IFN-γ release assay versus tuberculin skin test for monitoring TB infection in healthcare workers

Abstract

Healthcare workers (HCW) are a risk group for TB. Even in countries with low TB incidence, the risk of TB in HCW is elevated for a wide range of tasks in healthcare, and the prevention of nosocomial infection of HCW remains as a challenge. IFN-γ release assays (IGRA) facilitate the screening of HCW for latent TB infection. In comparison with the tuberculin skin test, the IGRA reduces the number of x-rays and the amount of chemoprevention needed. However, a borderline zone should be introduced for the interpretation of IGRA results in the serial testing of HCW. More data on disease progression depending on conversion and reversion in IGRA is needed and a better test, which is able to distinguish recent from remote latent TB infection, would be desirable in the future.

Keywords::

• healthcare worker
• IFN-γ release assay
• latent TB infection
• preventive treatment
• screening
• tuberculin skin test
• tuberculosis

Healthcare workers (HCW) are exposed to infectious agents. The excess death rate of HCW due to work-related infections was estimated to be between nine and 24 per million in the USA [1]. However, the increased risk of infection for HCW is not always easy to detect. This not only pertains to new, emerging infectious diseases like severe acute respiratory syndrome [2], but also to well-known diseases like TB. Working in healthcare was long considered a safe environment offering protection against TB. Only when the prevalence of the disease declined in the general population it become apparent that the rate of latent TB infection (LTBI) and active TB was high in those caring for tuberculosis patients [3]. With the further decline of TB in high-income countries, interest in it as an occupational disease dwindled. It was with the emergence of HIV that interest in TB as a co-infection inspired research and drew attention to the prevention of infectious diseases in HCW.

Several recent systematic reviews showed that HCW are at an increased risk for exposure to Mycobacterium tuberculosis [4–7]. In countries with low TB incidence, the relative risk of TB in HCW might not exceed the one in the general population, as HCW do not share the other risk factors for TB, for example, homelessness or drug addiction, and not all HCW are at increased risk of exposure [5]. However, the risk of exposure to TB is increased for a wide range of tasks or facilities in healthcare, for example, bronchoscopy, laboratory work, pneumology departments, emergency rooms and pathology [7]. In accordance with these findings, TB in HCW is on the list of occupational diseases compiled by the International Labor Organization [8] and different measures for the control of infection in healthcare have been proposed [9]. Early detection and isolation of infectious patients combined with the use of masks or respirators by patients and HCW as well as TB screening of HCW are the most important measures. In the USA, it has been well demonstrated that implementation of these infection-control measures reduces nosocomial infection. Between 1985 and 1993, several outbreaks of multi-drug-resistant TB were reported in nosocomial settings in the USA [6]. This led to recommendations for a comprehensive set of infection-control practices to protect HCW and reduce nosocomial transmission [9,10]. In the years following the publication of these recommendations, there was a strong decline in the burden of TB among HCW [9–11]. In Italy, Baussano et al. showed that introducing infection-control measures, including TB screening of HCW [12], led to a decrease in the annual rate of TB infection in HCW. In Portugal, Torres Costa et al. demonstrated that the introduction of systematic TB screening for HCW helped to improve awareness of HCW for infection control and dramatically reduced the incidence of TB in HCW [13,14]. In a recent tuberculin skin test (TST)-based screening program for HCW in Thailand, active TB was found in 6.3/1000 participants [15]. In China, 20 cases of pulmonary TB were detected among 3746 HCW tested with TST in 2005. The TB prevalence was 6.7/1000 among medical staff and 2.5/1000 among administrative/logistic staff [16]. The effectiveness of TB screening for HCW in China was therefore similar to that of screening in Thailand. As a result, systematic screening of HCW should be implemented in these countries [15,17].

TB screening for HCW is performed in order to prevent nosocomial transmission from HCW to patients through early detection of lung TB in HCW and in order to detect and treat recent LTBI in HCW [9]. Screenings can be performed as a pre-employment screening, routinely repeated screening, or as contact tracing after accidental contact with infectious patients or materials. It is therefore important that HCW undergo repeated TB screening and the interpretation of results in the serial testing of HCW is a major issue.

These screenings have been performed with the Mantoux TST for more than 100 years [18]. For several years now, two IFN-γ release assays (IGRA) have been commercially available: the ELISA-based QuantiFERON^®-TB Gold In-Tube (QFT; Cellestis, CA, USA) and the ELISPOT-based T-SPOT.TB^® (Oxford Immunotec, Oxfordshire, UK). Since their release, data on their performance in TB screenings for HCW have become available from different countries and they are currently being evaluated for use in serial TB screenings for HCW [19,20].

In this article, the performance of the IGRA in comparison with the TST in TB screenings for HCW will be discussed. Special emphasis will be placed on the IGRA results in the serial testing of HCW.

