Paris, France - a significant IGRA study

Objectives

We conducted a prospective and multicentre study between October 2005 and May 2007, involving 5 hospitals located in Paris and North of Ile de France Region to evaluate the intrinsic performance of IGRA in a French children cohort with latent TB infection or active TB disease, and to monitor the IFNγ responses in these children, receiving antituberculous prophylaxis or therapy.


Discussion

A key question regarding IGRA and one often asked by clinicians is whether IGRA might replace the TST. Longitudinal studies have correlated the size of induration with future risk of active TB. However, TST interpretation on a background of in excess of 90% of individuals with a BCG vaccine has triggered the need to develop a model for efficient screening to avoid an excessive number of false-positive subjects and subsequent overtreatment during contact tracing investigations. This problem is further emphasized when screening children, as they represent the most susceptible of the infected contacts for future development of active TB. This is shown in our prospective study where 46% of HC were TST positive as classified by clinicians, and 83% received chemoprophylaxis. By comparison only 9% of HC had a positive QF-TB-IT result.

Enrolment of our children started in October 2005, on a background of in excess of 90% of individuals with a BCG vaccine and at a time when IGRA were not yet licensed in France. However, discussion and proposals were underway to move towards a new TB control strategy with the abolition of BCG vaccination, and it was important to evaluate the impact of IGRA (in our study QF-TB-IT) on TB diagnosis in children. Up until this time, studies examining IGRA in children in both high or low burden regions of tuberculosis infection have demonstrated a higher accuracy of IGRA for TB except for the Connell et al and Nicol et al. studies.

Significantly higher IFNγ values were obtained from children with active TB as compared to LTBI children. However, there was an overlap in QF-TB-IT IFNγ values between these categories, not allowing a clear separation between both clinical presentations. But this represents a major issue for IGRA. In comparison, an adult pilot study using an ESAT-6 ELISPOT approach, showed that lymph node TB and self healed culture negative TB patients had higher frequencies of ESAT-6 specific CD4 T-cells than culture positive TB patients. This might indicate that replicating M. tuberculosis and its antigen burden finely control the level of T cell-IFN-γ responses. One might expect a lower burden of M. tuberculosis antigens in our LTBI children or in culture negative active TB children, and by correlation a higher IFN-γ response, as seen recently. Despite the fact that several culture negative active TB children had very high QF-TB-IT values, the geometric mean was no different when we compared it to the geometric mean calculated from culture positive active TB children IFN-γ values, although a trend was observed.

Another hypothesis might be that above a certain level of replicating mycobacteria, T-cell responses are not detectable in vitro as T-cells might be exhausted or sequestered at the site of infection. This argument was often used to explain the absence of a TST response in patients with advanced tuberculosis. It represents one of the difficulties encountered by IGRA, with negative results observed in culture confirmed tuberculosis. The dynamics of the T-cell response during active or latent TB was difficult to analyse and thereby to understand when only the TST was available, as repeated intradermal injections of tuberculin can caused a boosting effect, artificially enhancing the TST response. By comparison, IGRA, based on a simple venous puncture, might be repeated without interfering with the existing in vivo T-cell activity. Setting up a prospective study to investigate the effect of preventive and curative therapy on the IGRA response over time has allowed us to detect an increase in the IFN-γ response as soon as 10 days after institution of therapy, whether it was for LTBI or active TB.

This is the first time to our knowledge that the IFN-γ response has been analyzed so shortly after institution of treatment, and that such a rebound in the IFN-γ response is described in children. This has not been observed before in children mainly because the delay between the first and the second performed IGRA was too long; more than 1, 3, 6, 12 and 18 months depending on the protocol. However, an increase in the PPD, but not ESAT-6 or CFP-10 ELISPOT response was observed in children after 1 month treatment in the study by Nicol et al. By comparison, Wilkinson et al. demonstrated a 1.8 fold increased in the ESAT-6/CFP-10 ELISPOT response at 26 days in LTBI and treated adults. This was not observed in another study but the first sampling after starting treatment was 5 weeks. This rebound is clearly significant in our study when comparing QF-TB-IT values at day 0 and at day 10.

