The use of tumour necrosis factor (TNF) antagonists and their attendant increased risk of reactivation of latent Mycobacterium tuberculosis (TB) have raised concerns regarding the appropriate management of patients with rheumatoid arthritis (RA) who are found to have evidence of previous exposure and potential latent tuberculosis (LTB). Tuberculosis in the USA is mainly a disease of young adult members of the HIV-infected, immigrant and disadvantaged/marginalised populations. An upsurge of TB infection has been observed in industrialised cities populated by a large number of immigrants from developing countries.1 Thus, urban arthritis clinics face the question of the safety of combined treatment with two potentially hepatotoxic medications, methotrexate (MTX) and isoniazid (INH) either in patients with active TB infection, or with LTB exposure before initiating TNF antagonists. There are few studies reporting safety of the combination of MTX and INH, with one showing increased liver function tests (LFT) in two of five patients reported.2

The side effects associated with the use of MTX, in doses used to treat RA (usually <25 mg/week), are usually mild; however, some complications, particularly hepatic toxicity, can be serious.3 Recent work documented the adverse effects of MTX when used to treat 550 patients with RA.4 Abnormal LFT, defined as three times the upper limit of normal (ULN) on two or more occasions, were seen in 7.5% of patients treated with MTX. When MTX is used at higher doses to treat neoplastic disease, the reported incidence of elevated LFT is approximately 14%.4 These abnormalities usually resolve within a month of drug discontinuation. The mechanism by which MTX adversely affects the liver is unclear. MTX can cause hepatic steatosis, hypertrophy and fibrosis.5 Hepatic folate stores are also depleted by MTX, and supplementation with folic acid reduces the incidence of LFT elevation, suggesting that folate deficiency may be contributory in some cases of MTX-induced hepatic injury.6

INH has been used since 1952 as a frontline antimicrobial for TB. It is commonly used as monotherapy for prophylaxis in patients with LTB detected using a positive purified protein derivative skin test, in the absence of active disease, and/or a history of prior disease. Nine months of INH monotherapy has been recommended as the treatment of choice for patients with LTB.7 INH is used in combination with other antituberculosis treatment for active disease and in populations at high risk of exposure to drug-resistant strains. INH hepatotoxicity is a common complication that ranges in severity from asymptomatic elevation of LFT to hepatic failure requiring liver transplantation.8 9 Mild elevation of LFT (up to three times the ULN) is seen in as many as 10–33% of patients with INH with most patients asymptomatic; however, 10–16% of those with early mild elevations of LFT develop severe hepatitis that may progress to liver failure if the drug is not promptly discontinued. INH hepatotoxicity usually occurs during the first 3 months of treatment. The risk of developing hepatitis is age-related and is further increased by consumption of alcohol and baseline increased LFT. An increased risk of fatal hepatitis has also been noted among women, particularly black and Hispanic, and may also be increased during postpartum or immediately after pregnancy. The rate of hepatic elimination of INH has also been noted to be race-dependent; thus, 60% of African-American and white people eliminate INH slowly compared with only 10–20% of Asians.9^–11 The mechanism of INH-induced hepatotoxicity remains unclear. INH may produce reactive metabolites (eg, acetylhydrazine) that bind to, and damage cellular macromolecules in the liver. Drugs that induce cytochrome P-450 levels (including rifampin) might increase the risk of INH toxicity.12 Patients receiving INH may also be treated concomitantly with other potentially hepatotoxic drugs, such as pyrazinamide or protease inhibitors, for concomitant HIV infection.

Reactivation of TB is a significant problem with all available TNF antagonists used to treat RA, psoriatic arthritis, psoriasis and other inflammatory diseases; therefore, TB screening is recommended before initiating TNF antagonist treatment and before using abatacept (Orencia), a T cell co-stimulatory blocker approved for the treatment of RA. Thus far in clinical trials, cases of tuberculosis has not been reported with Rituximab (Rituxan) treatment for autoimmune diseases. The lifetime risk of reactivation TB in an immunocompetent individual is an estimated 5–10%, but this risk significantly increases with TNF antagonist treatment. Therefore antituberculous prophylaxis is especially important for patients with evidence of LTB when considering possible TNF antagonist treatment for concomitant inflammatory disease. A prophylaxis regimen of 9 months of daily INH treatment has been recommended.13 In addition untreated active TB is a contraindication to TNF antagonist treatment. Because MTX, a potentially hepatotoxic drug is needed for the treatment of many underlying inflammatory diseases, and because INH, another hepatotoxic treatment, may be needed to treat those patients with LTB and/or with a history of active TB, a critical question remains whether co-treatment with MTX and INH increases the risk of hepatic toxicity from these agents.

