To the Editors:
In their meta-analysis, Diel et al. [1] report a superior performance of interferon-γ release assays (IGRAs) for the diagnosis of latent tuberculosis infection (LTBI) compared with the tuberculin skin test (TST), the only other test available for diagnosing LTBI.
In the absence of a gold standard for LTBI, the proper assessment of the TST, IGRAs or any other future tests intended for the diagnosis of LTBI should focus on determining their capability for predicting the development of active tuberculosis (TB), i.e. the positive and negative predictive values for progression to active TB. Alternative evidence, such as the diagnostic accuracy of the test in active TB cases, is not appropriate for assessing the test’s capability to predict the development of TB.
Thus far, few studies have assessed the predictive capability of IGRAs and the TST; among them, only the study by Diel et al. [2] in Germany found IGRAs to be superior to the TST in predicting the development of TB (14.6% of IGRA-positive subjects versus 2.3% of TST-positive individuals developed TB during the 24-month follow-up period) [2]. All other studies comparing IGRAs and the TST, including two studies not included in the authors’ meta-analyses [3, 4], have reported a very similar rate of progression to active TB, ranging from 1.7 to 3.3% for IGRA-positive and 1.7 to 3.1% for TST-positive individuals, with follow-up periods from 19 to 24 months. The reasons for the discrepancy between these studies and the original study by Diel et al. [2] are not clear but could be due to different at-risk populations and, possibly, the type of tuberculin applied, as 0.1 mL Tuberculin-10-GT (Chiron Behring, Marburg, Germany) was used in an unknown proportion of subjects. The authors do not provide information on the equivalence of this tuberculin to the recommended purified protein derivative-S [5].
The main problem with most of these studies is the small number of study subjects and/or active TB cases developing among contacts: 6–15 TB cases were reported in all studies except one, which reported 26 cases [3].
Most studies that compared IGRAs and TST have reported low overall positive predictive values. Further studies are unlikely to show that IGRAs and the TST have better predictive capabilities because active TB develops in a minority of subjects who are considered to be infected, unless a different test or methodological approach is used.
In a large, longitudinal study in Canada that assessed the predictive value of the TST among untreated contacts of active TB cases, we found that 6.9% of household contacts and 48% of contacts 0–10 yrs of age with a positive TST developed TB [6]. These are much higher rates of progression to active TB than those reported in most studies for contacts with a positive IGRA.
Another potential limitation of IGRAs not considered by Diel et al. [1] is the high reversion rate (conversion from positive to negative) of 33–49% reported in studies where individuals were retested with IGRAs within a 1–3-month period [7, 8].
Until better tests are available, the usefulness of the TST and IGRAs, whether alone or in combination, should ideally be assessed by cost-effectiveness studies in different high-risk groups for developing TB: household contacts, contacts 0–10 yrs of age and immunosuppressed contacts of active TB cases [9]. Future studies are therefore needed to determine whether IGRAs, the TST or both are currently the best option in predicting the risk of developing TB.
Footnotes
Statement of Interest
None declared.
- ©ERS 2011
