Use of interferon gamma-based assay to diagnose tuberculosis infection in health care workers after short term exposure

In Norway the incidence of tuberculosis (TB) is generally low, and surveillance is primarily based on the detection and treatment of latent infection in risk groups and contact tracing following exposure. It is well established that most of the infected individuals will not progress to active disease, but will maintain a latent infection. Although a latent infection is clinically silent and not contagious, it can reactivate to cause highly contagious pulmonary tuberculosis, the most prevalent form of the disease in adults [1]. Several studies have concluded that transmission is usually caused by prolonged contact with an infectious case of TB, and the risk of being infected is dependent upon the amount of time spent sharing room air with the index case [2]. Still, contact tracing utilizing the fingerprinting methods IS6110-restriction fragment length polymorphism (RFLP) and spoligotyping indicate that a substantial proportion of TB cases acquire infection as a result of casual exposures of short duration [3]. Health care workers are constantly at risk of exposure, and reliable and specific diagnostic tools are essential to improve follow-up procedures. As most new infections result in latent TB infection, strain fingerprinting can not be used to trace transmission.

The current policy in Norwegian health institutions is to perform environmental screenings based on the Mantoux Tuberculin Skin Test (TST) after TB exposure. According to international guidelines, individuals with a TST > 5 mm in non-vaccinated persons and TST > 10 mm in persons vaccinated with Bacillus Calmette-Guérin (BCG) are considered to be infected with M. tuberculosis. In Norway, BCG vaccination is routinely offered to TST-negative children at the age of 14 years. In addition, it has been compulsory for all health care workers to be vaccinated with BCG if a positive TST has not been documented. Thus, in Norway the TST status of health care workers is usually known and the post-exposure TST is compared to the pre-exposure TST based on the increase in the size of the induration. In accordance with the national guidelines, the definition of post-exposure infection is an increase in TST induration of ≥ 10 mm or a TST of ≥ 15 mm if previous TST status is un-known [4]. Norwegian hospital personnel defined as infected are followed up with consultations and annual chest x-rays for three years. In addition, if an employee is likely infected, prophylactic treatment is offered. A major problem in many hospitals is that the same personnel have undergone multiple screenings, which result in boosting of their TST response. Thus, due to both low specificity and TST-induced boosting [5], a large proportion of the group defined as infected after exposure are probably false positives. This may lead to incorrect treatment, waste of resources, and unnecessary anxiety.

However, recently developed interferon-γ release assays (IGRAs), based on the Mtb-specific antigens ESAT-6 and CFP10, have contributed to improved specificity in TB screening. These RD1-encoded protein antigens are absent from all vaccine strains of M. bovis BCG and most non-tuberculous mycobacteria (except M. marinum, M. szulgai, M. Kansasii). Such tests can therefore distinguish Mtb infection from infections caused by other mycobacteria or previous BCG vaccination [5, 6]. Two commercially available and regulatory agency approved test systems can be used: T-SPOT.TB (Oxford Immunotec) is an ELISPOT assay based on the analysis of a defined number of isolated peripheral blood mononuclear cells, whereas Quantiferon-TB Gold (QFT) (Cellestis) is a whole blood in-tube ELISA-based test. The tests have comparable specificity (98–99%), but sensitivity is reported to be somewhat higher for the T-SPOT.TB test (97%) compared to the QFT assay (90%). This difference is primarily pronounced in immuno-suppressed persons and in children, where the frequency of indeterminate results is demonstrated to be higher for the QFT assay [7, 8].

The specificity of IFN-γ release assays has the potential to improve both the diagnosis of TB in infected individuals and the utilisation of public health resources for TB control. In this study, we have compared the T-SPOT.TB test with TST in hospital personnel exposed to TB. We have used the results to assess the role of IFN-γ release assays for improved screening of this target group, and determine the rate of TB transmission during short exposure in a TB low-endemic country.

