Two interferon-gamma release assays (IGRAs) are commonly used to detect TB infection in Japan: One is QuantiFERON-TB Plus (QFT-plus: QIAGEN, Hilden, Germany) and the other is T-SPOT.TB (T-SPOT: Oxford Immunotec, Abingdon, UK). QFT-plus is an enzyme-linked immunosorbent assay (ELISA)-based method and a fourth-generation QFT replacing QuantiFERON-TB Gold In-Tube (QFT-GIT). QFT-plus has TB1 and TB2 tubes. The TB2 tube can elicit IFN-γ production by both CD4 and CD8 lymphocyte responses [1] [2] . Therefore, it can broaden the immune response to the Mycobacterium tuberculosis (MTB) antigen, improving sensitivity to detect TB infection among immunocompromised patients with HIV, the elderly, and renal transplant recipients [3] - [5] . T-SPOT.TB is based on the enzyme-linked immunospot (ELISPOT) method, which measures the number of IFN-γ-producing T-cells stimulated by MTB-specific antigens.

Several studies have investigated the accuracy of IGRAs (QFT or T-SPOT.TB) for the diagnosis of TB infection. The frequency of indeterminate results of IGRAs in patients with active TB disease ranged from 1to 20% [6] - [9] and that of false-negative results ranged from 17 to 19% [10] - [12] . Moreover, because the production of IFN-γ is a part of the protective immune response against TB infection, it is difficult to understand the dynamics of the protective immune response against TB infection through IGRA results.

Because there have been few reports on the clinical evaluation of diagnostic tests for TB infection (IGRAs such as QFT or T-SPOT.TB) in tertiary hospitals that many elderly patients or patients receiving immunosuppressive treatment have visited, we prospectively elucidated the accuracy of IGRAs (QFT-plus and T-SPOT.TB) as diagnostic tests of TB infection in our tertiary hospital. We also compared the quantitative values of the IFN-γ response among patients with active TB disease, LTBI, old (cured) TB disease, and pulmonary nontuberculous mycobacterial (NTM) disease in order to confirm whether it is useful to differentiate active TB disease from other time-phases of TB infection.

1) Study population

Forty-five patients were prospectively enrolled from June 2019 to August 2023 in this study. They were diagnosed at Kawasaki Medical School Hospital (1070 beds, tertiary medical institution with emergency ward) from June 2019 and consisted of 14 patients with active TB disease (MTB smear and/or culture-positive results or PCR-positive for MTB from any clinical specimens), 4 patients with LTBI (they had a contact history with active TB patients but no clinical symptoms or abnormal findings on CT), 17 patients with old (cured) TB disease (they had a past history of TB disease and received antituberculous treatment within 1 to 30 years (average: 6.9 years)), and 10 patients with pulmonary NTM disease (causative microorganisms: Mycobacterium avium 5 and Mycobacterium intracellulare) (they satisfied the diagnostic criteria proposed by the American Thoracic Society (ATS) [13] ). As inclusion criteria, all patients except those with LTBI required the differentiation of pulmonary TB clinically. Patients with an obvious history of TB disease were excluded from patients with LTBI and pulmonary NTM disease to avoid the influence on the results of IGRAs.

This study was approved by the Ethical Committee of Kawasaki Medical School (No. 3520-01) and performed in accordance with the Declaration of Helsinki Informed consent was obtained from all patients in this study.

2) IGRAs (QFT-plus and T-SPOT.TB)

Blood samples for QFT-plus and T-SPOT.TB were collected at the same time. QFT-plus assays were conducted according to the manufacturer’s instructions and interpreted according to the manufacturer’s guidelines [14] . T-SPOT.TB samples were transported to a commercial laboratory (SRL, Ltd., Tokyo, Japan) within 32 hours following phlebotomy. After arrival, T-cell Xtend reagent was added to the samples and the assay was performed according to the manufacturer’s instructions [15] . The assay results were interpreted based on the US Food and Drug Administration-approved cutoffs: ≥8 spots were interpreted as a positive result, 5 - 7 as a borderline result, and ≤4 spots as a negative result. When the spot count in the positive control was <20 and/or the spot count in Nil was ≥10, the result was interpreted as invalid.

3) Statistical analysis

The data are reported as percentages for categorical variables (clinical findings and clinical evaluation of IGRAs) and as medians ± standard deviation (S.D.) or medians with interquartile ranges for continuous variables (clinical findings and the results of IGRAs of four groups). The chi-square test was used for categorical variables. The Mann-Whitney U test was carried out to calculate the difference between individual groups (active TB group, LTBI group, old (cured) Tb group, and NTM group) and group medians (>2) were compared using ANOVA. The positive response rate was defined as the number of positive cases as a percentage of the total number of cases tested excluding the indeterminate or invalid results, and borderline results were judged as negative. Statistical analysis was performed using Stat Flex version 7 software (Artec, Japan, 2020). A p-value of <0.05 was considered significant.