Although this result might be expected for active disease, it is more surprising for LTBI. This rebound during chemoprophylaxis observed in LTBI children might confirm the potential of the IGRA to detect recently acquired infection and the presence of low level of replicative bacteria. It could also demonstrate all the difficulties of classifying children either as latently infected or as presenting with active disease. Fourteen LTBI and active TB children with a negative QF-TB-IT response at day 0 rapidly developed a positive response after 10 days of preventive or curative therapy. In addition, two more produced a positive response after one month of preventive therapy. Strikingly, active TB children with a negative QF-TB-IT response at day 0, but a positive a day 10, were all culture positive. For each child showing an increase in IFNγ concentration at day 10 and day 30 as compared to day 0, the second test result was higher than 16% of the first QF-TB-IT IFNγ result, which is the threshold limit potentially defined as true conversion. As all children were enrolled and treated similarly, it is unlikely that previous intradermal injection of tuberculin, often performed several weeks before, would have been responsible for this increase. In fact (see below) several children classified as LTBI or with active TB had negative IFNγ values throughout the follow-up study despite being referred because of a TST switch or a TST induration diameter above 10 mm. Increases in IFNγ values due to previous intradermal injection of tuberculin are often low, as compared to the increase observed in our study due to treatment.

The results we present here either satisfy the hypothesis of T-cell sequestration or exhaustion and that killing bacteria lowers the antigenic burden and restores active T-cells detectable in vitro, as seen in TB paradoxical reactions; or indicate that killing bacteria release cytosolic stores full of ESAT-6 and CFP-10 antigens into the macrophage resulting in a transient increase in antigen presentation stimulating a larger T-cell response. In addition, a significant decline in IFNγ response was observed in LTBI and active TB treated patients, when comparing quantitative IFNγ level measured by QF-TB-IT at day 0 and 90 or 180 for LTBI and active TB children respectively. Similar declines were also observed at 3 and 6 months in two other monitoring studies, confirming the decrease of the antigenic and bacterial load due to preventive or curative therapy.

Such an early rebound might allow confirmation of the diagnosis, whether it is LTBI or active TB. It needs a prospective study with a treatment decision using IGRA values obtained between day 0 and day 30. In addition, such a decline in response may help clinicians to monitor the effect of preventive and curative therapy on latent or active infection.

Finally it demonstrates that the use of IGRA as a one-shot strategy similar to the TST approach might not be the correct strategy, as several LTBI or active TB children will be missed. Similar results were recently obtained in a high incidence area for TB. T-SPOT-TB, when compared to TST for the detection of active tuberculosis in children, demonstrated a poorer sensitivity as compared to TST for a combined group of culture-confirmed and clinically probable tuberculosis, indicating that T-SPOT-TB can not be used to exclude active disease. It emphasizes the need for a follow-up strategy as shown in our study, with an increase in QF-TB-IT sensitivity when combining day 0, 10 and 30 results.

In conclusion, our results have shown the importance of carefully recording the dynamics of the IFNγ response in children. We need to follow up on this study by implementing IGRA in any children case contact study or in high risk children population. In addition, repeating IGRA whether the first value is positive or negative will provide reassurance to the clinician in his treatment choice. This reinforces the hypothesis that high and/or rising levels of IFNγ in response to ESAT-6/CFP-10/Tb7.7 antigens might serve as a prognostic marker, thus allowing the targeting of those children in this category for treatment.

Full study HERE

Other studies by Hermmann include this (By following quantitative IFNγ values in each enrolled child with LTBI or active TB and receiving treatment, we were able to detect an increase in the IFNγ response at day 10 of treatment which might allow the confirmation of a diagnosis. In addition, a decline in IFNγ values during treatment makes it possible for clinicians to monitor the effect of preventive or curative therapy.) and this (is the TB skin test good enough?)