All individuals detected as T-SPOT.TB positive belonged to the TST positive group, and the percentage concordance between T-SPOT.TB and TST, including both previously and newly infected subjects, was 12% (5/42). Two out of these five individuals were born in a TB high-endemic country. Thus, according to the T-SPOT.TB results the frequency of latent TB in the total cohort was 3% (5/155), whereas the TB transmission rate in the actual TB exposure study was estimated to be 2% (3/155). The 48 control participants were all T-SPOT.TB negative, but 3 persons in the control group were TST positive.

The distribution of TST results, given by mm induration, within the exposed and control group is given in Figure 1. A higher frequency of exposed HCWs with an induration of more than 14 mm was seen. A majority of the healthy controls had an induration of less than 6 mm.

The average time between first exposure and testing was 11.5 weeks. Among the exposed individuals, 51 participants belonged to the "high exposure" group (> 8 hours), and 104 participants fell into the "low exposure" group (≤ 8 hours) (Table 1). There was no correlation between length of exposure and TST results. In addition, there was no correlation between T-SPOT.TB positivity and TST results. For the 3 T-SPOT.TB positive individuals infected after known exposure, two were exposed ≤ 8 hours and one was exposed > 8 hours.

A positive TST has previously been considered the gold standard in screening for M. tuberculosis infection. However, there is currently no definitive way to decide whether a person is latently infected with M. tuberculosis. A recent prevalence study by Soborg et al. in a Danish hospital confirmed earlier findings that TST is hampered by low specificity in BCG-vaccinated individuals [10, 11]. The authors found a 34% prevalence among TST-positive HCWs, but the only significant risk factor associated with a positive TST was prior BCG immunisation; no association was observed with other important risk factors, such as occupational exposure to TB or hospital staff position [10]. Other low-endemic country studies have also found that a positive TST test is primarily associated with prior BCG vaccination and the boostering effect of prior TST testing [12–15]. Nevertheless, how can we conclude that the low post-exposure rate of infection detected by T-SPOT.TB in our study represents the real situation if there is no gold standard? One possible strategy is to estimate the likelihood of having latent TB infection by calculating a contact score that quantifies exposure to and infectiousness of the index case, as was done by Shams et al. [16]. A range of other studies also provide extensive evidence that the IGRAs correlate better to exposure than does TST. Therefore, we based our conclusions regarding the prevalence of TB infections on the T-SPOT.TB test [5, 12–14, 16–21].

Of 155 exposed HCWs and 48 healthy controls, all but one had a visible scar from BCG immunisation, which has been compulsory in Norway at the age of 14 until recently. Norwegian legislation also demands that all HCWs are asked for a certificate of TST at the time of appointment, and if the existence of a recent TST cannot be documented then a new TST is performed. The high correlation between BCG vaccination and a positive TST and the high specificity of the T-SPOT.TB test for Mtb infection (98%) make it likely that these are mostly false positive TST reactions rather than false negative T-SPOT.TB results. In addition, as found by Nienhaus et al. [14], the boosting effect of repetitive skin testing in health personnel may also contribute to the somewhat surprisingly low concordance between infection status according to T-SPOT.TB and TST that was observed in this study (12%). Still, there is no diagnostic gold standard for latent TB and the fact that 88% of employees recently defined as infected by TST tested negative with the T-SPOT.TB test calls for further studies of kinetics and immune mechanisms in TB infection. In our study all five T-SPOT.TB-positive individuals also had a strong positive TST. Nienhaus et all found that 5.1% of the HCWs in three German hospitals had a negative TST and a positive QFT [14]. While the combination of a negative IGRA and a positive TST is largely explained by prior BCG immunisation and TST boostering, the other combination is not readily explained, and further research is needed.

In contrast to most other studies related to TB transmission within health institutions, we have in this work compared the performance of T-SPOT.TB and TST in a group of HCWs with well-defined short-term exposure to contagious TB patients in a hospital setting. In addition, the results have been used to evaluate the role of IGRAs in improving the surveillance of TB transmission to health personnel in a low-incidence country like Norway.