The clinical characteristics of the study subjects are presented in Table 1 . A total of 45 patients were enrolled during the study period. They consisted of 14 patients with active TB (active TB group), 4 with LTBI (LTBI group), 17 with old (cured) TB (old TB group), and 10 with pulmonary NTM disease (NTM group). There were no patients with human immunodeficiency virus (HIV) or receiving immunosuppressive treatment among the study subjects. In the LTBI group, IGRAs were performed from two to three months after contact screening with active TB patients. In the old TB group, the median period after the onset of active TB disease was 7.6 years (range: 1 - 30 years). The age was younger in the latent TB group than in the other groups. The patients in the active TB group showed significant lymphocytopenia, hypoproteinemia, and hypoalbuminemia in the laboratory findings compared with the other groups.

Clinical evaluation of QFT-plus and T-SPOT.TB excluding one case showed an indeterminate result of T-SPOT.TB, in Figure 1 and Figure 2 . Positive responses to QFT-plus were noted in 10 of 14 patients (71%) in the active TB group, 4 of 4 patients (100%) in the latent TB group, 9 of 17 patients (53%) in

*<0.05; COPD: chronic obstructive pulmonary disease.

the old TB group, and 0 of 10 patients (0%) in the NTM group. On the other hand, positive responses to T-SPOT.TB were noted in 8 of 14 patients (64%) in the active TB group, 4 of 4 patients (100%) in the latent TB group, 9 of 16 patients (56%) in the old TB group, and 0 of 10 patients (0%) in the NTM group. Borderline results of T-SPOT.TB were noted in 2 of 14 patients (14%) in the active TB group.

The median levels of IFN-γ using QFT-plus were as follows (median ± S.D.): 0.7 ± 1.3 IU/mL inTB1 and 1.2 ± 1.6 IU/mL in TB2 of the active TB group, 1.2 ± 1.2 IU/mL in TB1 and 1.6 ± 1.6 IU/mL in TB2 of the latent TB group, 0.2 ± 2.4 IU/mL in TB1 and 0.4 ± 3.0 IU/mL in TB2 of the old TB group, and <0.05 IU/mL in TB1 and <0.05 IU/mL in TB2 of the NTM group. Although they were significantly lower in the NTM group than in the other three groups, there were no significant differences among the active TB, latent TB and old TB groups ( Figure 3 ). Concerning the results of T-SPOT.TB, excluding one case that showed an indeterminate result, the median levels of T-SPOT.TB were as follows (median ± S.D.): 19.0 ± 26.7 spot in panel A and 8.0 ± 36.6 spot in panel B for the active TB group, 22.0 ± 10.6 spot in panel A and 9.5 ± 5.5 spot in panel B for the latent TB group, 14.0 ± 29.6 spot in panel A and 12.0 ± 72.0 spot in panel B for the old TB group, and 0 ± 0.7 spot in panel A and 0 ± 0.7 spot in panel B for the NTM group. Although they were significantly lower in the NTM group than in the other three groups, there were no significant differences among active TB, latent TB, and old TB groups ( Figure 4 ).

Among the 14 patients with active TB disease, one patient showed discordant results between QFT-plus and T-SPOT. This patient showed a false-negative response to QFT-plus, and we present her clinical findings in Table 2 . The clinical findings of four patients with active TB disease who showed false-negative results for both QFT-plus and T-SPOT.TB tests (two patients showed borderline results for T-SPOT.TB) are shown in Table 3 and Table 4 . They were all elderly patients and had many underlying diseases including malignant diseases. They also showed severe lymphocytopenia and hypoalbuminemia due to the underlying diseases. The concordance rates of QFT-plus and T-SPOT.TB tests were 93%

(13/14) in the active TB group, 100% (4/4) in the latent TB group, 88% (15/17) in the old TB group, and 100% (10/10) in the NTM group. There was good agreement (42/45; 93%) between QFT-plus and T-SPOT.TB test results among all patients included in this study.

DDH: DNA-DNA hybridization, PCR: Polymerase chain reaction; Extension: 1, within one-third of unilateral lung field; 2, within unilateral lung field; 3, over unilateral lung field.

COPD: Chronic obstructive pulmonary disease, ARDS: Adult respiratory distress syndrome; NTM: Nontuberculous mycobacteria, WBC: White blood cells.

DDH: DNA-DNA hybridization, PCR: Polymerase chain reaction; Extension: 1, within one-third of unilateral lung field; 2, within unilateral lung field; 3, over unilateral lung field.

We evaluated the accuracy of diagnostic tests for TB infection (IGRAs: QFT-plus and T-SPOT.TB) in our tertiary hospital, which many elderly patients or patients receiving immunosuppressive treatment have visited, to elucidate quantitative values of the IFN-γ response for patients with active TB, LTBI, old TB, and pulmonary NTM disease.

