Simultaneous and Longitudinal Comparison of Interferon Gamma Release Assay Data from Health Care Workers in Japan

Background: Tuberculosis is one of the serious occupational diseases among health care workers, especially those who work with patients suffering from respiratory disorders. It is important to assess latent tuberculosis infection status in such workers using interferon-γ release assays, including QuantiFERON-TB Gold and its successor, the QuantiFERON-TB Gold in-Tube test. Although the relative efficacies of these two methods have been evaluated in patients with tuberculosis, data from health care workers in Japan have not been extensively examined.


Introduction
The incidence of tuberculosis (TB) in Japan is less than 20 per 100,000 population and is continuing to decline. However, TB remains a major occupational disease of health care workers (HCW) [1,2]. Several reports have shown that HCW were at several-fold higher risk for TB than the general population [3][4][5][6]. One practical way to control TB is routine screening of HCW for latent tuberculosis infections (LTBIs) and administration of chemoprophylaxis to HCW suspected to have LTBIs. Therefore, evaluation of Mycobacterium tuberculosis (Mtb) infective status is crucial in HCW working in hospitals dedicated to TB patients [2,7,8]. In such contexts, screening of HCW should be routine and data should be evaluated longitudinally [2,9].
Recently, methods detecting Mycobacterium tuberculosis (Mtb)specific antigens have been developed. The target antigens include culture filtrate protein 10 kD (CFP-10), early secreted antigenic target 6 kD (ESAT-6), and TB7.7 [10][11][12]. The QuantiFERON-TB Gold (QFT-G) test uses CFP-10 and ESAT-6, whereas its successor, the QuantiFERON-TB Gold in-Tube test (QFT-GIT), additionally uses TB7.7 [13][14][15]. The results of the two assays cannot be directly compared, because not only do the stimulating antigens used vary, but some differences are evident in the methods used to stimulate lymphocytes. Thus, in the QFT-G tests, lymphocytes are separately stimulated by CFP-10 and ESAT-6 [16,17], whereas stimulation by a mixture of CFP-10, ESAT-6, and TB7.7 is commenced just after blood is drawn for the QFT-GIT assay [16,[18][19][20][21]. Several investigators have reported that the sensitivity of QFT-GIT was higher than that of QFT-G, but that the specificity values were similar [10,11,20,21]. One issue that must be considered is the intermediate results introduced in some countries including Japan, but not most developed countries [20]. In Japan, an intermediate result

Comparison of QFT-G and QFT-GIT assay data from 120 Japanese HCW
When comparing QFT-G and QFT-GIT assay data from 120 blood samples obtained at the same time, QFT-GIT yielded higher IFN-γ levels than did QFT-G in most samples ( Figure 1) subjects positive by QFT-G were also positive by QFT-GIT, although the IFN-γ levels yielded by the latter test were higher ( Figure 1). However, one of these 18 (6%) subjects was diagnosed as intermediate by QFT-GIT, although the IFN-γ level measured by QFT-G was just above the cut-off In the present study, we compared simultaneously derived QFT-G and QFT-GIT assay data from HCW in Japan. We also compared our results with those of QFT-G tests performed five years prior in some subjects.

Study design
This cross-sectional study involved 120 HCW of the Osaka Prefectural Medical Center for Respiratory and Allergic Diseases, Osaka, Japan. No subject had any underlying illnesses such as acute infection, autoimmune disorder, or any other chronic disease. No subject had an abnormal chest X-ray. The study protocol was approved by the Review Board of the Osaka Prefectural Medical Center for Respiratory and Allergic Diseases, and written informed consent was obtained from all participants.

QFT-G and QFT-GIT assays
The QFT-G and QFT-GIT assays were performed following the instructions of the manufacturer (Cellestis Limited, Carnegie, Australia). Blood samples were collected by normal phlebotomy into both evacuated 4 ml sterile sodium heparin tubes for QFT-G assay and into three 1 ml-volume QFT-GIT blood collection tubes [17][18][19]. When QFT-G tests were run, incubation with Mtb-specific antigens was initiated within 12 h of blood collection. After incubation for 24 h at 37°C, samples were centrifuged at approximately 500 g for 10 min to facilitate plasma collection. Plasma samples were stored at -70°C prior to conduct of ELISA detecting IFN-γ.