Methods

Recent reviews [19,20], and studies which will update these reviews, were identified in the PUBMED and EMBASE databases. For the keyword search, the same strategy was chosen as that used and described in the reviews by Zwerling et al. [19] and Ringshausen et al. [20].

TB screening for HCW

Routine TB screening for HCW is considered as a cornerstone for preventing TB in hospitals and it is performed in many high-income countries, for example, USA [9], Japan [21], France [22], Spain [23] and Switzerland [24]. In the UK, TB screening of HCW is mainly performed as a pre-employment screening [25]. Other countries, such as Thailand and China [15,17], are on the verge of introducing systematic screening of HCW.

The screening strategy should be developed depending on risk assessment [9]. Pre-employment screening is important if workforce recruits migrated from countries with high TB incidence. The risk of exposure for HCW can be categorized as follows: HCW with regular contact with infectious TB patients or infectious material belong in the high-risk group. They are HCW in TB wards and laboratories that analyze sputum or other fluids for M. tuberculosis. Depending on the TB incidence in patients, they might also be HCW in emergency rooms or HIV outpatient clinics. The medium-risk group comprises HCW with regular contact with patients who are not known to have active TB. Once again, the risk of infection here depends on the incidence of TB in the patients. The more patients diagnosed with TB, the higher the infection risk for the HCW. However, delayed diagnosis of TB in a hospital with few TB patients and limited experience in the diagnosis of TB might result in a higher infection risk for HCW than expected [7]. All HCW who do not have regular contact with patients or potentially infectious materials belong in the low-risk group. The HCW of the high-risk group should be screened routinely on an annual, bi-annual or tri-annual basis, depending on risk assessment. In countries with low TB incidence and high hygiene standards, bi- or tri-annual screenings may suffice instead of annual screening if the conversion rate in HCW is low. For the medium-risk group, it can be considered whether TB screening is performed routinely or is reduced exclusively to contact tracings. Again, this depends on the above-mentioned risk assessment. HCW in the low-risk group should not be routinely screened as this would reduce the positive predictive value of the screening test. They should only be screened after accidental close contact with infectious patients or materials.

Immunological test for TB screening

Until recently, TB screening was performed using a Mantoux TST. As with the IGRA, TST measures the cell-mediated immune response to antigens specific for M. tuberculosis. Therefore, both TST and IGRA do not measure the presence of viable M. tuberculosis but merely the immunological footprints left behind by a remote or a recent infection. This is important to bear in mind when the variability of IGRA is discussed later. LTBI is considered to be given when the immunological tests are positive and active TB is ruled out by way of x-ray, bronchoalveolar lavage or sputum-smear microscopy.

For the TST, tuberculin is injected strictly intradermally on the inner side of the forearm. The injection is successful when a wheal becomes visible on the skin. In order to read the test, a second appointment is needed 48–72 h after the application of the tuberculin. The largest diameter of the induration needs to be measured [26]. The erythema, which can be much larger, should not be confused with the induration. Test interpretation depends on the circumstances and country. For example, a diameter of 5+ mm is considered positive in Germany if the HCW was recently exposed [27]. Otherwise, a diameter of 10+ mm is considered positive in unexposed persons. Following ATS guidelines, a transversal induration diameter of 5+ mm is considered a positive test result [28]. In most countries, a diameter of 10+ mm is considered as a positive result, for example, Portugal [13], France [22] and Vietnam [29]. In France, recent infection is considered likely if the diameter is larger than 15 mm [22]. In addition, an increase of 6+ or 10+ mm compared with an earlier negative TST (<10 mm) is considered indicative of a recent LTBI [26,30].

Tuberculin is a purified protein derivate from different strains of M. tuberculosis and it contains a mixture of many antigens [31]. These antigens are shared with the attenuated strains of Mycobacterium bovis used for the BCG vaccine and by nontuberculous mycobacteria such as Mycobacterium avium. Cross-reactivity with a BCG vaccination and sensitization, the so-called booster phenomena, due to intradermal application result in a rather low specificity of the TST when screening HCW [13,19,27]. As different strains of M. tuberculosis are used for the production of tuberculin and different amounts of tuberculin are injected intradermally, comparison of study results from different countries is hampered.

The commercially available IGRA use two antigens of the region of difference I of M. tuberculosis: early secreted antigen target 6 (ESAT-6) and culture filtrate protein 10 (CFP-10). In addition, the QFT uses the antigen TB7.7, encoded in region of difference 11 [32]. With the ELISPOT-based technique for the T-SPOT.TB, stimulation with the two antigens (ESAT-6 and CFP-10) is performed separately. Alternatively, stimulation with the three antigens (ESAT-6, CFP-10 and TB7.7) is performed simultaneously with the ELISA technique for the QFT. The antigens used by the IGRA are not shared by the strains of M. bovis used for BCG vaccination. In systematic reviews, the specificity of the IGRA was estimated to be higher than the specificity of TST [33–35].