The positive response of patients with active TB disease showed a lower rate (71% on QFT-plus and 64% on T-SPOT.TB) compared with recently published reports of systematic reviews [16] - [19] . As a reason, false-negative responses to QFT-plus as well as T-SPOT.TB were recognized in elderly patients and patients with an immunosuppressed condition due to malignant disease. All patients with false-negative responses presented with lymphocytopenia (<1000/μL) and hypoalbuminemia (<3.0 g/dL) in the laboratory findings at the time of definite diagnosis. The frequency of false-negative results of IGRAs for patients with active TB disease ranged from 17% - 19% in recent studies, being significantly correlated with: elderly, female, acid-fast bacilli smear-negative, HIV co-infection, and inflammatory disease [10] - [12] . In order to overcome the false-negative results of IGRAs, we need to discover new specific biomarkers of active TB disease other than IFN-γ. For example, the measurement of IFN-γ-induced protein 10 (IP-10) or MIG using QFT supernatants was useful for the prediction of TB infection in patients with borderline QFT results (≥0.15 and <0.35 IU/mL) to reduce the false-negative results of IGRAs [20] . Concerning the comparison of IFN-γ concentrations of TB1 and TB2 in patients with active TB disease, that of TB2 was higher than that of TB1, but there was no significant difference. Pourakbari et al. also reported that higher IFN-γrelease in TB2 in the diagnosis of active TB was noted in the majority of patients with active TB disease compared with TB1, being consistent with the results of this study [18] .

Regarding evaluation of the identification of active TB, LTBI, old TB, and NTM groups, there were no significant differences excluding between the NTM group and the other three groups. Although IFN-γ release in active TB and LTBI groups was higher than that in the old TB group, there were no significant differences between individual groups. Actually, because the production of IFN-γ is partly involved in the immune response against TB infection, it is difficult to identify the dynamics of immune responses against TB infection through IGRA results. In fact, IFN-γ levels using IGRAs were higher in patients with active TB than in those with LTBI, but many patients in both groups showed overlapping IFN-γ levels [21] [22] . To distinguish between active TB and LTBI, Mamishi et al. reported that interleukin-2 (IL-2) release stimulated by TB-specific antigens (ESAT-6, CFP-10, and TB7.7) was significantly higher in patients with LTBI compared with patients with active TB. IL-2 is a potential biomarker for discrimination between active TB and LTBI instead of IFN-γ [23] . In a previously reported meta-analysis, the sensitivity of IL-2 following stimulation with TB antigen of QFT-GIT for diagnosing LTBI ranged from 43 to 100%, and the specificity of this biomarker ranged from 89% to 100% [24] . Akashi et al. reported that IL-1ra in the QFT-GIT supernatant may be a good biomarker for the discrimination of active TB and LTBI [25] . To date, because there have been no large-scale studies on many biomarkers that can discriminate between active TB and LTBI, it is necessary to identify new biomarkers that may be useful for the differentiation of active TB and LTBI.

Regarding the transitional change of IGRAs after the initiation of antituberculous treatment, Pourakbari et al. reported that the positive rate of IGRAs (QFT-GIT and T-SPOT.TB) following antituberculous treatment was estimated at 76% and there was no difference on comparing with the positive rate of IGRAs before the initiation of treatment, which was 76% [26] . Although the median period after the onset of active TB disease was 7.6 years in patients in the old TB group, the positive rate of QFT-plus was 53% and that of T-SPOT.TB was 56%, with no significant difference between the estimation of QFT-plus and T-SPOT.TB regarding both the positive rate and median levels in the old TB group and those in the active TB group. Because the immune response of IFN-γ continues after the end of antituberculous treatment, it is not always useful to judge cured (old) TB infection by IGRAs, and so the development of new biomarkers instead of IFN-γ that are be able to discriminate active TB and cured (old) TB is eagerly anticipated.

This study had several limitations. First, the small sample size in each group, especially in the LTBI group, wasa major limitation. Therefore, there might not be significant differences in IFN-γ levels for TB-specific antigens among three groups (active TB, LTBI, and old TB groups). The number of patients with LTBI visiting our hospital has recently decreased. Secondly, because this study was performed with patients with various severe underlying diseases in a tertiary hospital-based population suspected of having active TB disease, the positive response rates of IGRAs might be lower than in other previous studies. However, it is noteworthy that this study assessed such high-risk populations. Thirdly, because this study was geographically restricted to a small area of Japan, we should perform a large-scale nationwide study in Japan in the future.

The purpose of this study was to evaluate the usefulness of two IGRAs (QFT-plus and T-SPOT.TB) for patients suspected of having TB infection in a tertiary hospital. Although we could describe the usefulness of the two IGRAs between NTM and three other groups, we could not demonstrate the clinical usefulness among the three groups (active TB, LTBI, and old TB groups) using them. In the future, new biomarkers other than IFN-γ need to be discovered to discriminate among active TB, LTBI, and old TB. Otherwise, because false-negative responses of the two IGRAs were recognized in elderly patients and patients with immunocompromised conditions through this study, we think that it is necessary to carefully consider the diagnosis of TB infection using IGRAs in patients with severe underlying diseases in a tertiary hospital.

This research received external funding from Boehringer Ingelheim GmbH.