

Exposure and outcome

Baseline was the index date that febuxostat or benzbromarone was first prescribed. Covariates were age, sex, body mass index (BMI), alcohol drinking status, smoking status, duration of gout, serum biochemical variables (serum urate, AST, ALT, fasting blood glucose, triglyceride, cholesterol, and serum creatinine levels), and comorbidities (self‐reported hypertension, diabetes, hepatosteatosis, cardiovascular disease, nephrolith, renal cyst, or renal insufficiency). Follow‐up commenced and continued until December 2023, unless patients were censored for the following reasons: episode of hepatotoxicity, loss to follow‐up, or discontinued the study drug.

The primary outcome was hepatotoxicity, which was defined in this analysis as the first event of either ALT or AST above three times the upper limit of normal (ULN). ¹⁷ Severity of LFT abnormality was also assessed according to Common Terminology Criteria for Adverse Events (CTCAE) version ¹⁸ 5.0. CTCAE grades AST increase and ALT increase as grade 1: above the ULN to 3.0 × ULN if baseline was normal, 1.5 to 3.0 × baseline if baseline was abnormal; grade 2: >3.0 to 5.0 × ULN if baseline was normal, >3.0 to 5.0 × baseline if baseline was abnormal; grade 3: >5.0 to 20.0 × ULN if baseline was normal, >5.0 to 20.0 × baseline if baseline was abnormal; and grade 4: >20.0 × ULN if baseline was normal, >20.0 × baseline if baseline was abnormal. The CTCAE grade was determined based on either the highest AST or ALT value during the follow‐up period.

Statistical analysis

To reduce the effect of potential confounders, we performed 1:1 matching with a caliper of 0.1 pooled SDs using nearest neighbor matching of the two groups for age, sex, ALT, and AST. Comparisons between the two groups before and after propensity score matching (PSM) were explored with a standardized mean difference (SMD). Crude rates of hepatotoxicity were calculated per 1,000 person‐years for all patients. Kaplan‐Meier analysis and robust Cox regression (univariately or multivariately adjusted) were used to estimate the risk of hepatotoxicity episode using the Survival package, with hazard ratios (HRs) and 95% confidence intervals (CIs) calculated. Subgroup analyses including prespecified subgroups of age, BMI, duration, and comorbidities were used to examine how the risks of hepatotoxicity differed. Sensitivity analyses were performed in those with different follow‐up periods and selection criteria. The inverse probability of treatment weighting (IPTW) using propensity score (based on age, sex, BMI, serum urate, AST, ALT, fasting blood glucose, triglyceride, and estimated glomerular filtration rate) was performed to balance observable characteristics between groups. A P value <0.05 was considered statistically significant. All statistical analyses were conducted using R (version 4.2.1).

RESULTS

Of the 8,952 patients with gout who had data available (all were Chinese patients), 1,844 patients and 494 patients started febuxostat or benzbromarone, respectively, with at least three AST or ALT measurements per year in the cohort. Comparison between included patients and those excluded due to lack of three tests of ALT or AST were shown in Supplementary Table S1, showing balanced baseline ALT or AST. Among them, 488 patients were matched based on age, sex, and baseline AST and ALT. Baseline characteristics of these two groups were compared (Table 1). The patients were, on average, 44 to 45 years of age and were predominantly male participants. The two groups were similar in many respects including serum urate, cholesterol, and fasting blood glucose levels. However, some variables were not balanced with the SMD above the threshold of 0.1 even after PSM. To make more variables balanced (SMD < 0.1), further IPTW was performed in the sensitivity analysis. Febuxostat and benzbromarone usage patterns are summarized in Supplementary Table S2.

Table 1.