IGRA are in vitro tests. As a result, the problem of boosting in serial testing is circumvented. IGRA results are available within 24 h compared with the 48–72 h for the TST. No second appointment is needed for reading the test. A further advantage of the IGRA is the use of mitogen and unstimulated controls. The mitogen control tests the immune competence of the patient and helps to avoid false negative results due to a compromised immune system. The unstimulated (Nil) control correct for unspecific reactions.

In short, a QFT is considered positive if the INF-γ concentration is ≥0.35 IU/ml after subtraction of the INF-γ concentration of the Nil control. In addition, the INF-γ concentration of the specific antigen control needs to be 25% higher than the concentration in the Nil control. The test is considered indeterminate if the mitogen control is <0.5 IU/ml and the difference between antigen tube and Nil control is <0.35 IU/ml. In short, T-SPOT.TB is considered positive if the number of spot-forming cells (SFC) in at least one of the two antigen-stimulated wells is above six after subtracting the SFC of the Nil control. In addition, the spots of the specific antigen wells need to be twice those in the Nil control wells. The T-SPOT.TB is considered indeterminate if the number of SFC in the positive control well is less than 20 and if the antigen-specific wells are negative. In addition, it is recommended to use a borderline zone between 5 and 7 SFC for the T-SPOT.TB [31,36]. For the QFT, no comparable recommendation regarding a borderline zone exists so far. The manufacturers provide software for the interpretation of the IGRA. However, test results should be reported qualitatively and quantitatively.

In contact tracings, IGRA correlated better than TST with exposure to infectious patients [35,37]. Furthermore, in countries with low TB incidence and high average income, IGRA have a higher predictive value for disease progression [38–41]. Restricting the analysis to high-risk groups, a progression rate of 6.8% after a positive IGRA was observed in a meta-analysis [40]. This was higher than for TST (2.4%) and differences were statistically significant. So far, the highest progression rate after a positive IGRA was observed in German close contacts (12.9%) [39]. Surprisingly, the progression rate was three-times higher in German-born contacts than in contacts of non-German heritage (5 vs 15.2%, calculated from [39]), while the prevalence of positive IGRA was higher in non-German contacts (OR: 2.4). Most likely, this reflects different proportions of remote and recent infections in the subgroups. The progression rate is diluted by remote LTBI with a low risk of progression. This might also be the reason why Rangaka et al. did not observe a different progression rate for TST and IGRA for countries with high TB incidence in their meta-analysis [41].

When the performance of the two commercially available IGRA is compared, the sensitivity of the T-SPOT.TB seems to be higher than the sensitivity of the QFT, while the specificity of the QFT seems to be superior to the specificity of the T-SPOT.TB [37]. However, in a head-to-head comparison in a contact tracing in a country with low TB incidence, the agreement between the two IGRA was strong [42].

As the knowledge about the performance of IGRA is growing rapidly, the use of IGRA in contact tracing or HCW screening is endorsed in different national guidelines [9,36,43,44]. However, interpretation of IGRA results in the serial testing of HCW must still be clarified and a consensus on its interpretation needs to be found [19,20]. Since the performance of IGRA in serial testing is not yet well understood, serial testing of HCW with IGRA is not recommended in Canada [45].

The effect of introducing IGRA in TB screening of HCW

Owing to the higher specificity, the introduction of IGRA in TB screening of HCW is likely to reduce the number of x-rays and the number of preventive treatments that would be needed if a decision were made based on the TST. This should be particularly true for countries in which BCG vaccination is still performed or was performed until recently. However, in the systematic review of Zwerling et al. [19], this effect was well documented for countries with low and intermediate TB incidence but not for countries with high TB incidence.

In addition to Zwerling et al. [19], who covered studies until October 2010, the authors identified seven further cross-sectional studies that compared TST and IGRA performance in HCW [30,46–51]. Therefore, the number of cross-sectional studies comparing TST and IGRA in HCW increased from 27 to 34 [19]. Among 32 cross-sectional studies from countries with low and intermediate TB incidence, all but one reported a lower prevalence of positive IGRA than positive TST [52]. The difference in the prevalence of positive TST and IGRA was not statistically significant in seven studies [25,48,53–57]. In the other 24 studies, the difference was statistically significant [13,21,23,27,30,46,47,49–51,58–70]. However, there was no clear trend that TST-positive/IGRA-negative discordant results increased with an increasing proportion of BCG-vaccinated HCW in the study populations. Different vaccines, different age at vaccination and different revaccination schemas might be responsible for this lack of correlation. Most of these studies were performed with QFT, but six studies used the T.SPOT.TB and showed basically the same results [23,51,52,62,64,70]. All but one showed lower proportions of positive T-SPOT.TB than positive TST and the differences were statistically significant [52].

