European Respiratory Society

To the Editors:

We read with interest the recent study by Latorre et al. 1 entitled “Evaluating the non-tuberculous mycobacteria effect in the tuberculosis infection diagnosis”. We agree that discordance between tuberculin skin tests (TSTs) and interferon (IFN)-γ release assays (IGRAs) presents physicians with a considerable management dilemma when evaluating children for latent tuberculosis (TB) infection (LTBI) in routine clinical practice. We have previously urged caution in the interpretation of discordant results and have highlighted this area as a research priority 2, 3. We therefore commend the authors for investigating a potential underlying cause of discordance. However, we believe that the interpretation of the data presented by Latorre et al. 1 is based on erroneous assumptions, and that as a result the conclusions are overstated. We suggest that a more cautious and contextualised interpretation of the study findings is warranted.

As indicated by Latorre et al. 1, previous bacille Calmette–Guérin immunisation and exposure to non-tuberculous mycobacteria (NTM) are frequently cited as the primary factors underlying discordance between TSTs and IGRAs, although convincing data to support these concepts are currently lacking. In the study by Latorre et al. 1, children with suspected LTBI were assessed with a TST, a commercial IFN-γ ELISpot assay (the T.SPOT.TB assay, incorporating early secretory antigenic target 6 (ESAT-6) and culture filtrate protein 10 (CFP-10)) and an in-house IFN-γ ELISpot assay using Mycobacterium avium sensitin (MAS) as the stimulating antigen. In the subgroup of children that had TST+/T.SPOT.TB- discordance, 47.6% showed a “positive” response in the IFN-γ ELISpot using MAS as the stimulant (contrasting with the absence of response to ESAT-6 and CFP-10). The authors interpret this observation as evidence that previous NTM sensitisation in these children resulted in a false-positive TST result and thereby discordance.

While we agree that this is one possible explanation, there is an alternative explanation that would equally account for these observations. Importantly, significant cross-reactivity between different mycobacterial sensitins has been previously consistently shown in animal models 4. Furthermore, more than a decade ago, Lein et al. 5 convincingly demonstrated that T cell assays incorporating MAS cannot reliably distinguish between M. tuberculosis and M. avium complex (MAC) infection in humans. In that study, the authors used the same MAS preparation as Latorre et al. 1 to assess T-cell responses in adults with culture-confirmed TB (n = 27) or MAC (n = 10) infection. Somewhat unexpectedly, higher mean IFN-γ concentrations were observed in supernatants from peripheral blood mononuclear cells stimulated with MAS in patients with TB than those with MAC infection. In addition, MAS-sensitised T cells were detected in the majority of patients with TB. These data strongly suggest that there is considerable cross-reactivity between antigens encountered by the human immune system during M. tuberculosis infection and antigens contained in MAS. We also note that in the study by Latorre et al. 1, in the subgroup of children that were TST+/T.SPOT.TB+ (and therefore highly likely to have LTBI), 50% showed a “positive” response to MAS in the in-house IFN-γ ELISpot assay, which further questions the ability of MAS-based assays to discriminate between TB and NTM infection, or alternatively exposure.

The limited ability of MAS to distinguish between different mycobacterial infections is not surprising. Unlike the well-defined peptides ESAT-6 and CFP-10, which are thought to be relatively M. tuberculosis-specific (despite orthologues of these proteins being present in several other mycobacterial species including M. kansasii, M. marinum and M. szulgai), MAS is a mixture of heterogenous mycobacterial antigens, analogous to the purified protein derivative used in the TST 6. Cross-reactivity with other mycobacterial species is therefore likely to occur, as indicated by the manufacturer’s warning mentioned by Latorre et al. 1, that is particularly likely to be the case with M. intracellulare and M. scrofulaceum.

Taken together, these facts make it questionable whether the observations by Latorre et al. 1 in the subgroup of children with TST+/T.SPOT.TB- discordance truly reflect previous NTM exposure. An alternative explanation is that the assays using MAS detected T-cell sensitisation resulting from previous M. tuberculosis exposure and/or LTBI (i.e. confirming the positive TST), while the T.SPOT.TB produced a false-negative result. Published data show that up to one-third of children with culture-proven active TB have negative or indeterminate T.SPOT.TB assay results 7, which highlights the limitations of these assays and lends support to the latter explanation. In the absence of a gold standard for LTBI, neither hypothesis can be tested with certainty. However, given these uncertainties we believe it is premature of the authors to suggest that chemoprophylaxis could be safely withheld in these patients. Contrary to the authors’ assertions, we believe their study does not provide “enough evidence” to justify changes in clinical practice.

We concur with Latorre et al. 1 that there remains an urgent need to explore the immunology of underlying discordance between TSTs and IGRAs in greater detail. However, in view of the comparatively poor performance of IGRAs in children and the uncertainties surrounding their interpretation we, and other researchers in this field, firmly believe that research to identify better biomarkers and immunological correlates of TB infection remains crucial 3.


  • Statement of Interest

    None declared.