Three-year outcomes of tumor necrosis factor alpha inhibitor therapy in rheumatoid arthritis patients with elevated liver enzymes

Abstract

BACKGROUND

Elevated liver enzymes in rheumatoid arthritis (RA) are often attributed to multiple factors including disease activity and treatment-related adverse effects. Tumor necrosis factor inhibitors (TNFi) have shown mixed effects on liver function, with varying safety profiles among agents.

AIM

To evaluate the hepatic safety of TNFi therapy—etanercept and adalimumab—in RA patients with elevated liver enzymes.

METHODS

A retrospective chart review was conducted for RA patients with elevated liver enzymes receiving TNFi at a single center between January 1, 2019, and September 30, 2024. Out of the patients screened, 9 met the inclusion criteria. Trends in liver enzymes, fibrosis-4 (FIB-4) score, and changes in the Child-Pugh class were analyzed at 1-year and 3-year follow-up periods.

RESULTS

Among 9 patients (4 on adalimumab, 5 on etanercept), the median age was 56 years [interquartile range (IQR): 49.5–64.5 years], 77.8% were female, and the median body mass index was 36.99 kg/m² (IQR: 30.95–43.43 kg/m²). Median baseline FIB-4 was 1.25 (IQR: 1.02–1.65), with no cirrhosis observed at baseline. Aspartate aminotransferase, alanine aminotransferase, and alkaline phosphatase levels declined consistently, with significant reductions from baseline to 3 years (P = 0.003). FIB-4 scores also significantly decreased (P = 0.003), while albumin, bilirubin, and Child-Pugh class remained stable at the 3-year follow-up. At 3 years, 66.7% achieved RA remission (P = 0.03).

CONCLUSION

TNFi therapy (adalimumab or etanercept) was associated with significant improvement in liver enzymes and FIB-4 without hepatic decompensation, supporting its safety in our cohort of RA patients with liver involvement. Larger prospective studies are warranted to further validate these findings.

Key Words: Elevated liver enzymes; Rheumatoid arthritis; Etanercept; Adalimumab; Fibrosis-4 scores

Core Tip: Rheumatoid arthritis (RA) patients often present with elevated liver enzymes, complicating the safe administration of immunosuppressive therapies. This retrospective study evaluated the hepatic safety of tumor necrosis factor inhibitors (TNFi)—etanercept and adalimumab—in RA patients with persistently elevated liver enzymes. Over a 3-year follow-up, TNFi therapy led to statistically significant improvements in aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and fibrosis-4 scores, without hepatic decompensation or changes in Child-Pugh class. These findings suggest that TNFi therapy may be hepatologically safe and potentially beneficial in liver enzyme normalization among this high-risk population.

INTRODUCTION

Liver involvement in rheumatoid arthritis (RA) is not widely recognized as a possible extra-articular manifestation, yet abnormalities in liver function tests (LFTs) have been reported in a significant proportion of patients, ranging from 5% to 77%[1]. While RA itself rarely causes direct liver damage, hepatotoxicity is frequently observed as a consequence of long-term pharmacotherapy or an inflammatory state. Methotrexate (MTX) a first-line treatment for RA and nonsteroidal anti-inflammatory drugs (NSAIDs) have been associated with hepatotoxicity. Although MTX is generally well tolerated, it carries potential risks, including hepatotoxicity and cytopenia[2,3]. Leflunomide has also been linked to elevated liver enzymes, particularly within the first six months of therapy, with abnormal liver function occurring in approximately 14%–22% of patients[4]. Similarly, hydroxychloroquine sulfate has been associated with transaminase elevations in isolated cases[5]. In comparison, biologic and targeted synthetic agents tend to have a lower risk of hepatotoxicity, though liver enzyme elevations, ranging from mild to severe, have been documented[6].

