• Tuberculosis

The study reported by Lucas et al1 is a valuable addition to recent publications that have compared the performance of commercial interferon-γ release assays (IGRAs) with that of the tuberculin skin test (TST) for the diagnosis of latent tuberculosis infection (LTBI) in high-risk children.2 3 However, we believe that the principal conclusions are not supported by the data provided and that a more guarded interpretation is warranted.

In agreement with previous studies in children,3–5 Lucas et al found significant discordance between the results of IGRAs and TST. Specifically, of 420 T-SPOT.TB and 460 QuantiFERON-TB Gold I Tube (QFT-GIT) assays that were successfully completed, results were positive in 38 (9%) and 45 (10%) children, respectively. In contrast, of 304 children in whom the TST was read, the result was positive in 54 (18%). The authors contend that the overall low level of agreement between TST and IGRAs (κ =0.45–0.46) in their study ‘reflects the likely superior specificity of the IGRAs resulting from the use of antigens not found in either BCG or NTM’. However, the authors' own data suggest that previous BCG immunisation did not significantly impact on the results of either TST or IGRAs. Specifically, table 3 shows that the odds ratio of a positive TST result in children with a history of BCG immunisation (compared with those without) was 1.7 (95% CI 0.8 to 3.5), almost identical to those for the T-SPOT.TB (OR 1.8; 95% CI 0.8 to 4.0) and the QFT-GIT (OR 1.7; 95% CI 0.8 to 3.6). Therefore, prior BCG vaccination had no greater influence on the TST result than on the result of either IGRA. In addition, contrary to the authors' statement, orthologues of early secretory antigenic target and culture filtrate protein, which form the basis of both IGRAs, are also present in some non-tuberculous mycobacteria.6 7 The clinical relevance of this is shown by a recent publication, which reported positive QFT assays in more than half of patients with M kansasii or M marinum infection.8 This illustrates that commercial IGRAs are not entirely species-specific for M tuberculosis infection.

On the basis of the observation that nine refugee children had a positive IGRA result (either T-SPOT.TB or QFT-GIT) while being TST negative (ie, IGRA+/TST– discordant), the authors conclude that the ‘TST has inferior sensitivity for demonstrating LTBI’. However, this ignores the data in table 3 and figure 3 that suggest that TST may have had superior sensitivity to either of the two IGRAs in household TB contacts (ie, those at highest risk of LTBI). Notably, children in this subgroup were considerably more likely to have a positive TST than a positive IGRA (TST, OR 4, 95% CI 1.7 to 9.5; T-SPOT.TB, 2.4 95% CI 0.9 to 6.5; QFT-IT, 2.4 95% CI 1.0 to 5.8). Therefore, this raises the possibility that the IGRA results in these children represented false negatives. However, in the absence of a gold standard for LTBI, the superiority of one test over another cannot be conclusively established.

The authors highlight that a considerable proportion (11%) of the ‘TSTs could not be completed, reflecting the logistical difficulties frequently encountered in reading this test’. However, the authors did not mention that the proportion of IGRA results without interpretable results in their study was even higher (21.4% for T-SPOT.TB and 25.6% for QFT-GIT) (table 1). The ‘real-life’ difficulties in obtaining sufficient blood for an IGRA in the paediatric setting is further highlighted by a previous study that reported a failure rate of 15% related to phlebotomy.9 Notably, all patients with failed phlebotomy in that study had an interpretable TST. Obtaining sufficient blood volume for both IGRAs to increase sensitivity, as suggested by Lucas et al, is therefore likely to be a rather unrealistic goal in routine clinical paediatric practice. The relatively high proportion of indeterminate IGRA results reported in the study by Lucas et al and other studies in children further limit the usefulness of these tests in the paediatric age group.5 10 11

Important strengths of the study by Lucas et al include the size of the cohort and the novel information provided on the potential influence of helminth infection on the performance of IGRAs. Nevertheless, the data provided do not support the statement that ‘IGRAs are the preferred screening tool for LTBI in refugee children’ nor the conclusion that a primarily TST-based screening approach is ‘inferior to a solely IGRA-based approach’. The TST retains an important place in the evaluation of children with suspected LTBI, particularly as there remains a paucity of data on the sensitivity and predictive value of IGRAs in children. Published data by our group and others suggest that the performance of IGRAs is compromised in young children, particularly those under the age of 5 years.2 10 11 This suggests that the ‘decreased positivity’ (ie, sensitivity) of IGRAs in young children probably reflects intrinsic limitations of the assays, rather than ‘less time for potential exposure to M tuberculosis’. Until more convincing evidence becomes available, we, and others in this field, maintain that children at high risk of LTBI with an IGRA–/TST+ discordant result should be given preventive treatment, precluding a solely IGRA-based approach.12 We agree with the authors that ‘both IGRAs have methodological and performance characteristics that limit their usefulness in refugee children’ and eagerly await the development of improved immunodiagnostic tools for the diagnosis of TB in children.