This study suggested that treatment with TNF-α antagonists does not affect the QTF test. The QTF test was positive in 34% of the patients before treatment and in 42%, 3.6 months later. It was previously shown that IFN-γ production in response to phytohemagglutinin is impaired in RA patients and is more pronounced in RA patients with high inflammatory activity compared with those with low activity [22
]. The fact that the IFN-γ level in the positive-control well increased significantly after treatment with TNF-α antagonists, probably because of an improvement in the immunocompromised state of these patients, whereas the IFN-γ level in the tuberculosis-antigen well did not change, also supports the notion that results of the QTF test are not affected by treatment with TNF-α antagonists. This is important for showing the utility of this test during treatment with TNF-α antagonists.
Several mechanisms are thought to play roles in the impairment of immunity against tuberculosis in patients using TNF-α antagonists [24
]. One explanation would be that these drugs interfere with the secretion of IFN-γ by memory T cells [24
]. This raises concern that IGRAs may not be useful in patients who are already using TNF-α antagonists. An ex vivo
study of patients treated with TNF-α antagonists showed that the number of IFN-γ-releasing lymphocytes and immediate release of IFN-γ after challenging with mycobacterial antigens was significantly decreased after 14 weeks of treatment with these drugs, compared with the initial values [25
]. The authors concluded that ELISPOT assays are not reliable to diagnose previous or latent tuberculosis in patients using TNF-α antagonists. However, and in line with what we observed here, a clinical study with serial ELISPOT testing before and after treatment with infliximab suggested that ELISPOT is a reliable tool for monitoring tuberculosis in these patients [26
]. Similarly, a cross-sectional study that looked at the QTF-Tb Gold test in patients who were already using TNF-α antagonists showed that the number of patients with a positive test was lower in the group treated with TNF-α antagonists and suggested that QTF-Tb Gold positivity decreased with TNF-α antagonist use [27
]. However, QTF-Tb Gold assays were not available before treatments with TNF-α antagonists were started in these patients. In contrast, our results provide evidence that the QTF-Tb Gold test is not affected by TNF-α antagonists. Thus it is important to test how IGRAs perform in different immunosuppressed patient populations with serial testing before and after treatment with these agents, especially in settings with a high tuberculosis prevalence.
The lower QTF and TST positivity among RA patients who are prescribed TNF-α antagonists, compared with those who are not, is probably related to the higher disease activity in the first group. The demographic features and the drugs used when the first QTF test was performed were similar between the two groups. However, the first group had a higher disease activity, causing them to be prescribed TNF-α antagonists. A formal analysis of disease activity was not performed for the purposes of this study, but a minimum DAS28 score of 5.1 is required for prescribing TNF-α antagonists in Turkey.
Our study showed that the positivity rate of QTF was higher than the TST among RA patients and lower than the TST in healthy controls. However, because no gold standard exists for diagnosing latent tuberculosis, it is impossible to know definitely which test is more accurate. The QTF test was positive in 47% of healthy controls, and the TST was positive in 68%. The estimated prevalence of latent tuberculosis, defined as TST positivity among those who did not have a BCG vaccination, is 25% in Turkey [28
]. Thus, assuming that our healthy controls reflect the genreal population in Turkey, we can speculate that the false-positivity rate of QTF may be lower than that of TST, but it is probably still quite high.
Patients with leprosy had the highest QTF positivity rate and mean IFN-γ levels in tuberculosis-antigen wells among the groups we studied, whereas their IFN-γ levels in positive-control wells and TST positivity rate were not higher than those in the other groups. This finding suggests that the QTF test may give false-positive results in leprosy patients because of cross-sensitivity with M. tuberculosis antigens. As far as we know, this is the first study to look at the QTF test in leprosy patients. It may be true that patients with leprosy live in worse socioeconomic conditions, spend more time in hospitals and institutions, and may be more exposed to M. tuberculosis. However, none of the patients we studied had a previous personal history or history of contact with M. tuberculosis. Thus it is possible that infection with M. leprae might be causing positive QTF results.