Only two cross-sectional studies comparing TST and IGRA are available for HCW in countries with high TB incidence and the difference between TST and IGRA-positive HCW is less obvious [71,72]. TST and IGRA positivity rates were high in both studies (40–66%) and a significant difference between TST and IGRA was found only in the Vietnamese study, even though BCG vaccination was more often performed in India than in Vietnam (71 vs 37%). Again, different vaccination schemas and vaccines might explain these divergent results.

In BCG-vaccinated HCW, the prevalence of positive TST is two- to five-times as high as in unvaccinated HCW [13,22,27,46,47]. No clear relationship between time of last BCG vaccination and TST is observed [13].

The prevalence of a positive IGRA increases with the diameter observed in the TST. Therefore, a more stringent definition for TST will reduce the number of TST-positive results not confirmed by IGRA. However, this will increase the number of TST-negative/IGRA-positive discordant results [59,60,67]. In addition, even if the cutoff for the TST is increased to 15 mm, only every second positive TST is confirmed by IGRA [13,22]. Thus, unspecific reactions of TST cannot be excluded even with very high cutoffs for the TST.

Table 1 lists studies with a head-to-head comparison of TST and IGRA results in HCW that allow the display of the concordant and discordant results. The results of 11 studies comprising a total of 3733 HCW are provided [13,22,23,25,27,50,53,55,59,62,67]. Concerning discordant results, the proportion of IGRA-negative/TST-positive discordance is ten-times larger than IGRA-positive/TST-negative discordance (29.5 vs 2.7% of all HCW tested). The highest proportions of IGRA-negative/TST-positive results were observed in Japan (57.9%) and France (50%). In both studies, a diameter in TST of 10 mm and more was considered positive and BCG vaccination is universal [22,67]. In general, the proportion of IGRA-negative/TST-positive results is low when the proportion of positivity in both tests (TST and IGRA) is low as well. For 62.3% of the HCW who are positive in TST, further evaluation in order to rule out active lung TB by x-ray can be spared, as long as there is no evidence that the prevalence of active TB or progression to active TB is increased in HCW with a positive TST and a negative IGRA [38].

The TST is negative in 13.1% of the HCW with a positive IGRA (calculated from Table 1). However, this proportion seems to be much larger when the prevalence of LTBI is low; for example, 33, 23.8 and 40% of the HCW with a positive IGRA in the USA, Australia and Germany, respectively, were affected [27,50,59]. In a two-step, screening – IGRA for the verification of a positive TST – the prevalence of a LTBI would be largely underestimated. However, this problem is of a smaller scale in countries with high prevalence of LTBI in HCW; for example, 16.3, 6.5 and 2.3% of the HCW with a positive IGRA in Georgia, Portugal and Spain, respectively, had a negative TST [13,23,53]. The second study from Spain comprising young, unexposed students found a low prevalence of LTBI and a high proportion of TST-negative healthcare students (37.5%) in those with a positive IGRA [55]. Therefore, two-step screening cannot be recommended for a country or a population with a low incidence of TB, while, for a country with a higher incidence of TB, the problem of underestimation seems to be on a scale that might be acceptable. In this context, it needs to be mentioned that little is known so far about the meaning of TST-negative/IGRA-positive discordant combinations. These combinations are partly explained by decreasing sensitivity of the TST with age [73]. In the sole HCW study that observed active TB during follow-up, only concordant TST and IGRA-positive HCW developed active TB [38].

When switching from annual TST screening to annual IGRA screening in a country with low risk of exposure to infectious TB patients or materials and low BCG vaccination rate, discordant test results, that is, actual positive IGRA and negative TST results in the past, might occur as described by Gandra et al. [74]. Out of 6530 HCWs tested with IGRA, 287 (4.4%) were positive and, out of these 287 HCW, 164 (57.1%) had a negative TST in the past. When retesting these 164 HCW with TST and IGRA, within 4 weeks after the first IGRA, only 69 (42.1%) remained IGRA-positive and two HCW were positive in TST (1.2%). As only two of these HCW reported exposure to TB patients in the past, it remains unexplained why the IGRA was, and remained, positive in 67 HCW while the TST was negative. However, it does not seem likely that these HCW will be at an increased risk for developing active TB. Disregarding the potential for unknown contacts with non-diagnosed TB patients, or sensitization with nontuberculous mycobacteria other than M. avium, the data reported translates into a specificity of the IGRA (QFT) of 97.5% (164/6530) or 99% (67/6530), which is well within the expected range [35,37]. A high specificity of IGRA (T.SPOT-TB) was also found among 124 unexposed students of a college healthcare center (100%). As no student was BCG-vaccinated, agreement with the TST was good (97%) [48].

