Numbers needed to treat to prevent tuberculosis

To the Editor:

A new action framework advocacy document for tuberculosis elimination in low-incidence countries has recently been published in this journal [1]. We completely agree with the importance of addressing the goal of elimination of tuberculosis, which may be achievable in regions such as Western Europe and North America, but we are concerned that this goal may be overly ambitious at this point in time. As protective immunity can, unfortunately, not be achieved by a vaccine or other immune-based intervention, tuberculosis control still relies on early identification and treatment of active tuberculosis, and preventive chemotherapy of individuals with latent Mycobacterium tuberculosis infection (LTBI). However, both the tuberculin skin test (TST) and interferon-γ release assays (IGRAs) (the ELISA-based QuantiFERON-TB Gold In-Tube assay (QIAGEN, Venlo, the Netherlands) or the ELISPOT-based T-SPOT.TB assay (Oxford Immunotec, Oxford, UK)) that are recommended to diagnose LTBI have a low positive predictive value for the future development of tuberculosis, and are therefore insufficient as biomarkers to identify individuals developing active disease. With tuberculosis being a rare event, the vast majority of individuals with positive test results will never develop disease [2]. Moreover, in immunocompromised individuals, especially in patients with HIV infection, the negative predictive value of an immunodiagnostic test is insufficient to rule out the future development of tuberculosis, as was recently demonstrated in a large study in Europe [3]. Thus, systematic testing and treatment of LTBI of risk groups [1] is considered difficult to implement.

In order to optimise allocation of the available resources for tuberculosis elimination, the first step would be an improvement in the knowledge of risk groups for tuberculosis in low-incidence countries. More recently, it has become evident that the risk of the development of tuberculosis among patient groups classically considered at risk is highly variable, and is not necessarily associated with the results of immunodiagnostic testing by the TST or IGRAs [2–5]. More detailed knowledge of groups at actual risk of tuberculosis and on immunodiagnostic test performance in these risk groups may provide better guidance for physicians on priorities for targeting tuberculosis testing and LTBI treatment in low-incidence countries.

The decision to administer prophylactic treatment to a certain risk group should depend on the estimated number needed to treat to prevent one case of tuberculosis. Although not selected by a systematic approach, we made use of the fact that several studies with large sample sizes from low-incidence countries of tuberculosis were recently published from three major groups classically considered at risk for progression. In these studies, tuberculosis contacts [2, 4, 5], immunocompromised hosts [3] and healthcare workers [6–8] were evaluated for LTBI and subsequent progression. We have calculated the number needed to treat to prevent one case of tuberculosis (table 1), and found that the number needed to treat was very similar across three studies evaluating 1579, 5020 and 4774 tuberculosis contacts, respectively, with a positive TST or IGRA, ranging from 30 to 37 [2, 4, 5] (table 1). In immunocompromised individuals with a positive test (n=1537) [3], the number needed to treat to prevent a case was higher and ranged from approximately 50 to 80 (table 1). Notably, as incident cases of tuberculosis were mainly HIV positive, the number of HIV-infected persons with a positive test needed to treat was lower (range 14.0–25.5) and tuberculosis cases were only found in patients with ongoing viral replication [3]. It should be emphasised that incident tuberculosis cases also occurred in individuals with negative test results [2, 3, 5], especially among patients with HIV infection. In patients with chronic renal failure (n=270) or rheumatoid arthritis (n=199), stem cell (n=103) or solid organ transplant recipients (n=197) [3], and healthcare workers (n>15 000) [6–8], the number of individuals with LTBI needed to treat to prevent one case could not be calculated because no tuberculosis cases occurred during 2 years of follow up in these cohorts.

In the absence of better biomarkers [9], comparing the number needed to treat to prevent one case is helpful to define groups with highest risk of developing tuberculosis and to prioritise existing prevention strategies. These strategies should particularly focus on HIV-infected individuals, preferentially with ongoing viral replication, and on close contacts of infectious cases of tuberculosis, although higher numbers are needed to screen and treat among contacts. In contrast, given the low number or even absence of incident cases in other groups classically considered at risk, screening approaches more closely targeted to individuals with additional risk factors would seem more effective to identify candidates who will most likely benefit from preventive chemotherapy [10]. Finally, some groups formerly considered at risk of tuberculosis (e.g. healthcare workers) are no longer risk groups in low-incidence countries [6–8], unless having had substantial contact to a patient with tuberculosis. It should be stated that given the inadequate test characteristics of the current diagnostics, not all tuberculosis will be prevented by these strategies, since active disease will continue to occur in those tested negative.