Provided that the T-SPOT.TB results are the most reliable compared to TST results, our study indicates that the risk of infection among health care workers after short-term exposure to TB patients in a hospital setting is low (2%). This somewhat contradicts the findings of a Swiss long-term institutional study in which 15% of contacts were T-SPOT.TB-positive after prolonged unprotected exposure [13]. Although both studies were performed in health care institutions in low incidence countries, the exposure time may account for the observed differences in transmission. A study from Denmark also reported a low proportion (1%) of latent TB among HCWs as detected by the QFT test [10]. However, these data were not based on recent exposure, but rather represent the general prevalence level among hospital personnel working in departments with TB patients. We found a prevalence level of 3% in our cohort. Not surprisingly, these results are in contrast to findings from a high-endemic country like Russia, where a study utilizing QFT revealed a prevalence level of 41% among hospital staff working with infectious diseases [22]. Several reports based on TST conversion indicate that the risk of being infected may be high, even within a limited time frame of exposure [23, 24]. This has also been confirmed by a T-SPOT.TB study in Italy in which 32% of the staff in a maternity ward became positive for TB after a mean exposure time of 6 hours [19]. Compared to these findings, our study detected a low degree of transmission. However, it should be noted that the majority of the individuals in this study had been exposed for less than 8 hours before precautions were taken. Still, among the three persons with positive T-SPOT.TB tests (believed to be recently infected), two were exposed for less than 8 hours. The absence of a statistically significant correlation between exposure time and both TST and T-SPOT.TB results is probably due to both the low exposure time and the small number of participants, since these correlations have been demonstrated in many other studies [19, 23, 24].

Importantly, the results indicate that using IGRAs as an alternative to the present follow-up strategy based on TST results could save substantial resources. Although TST by itself will be less expensive than a comprehensive laboratory procedure, the utilization of Mtb-specific blood tests has the potential to save major resources as the number of persons who must be followed up for 3 years can be reduced by up to 88%. In this context, the possibility of avoiding unnecessary and costly treatment, including serious side effects, is also of considerable importance. In addition, most exposed health care workers will avoid long-term anxiety by obtaining a negative result at a very early investigational stage. Finally, the small number of infected persons who require treatment can be identified immediately.

Oxlade et al. performed a 20-year cost-benefit analysis that used Markov modelling to compare the costs of TB screening with different strategies among hypothetical cohorts of foreign-born immigrants to Canada and contacts of TB cases. Model inputs were derived from published literature and utilization of the QFT test. For entering immigrants, screening with Chest X-Ray would be the most cost-effective and QFT the least cost-effective strategy. Sequential screening with TST followed by QFT was more cost-effective than either QFT or TST alone. In contact tracing after exposure, however, screening with TST followed by QFT, if positive, was more cost-effective than any other strategy. This was largely because TST alone was not effective if the exposed group had been vaccinated with BCG after infancy [25]. These findings were also confirmed by a Swiss study by Wrighton-Smith et al., estimating the costs of screening a cohort of 1000 individuals for latent tuberculosis; screening with TST alone followed by Chest X-Ray and clinical follow up of the positive cases was estimated to €695820; T-SPOT.TB alone was estimated to €387135; TST followed by T-SPOT.TB of the positives was estimated to €342563, i.e. the less costly [26].

Due to the fact that BCG immunization has been administered routinely in Norway, specific blood tests should be introduced in all post-exposure contact tracing situations. Because TB transmission to health personnel in Norway seems to be rather low, the two-step screening approach (TST followed by IGRA) might be attractive. However, some studies indicate an unacceptable low sensitivity of TST; Nienhaus et al. found that 40% of the HCW with latent TB infection according to IGRA results had a negative TST, and would have been missed utilizing TST followed by QFT of the TST positives. There is also evidence that a positive IGRA is a much better predictor of future reactivation than a positive TST [27], which is of particular interest because the Norwegian guidelines recommend prophylactic treatment in all documented cases of latent TB infection aged < 35 years. Further research is obviously needed to define improved screening strategies in low-endemic settings, both in the light of sensitivity and cost-effectiveness.

It should also be noted that the QFT method, although less sensitive in immuno-suppressed individuals, has both logistic and economic advantages compared to the T-SPOT.TB assay, and implementation of the QFT test has recently been suggested in Norway's national guidelines. The introduction of specific T-cell based assays for post-exposure screening and subsequent prophylactic treatment will become a rational and important component of the national TB control strategy.