Progression from LTBI to active TB in HCW

The risk of progression from LTBI to active TB in HCW seems to be lower than in close contacts in the general population since data on disease progression after a positive IGRA are only reported from one cohort of HCW and the progression rate was low. Based on four predicted cases of TB, 0.4% of IGRA-positive and 0.2% of TST-positive HCW progressed from LTBI to active TB [38]. Neither the TST nor the IGRA are able to distinguish between a remote and a recent infection. Taking the high prevalence of positive IGRA in some HCW studies into consideration (Table 1), it is to be assumed that most positive IGRA are due to a remote infection. As progression to active TB is highest during the first 2 years after infection, it does not seem to be useful to offer preventive treatment to all HCW with a positive IGRA.

So far, the variability of the IGRA in serial testing is not well understood. The IGRA are ex vivo tests, therefore other than with the TST, test variability is not influenced by previous in vivo application of the antigens. Whether a previous TST might boost the result of an IGRA is debated. However, the effect, if any, seems to be on a minor scale [20]. Reasons for test variability of IGRA may be based on variations of the handling and reading of the tests, variability of the immune system independent of infection status, and differences in the activity of the M. tuberculosis infection, that is, transient infection, low replication of M. tuberculosis with no stimulation of the immune system, high replication of M. tuberculosis with stimulation of the immune system and uncontrolled replication causing active TB [75,76]. In addition, reversions are partly explained by the statistical mechanism of regression toward the population mean [20]. This effect is expected to be higher in countries with low LTBI prevalence as it is confirmed by the analysis below.

IGRA variability & serial testing of HCW

Three reviews have covered the topic of IGRA variability in the serial testing of HCW so far [19,20,77]. All three came to the conclusion that reversion of positive IGRA results to negative results occurs more often than conversion from negative IGRA results to positive ones. In addition, more importantly, the probability of conversion or reversion depends on the quantitative results of the first IGRA. Therefore, a borderline zone might be helpful in order to separate real conversions and reversions from variation caused by chance. However, so far a borderline zone has only been proposed for the T-SPOT.TB [31,36] based on two rather small studies [76,78] and no consensus has been reached regarding the definition of such a borderline zone for the QFT.

Table 2 gives an updated overview of the studies available so far on the serial testing of HCW using IGRA. Studies investigating short-term variability in small samples of HCWs were disregarded. For an optimal assessment of the variability of IGRA in serial testing, studies should be performed in low exposure settings in order to estimate the natural variability of the immune response to M. tuberculosis. A total of 15 studies on serial testing are available: one study from a country with high TB incidence [79], five studies from intermediate countries [14,80–83] and nine studies from countries with low TB incidence [74,84–91].

In all studies, the reversion rate was higher than the conversion rate. There is a tendency of lower conversion rates in countries with low TB incidence, but the ranges of the conversion rates for the studies categorized by TB incidence are overlapping in countries with high (10.1%), intermediate (2.1–14.4%) and low TB incidence (0.7–7.1%). The reversion rate was lower in the country with high TB incidence (23.7%) than in countries with low TB incidence (31.0–57.9%). Only one study analyzed conversion and reversion rates of T-SPOT.TB [92] and the results were comparable with the ones observed for QFT.

In Table 3, the results of a second IGRA are given, which were dependent on the qualitative results of the first IGRA for a combined cohort of Portuguese and German HCW [14,88,89]. Conversions occurred most frequently when the concentration of IFN-γ was between 0.2 and <0.35 IU/ml. Reversions occurred most frequently when the first positive IGRA showed IFN-γ concentrations of between 0.35 and 0.7 IU/ml. Therefore, 0.2 and 0.7 IU/ml are potential candidates for the lower and upper limits of a borderline zone. Applying this borderline zone reduces the conversion rate to 5.2% and the reversion rate to 10.9%. However, conversions were also overrepresented when the concentration in the first negative IGRA was between 0.1 and 0.2 IU/ml, and reversions were also high when the first positive IGRA showed concentrations of between 0.7 and 1.0 IU/ml. Therefore, the borderline zone might be extended to cover the range between 0.1 and 1.0. However, this would reduce the conversion rate to 2.8% but would influence the reversion rate only slightly (9.7 vs 10.9%). In addition, the number of HCW falling into the borderline zone with their IGRA results would increase from 8.4 to 19.2%. Therefore, a borderline zone from 0.2 to 0.7 IU/ml is best supported by this data. It has been suggested to extend the borderline zone to 2.0 IU/ml as all reverters had concentrations below 2.0 IU/ml in the first IGRA in the respective study [86]. In the absence of any TB risk, this might be a wise decision; however, this borderline zone seems to be too large for exposure-guided screenings. Alternatively, to a borderline zone for the IGRA, a minimal change of the IGRA results might be defined and results below this minimal change could be considered equal. This concept was already followed for the TST. TST results with less than 6 mm difference should be considered identical [26]. For QFT a minimal change of 0.35 IU/ml for a conversion or reversion is discussed [20]. However, this definition is less stringent than introducing a borderline zone from 0.2 to 0.7 IU/ml for the QFT. In addition, it does not distinguish between those results close to the cutoff and those further away from the cutoff. A borderline zone, between 5 and 7 SFC for the T-SPOT.TB is already endorsed by the European Centre for Disease Prevention and Control and CDC [31,36]. This further supports the introduction of a borderline zone for the QFT.