METHODS

To investigate the toxicity of MTX and INH treatment in patients with RA, we performed a retrospective chart review of patients seen at the Bellevue Hospital Arthritis Clinics in New York City between 2002 and 2006. Patients that were concomitantly treated with both MTX and INH were included. All patients were diagnosed with either LTB or remote active TB (defined as previous active disease, and diagnosed by history and typical prior TB radiographic findings). RA disease characteristics, duration of treatment, concomitant medications and LFT were followed. Additional inclusion criteria were: diagnosis of RA according to American College of Rheumatology criteria, treatment with low to intermediate stable doses of MTX (10–25 mg/week, orally or subcutaneously), supplementation with folic acid (1 mg/day) and treatment with INH (300 mg/day) combined with pyridoxine (50 mg/day). Patients were started on INH after LFTs were normal on at least two separate occasions. As alcohol abuse is an important contraindication to MTX treatment, all patients subsequently started on INH were advised to limit their alcohol intake and this information was communicated and confirmed with patients and their proxies at each visit. The primary outcome investigated was increase in LFT. Secondary clinical outcomes included hair loss, gastrointestinal upset, skin discoloration, and neuropathy. Arthritis flares and respiratory symptoms were also documented. Laboratory values were recorded at baseline and routine periodic intervals, typically every 4 weeks. These included, in addition to LFT (aspartate amino transferase, amino alanine transferase, alkaline phosphatase, total protein, albumin, bilirubin), complete cell counts, erythrocyte sedimentation rate, and C-reactive protein. Chest x-ray (CXR) results were recorded for all patients. All data were entered into a research database.

Several patients in our study had a known history of prior active TB. Despite which their RA on MTX was not well controlled and required an advance in treatment to TNF antagonists. Therefore, the second goal of the study was to investigate the risk of TB reactivation particularly among patients with prior active TB during treatment with TNF antagonist treatment. Patients with RA that reported history of active TB and demonstrated imaging findings suggesting granulomatous disease were included. These patients were referred to the Bellevue Hospital Chest/Tuberculosis clinic where they also underwent a chest computed tomography scan. In the absence of contraindication, treatment with INH, rifampicin, pyrazinamide and ethambutol (IRPE) was started in compliance with a Direct Observed Treatment (DOT) protocol. Treatment duration as per standard care was at least 4 months, prior to initiating TNF antagonist treatment. Patients were monitored during the DOT period, and after 4 months they were started on TNF antagonist treatment. Signs and symptoms of TB reactivation were evaluated and recorded every visit. In addition to serial CXR, a chest computed tomography scan was repeated after 4 months.

RESULTS

Between 2002 and 2006, 44 patients with RA were seen in the Arthritis Clinic at Bellevue Hospital who received MTX together with INH. Most patients were female, 68% (30 of 44), with a mean age of 51.4 (SD 10.8) years old. Average RA disease duration was 5.7 (4.3) years. All patients fulfilled American College of Rheumatology criteria for RA. Among them 80% (35 of 44) had positive rheumatoid factor. All patients had active disease with more than eight swollen and tender joints (of 28 recorded), despite MTX treatment.

The mean dose of baseline MTX was 16.8 (4.2) mg/week. All patients received folic acid (1 mg/day). Patients were on stable doses of MTX for at least 3 months before INH was initiated. The mean duration of INH co-treatment with MTX was 7.5 (3.0) months. All patient received INH at dose of 300 mg/day and pyridoxine at dose of 50 mg/day. During the study period, no patients developed signs or symptoms of TB reactivation or an abnormal CXR. LFT were monitored at baseline, and then every 4 weeks. Transient increases in LFT were seen in 11.3% (five of 44) of patients, but in no case was this more than twice the ULN. All abnormal LFTs resolved spontaneously without intervention (table 1). One patient developed peripheral neuropathy that resolved 3 weeks after subsequent switch to rifampin from INH. There were no other recorded adverse events.