Baseline characteristics of study participants (before and after PSM)^*

Characteristic	Before matching			After PSM
	Febuxostat	Benzbromarone	SMD	Febuxostat	Benzbromarone	SMD
Participants, n	1,844	494		488	488
Sex, n (%)			0.055			0.071
Female	44 (2.4)	8 (1.6)		13 (2.7)	8 (1.6)
Male	1,800 (97.6)	486 (98.4)		475 (97.3)	480 (98.4)
Age, mean (SD), yr	46.03 (14.74)	43.20 (15.05)	0.19	44.80 (14.70)	43.52 (14.85)	0.087
BMI, mean (SD), kg/m2	27.49 (3.65)	26.70 (3.34)	0.224	27.45 (3.72)	26.75 (3.31)	0.197
Duration of gout, mean (SD), yr	7.46 (6.31)	5.94 (5.98)	0.247	7.57 (6.93)	6.01 (5.99)	0.241
Smoker, n (%)	636 (47.3)	135 (38.1)	0.185	156 (41.5)	134 (38.2)	0.068
Drinker, n (%)	753 (82.7)	189 (78.1)	0.115	207 (83.5)	188 (78.3)	0.131
ALT, mean (SD), U/L	34.62 (20.70)	34.67 (21.55)	0.002	34.83 (20.90)	34.95 (21.53)	0.006
AST, mean (SD), U/L	23.61 (9.08)	23.52 (8.81)	0.01	23.64 (8.97)	23.55 (8.84)	0.009
FBG, mean (SD), mmol/L	5.73 (0.92)	5.57 (0.68)	0.199	5.70 (0.96)	5.58 (0.68)	0.15
TG, mean (SD), mmol/L	2.30 (1.87)	2.01 (1.23)	0.187	2.29 (1.69)	2.02 (1.23)	0.183
Serum urate, mean (SD), μmol/L	503.06 (129.19)	480.66 (107.31)	0.189	501.51 (130.14)	480.46 (107.14)	0.177
Total cholesterol, mean (SD), mmol/L	4.98 (1.00)	4.91 (0.96)	0.064	5.01 (0.97)	4.93 (0.95)	0.079
eGFR, mean (SD), mL/min/1.73m2	89.71 (21.71)	94.94 (18.48)	0.26	89.75 (21.12)	94.65 (18.35)	0.248
Hypertension, n (%)	810 (44.0)	148 (30.1)	0.292	197 (40.4)	148 (30.5)	0.208
Diabetes, n (%)	109 (5.9)	21 (4.3)	0.075	29 (5.9)	21 (4.3)	0.073
Cardiovascular disease, n (%)	71 (3.9)	27 (5.5)	0.078	13 (2.7)	27 (5.6)	0.147
Dyslipidemia, n (%)	1,092 (59.3)	266 (54.2)	0.104	277 (56.8)	265 (54.6)	0.043
Renal disease, n (%)	500 (27.2)	31 (6.3)	0.582	119 (24.4)	31 (6.4)	0.514
Liver disease, n (%)	579 (31.5)	156 (31.8)	0.008	153 (31.4)	156 (32.2)	0.019

^*Dyslipidemia was hypertriglyceridemia or hypercholesteremia. Renal disease was defined as those with nephrolith, renal cyst, or renal insufficiency. Liver disease was defined as those with hepatosteatosis or abnormal liver function tests (AST or ALT is greater than the upper limit of normal) at baseline. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; eGFR, estimated glomerular filtration rate; FBG, fasting blood glucose; PSM, propensity score matching; SMD, standardized mean difference; TG, triglyceride.

The mean follow‐up was longer in febuxostat users (1.34 years) compared with benzbromarone users (1.09 years). There were no fatal hepatic adverse events during treatment. The reasons for discontinuing the study drug were mainly low medication adherence, with other reasons including elevated transaminases, COVID‐19, skin allergies, chest pain, diarrhea, dizziness, and muscle soreness in the febuxostat group; and elevated transaminases, dizziness, COVID‐19, rash, heartburn, and asthenia in the benzbromarone group. Overall, 864 people experienced a CTCAE ALT or AST increase, with CTCAE grade 1 (38.0% of patients), grade 2 (1.3% of patients), and grade 3 (0.4% of patients) in the febuxostat group, and CTCAE grade 1 (22.1% of patients), grade 2 (0.6% of patients), and grade 3 (0.2% of patients) in the benzbromarone group (Table 2). No CTCAE grade 4 elevations were observed. For the primary analysis of hepatotoxicity, defined as >3 × ULN, a total of 86 events occurred. Of note, all grade‐2 to −3 individuals were those experiencing hepatotoxicity, whereas most grade‐1 events represented mild liver abnormalities. In the matched cohort, the incidence of hepatotoxicity was lower among benzbromarone users (16.8 of 1,000 person‐years) versus febuxostat users (39.6 of 1,000 person‐years) (Table 3). The reverse Kaplan‐Meier survival plots are shown in Figure 2 and demonstrated that the hepatotoxicity risk remained different throughout the follow‐up period. After adjustment for age and sex, febuxostat use was associated with a greater risk of hepatotoxicity (adjusted HR 2.75, 95% CI 1.28–5.91) compared to benzbromarone.

