Biological agents and tuberculosis: risk estimates and screening strategies

The introduction of biological agents has revolutionized the management of rheumatic diseases. However, the immunomodulatory effect of these drugs is associated with an increased risk of infection, especially an increased risk of tuberculosis (TB) reactivation – a fact of which the medical community has been well aware ever since a study on TB case reports after infliximab therapy was published in 2001.1 The recognition of an increased risk of TB reactivation in patients on tumor necrosis factor-alpha (TNF-α) inhibitor treatment has led to the publication of numerous guidelines on screening and treatment for latent TB infection (LTBI) in these patients, but many uncertainties remain.2, 3 This issue of the International Journal of Rheumatic Diseases contains two retrospective cohort studies from South Korea that contribute to the discussion about the risk of TB associated with TNF-α inhibitor treatment and the optimal LTBI screening strategy in patients due to commence treatment with biological agents, respectively.4, 5 Studies have shown that the risk of TB is significantly higher in patients treated with anti-TNF-α monoclonal antibodies (e.g., adalimumab and infliximab) than in patients receiving soluble TNF-α receptor therapy (etanercept).6-9 The study by Jung et al.4 in this issue, conducted in over 8000 patients with either rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis or inflammatory bowel disease who were treated with TNF-α inhibitors, confirmed this finding. However, in contrast to previous studies which found that adalimumab was associated with a higher risk of developing TB than infliximab,6, 7 in the study by Jung et al. this risk was higher with infliximab. As the current study did not adjust for potential confounders, the higher risk associated with infliximab than with adalimumab was likely caused by the lower proportion of preventive TB treatment in the infliximab group (14% vs. 23%). In a prospective French registry study estimates for the relative risk of TB based on standardized incidence ratios were 29.3 (95% CI 20.3–42.4) for adalimumab, 18.6 (95% CI 13.4–25.8) for infliximab and 1.8 (95% CI 0.7–4.3) for etanercept, compared to the general population.6 A systematic review of randomized controlled trials found that the risk for TB is significantly higher when TNF-α inhibitors are combined with methotrexate or azathioprine compared to TNF-α inhibitor monotherapy (24 cases of TB/4241 patients vs. 2/5769; odds ratio [OR] 13.3; 95% CI 3.7–100).10 Recent pooled data from randomized controlled trials suggest that there is no definite increase in TB risk with the newer TNF-α inhibitors golimumab11 and certolizumab pegol12 compared to placebo, but caution is warranted. At this point in time it seems reasonable to apply guidelines for LTBI screening and treatment in patients due to commence treatment with a TNF-α inhibitor to the newer TNF-α inhibitors as well. Little is known about the risk of TB reactivation in other biological agents such as rituximab which targets the B-cell specific CD20 antigen, the T-cell co-stimulation inhibitor abatacept and the interleukin-6 receptor inhibitor tocilizumab. It appears that the risk of serious infections is significantly lower with these agents compared to TNF-α inhibitors,13 and to date there is no clear association between them and an increased risk of TB reactivation, but longer follow-up is needed.14, 15 The relative risk of TB in rheumatoid arthritis associated with the disease itself and the use of non-biological medications ranged between 2 and 16 in studies that used standardized risk rates.2 Information on the relative risk of TB associated with rheumatic diseases other than rheumatoid arthritis is lacking. The current study by Jung et al. does not allow any conclusions to be drawn about the relative risk of TB associated with different diseases per se, as all patients were on treatment with TNF-α inhibitors and the analysis was not adjusted for the type of TNF-α inhibitor treatment used in different diseases. Relative risks of TB in other conditions associated with immunosuppression have been described. The relative risk for TB in human immunodeficiency virus (HIV) infection was 36.7 (95% CI, 11.6–116) in countries with a low prevalence of HIV infection;16 in patients with chronic renal failure on hemodialysis it was 7.8 (95% CI 3.3–18.7);17 and in diabetes on insulin it was 2.3 (95% CI 1.4–3.7).18 Noticeably, the relative risk of TB associated with etanercept treatment appears to be lower than the relative risk of TB in patients on dialysis, who are not targeted for LTBI screening and treatment to the same degree as patients on TNF-α inhibitor treatment. The study by Kim et al.5 in this issue describes screening and treatment practices for LTBI and the subsequent risk of developing TB in a cohort of over 800 Korean patients with rheumatic disease prior to therapy with a biological agent. The optimal LTBI screening strategy in patients due to commence treatment with TNF-α inhibitor remains unclear, and recommendations vary between different countries and different specialty medical societies. Generally, there is consensus that active TB should be excluded by taking a medical history and performing a chest X-ray. LTBI screening recommendations include assessment of the risk of developing TB and/or a combination of interferon-gamma-release-assay (IGRA) tests such as the QuantiFERON-TB Gold in tube (QFT-GIT; Cellestis, Melbourne, Victoria, Australia) or ELISPOT assay (T-SPOT.TB; Oxford Immunotec, Abingdon, UK) and tuberculin skin testing (TST).2, 3, 19, 20 The British Thoracic Society guidelines recommend the use of TB risk stratification tables based on age, ethnicity and year of UK entry to guide the LTBI treatment decision without additional TST or IGRA tests (because they may be false-negative).20 However, a British study showed that 35% of patients classified as being at low risk of having LTBI based on risk stratification tables had a positive TST and/or IGRA, and would not have received treatment for LTBI according to the British Thoracic Society guidelines.21 American guidelines from the Centers for Disease Control and Prevention recommend combining a clinical history of TB risk factors with a TST or IGRA.3 The Korean Practice Guidelines for the Control and Management of TB relevant to the current study by Kim et al. recommend TST alone or IGRA alone or TST/IGRA dual screening as equally valid screening methods. Prediction of the risk of developing active TB depends on the test characteristics (sensitivity and specificity) and on the pre-test probability of having TB infection and progressing to active disease. Test sensitivity and specificity cannot be measured for the presence of LTBI (due to the absence of a gold standard), and are thus usually referring to the risk of developing TB. However, to date, the positive predictive value of a positive TST or IGRA for the development of TB in patients undergoing screening before commencing treatment with a TNF-α inhibitor is not known. It is well documented that IGRAs increase specificity for identifying true LTBI in bacille Calmette–Guérin (BCG)-vaccinated individuals compared to the TST.22 However, in the context of immunosuppression or other high-risk situations for developing TB, test sensitivity becomes increasingly important. In most studies in immunosuppressed patients without a history of BCG vaccination, slightly more patients had a positive IGRA result than a positive TST result, possibly suggesting superior sensitivity for IGRAs.23-25 Using IGRAs alone though, appears insufficient to identify all patients at risk of developing TB, and combining IGRA and TST has been shown to increase sensitivity.26, 27 Thus, it seems reasonable to have a screening algorithm that takes into account the pre-test probability of having LTBI as well as current immunosuppression in patients to be tested (and thus the risk of obtaining a false-negative result) to determine the optimal test strategy.28 A negative IGRA result alone could then be considered sufficient evidence in patients with low pre-test probability for LTBI and lack of immunosuppression, whereas a combination of IGRA and TST to increase sensitivity would be used in patients with increased pre-test probability of having LTBI and/or immunosuppression. In the latter scenario, tests ideally should be performed sequentially, which means that the second test would only be performed if the first test was negative. Such a screening strategy has the potential to increase the net-benefit of LTBI treatment in patients about to embark on treatment with a TNF-α inhibitor by reducing unnecessary treatment (and possible adverse events) in patients with low pre-test probability for LTBI, while ensuring that patients at risk are appropriately identified and receive treatment for LTBI.