All of our patients had a positive purified protein derivative (defined as induration >5 mm in orthogonal diameter after 72 h). LTB was diagnosed in 86% (38 of 44) of the patients based on no previous evidence of active TB with a normal CXR, and deemed appropriate for treatment with INH. Our data were collected prior to completion of INH treatment by some patients (11 patients), therefore the average duration of INH treatment reported was 7.5 (3.0) months (2–16 months). All the patients with LTB were treated with INH, and 66% (25 of 38) of them were subsequently started on TNF antagonist treatment. Thirty-four per cent of our patients (13 of 38) were unable to obtained TNF antagonists because of lack of medical coverage. Twenty patients (80%, 20 of 25) received entercept (Enbrel) and five patients received adalimumab (Humira) (20%, five of 25). TNF antagonists were initiated in 72% (18 of 25) of patients while INH was also being administered, while 28% (seven of 25) started TNF antagonists after completion of 9 months of INH. To date, while the duration of follow-up has been variable (28.7 (14.1) months), no patient has developed signs or symptoms of TB reactivation, abnormal CXR, or other related toxicity (table 2).

Previous active TB was detected in 13.6% (six of 44) of patients based on history and/or typical prior TB radiographic findings. Of these, 100% (6/6) were treated with four anti-TB drugs (IRPE) based on previous active disease. Average duration of treatment was 4.3 (0.8) months. All patients were started on anti-TNF treatment after completing the anti-TB four-drug protocol. Five patients (83%, five of six) were started on entercept (Enbrel) and one patient was started on adalimumab (Humira) (17%, one of six). With an average follow-up of 16 (12.6) months (range 5–38) to date on TNF antagonist treatment, no patient has developed reactivation of TB (table 3).

DISCUSSION

The NYU Arthritis Clinic at Bellevue Hospital cares for a predominantly urban patient population with a large number of recent immigrants from many countries in which the risk of TB is higher than that of the general population in the USA. As we have detected an almost 30% rate of LTB exposure in our RA patient population (Greenberg et al, manuscript under review), we have faced the very real question regarding safety of the combination of MTX and INH.

In this study we evaluated the hepatotoxicity of INH combined with MTX in 44 patients with RA. We did not see any cases of severe hepatotoxicity or hepatic failure. The incidence of mild reversible abnormalities in LFTs was low, occurring in 11% of patients, similar to rates previously reported for INH in the absence of other hepatotoxic co-treatment.9 These patients were asymptomatic and their LFT increased to a maximum of twice the ULN. In all cases, LFT elevation resolved spontaneously without interruption of medications or other interventions.

When INH and MTX were taken in combination, the incidence of abnormal LFT was not increased above the level seen in patients administered MTX or INH alone.2 9 This is interesting as one would predict their hepatotoxicity to be additive as seen, for example, when INH is taken together with pyrazinamide.2 One explanation might be selection bias. We selected patients as appropriate for INH treatment only if they had normal LFT when taking MTX, thus excluding those with abnormal LFT from MTX. All our patients were treated with MTX in combination with folic acid. It might be that folic acid supplementation protects patient from developing significant hepatotoxicity.6 Whether the inclusion of patients with a pre-existing elevation of LFT while on MTX or those not taking supplemental folic acid would have shown more combined hepatotoxicity is unknown.

Given the size of the study population and the lack of significant sustained elevations in LFT, we were unable to identify any subset of patients with an increased risk of developing abnormal LFT during treatment with both drugs.

We did not identify any patients with reactivation of TB using a regimen of INH as monotherapy for LTB with average treatment duration of 7.5 months. In addition, we did not see reactivation of TB after an average of 4.3 months of anti-TB regimen (with IRPE) before initiating TNF antagonist treatment for the six patients with history of active TB. It should be emphasised that these patients were closely monitored for symptoms in conjunction with TB specialists. Additional studies are warranted to determine the optimal and safest regimen for the treatment of active TB in the setting of initiating TNF antagonist therapies. Prompt identification of active TB, referral to specialised clinics, and clinical surveillance are strongly recommended.

In summary, the use of INH for LTB was well tolerated in patients with RA treated on a background regimen of MTX with a low occurrence (11%) of spontaneously reversible LFT abnormalities. While the risks and benefits of all treatment must always be considered, in our experience the additive risk of INH to MTX in terms of hepatotoxicity in patients with RA was low. We did not see any sustained abnormal LFT requiring cessation or discontinuation of these therapies. None the less it is prudent to follow LFT in patients closely when taking this combination.