Table 2.

Proportion of LFTs abnormalities and hepatotoxicity events^*

	Events of LFT abnormalities, n (%)	Events of hepatotoxicity, n (%)
Febuxostat	751 (39.7)	76 (4.0) a
Grade 1	720 (38.0)	31 (1.6)
Grade 2	24 (1.3)	24 (1.3)
Grade 3	7 (0.4)	7 (0.4)
Benzbromarone	113 (22.9)	9 (1.8) b
Grade 1	109 (22.1)	4 (0.8)
Grade 2	3 (0.6)	3 (0.6)
Grade 3	1 (0.2)	1 (0.2)

^*ALT, alanine aminotransferase; AST, aspartate aminotransferase; CTCAE, Common Terminology Criteria for Adverse Events; LFT, liver function test.

^a There were 14 individuals who met the criteria for hepatotoxicity but not the CTCAE grade 1 to 4 for ALT or AST elevation due to elevated baseline ALT or AST levels.

^b There was one individual who met the criteria for hepatotoxicity but not the CTCAE grade 1 to 4 for ALT or AST elevation due to elevated baseline ALT or AST levels.

Table 3.

Propensity score–matched association of febuxostat or benzbromarone with the hazard of incident hepatotoxicity^*

	Events, n	Person‐ years	IR	Unadjusted HR (95% CI)	Adjusted for age and sex HR (95% CI)
Before matching	–	–	–	2.11 (1.06–4.21)	2.28 (1.14–4.56)
Febuxostat	77	2,474	31.1	–	–
Benzbromarone	9	539	16.7	–	–
After PSM	–	–	–	2.59 (1.21–5.56)	2.75 (1.28–5.91)
Febuxostat	25	632	39.6	–	–
Benzbromarone	9	535	16.8	–	–
With IPTW	–	–	–	2.11 (1.03–4.33)	2.12 (1.03–4.35)

^*CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting; IR, incidence rate per 1,000 person‐years; PSM, propensity score matching.

Figure 2.

Risk of hepatotoxicity between Ben and Feb groups in the matched cohort. Ben, benzbromarone; Feb, febuxostat.

In subgroup analysis, the greater risk of hepatotoxicity associated with febuxostat in comparison with benzbromarone was similar among each of these stratified subgroups including baseline liver disease, with all P for interaction >0.05 (Figure 3). Risk factors were explored in patients taking febuxostat or benzbromarone separately (Supplementary Table S3). Individuals with previous liver disease (hepatosteatosis or abnormal LFTs) were more likely to develop hepatotoxicity. Further, we performed a subgroup analysis for those stratified ALT or AST, showing that hepatotoxicity appeared to occur with increasing levels of transaminases, especially above 40 U/L (Supplementary Table S4).

Figure 3.

Subgroup analysis according to baseline variables of hepatotoxicity risk associated with febuxostat or benzbromarone in the matched cohort. Liver disease included hepatosteatosis or abnormal liver function tests. BMI, body mass index; CI, confidence interval; HR, hazard ratio; sUA, serum uric acid.

We performed several sensitivity analyses. To include as many patients as possible, the IPTW was performed, and the conclusions were stable (Table 3, Supplementary Table S5). When limiting the follow‐up period from baseline to 180 or 365 days in the matched cohort, febuxostat use was associated with more hepatotoxicity episodes compared to benzbromarone (HR 3.75, 95% CI 1.50–9.30, P = 0.005 for 180 days of follow‐up; 3.34, 95% CI 1.41–7.90, P = 0.006 for 365 days of follow‐up). Furthermore, only the patients in the overall cohort having at least five liver tests in a year were included, and the result stayed significant (HR 2.11, 95% CI 1.06–4.21, P = 0.034).

