We read with interest the letter by Provenzano et al¹ on TB screening and anti-TNFα treatment and wish to comment on this highly topical subject.

Latent TB infection (LTBI) was diagnosed in 24.6% of the 69 rheumatological patients undergoing evaluation for anti-TNFα treatment (n = 17), six of whom received anti-TNFα treatment and TB chemoprophylaxis. The ethnicity and place of birth was not given, which may have had some bearing on this apparently high incidence of LTBI. Previous BCG vaccination was not reported in the cohort, particularly in those (8.7%) with a positive Mantoux test, which could give rise to false positive results. Before Mantoux testing steroids were stopped (for 1 week) but no comment is made regarding other immunosuppressive treatments which might interfere with the accuracy of tuberculin skin testing² and would account for the poor sensitivity of the Mantoux test in this cohort (sensitivity 35%). It is also unclear whether the two patients with a previous history of TB had received appropriate treatment at the time of initial diagnosis or whether they were subsequently included in the six patients who received chemoprophylaxis. Recent BTS guidelines recommend that patients who have previously been adequately treated should be monitored rather than receive chemoprophylaxis.² Four of the six patients who received isoniazid chemoprophylaxis were required to stop the drug due to hepatotoxicity. The authors did not comment on whether these patients had abnormal liver function tests before receiving isoniazid, nor on the degree of hepatotoxicity required to discontinue the drug.

Our experience is with similarly small numbers. Nine out of 50 (18%) rheumatological patients screened for anti-TNFα treatment were referred to our TB clinic after they were found to have either risk factors for TB (ethnicity and place of birth, n = 7), positive TB skin tests after recent TB exposure (n = 1), or a history of previous adequately treated TB (n = 1). Our patients had a mean age of 55 years, identical to that reported by Provenzano et al, although they did not report whether the mean age of those receiving isoniazid was similar to that of the entire cohort. All nine of our patients were on immunosuppressive therapy including steroid therapy at the time of screening. One patient had abnormal liver function tests thought to be secondary to methotrexate, so TB chemoprophylaxis has been deferred in this patient and methotrexate has been withdrawn awaiting normalisation of liver function tests. Six patients with normal liver function tests have commenced 6 months of treatment with isoniazid before starting anti-TNFα treatment after a risk assessment according to BTS guidelines. One patient with rheumatoid arthritis receiving hydroxychloroquine and prednisolone developed an isolated raised ALT (>200) with no symptoms and isoniazid was discontinued in accordance with previously published recommendations.³

While these numbers are small, they suggest that the high level of hepatotoxicity reported by Provenzano et al is not universal. We agree that the additive effects of concurrent therapy for active rheumatological disease and rheumatological disease per se might increase the rates of liver toxicity in patients treated with TB chemoprophylaxis. We suggest that further studies are needed in this patient population to assess whether the incidence of significant hepatotoxicity related to TB chemoprophylaxis is associated with identifiable risk factors such as age, ethnicity, co-morbidity, or medications such as immunosuppressive therapy. In addition, further research is required to determine the value of interferon γ assays for the diagnosis of LTBI in this patient population, given the limitations of TB skin tests and risk assessments.