Expert commentary

TB in HCW will remain an unresolved issue over the coming years, even in countries with a low incidence of TB and a high average income. This will be the case all the more in low-income countries with a high incidence of TB. The introduction of IGRA into TB screenings for HCW changed our knowledge about LTBI in HCW. In particular, it seems safe to say that the prevalence of LTBI is lower than had been previously thought based on studies using the TST. For HCW, this is good news as the medical evaluation after a positive test can be spared for many HCW who are thought to be infected based on the TST. In addition, the once-positive-always-positive approach used in HCW screenings with TST can be abandoned in many countries. Because of the unexpectedly high reversion rate, it is reasonable to retest a HCW with IGRA in serial testing even though the IGRA has been positive before. Once again, this will spare x-ray exposure for the HCW because no further medical evaluation is needed if the IGRA reverted to negative and no clinical signs of active TB are apparent. How oscillating HCW, who change from positive to negative and back again, should be treated remains to be discussed. A simple approach would be to rule out pulmonary TB whenever they test positive in IGRA and not to do any further medical evaluation if they test negative and no clinical symptoms are apparent.

The introduction of a borderline zone will further reduce medical evaluations, as the number of conversions will decrease. However, the subject gets a little cloudy here, as we do not have data on disease progression depending on reversions and conversions of IGRA. Even with the classical negative/positive approach, data on disease progression are sparse. Critics might wonder whether we are testing the wrong HCW and it is not easy to prove these critics wrong. The only study that reported progression to active TB after a positive IGRA in HCW found the progression rate to be 0.4% [38]. Screening for these HCW was performed following risk assessment and the HCWs who progressed to active TB pertained to the medium-risk group. However, it needs to be discussed whether the risk categories explained above are defined well enough to facilitate the detection of recent LTBI with increased risk of disease progression. Using an exposure score which reflects the infection risk during the last 6 months might be helpful in order to differentiate remote from recent infection.

IGRA are good news for epidemiologists as well. Due to the high specificity of both IGRA (QFT and T-SPOT.TB), studies on infection risk and surveillance programs can be conducted without struggling with high probabilities of misclassification of the outcome. Since non-differential misclassification of LTBI will dilute any association between exposure and outcome, here LTBI, the quality of epidemiologic studies investigating occupational risk factors for LTBI will improve with the use of the more specific IGRA instead of the TST.

Replacement of TST with IGRA should be performed with care, especially in HCW with low TB risk. As specificity of IGRA is not 100% and sensitivity of TST is less than perfect, changing from TST to IGRA screening will reveal a group of TST-negative/IGRA-positive HCW. Retesting these HCW within 4 weeks will separate those who are IGRA-positive owing to an unspecific reaction from those who are TST-negative owing to the imperfect sensitivity of TST but truly IGRA-positive. However, this separation will never be perfect and we need better data in order to decide whether HCW with a history of negative TST but positive IGRA will need medical evaluation and preventive treatment regardless of the recent exposure situation.

TB screening of HCW using IGRA instead of the TST will be useful in countries rich in resources. However, a test that can distinguish between remote and recent infection is desirable and more studies on the interpretation of IGRA results in serial testing are needed. Data on progression to active TB depending on the variation of IGRA results are most valuable in this respect.

Amid growing evidence that TB screening for HCW with IGRA has advantages over screening with TST, it needs to be kept in mind that TB screening for HCW with TST is better than no screening at all. Especially in countries with limited resources and lack of established TB screenings for HCW, the TST will remain an important tool in occupational medicine for HCW due to its low price and the easy usability without the need for a high-tech laboratory. However, it would be desirable to replace the tuberculin used for the TST with more specific antigens, e.g. with those used by the IGRAs. In a recent animal study, this was shown to be feasible and is currently being tested for humans [93].