DISCUSSION

This study has shown that episodes of hepatotoxicity (defined as ALT or AST >3 × ULN) are significantly more common for patients with gout starting febuxostat compared with benzbromarone. These episodes are mostly mild‐to‐moderate elevations in transaminases, with very few episodes of grade 3 or 4 ALT or AST elevation.

Regular monitoring of liver function during ULT is recommended to identify abnormal liver tests. The results of this analysis align with short‐term clinical trials have shown that febuxostat has a more significant effect on increasing transaminases than benzbromarone. ⁷ , ¹⁹ Previous studies have been limited by short‐term observation period of change of AST or ALT, and not as the main outcome. In this study that focused on long‐term hepatic safety, mild LFT abnormalities were common in both groups, with a higher proportion of any grade 1 to 3 elevations in the setting of febuxostat use. It remains unclear if hepatocyte damage would persist or progress with long‐term administration of ULT.

Febuxostat is metabolized via liver, the hepatotoxicity primarily resulting from glucuronidation and to a lesser extent via cytochrome (CYP) 450 system. ¹² Although preclinical studies indicated beneficial effects including attenuation of insulin resistance and lipid peroxidation of febuxostat in an animal model of MASH, ²⁰ in the original phase 2 trials of febuxostat, more AST or ALT elevations were observed compared with placebo. ²¹ , ²² Subsequently, diabetes, colchicine use, and pre‐existing liver disease were significantly associated with increased risk of hepatotoxicity while taking febuxostat. ²³ , ²⁴ In this study, we found those with baseline LFT abnormality were more likely to have abnormal LFTs during treatment, independent of other confounders, similarly when taking febuxostat or benzbromarone.

Benzbromarone remains markedly restricted due to concerns about rare episodes of severe hepatotoxicity. ²⁵ Identifiable susceptibility factors include CYP2C9 polymorphisms, metabolic epoxidation, and inactivation of cytochrome P450 3A4 for hepatotoxicity. ²⁶ , ²⁷ , ²⁸ However, our study indicates liver toxicity with benzbromarone is very rare, supporting it as a useful agent that has less hepatoxicity than febuxostat. At present, there is a deep pipeline of benzbromarone analogs in clinical development, with several currently in clinical trials. ²⁹ These benzbromarone analogs have been designed to further limit hepatotoxicity while preserving and, in some cases, enhancing urate lowering. In the future, the case for choosing a benzbromarone analog over febuxostat as a urate‐lowering drug option may only be strengthened.

There are limitations in this study. First, this finding applies to a specific cohort from a single center in China. It may not be applicable to other countries or populations. In the cohort, the drug use was retrieved from patient records, so medication adherence was not verified, and a drug dose‐dependent risk was not assessed. Additionally, selection bias may be existing while we validate it in another cohort with more LFTs. Although PSM and IPTW was performed to control for potential confounders, residual confounding remained including concomitant drugs and consumption degree due to incomplete records. In addition, very few participants with age 60+ years or diabetes developed an event in this cohort, which deserves further study in these specified groups. The analysis also did not include other measures of liver function (such as bilirubin, serum albumin, and prothrombin ratio), so a full assessment of liver function was not undertaken.

Hepatic safety, particularly in those with pre‐existing liver disease, is an important consideration in prescribing either febuxostat or benzbromarone. Despite the widespread limitation of benzbromarone due to concerns about severe hepatotoxicity, these events are rare, and benzbromarone has a lower risk of hepatotoxicity than febuxostat.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Li confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

Supporting information

Disclosure form.

Supplementary Table S1: Baseline characteristics between study subjects included in the study and those excluded due to lack of 3 tests of ALT or AST

Supplementary Table S2: Febuxostat and benzbromarone doses

Supplementary Table S3: Risk factors associated with hepatotoxicity

Supplementary Table S4: Association between baseline transaminases and occurrence of hepatotoxicity

Supplementary Table S5: Baseline characteristics of the study subjects after inverse probability of treatment weighting.