Tumor necrosis factor inhibitors (TNFi), including infliximab (IFX) (Remicade^™, Janssen Biotech, Inc.), etanercept (Enbrel^™, Amgen Inc.), adalimumab (Humira^™, AbbVie Inc.), golimumab (Simponi^™, Janssen Biotech, Inc.), and certolizumab pegol (Cimzia^™, UCB, Inc.), are approved for the treatment of immune-mediated diseases, including RA[7]. While some of the TNFi are monoclonal antibodies, others are recombinant proteins. The safety of TNFi in patients with liver disease is unclear. Case reports of hepatotoxicity with TNFi have been reported[8,9]. Potential differences between different TNFi were also reported with some improving upon changing to a different tumor necrosis factor (TNF) medication (adalimumab vs IFX vs etanercept)[7]. Interestingly, despite the reported side effects of hepatotoxicity, anti-TNF-α agents have been investigated as potential treatments for severe hepatic conditions such as alcoholic hepatitis (AH) and metabolic dysfunction-associated steatotic liver disease (MASLD)[10,11]. Evidence from observational studies suggests that TNFi may serve as rescue therapy in difficult-to-treat autoimmune hepatitis (AIH) and primary biliary cholangitis[12,13]. The long-term effects of TNFi in RA patients with liver enzymes are also lacking.

This retrospective study aims to evaluate the long-term (3 years) impact of TNFi (etanercept or adalimumab) on liver enzymes and hepatic safety in RA patients with elevated liver enzymes. By addressing a crucial knowledge gap, our study provides important insights into the evolving hepatic safety profile of these therapies in RA patients.

MATERIALS AND METHODS

We conducted a retrospective chart review of patients with RA [International Classification of Diseases (ICD)-10 code M 06.9, M 05.9] with concurrent elevated liver enzymes (ICD-10 code R 94.5, R 74.8) who were started on etanercept or adalimumab in the Texas Tech Rheumatology clinic from Jan 1, 2019 to September 30, 2024. Ethics committee approval was obtained from the institutional review board at Texas Tech University Health Sciences Center, El Paso, Texas. We conducted a retrospective cohort review of patients with RA in the age group 18-89 years, who had persistently elevated liver enzymes [aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP)] prior to initiation of TNFi (etanercept or adalimumab). Patients who completed at least 3 years on subcutaneous TNFi (etanercept or adalimumab) were included. Patients were excluded due to follow-up of less than 3 years on TNFi, insufficient chart information, or normalization of LFTs before initiation of TNFi (single abnormal ALT value that normalized at the next visit). A total of 9 patients were included in the study.

We collected baseline patient demographics [age, race, gender, weight, body mass index (BMI), alcohol use, and smoking status], along with relevant comorbid conditions such as diabetes, fatty liver disease, and hypertension. Data was collected at three time points: (1) At the initiation of TNFi therapy (visit 1); (2) 1 year; and (3) 3 years post-initiation. Medication history was recorded, including rheumatic drugs (MTX, leflunomide, steroids, NSAIDs, and over-the-counter medications) as well as other medications such as duloxetine, gabapentin, antidiabetic medications, and antihypertensive drugs.

The presence of cirrhosis, interval ascites, interval encephalopathy, and disease activity were evaluated at each visit. Laboratory values at the first, second (after one year) and third (after three years) clinic visits, including liver enzymes (AST, ALT, ALP), renal function markers (serum creatinine), serum albumin, platelet count, total serum bilirubin (TSB), and international normalized ratio were collected.

The primary outcome measures included changes in liver enzymes, fibrosis-4 (FIB-4) score (a validated non-invasive marker in steatosis for predicting progression to liver fibrosis calculated using age, AST, ALT, and platelet count), presence or absence of cirrhosis, and Child-Pugh class at 3 years.

Statistical analysis

A descriptive statistical analysis was done. The normality of variables was checked with the use of the Shapiro–Wilk test (P > 0.05). Variables are expressed as median with interquartile range (IQR). Wilcoxon signed-rank test was used for paired quantitative variables (AST, ALT, ALP, serum albumin, TSB, platelet count, and FIB-4 score) as our sample size is small and some parameters like ALP did not fulfill normality by Shapiro-Wilk test. Fisher’s exact test was used to assess the association between categorical data (disease activity). The results were considered statistically significant at a P value less than 0.05

RESULTS

Between January 1, 2019, and September 30, 2024, medical records of patients with RA and elevated liver enzymes receiving TNFi were screened. Nine patients with complete 3-year follow-up data met the inclusion criteria and were analyzed in this study. A total of 4 patients were treated with adalimumab, while 5 patients received etanercept.