The discordance rate between the QTF test and the TST was 29% in our study. In another study from Turkey, which looked at the agreement between QTF and TST results among RA and AS patients, the QTF was positive among 35%, the TST was positive among 66% of the patients, and the discordance rate was 39% [29
]. Another study from Turkey, this time among health-care workers, showed a discordance rate of 37% [30
]. The discordance rate is usually lower in reports from countries where tuberculosis prevalence is low, because most patients are both QTF and TST negative [31
]. In countries where tuberculosis is endemic, discordance rates are usually higher [32
]. It is suggested that BCG vaccination is one of the reasons for discordance when the TST is positive and the QTF is negative. However, in our population, this was not the case. The number of patients with a BCG scar was similar among TST(+) QTF(-) patients and the rest of the patients. Conversely, 14% of our patients were QTF(+) and TST(-). These patients usually had low IFN-γ levels. This suggests that the current low cut-off value for QTF positivity, as advised by the manufacturer, may be reponsible for some of the QTF(+)TST(-) results, thus increasing the sensitivity, but decreasing the specificity of the test. In our study population, increasing the cut-off for QTF positivity to 0.85 IU/ml would decrease the frequency of QTF(+)TST(-) discordant results from 14% to 9%. The problem with the cut-off value was emphasized by another study, which showed that reproducibility is low among patients with IFN-γ levels close to the cut-off point [33
]. Among patients with inconsistent results between two consecutive tests, maximum IFN-γ response in either test was 0.68 IU/ml, compared with 4.99 IU/ml in persistently positive patients [33
]. Thus, it is important to interpret QTF results, not just as negative or positive, but quantitatively, together with IFN-γ levels in the positive-control and tuberculosis-antigen wells, especially in immunosuppressed populations.
We had only two rheumatology patients with active tuberculosis at the time of this study. The other patients who had had tuberculosis in the past had received sufficient antituberculosis treatment. This may be the reason for the low QTF positivity in this group. However, QTF positivity rates of as low as 64% have previously been reported among patients with active tuberculosis [34
The main limitation of our study is the same with all QTF and, for that matter, TST studies. No gold standard exists for diagnosing latent tuberculosis. In most studies about the sensitivity of the QTF test for latent tuberculosis, surrogates (such as history of exposure to a patient with tuberculosis, sequelae on chest radiograph, or active tuberculosis) have been used [10
]. Because of this uncertainty, we did not attempt to calculate sensitivities and specificities of the QTF and TST tests. We compared the results of patient groups with healthy controls instead.
The other limitation of our study was that it lacked long-term follow-up. Although the prospective nature of our study among patients using TNF-α antagonists gave us an idea about the effect of anti-TNF treatment on the QTF test, we do not know whether being QTF positive shows a risk for activation of latent tuberculosis in the long term. In our limited experience, we can point out that among the eight patients who were QTF positive and TST negative initially, and who did not use isoniazid, none has developed tuberculosis. It is reasonable to postulate that these same patients would be expected to develop activation within the first few months of TNF-α antagonists use, had they had latent tuberculosis. However, data on more patients are needed to comment on the safety of not treating such patients, as well as patients who had QTF conversion from negative to positive during treatment with TNF-α antagonists.
When considering QTF as a means for screening latent tuberculosis in large populations, it has advantages, such as the lack of boosting and no need for return visits, but it also has several problems. It is expensive and requires specialized laboratories. Its positivity rate in other mycobacteria and its reproducibility and reliability must be studied further in immunocompromised populations.
Previous studies comparing QTF and TST for screening latent tuberculosis before starting TNF-α antagonists suggested that these tests show good concordance in populations with low tuberculosis prevalence [35
]. In such settings, the QTF test may be used either to confirm positive TST results or when false-negative TST results are suspected [12