Assessing risk of liver enzyme elevation in patients with immune-mediated diseases and different hepatitis B virus serostatus receiving anti-TNF agents: a nested case-control study

Tumour necrosis factor inhibitors (anti-TNF), including infliximab, etanercept, adalimumab, and golimumab, are widely used to treat immune-mediated diseases, including rheumatoid arthritis (RA), ankylosing spondylitis (AS), psoriasis (PsO), and psoriatic arthritis (PsA). Cases of liver enzyme elevation have been reported after anti-TNF use, and possible causes include reactivated hepatitis B virus (HBV) infection [1], idiosyncratic hepatotoxicity related to concomitant immunosuppressive agents (for example, methotrexate (MTX) [2]), chronic liver diseases such as fatty liver or alcoholic liver disease [3], and autoimmune hepatitis [4].

Patients with chronic HBV infection—i.e. HBV surface antigen seropositive (HBsAg⁺)—are at risk of HBV reactivation consequent to immunosuppression, and should therefore be given prophylactic antiviral treatment before receiving conventional or biologic immunosuppressant therapies [5]. However, the absolute risk of liver damage in patients with differing HBV infection status is uncertain. Although heightened risk of hepatitis among HBsAg⁺ patients receiving anti-TNF agents has been reported in individual cases and small clinic-based cohorts, the risk estimates provided by such studies are imprecise [1, 6‐11]; for example, the risk may be influenced by sex, age, type of immune-mediated disease and, especially, concomitant hepatotoxic agents [12]. No published studies have investigated the correlation between HBsAg⁺ status and liver enzyme elevation after controlling for the aforementioned factors. Although HBV surface antigen-negative/HBV core antibody-positive (HBsAg^–/HBcAb⁺) patients may also be at risk, the risk appears to be relatively low with inconsistent findings. In particular, some investigators have reported cases of hepatitis due to HBV reactivation [1, 13], whereas others report no cases [7, 14]. Concomitant immunosuppressive therapy, especially with hepatotoxic agents, is associated with the risk of hepatitis consequent to HBV reactivation in HBsAg^–/HBcAb⁺ patients [15].

Without appropriately controlling for potential risk factors, it is impossible to compare the risk of developing liver enzyme elevation in patients with differing HBV infection status receiving anti-TNF therapy. This clinical question is important in Taiwan, where HBV infection is hyperendemic; up to 15–20% of the population is HBsAg⁺ and seroprevalence of HBcAb before universal HBV vaccination commenced in 1984 was 80–90% [16‐18]. We identified a large cohort of hospital patients and conducted a nested case-control study to evaluate the associations between differing HBV serostatus and risk of alanine aminotransferase (ALT) elevation in patients receiving anti-TNF therapy.

This study, which was conducted in an HBV endemic region, is the first to have determined the relative risk of liver enzyme elevation in large cohorts of anti-TNF users with differing HBV serostatus, controlling for potential risk factors. The influence of HBV infection status on such a risk has never been estimated accurately; the results of published studies may have been influenced by concomitant hepatotoxic pharmacotherapy, underlying disease, and other patient characteristics. Moreover, this study reflects the precise risk of ALT elevation in HBV-infected patients because most were prescribed anti-TNF agents before formal risk management guidelines were issued in Taiwan and, therefore, received no prior antiviral prophylaxis [28]. HBsAg⁺ anti-TNF users had an almost eight-fold higher likelihood of ALT elevation than uninfected patients. MTX use without supplementary folic acid, and previous history of elevated ALT, were associated with a more than ten-fold higher risk of ALT elevation.

In previous studies, the major concern when HBsAg⁺ anti-TNF users developed liver enzyme elevation was HBV reactivation; however, the rates reported in these small studies differed considerably. In two studies conducted in Taiwan [1, 10], HBV reactivation incidence rates were 62.5% (n = 5/8) and 50% (n = 3/6), respectively; however, both studies investigated patients with only one immune-mediated disease (RA or PsO/PsA). Two Korean studies in AS patients reported relatively lower HBV reactivation rates of 6.9% (n = 2/29) and 12.5% (n = 1/8), respectively [6, 11]. Therefore, different types of immune-mediated diseases appear to influence the likelihood of HBV reactivation. Patient demographics differ between diseases [1, 6, 11]; for instance, most RA patients are female, whereas male AS patients predominate. Patients with different diseases are also likely to have proportionally different exposure to hepatotoxic immunosuppressant drugs—MTX is often given to RA and PsO/PsA patients but less often to AS patients. Moreover, previous studies showed that each type of immune-mediated disease had a different incidence of hepatotoxicity, despite treatment with the same immunosuppressant drugs [29]. Furthermore, it is likely that studies from different countries used different treatment strategies and doses of hepatotoxic immunosuppressants [30‐32]. Therefore, analysis of the risk of hepatitis in HBsAg⁺ subjects must take into account many potential risk factors, including sex, age, their underlying immune-mediated disease, and concomitant use of hepatotoxic drugs. For these reasons, we controlled for such potential risk factors when evaluating the risk of liver enzyme elevation. Since no cases or controls received antiviral prophylaxis before the event date or index date, the role of antiviral prophylaxis was not evaluated.