The median age was 56 years (IQR: 49.5–64.5 years), with a predominance of females (77.8%, n = 7). The median BMI was 36.99 kg/m² (IQR: 30.95-43.43 kg/m²).

Five patients (55.5%) reported alcohol consumption of fewer than 5 drinks per week, and 2 patients (22.2%) had a history of tobacco use disorder. The median FIB-4 index at baseline was 1.25 (IQR: 1.02–1.65). While none of the patients had cirrhosis at baseline, one presented with the diagnosis of hepatic steatosis. The disease activity as reported by the rheumatologist was low in 11.1% (n = 1) cases, moderate in 77.8% (n = 7) cases, and high in 11.1% (n = 1) cases at baseline. Baseline characteristics of the study population are summarized in Table 1.

Table 1 Baseline characteristics of the patients, n (%).

Characteristics	Study participants (n = 9)
Age (years)	56 (49.5-64.5)
Females	7 (77.8)
Hispanic	9 (100)
Body mass index (kg/m2)	36.99 (30.95-43.43)
Comorbidities
Hypertension	4 (44.4)
Type 2 diabetes mellitus	1 (11.1)
Fatty liver	1 (11.1)
Low disease activity	1 (11.1)
Moderate disease activity	7 (77.8)
High disease activity	1 (11.1)
Medications
Statins	3 (33.3)
Methotrexate	7 (77.8)
Leflunomide	3 (33.3)
Sulfasalazine	2 (22.2)
Steroids	6 (66.7)
Non-steroidal anti-inflammatory drugs	8 (88.9)

Values are reported as median (interquartile range) or n (%).

At 3-year follow-up, significant improvement in AST levels was observed (P = 0.003). Of the 7 patients with elevated AST at baseline, normalization was observed in all 7 patients, while 2 patients with normal baseline AST levels remained stable on treatment. At 1-year follow-up, normalization was observed in 5 patients, 1 patient exhibited an increase in AST levels and 1 patient had no change with persistent elevation. Two patients with normal baseline AST levels remained stable on treatment. Overall, no significant change in AST levels was observed from baseline to the 1-year follow-up (P = 0.16). Significant reduction in ALT was noted at 3 years (P = 0.003). Of the 7 patients with elevated ALT at baseline, 5 patients demonstrated normalization ALT levels, while 2 patients showed a declining trend that remained above the normal range. AST and ALT levels exhibited a declining trend from baseline to 1-year and from 1-year to 3-year follow-ups; a significant reduction was observed in both enzymes between baseline and 3-year follow-ups (P = 0.003). No significant change in ALT levels was observed between baseline and 1-year follow-up (P = 0.129). One patient receiving adalimumab experienced a transient elevation in liver enzymes at the 1-year follow-up, attributed to a viral illness. The liver enzyme levels returned to normal without intervention to within normal limits at a 3-year follow-up. ALP levels significantly declined from baseline to 3-year follow-up (P = 0.003) as well as between 1-3 years (P = 0.01). Of the 3 patients with elevated ALP at baseline, 1 patient normalized, 1 patient had persistent elevation and 1 patient showed improvement with decline to almost normal levels.

While TSB and albumin did not show significant change at the 3-year follow-up, notably, FIB-4 scores showed a consistent decline from baseline to 1 year and from 1 year to 3 years, with a significant reduction observed from baseline to the 3-year follow-up (P = 0.003). At 1 year follow up no significant changes were observed in albumin, TSB, platelet counts, or FIB-4 scores across the study participants. None of the patients developed cirrhosis at 1-year and 3-year follow-up. All patients remained in Child-Pugh class A. Trends of liver enzymes, TSB, serum albumin, platelet levels, and FIB-4 index have been summarized in Table 2 and Figure 1.

