Predictors for false-negative QuantiFERON-TB Gold assay results in patients with extrapulmonary tuberculosis

Despite diagnostic and therapeutic advances, tuberculosis (TB) remains an important cause of morbidity and mortality. Early diagnosis and prompt treatment are crucial to reduce mortality and spread of TB. Although the incidence of TB in Korea has been decreasing over time, the incidence in 2016 was still estimated at 77 per 100,000 population [1]. In Korea, 6196 cases of extrapulmonary TB (EPTB) were reported among 30,892 new cases of TB at the end of 2016 [2]. EPTB is a heterogeneous disease involving various organs such as lymph nodes, bone, central nervous system (CNS), gastrointestinal system and genitourinary system, and the clinical manifestations of EPTB are varied. EPTB can be diagnosed by histopathologic features or clinical findings as well as microbiologic test. Diagnosis of EPTB is sometimes difficult because physicians do not always suspect the disease, have difficulty obtaining specimens, or do not perform appropriate diagnostic methods such as smear, polymerase chain reaction (PCR) or culture. The interferon (IFN)-γ release assay (IGRA) detects IFN-γ response to TB-specific antigens, early secretory antigen target-6 and culture filtrate protein-10, and has been used to support the diagnosis of latent or active TB [3, 4]. However, there is some controversy around the usefulness of IGRA for diagnosis of active TB, and it has not been clarified whether IGRA is useful for the diagnosis of EPTB [4]. Identification of factors associated with negative IGRA results in patients suspicious for EPTB may be helpful to optimize use of the assay in clinical practice. In this study, we investigated the diagnostic performance of QuantiFERON-TB Gold assay (QFT-GIT) according to categories and anatomic sites of EPTB, and identified predictors for false-negative assay results.

In Korea, the proportion of EPTB was 20.1% (n = 6196) of all cases of reported TB (n = 30,892) in 2016, which has slightly increased compared with 2009 (19.3%, 6923/35824) [2]. One study in the United States also showed that the proportion of EPTB increased from 1993 to 2006, and 18.7% were EPTB [9]. IGRA is a new immunologic diagnostic tool for active or latent TB based on T-cell immune response to TB antigen, which is not associated with non-tuberculous mycobacteria or BCG-vaccination [10, 11]. QFT-GIT assay uses an enzyme-linked immunosorbent assay to measure antigen-specific production of IFN-γ by circulating T cells in whole blood. The other test, the T-SPOT.TB (Oxford Immunotec, Oxford, United Kingdom), measures IFN- γ -spot-forming cells [11]. A meta-analysis showed that the sensitivity of QuantiFERON TB Gold and T-SPOT.TB for diagnosis of TB was 0.842 (95% CI 0.811–0.870) and 0.840 (95% CI 0.814–0.864)), respectively; specificity was 0.745 (95% CI 0.715–0.775) and 0.658 (95% CI 0.621–0.693), respectively; positive likelihood ratio was 3.652 (95% CI 2.180–6.117) and 2.196 (95% CI 1.727–2.794), respectively;and negative likelihood ratio was 0.212 (95% CI 0.109–0.414) and 0.246 (95% CI 0.161–0.377) respectively [12]. The role of IGRAs for diagnosis of active TB remains unclear. However, in real practice, IGRA may have an adjunctive role to diagnose TB or rule out active TB, especially if microbiological or pathological clues for TB cannot be found in patients with high clinical suspicion of TB. TB can be diagnosed by clinical features, radiology, pathology and microbiology. However, diagnosis of EPTB is more elusive than that of pulmonary TB because EPTB is not always suspected, and there is difficulty obtaining pathological evidence, and failure to perform appropriate diagnostic tests. Our study of EPTB cases showed that 28.8% of all cases and 9.8% of proven TB cases had negative QFT-GIT results. Among the proven and probable TB cases, there was a higher percentage of negative QFT-GIT result for pleural and skeletal TB. Our findings suggest that physicians should not rule out the possibility of TB even though suspicious skeletal or pleural TB cases show negative QFT-GIT results.

Previous studies have shown that the quantitative response of IGRA is higher in active TB than in latent TB, and generally falls during anti-TB medication [13, 14]. It is not clear whether a high response of IGRA is associated with disease burden and is useful for diagnosis of active TB [15]. In our study, quantitative values for QFT-GIT were higher in proven than in possible cases of TB, although the difference was not significant. All cases of disseminated TB had positive QFT-GIT results. This supports the diagnostic utility of quantitative values of IFN-γ for patients with severe TB infection, however, further, large studies are needed to confirm this.

Patients with CNS involvement showed a higher percentage of negative QFT-GIT assay results than those with TB involving other sites. In cases of TB meningitis, sensitivity of conventional diagnostic tests such as acid-fast bacilli stain are reported to be as low as 30%, and PCR-based assays have the disadvantage of low sensitivity [16, 17]. Mycobacterial culture of CSF provides definitive diagnosis for patients with TB meningitis, however, culture requires long incubation of up to 8 weeks. Because of high mortality and neurologic sequelae, especially in highly TB endemic areas, clinically suspicious TB meningitis should be treated promptly before a definitive diagnosis is obtained [18]. In our practice, possible cases accounted for 80.7% of CNS TB patients, and this may have affected our finding that possible TB was independently associated with false-negative QFT-GIT assay results. Some studies have shown that sensitivity of IGRA is affected by HIV status, diabetes mellitus, neutropenia and immunosuppression [15, 19–22]. Our present study also showed that patients with positive QFT-GIT assay were older than those with negative results, although this difference was not significant. In Korea, which is an intermediate TB endemic area, older people have more chance of being exposed to TB, and the IGRA positive rate increases with age [23, 24]. Mori et al. reported that the higher QFT-GIT positivity rates were found in elderly people, and specific IFN- γ response may wane considerably with time after infection [25]. Our study included many elderly people of 33.7% of above 65 years old, and this age distribution may underestimate the false negative QFT-GIT results. In contrast to previous studies, we did not identify underlying disease or neutropenia as a risk factor for false-negative IGRA results.

Our study had some limitations. First, the number of patients according to EPTB organs or categories was not even because this was a single center retrospective study. Nevertheless, our study included relatively a large number of EPTB cases and analyzed performance of QFT-GIT according to EPTB site or diagnostic category. Second, some may argue probable EPTB cases, however our study can reflect real practice in intermediate TB endemic area.

In conclusion, false negative QFT-GIT results were found in 28.8% of patients with EPTB, and the anatomic site of TB may affect the positive rate of QFT-GIT results. The possibility of TB should be considered especially in patients with clinical suspicion of pleural or bone TB although QFT-GIT results are negative. Prudent interpretation of QFT-GIT might be needed according to site of involvement and diagnostic criteria in patients with high suspicion of EPTB.