In this study, we examined the diagnostic value of HBHA, a novel mycobacterial antigen, for the detection of LTBI in a low-incidence country, where infection by NTM is rare and where BCG vaccination is not routinely used. We found that the in vitro
HBHA-specific IGRA yielded a sensitivity of 92.06% and a specificity of 93.88% for the detection of LTBI, a substantially better sensitivity in this population than an ESAT-6-specific IGRA (40.74%, with a 90.91% specificity) and a substantially better specificity than a PPD-specific IFN-γ test (70%, with a similar sensitivity, 90.0%). In contrast, the ESAT-6 and HBHA responses were not significantly different for the diagnosis of acute TB. HBHA performed relatively poorly for the diagnosis of acute TB, as shown previously 
. This is due at least partly to the induction of HBHA-specific regulatory T cells during active TB 
. Combining the ESAT-6 results with the HBHA results did not improve sensitivity for LTBI. Low sensitivity of the ESAT-6 IGRA for the detection of LTBI, especially if carried out a long time after the point-source exposure has already been reported as being possibly related to the declining antigenic and bacterial load 
. Similarly, although tested here only on 20 subjects, the QFT-IT test detected only 50% LTBI, compared to 85% detected in this series by the HBHA IGRA. The difference of performance between the two tests was especially pronounced when LTBI presumed to result from exposure in the remote past were diagnosed. In contrast, both tests were equally effective in recent TST converters.
An important consideration in this study is obviously the definition of the LTBI subjects enrolled. We have chosen rather stringent TST positive criteria, so that it is unlikely that the TST+
subjects were false positives because of NTM infection, although occasional mis-diagnosis of TST positivity due to NTM infection can obviously not formally be excluded. In addition, the study was carried out in Belgium, a country with a low TB incidence (10.9/100,000; www.fares.be
). True LTBI resulting from remote exposure can only be studied in low- or intermediate-incidence areas, where re-exposure to M. tuberculosis
The influence of a previous BCG vaccination that might have given rise to false-positive TST results is also highly improbable, since 31 out of the 65 LTBI subjects were identified by the Occupational Medicine Department because of a TST conversion in the absence of a BCG vaccination at the time of the TST conversion. Furthermore, TST reactivity after BCG vaccination rarely lasts for more than 10 years in the absence of subsequent M. tuberculosis
, making it also unlikely that the TST positivity of the remotely infected subjects was due to BCG vaccination in their childhood. In Belgium BCG vaccination is not routinely used, and only 7 of the subjects had been previously vaccinated, more than 15 years before their enrolment in the study. The contention that the HBHA response was not confounded by BCG vaccination is consistent with the fact that in the control group, no significant difference in the numbers of HBHA-responders was found between the vaccinated and the non-vaccinated subjects. In contrast, a higher number of positives were found in the vaccinated control group (10 out of 22) compared to the non-vaccinated group (5 out of 26) when the PPD responses were examined, probably accounting for the lower specificity of the PPD IGRA compared to the HBHA IGRA. The absence of influence of a previous BCG vaccination on the HBHA IGRA might be surprising, as HBHA is produced by all members of the M. tuberculosis
complex, including M. bovis
. Although the reasons for this difference between BCG vaccination and M. tuberculosis
infection are not yet known, there are preliminary indications that the expression levels of the HBHA encoding gene are much lower in BCG than in M. tuberculosis
, especially within cells in which the mycobacteria reside during latency (C. L., unpublished).
In this study, only the RT23 PPD was used for the TST, as this is the most widely used PPD in Europe. Previous studies have shown that the pattern of reactivity obtained with 2 TU of RT23 PPD is similar to that obtained with 5 TU of a different PPD, such as Tubersol 
. Therefore, the findings reported here are expected to be similar if PPD preparations other than RT23 would have been used in the TST.
A major advantage of the HBHA IGRA reported here is the ability to detect LTBI resulting from a remote M. tuberculosis
exposure. This may have practical consequences for TB control strategies. Although diagnosis and treatment of recently infected subjects deserves the utmost priority, because these subjects are at highest risk of developing active TB, detection of LTBI resulting from a remote infection may also be important. This is the case especially for health care workers 
and for immunocompromised individuals, as immunosuppression because of either infection with HIV or of immunosuppressive treatment substantially increases the risk of TB reactivation 
. Finally, in the long run, if eradication of TB is considered to be a realistic goal, especially in low-prevalence countries, LTBI subjects will have to be effectively diagnosed at a population-wide level 
A limitation of the present study is the relatively small sample size in the cross-sectional study design. However, a longitudinal follow-up performed for 17 LTBI subjects suggests that the subjects categorized as LTBI resulting from a remote infection remained positive and were thus true LTBI cases.
In conclusion, we provide here evidence that a HBHA-based IGRA may be useful for the detection of LTBI and is both substantially more sensitive for LTBI than an ESAT-6-based test, and substantially more specific than a PPD-based test. In addition, it is able to detect LTBI subjects with a remote infection that were missed by the QFT-IT test, a test proposed by the US Centers for Disease Control to replace the TST 
. These LTBI subjects are important to detect in view of the increasing number of patients that require immunosuppressive therapy or that are infected with HIV and should therefore receive preventive anti-tuberculous treatment with high priority 
. Thus, the current ESAT-6- and/or CFP-10-based IFN-γ IGRAs may underestimate LTBI, and our results indicate that the replacement of the TST by IGRAs should be considered with caution. The use of a HBHA IGRA may provide an attractive solution, by combining the operational advantages of an in vitro
IFN-γ test with high sensitivity and specificity for LTBI.
However, one of the current challenges is that HBHA is a methylated antigen, and that the methylation is important for the HBHA-specific T-cell responses 
. When recombinant, non-methylated HBHA is used instead of native, methylated HBHA, the sensitivity of the IGRA for the detection of LTBI decreases to 69.44% with a 88.46% specificity (data not shown). HBHA is thus not yet commercially available. However, a product development plan is currently being worked out, which should also allow us to test the diagnostic potential of this antigen as a skin test reagent. Furthermore, a prospective head-to-head comparison of the HBHA IGRA and the TST for the detection of LTBI among health care workers in Belgium will begin soon, as well as studies in immuno-suppressed populations, such as HIV-infected patients and patients starting infliximab therapy.