Figure 1 Declining trends in the liver enzymes (aspartate aminotransferase, alanine aminotransferase and alkaline phosphatase) and fibrosis-4 index over a 3-year period. Box and whisker plot with values indicating the median at baseline, 1 year and in 3 years and one outlier as observed. ALT: Alanine aminotransferase; ALP: Alkaline phosphatase; AST: Aspartate aminotransferase; FIB-4: Fibrosis-4 index.

Table 2 Trends in liver parameters at baseline, 1-year and 3-year follow-up.

Parameter	Baseline	1-year follow-up	P value (baseline to 1-year follow-up)	3-year follow-up	P value (1-year to 3-year follow-up)	P value (baseline to 3-year follow-up)
Aspartate aminotransferase (IU/L)	55 (43-63.5)	28 (25.5-43.5)	0.161	26 (20.5-28.5)	0.027	0.003
Alanine aminotransferase (IU/L)	78 (42-90)	38 (21.5-61.5)	0.129	25 (20.5-42)	0.129	0.003
Alkaline phosphatase (IU/L)	110 (93.5-161.5)	97 (88-151)	0.05	78 (70-117.5)	0.01	0.003
Serum albumin (g/dL)	4 (3.8-4.15)	4 (3.6-4.2)	0.204	3.9 (3.75-4)	0.732	0.207
Serum total bilirubin (mg/dL)	0.6 (0.45-0.85)	0.6 (0.4-0.7)	0.115	0.6 (0.45-0.75)	0.458	0.228
Platelet count (× 103/UL)	260 (251-303.5)	257 (240.5-284)	0.09	250 (240-282.5)	1.00	0.734
Fibrosis-4 index	1.26 (1.02-1.65)	1.22 (0.80-1.54)	0.25	0.92 (0.77-1.11)	0.09	0.003

Values are reported as median (interquartile range). Wilcoxon signed rank test was used for paired data. P value less than 0.05 was considered significant.

No significant changes were observed for serum albumin levels, except for 1 patient with a high BMI, who demonstrated a decline in albumin from baseline to 3-year follow-up.

At the 1-year follow-up, 11.1% (n = 1) of patients achieved remission in RA disease activity, while 55.6% (n = 5), 22.2% (n = 2), and 11.1% (n = 1) had low, moderate, and high disease activity respectively. At the 3-year follow-up, 66.67% (n = 6) achieved remission, 22.22% (n = 2) showed low, and 11.1% (n = 1) showed moderate disease activity. A significant proportion of patients achieved remission after 3 years of TNF-α inhibitor treatment (P = 0.03), with sustained improvement in disease activity observed in all patients at the 3-year follow-up.

Of the 2 patients with elevated ALT at 3 years, 1 patient had a prior diagnosis of fatty liver, had a BMI of 47, and was in low disease activity (improved from prior moderate disease activity), the other had a BMI of 27, had hypertension, was in RA remission and was taking diclofenac. Supplementary Tables 1, 2, 3, and 4 describes the trends of AST, ALT, ALP and FIB-4 observed for all patients at 1 year and 3 years.

DISCUSSION

In this single-center retrospective study of RA patients with elevated liver enzymes started on TNFi, we observed significant improvements in liver enzymes with normalization in the majority over a three-year follow-up period. Patients demonstrated a consistent decline in AST, ALT, and ALP levels from baseline to 1 year, with a statistically significant reduction from baseline to the 3-year follow-up. No significant changes were observed in serum albumin and bilirubin levels, supporting preserved hepatic synthetic function throughout the study period. Additionally, a significant reduction in the FIB-4 index from baseline to 3 years was observed. None of the patients developed cirrhosis, and TNFi therapy was well-tolerated, with no major adverse events reported. Most patients achieved remission, with sustained improvement in disease activity at the 3-year follow-up. Our study highlights the long-term efficacy and hepatic safety of TNFi (Etanercept and Adalimumab) in the management of patients with RA.