Five-year view

Within the next several years, the concept of LTBI will be better understood and data on disease progression depending on conversion or reversion in IGRA will become available, which will help decision-making regarding preventive treatment. However, in order to achieve this goal, multicenter studies and maybe even multi-country collaborations are needed as the progression rate in HCW with reversions in the IGRA is most likely low and large samples of HCW are therefore needed. An example of this approach is the French–Portuguese–German-collaboration on IGRA evaluation in HCW, which should be further elaborated [94].

More countries will have developed and implemented comprehensive programs for the prevention of nosocomial transmission and more data will be available on the effectiveness and efficiency of these programs within the next several years. This will help other countries shape their own preventive programs and it will further strengthen our knowledge of LTBI and disease progression in HCW.

It is hard to predict whether a test will be available within 5 years that will be able to distinguish remote from recent infection with M. tuberculosis so that patients profiting most from preventive treatment can be better targeted. Studies that try to identify additional biomarkers for LTBI are currently being performed [95]. Whether these potential new biomarkers will be able to distinguish recent from remote LTBI remains to be seen.

As only a small portion of those who are infected will develop active TB, it is important to better understand the protective immunity in individuals after being infected. This understanding would allow for a more targeted preventive treatment of infected patients. This way the number of patients to treat in order to prevent active TB in one patient could be reduced and the acceptance of preventive treatment in physicians and patients would improve.

Currently, the antigens used in IGRA are protected by patent. This will expire within the next several years. Therefore, new competitors in the market might offer a version of the IGRA at a lower price. This would be most desirable in order to increase the availability of IGRA in countries with few resources but high TB incidence. But, even in countries with high average income, the healthcare system often has severe budget restraints and a less expensive tool for effective TB screening is desirable.

Preventive treatment of LTBI is currently performed with Isoniazid for 6 or 9 months. This hinders doctor’s enthusiasm and patient’s acceptance of preventive treatment of LTBI. Data on an alternative short-term plan for preventive treatment as effective as isoniazid monotherapy and with fewer unwanted side effects was recently published [96]. It would be very helpful if this particular short-term treatment or other effective short-term treatments [97] could replace the long-term treatment of LTBI with isoniazid and increase the acceptance of preventive treatment of LTBI by doctors and patients in years to come.

Table 1. Head-to-head comparison of the IFN-γ release assay and the tuberculin skin test in healthcare workers from different countries.

Study (year)	Country	Sample size	IGRA+; TST+	IGRA-; TST+	IGRA+; TST-	IGRA-; TST-	Ref.
		n	n (%)	n (%)	n (%)	n (%)
Hotta et al. (2007)	Japan	202	3 (1.5)	117 (57.9)	0	82 (40.6)	[67]
Mirtskhulava et al. (2008)	Georgia	265	133 (50.2)	44 (16.6)	26 (9.8)	62 (23.4)	[53]
Nienhaus et al. (2008)	Germany	261	15 (5.7)	48 (18.4)	10 (3.8)	188 (72.4)	[27]
Alvarez-León et al. (2009)	Spain	123	5 (4.1)	4 (3.3)	3 (2.4)	111 (90.2)	[55]
Casas et al. (2009)	Spain	145	42 (30.0)	59 (40.1)	1 (0.7)	43 (29.7)	[23]
Girardi et al. (2009)	Italy	115	25 (21.7)	36 (31.3)	4 (3.5)	50 (43.5)	[62]
Khanna et al. (2009)	UK	148	9 (6.1)	15 (10.1)	3 (2.0)	121 (81.8)	[25]
Torres Costa et al. (2009)	Portugal	1218	371 (30.5)	532 (43.7)	26 (2.1)	289 (23.7)	[13]
Tripodi et al. (2009)	France	148	23 (15.5)	74 (50.0)	5 (3.3)	46 (31.1)	[22]
Vinton et al. (2009)	Australia	341	16 (4.7)	98 (28.9)	5 (1.5)	222 (65.5)	[59]
Khoury et al. (2011)	USA	611	8 (1.3)	42 (6.9)	4 (0.7)	557 (91.1)	[50]
Total		3733	668 (17.9)	1103 (29.5)	101 (2.7)	1597 (42.8)

+: Positive; -: Negative; IGRA: IFN-γ release assay; TST: Tuberculin skin test.

Table 2. Characteristics and results of studies on serial testing using IFN-γ release assay.