In contrast with HBsAg⁺ status, the HBsAg^–/HBcAb⁺ serotype was not associated with increased likelihood of ALT elevation in patients receiving anti-TNF therapy. A recent meta-analysis of nine studies that involved 486 HBsAg^–/HBcAb⁺ subjects who received anti-TNF therapies reported that HBV reactivation incidence was only 1.7% (n = 8/468) [15]. Given such a low incidence of HBV reactivation and the moderate sample size of our current study, it would be difficult to detect a significant association between HBsAg^–/HBcAb⁺ and risk of ALT elevation. Furthermore, we used serum ALT as an indicator which will underestimate the incidence of HBV reactivation in HBsAg^–/HBcAb⁺ subjects because this is defined by the increase in the viral DNA level or conversion from HBsAg^– to HBsAg⁺; thus ALT levels may remain normal during HBV reactivation [13, 33].

As in prior studies, we found that MTX use without folic acid was positively associated with the risk of ALT elevation [34]; however, no study has compared the risk of MTX-related hepatitis in anti-TNF users with differing HBV infection status. Although we found that MTX use was associated with a higher risk of ALT elevation than HBsAg⁺ status, previous reports suggested that cases of MTX-related hepatitis were mostly mild [35]. Besides, our observation that ALT levels normalised after withdrawing MTX in most cases may indicate that ALT elevations were due more to MTX hepatoxicity than other factors; this may explain why our findings differ from those of a recent Japanese study that reported a lower risk of HBV reactivation with concomitant MTX and a higher risk with prednisolone [36]. However, as we did not directly evaluate HBV reactivation, patients in either study probably had different risk factors.

We also found that a history of elevated ALT was a risk factor for ALT elevation in patients receiving anti-TNF agents, as was also shown by another study in new MTX users [34]. This finding suggests that ALT elevations may be due to other patient-related risk factors, including excessive alcohol consumption, unhealthy lifestyle, or diseases such as fatty liver disease or cirrhosis [3]. Risk of ALT elevation declined with advancing age, which is consistent with other reports that ALT levels correlate negatively with age [37]; epidemiologic studies of non-alcoholic fatty liver disease have consistently shown an inverse association between age and steatosis [38‐40].

These study findings should be considered in light of notable limitations, principally arising from the retrospective design, which makes it difficult to draw firm conclusions; in particular, we accrued only 30 cases (22 HBV seropositive) and several clinical data were incomplete. First and foremost, ALT elevation does not necessarily reflect HBV reactivation, and lack of viral load data in most cases with elevated ALT is a major limitation to interpreting its clinical significance; for example, chronic liver cirrhosis is another possible cause of ALT elevation in HBsAg⁺ versus uninfected subjects. Second, we have no data on other common causes of ALT elevation such as obesity or alcohol intake, although a history of elevated ALT in some patients (Additional file 1: Table S1) might be due to such underlying causes. Third, although we investigated the effects of HBV serostatus and immunosuppressants as risk factors, we did not specifically evaluate the risk of immunosuppressant use in HBsAg⁺ patients, which is known to increase the risk of HBV reactivation [5]. Fourth, although viral load and HBsAb status may influence the risk of HBV reactivation in HBsAg^–/HBcAb⁺ patients, we lack viral load data for these patients, and had HBsAb results for only 10/224, among whom nine, including three cases, were positive, and one negative. Fifth, although patients had four liver enzyme assessments on average within 12 months after anti-TNF therapy commenced, 13/407 patients had no test (Fig. 1). Lastly, infliximab, which is reported to have the highest risk of HBV reactivation, is not available in Taiwan and so was not included.

This hospital-based nested case-control study demonstrates that HBsAg⁺ patients with immune-mediated diseases receiving anti-TNF therapies are approximately eight times more likely than uninfected patients to develop ALT elevation, whereas those with HBsAg^–/HBcAb⁺ serostatus have similar risk to that of uninfected patients. MTX monotherapy without folic acid and a history of elevated ALT also correlate with ALT elevation.