Elevated TNF-α levels are observed in various inflammatory disorders, including liver diseases like AH, where they have been hypothesized to drive cytokine-mediated tissue damage. While therapies targeting TNF-α showed potential promise in initial uncontrolled studies, subsequent randomized controlled trials found worse outcomes in patients with AH treated with IFX compared to controls, reflecting the dual nature of TNF-α inhibition[14,15]. Additionally, increases in liver enzymes during TNFi therapy are often attributed to concomitant medications such as NSAIDs or disease-modifying anti-rheumatic drugs (e.g., MTX) or reactivation of latent hepatitis B or C infections[7]. Among TNFi agents, IFX is most frequently associated with drug-induced liver injury (DILI)[16-18]. IFX-induced DILI is idiosyncratic or immune-mediated, presenting with marked hepatocellular injury and positive autoimmune serologic markers.

Transient elevations in serum aminotransferases may arise after 2–5 infusions, whereas an AIH-like phenotype tends to typically develop more than 6 months after initiating therapy[16,19]. The prognosis is generally favorable after discontinuation of the drug, with some cases requiring brief corticosteroid therapy[18,20,21]. Several reports have documented successful switching from IFX to an alternative TNFi, such as etanercept or adalimumab, without recurrence of liver injury, suggesting a lack of cross-reactivity[16,17,22-24]. Moreover, data from a national database indicated that the risk of liver enzyme elevation in RA patients on TNFi is low, with etanercept showing a lower likelihood of causing liver enzyme abnormalities compared to IFX or adalimumab[25]. Although IFX, adalimumab, and etanercept target the TNF pathway, they differ in molecular structure, effectiveness, and profiles of adverse events. This indicates that developing an adverse event with one TNFi does not necessarily predict a comparable response with another TNFi, suggesting a lack of class effect[26]. Our patients demonstrated a declining trend in AST and ALT levels, with a significant reduction observed between baseline and 3-year follow-up. One patient experienced a transient elevation in liver enzymes, which spontaneously resolved during subsequent follow-up.

Cholestatic pattern of liver injury induced by TNFi agents has been reported in the literature[18,27,28]. Menghini et al[27] described a case of a 44-year-old woman with Crohn’s disease who developed jaundice and cholestatic liver injury after a single infusion of IFX. However, the patient was also taking oral contraceptives and hepatotoxic medications such as mercaptopurine. Similarly, Ierardi et al[28] reported a case of steroid-refractory ulcerative colitis patient who developed cholestatic liver injury after a single infusion of IFX, which resolved spontaneously within 6 weeks. Autoantibodies and immunoreactive features were not reported in either case. The exact pathogenesis of IFX-induced cholestatic hepatitis remains unclear, and existing literature is limited to sporadic case reports, with no such data available for etanercept or adalimumab. Our patients demonstrated a declining trend in ALP levels, with a significant reduction observed at the 3-year follow-up. None of the study participants experienced cholestatic liver injury during the study period.

Serum albumin concentrations hold prognostic significance in cirrhosis and chronic liver disease[29,30]. In RA, a negative correlation has been reported between albumin levels and disease activity[31,32]. Additionally, low serum albumin concentrations have been associated with reduced serum IFX levels, increased drug clearance, and predicted non-response in patients with inflammatory bowel disease[33]. Serum bilirubin is another key prognostic marker, integral to scoring systems such as the Child-Pugh and Model for End-Stage Liver Disease scores[34], with elevated levels indicating severe liver dysfunction and poor outcomes in chronic liver disease. In the context of DILI, "Hy’s law" suggests that drug-induced hepatocellular injury accompanied by jaundice carries a high risk of progression to acute liver failure[35]. Moreover, a recent meta-analysis highlighted circulating bilirubin’s anti-inflammatory and antioxidant properties in rheumatic diseases, with TSB levels negatively correlating with disease activity and inflammatory markers in RA[36,37]. Our study showed no significant changes in serum albumin or TSB levels from baseline to the three-year follow-up, indicating preserved hepatic synthetic function throughout the study period. An exception was a patient with morbid obesity (BMI: 43.45 kg/m²), who exhibited a slight decline in albumin levels from 3.8 g/dL at baseline to 3.4 g/dL at 3-year follow-up, alongside elevated ALP levels at baseline and one year, which normalized at 3-year follow-up. No other specific etiology was identified, and these findings were attributed to morbid obesity, consistent with previous studies[38,39]. Despite these changes, this patient tolerated adalimumab well and achieved disease remission at the one-year follow-up.