Study (year)	Country	IGRA	Months between IGRA	n^†	Initially positive IGRA, n (%)	IGRA reversions, n (%)	IGRA conversions n/N^‡ (%)	Ref.
Studies in countries with a high incidence of TB (≥100 per 100,000 population and year)
Pai et al. (2006)	India	QFT	18	216	38 (17.6)	9 (23.7)	18/178 (10.1)	[79]
Studies in countries with an intermediate incidence of TB (≥ 20 and <100 per 100,000 population and year)
Chee et al. (2009)	Singapore	T-SPOT	12	182	–	–	9/182 (4.9)	[80]
Lee et al. (2009)	South Korea	QFT	12	169	23 (13.6)	–	21/146 (14.4)	[81]
Yoshiyama et al. (2009)	Japan	QFT	24–48	311	30 (8.7)	13 (43.3)	6/286 (2.1)	[82]
Torres Costa et al. (2011)	Portugal	QFT	12	670	208 (31.0)	46 (22.1)	51/462 (11.0)	[14]
Rafiza et al. (2012)	Malaysia	QFT	12	769	64 (8.3)	19 (29.7)	69/704 (9.8)	[83]
Studies in countries with a low incidence of TB (<20 per 100,000 population and year)
Pollock et al. (2008)	USA	QFT	1–7	43	–	–	2/43 (4.7)	[91]
Zwerling et al. (2009)	Canada	QFT	12	60	4 (6.7)	2 (50.0)	4/56 (7.1)	[84]
Belknap et al. (2010)	USA	QFT	6	1281	50 (3.9)	20 (40.0)	44/1169 (3.8)	[92]
		T-SPOT			68 (5.3)	36 (52.9)	44/1117 (3.9)
Gandra et al. (2010)	USA	QFT	1	164	–	95 (57.9)	–	[74]
Ringshausen et al. (2010)	Germany	QFT	5	182	18 (9.9)	6 (33.3)	3/162 (1.9)	[87]
Schablon et al. (2010)	Germany	QFT	12	287	42 (14.6)	13 (31.0)	15/245 (6.1)	[88]
Schablon et al. (2011)	Germany	QFT	12	154	2 (1.5)	1 (50.0)	1/152 (0.7)	[89]
Fong et al. (2012)	USA	QFT	12	1857	–	–	52/1857 (2.8)	[90]
Joshi et al. (2012)	USA	QFT	1	45	–	18 (40.0)	–	[86]

^†Number of subjects with IGRA test repeats in the course of the study.

^‡Number of subjects with initially negative IGRA.

IGRA: IFN-γ release assay; QFT: QuantiFERON-TB Gold or Gold In-Tube; TB: Tuberculosis; T-SPOT: T-SPOT.TB.

Table 3. Rates of divergent IFN-γ release assay results depending on the IFN-γ concentration in the first IFN-γ release assay (n = 1787).

First IGRA	Second IGRA-positive		Total
	n	Row (%)	n	Col (%)
<0.1 IU/ml	76	6.9	1095	80.8
0.1 to <0.2 IU/ml	30	20.5	146	10.8
0.2 to <0.35 IU/ml	43	37.7	114	8.4
Negative	149	11.0	1355	100/75.8^†
First IGRA	Second IGRA-negative		Total
	n	Row (%)	n	Col (%)
0.35 to <0.7 IU/ml	63	52.5	120	27.8
0.7 to <1.0 IU/ml	9	20.5	44	10.2
1.0 to <2.0 IU/ml	18	29.5	61	14.1
2.0 to <3.0 IU/ml	9	20.0	45	10.4
3.0 to <4.0 IU/ml	8	27.6	29	6.7
4.0 to <5.0 IU/ml	3	14.3	21	4.9
≥5.0 IU/ml	9	8.0	112	25.9
Positive	119	27.5	432	100/24.2^†

^†Of the total sample (n = 1787).

Key issues

• • Healthcare workers (HCW) are at increased risk for TB, and early detection of pulmonary TB in HCW in countries with high and intermediate TB burdens will have an important impact on TB incidence in the respective country.

• • IFN-γ release assays (IGRA) are more specific than the tuberculin skin test (TST) as they are in vitro tests and use antigens not shared by the strains used for BCG vaccination. Therefore, using IGRA in TB screenings of HCW will reduce the need for medical evaluation (x-ray) and preventive treatment.

• • IGRA and TST do not allow for the distinction of remote or recent latent TB infection. Therefore, information on recent exposure to M. tuberculosis is crucial for selecting HCW for TB screening.

• • Since IGRA are expensive and require a high tech laboratory, TST will remain an important screening tool in countries with limited resources and infrastructure.

• • IGRA results should be reported qualitatively and quantitatively because the conversion and reversion rates of a latter IGRA depend on the quantitative results of the former IGRA.

• • The reversion rate of IGRA in serial testing is higher than expected. A borderline zone needs to be defined for both IGRA, in order to reduce variability by chance alone. For T-SPOT.TB, the borderline zone of 5–7 spot-forming cells has already been endorsed by CDC and European Centre for Disease Prevention and Control but needs to be evaluated. For QuantiFERON^®-TB Gold In-Tube, a borderline zone from 0.2 to 0.7 IU/ml is proposed in this paper.

• • Better data on disease progression depending on IGRA results are needed. In particular, data on disease progression depending on conversion or reversion in IGRA are also desirable.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.