QFT ELISA assay
The concentrations of IFN-γ in plasma samples were determined via ELISA according to the manufacturer's protocol. All ELISAs were performed by the same trained staff. QFT-G and QFT-GIT test responses were automatically calculated using QFT-G ELISA Analysis software (Cellestis Limited) after input of ELISA plate optical density values. QFT-G and QFT-GIT test data were interpreted as suggested by the manufacturer. An IFN-γ response to Mtb-specific antigens that was at least 0.35 IU/ml and greater than the nil control value was considered positive. Samples with 0.10 ≤ IFN-γ level <0.35 IU/ml were regarded as intermediate, according to guidelines of Committee for Prevention of the Japanese Society of Tuberculosis [17,19]. Mitogen stimulation was used to positively control the quality of both blood samples and laboratory technique. If a sample IFN-γ value was <0.35 IU/ml when the positive control value (upon mitogen stimulation) was ≥ 0.5 IU/ml, the test result was considered to be negative.

Data analyses
The extent of agreement between data yielded by the two tests was evaluated using the McNemar approach, and agreement was expressed in terms of both a kappa coefficient and the level of overall agreement (the proportions of samples yielding positive or negative results in both tests). Non-parametric statistics (Wilcoxon's ranked sign test and Spearman's ranked correlation coefficient) were employed when the means of IFN-γ measurements obtained using either method were compared. A difference associated with a p value <0.05 was considered to be statistically significant.

Discussion
In the present study, we simultaneously compared QFT-G and QFT-GIT test results and also performed a longitudinal comparison of recent data with earlier QFT-G results obtained 5 years prior, because such a comparison should be helpful in terms of early diagnosis of LTBI in HCW, especially staff of dedicated tuberculosis wards. We confirmed that the QFT-GIT test was more sensitive than the QFT-G test when used to screen HCW, similar to the results of a previous study in TB patients [20]. However, discrepancies were evident between the results of recent QFT-GIT and QFT-G tests and also when these data were compared with QFT-G results obtained 5 years prior. Differences between the results of the QFT-GIT and QFT-G tests may be attributable to differences in the antigens used to stimulate interferon secretion; such antigens may exert synergistic effects.
Differences between recent test results and older data may be attributable to functional regression of Mtb-specific T lymphocytes [22]. In BCG-vaccinated newborns, the levels of BCG-specific T lymphocytes peak 10 weeks after vaccination [23]. The levels of Mtbspecific IFN-γ-secreting T cells declined in both TB and LTBI patients during anti-TB treatment, suggesting that the IGRA results reflect the mycobacterial load [24,25]. In the present study, the QFT-G test values fell over the 5-year interval in most QFT-G-positive subjects (90.9%), and almost half became negative (54.5%). However, when the earlier results were compared with recent study QFT-GIT data, 10 of 38 (26.3%) previously negative subjects were of intermediate status and 1 (2.6%) was positive; this positive case was negative upon recent QFT-G testing. These data thus also suggested that the sensitivities of the two types of test differed, and that QFT-GIT testing was superior in terms of sensitivity. However, the results of the two tests conducted on the same subjects tended to differ, with statistical significance, suggesting that data from either test should not be compared.
Our present results using the QFT-G test are similar to those of a previous work with HCW of a Japanese tuberculosis referral hospital; approximately 10% were positive [26]. However, use of the more sensitive QFT-GIT test in the present study diagnosed 25% of subjects with LTBI; this proportion is very much higher than that of the general population [22]. A recent systematic review of the utility of IGRA assays used to screen HCW for TB showed that the initial positivity rate was 1.3-31.0%; the analysis included data from countries where the incidence of TB is high [7]. Such variation is attributable to differences in sample sizes and the backgrounds of HCW among the studies; the extent of involvement and contact with TB patients, administration of chemo-prophylaxis, and years of work with TB patients, will have varied significantly among the studies.
In conclusion, our simultaneous and longitudinal study of the same HCW including medical staff in TB-specific wards in a hospital in Japan suggests that QFT-GIT should be used for the screening of Mtb infection in HCW in Japan.

Longitudinal analysis of HCW data
The QFT-GIT and QFT-G tests, performed simultaneously, yielded different results. It was thus important to conduct a longitudinal comparison of the extent of latent Mtb infections in HCW tested 5 years prior using the QFT-G test. Of 120 participants in the present study, results of QFT-G measured 5 years prior were available in 58 subjects (48.3%). Of these, 11 (19.0%) had been previously diagnosed as positive, 9 (15. We further compared earlier QFT-G results with our QFT-GIT data ( Figure 2B). Of 38 previously negative subjects, 10 (26.3%) were