Recent evidence supports the use of the FIB-4 index as a valuable noninvasive marker for early detection and monitoring of liver fibrosis in RA patients. Miyata et al[40] demonstrated that in RA patients undergoing long-term MTX therapy, elevated FIB-4 values correlated with histologic evidence of steatohepatitis and fibrosis, even when serum aminotransferase levels were near-normal, and that FIB-4 values decreased following MTX discontinuation alongside histologic improvement. In contrast, Namura et al[41] investigated short-term MTX exposure and found that FIB-4 values increased significantly within 12 months of MTX initiation, independent of cumulative MTX dose. Instead, higher RA disease activity at baseline was the principal factor associated with FIB-4 elevation over time[41]. Our study showed consistent decline in FIB-4 index as the RA disease severity decreased. Our findings align with a recent large cohort study which confirmed that FIB-4 serves not only as a diagnostic tool but also as a prognostic marker for liver events, cardiovascular events, and mortality. Like their findings, we observed significant reductions in FIB-4 scores over 3 years, suggesting that TNFi therapy may help stabilize or improve liver fibrosis risk in RA patients with elevated liver enzymes[42].

TNF-α plays a crucial role in the pathogenesis of liver inflammation and fibrosis by promoting hepatic stellate cell survival, inducing hepatocyte death, and activating immune cells, all of which contribute to the progression of liver fibrosis[43]. In preclinical models, mice fed a high-fat diet and treated with an anti-TNF receptor 1 (TNFR1) antibody for 8 weeks showed significant reductions in liver steatosis, triglyceride content, apoptotic liver injury, MASLD activity, ALT levels, and liver fibrosis, highlighting TNFR1 inhibition as a potential therapeutic target for MASLD[44]. In our study, none of the patients developed cirrhosis or fulminant liver failure during TNFi treatment. While all patients had a low FIB-4 index at baseline, we observed a significant reduction in the FIB-4 index from baseline to the three-year follow-up. As a surrogate marker of liver fibrosis, the improvement in FIB-4 index further supports the potential protective effect of TNFi on liver fibrosis in this patient subset.

This study is unique to provide 3-year follow-up data, offering valuable insights into the long-term hepatic outcomes and impact of subcutaneous TNFi (adalimumab or etanercept) on liver function and fibrosis markers in a real-world setting. Using the FIB-4 index alongside detailed liver function analyses, we conducted a comprehensive evaluation of hepatic safety. The absence of serious liver-related adverse events, such as cirrhosis or fulminant liver failure, supports the favorable hepatic safety profile of TNFi, adding valuable data to the limited literature on this topic. This study has some limitations. The small sample size limits generalizability and statistical power, while the single-center, retrospective design introduces potential biases. Also, concurrent use of hepatotoxic medications, alcohol consumption and presence of metabolic diseases contributes to confounding factors. Additionally, the lack of a control group makes it difficult to attribute changes in liver function and FIB-4 index solely to TNFi therapy due to potential confounders. Larger, prospective, multicenter studies with control groups are needed to confirm these findings.

CONCLUSION

In a single-center retrospective study of patients with RA with elevated liver enzymes, TNFi therapy, resulted in normalization of liver enzymes in the majority over a 3-year follow-up period. Despite a small sample, reductions observed in AST, ALT, ALP, and FIB-4 over 3 years was statistically significant (P < 0.01). No significant changes were observed in serum albumin and TSB levels or the Child-Pugh class highlighting the long-term efficacy and safety of TNFi, etanercept, and adalimumab in RA patients with elevated liver enzymes. Larger prospective controlled studies are necessary to determine the safety profile of TNFi in RA patients with baseline elevated liver enzymes.