Performance of whole-blood interferon-gamma release assay in patients admitted to the emergency department with pulmonary infiltrates

The whole-blood interferon-gamma release assay (IGRA) is a new type of rapid, immune-based blood test that has significantly aided the diagnosis of tuberculosis (TB) infection [1–3]. However, since the test measures an immunological response, its performance is likely to be affected by differences in mycobacterial activity and factors that suppress the host's immune system, in particular cellular immunity. Indeed, we often encounter patients whose IGRA has returned an indeterminate result. The high rate of indeterminate results has led to concern about using new IGRAs, especially when dealing with patients whose cellular immunity is likely to be impaired [4, 5]. However, to date, few studies have sought to identify the factors that are associated with the indeterminate results of IGRA test [6–10].

To address these issues, this study was conducted to assess whether the performance of IGRA test would be affected by the manner of presentation to the hospital. The emergency departments (ED) of Korean hospital often receive patients with pulmonary infiltrates associated with acute derangement. Since Korea is an area with an intermediate burden of active tuberculosis [11], it is frequently challenging for clinicians to distinguish between TB and non-TB. In such clinical situation, IGRA test for the diagnosis of TB infection may be helpful as a supplemental tool for the diagnostic exclusion of active TB [12]. Thus, we investigated the performance of IGRA test in these ED inpatients and compared it with its performance in outpatients with pulmonary TB. In addition, we sought to identify the factors that influence the performance of IGRA test in both inpatients and outpatients with active pulmonary TB.

The main findings of this study were: 1) Inpatients admitted to the ED with either pulmonary TB or non-TB diseases were significantly more likely to have indeterminate QFT results than outpatients with pulmonary TB; 2) The sensitivity of QFT test for the diagnosis of pulmonary TB showed significant difference between inpatients (58.0%) and outpatients (82.8%); 3) Independent factors that were associated with indeterminate QFT results in inpatients were lymphocytopenia, lower albumin concentrations, and higher CRP levels; 4) The inpatients with TB differed from the outpatients in terms of the risk factors that were associated with non-positive QFT results.

Previous large studies have demonstrated that the QFT, when used routinely in hospitals to diagnose TB infection, often yields indeterminate results [6, 7, 9, 10]. However, these studies only analyzed cohorts with a high proportion of immunosuppressed individuals [6, 7], children [9], or preselected patients with suspected TB and latent TB infection [10]. In addition, the proportion of patients with active TB in these studies was small. Moreover, these studies did not discriminate between inpatients and outpatients. Therefore, the results of these studies may not adequately reflect the performance of the QFT in patients who present at an ED with pulmonary infiltrates. Our study demonstrates the performance of the QFT in these inpatients. Our results suggest that the performance of the QFT in an ED can vary depending on the disease that is involved and may be significantly worse than its performance in an OPD setting. This is the first report of the performance of the QFT when it is applied routinely in an ED to differentially diagnose patients who present with acute respiratory symptoms and pulmonary infiltrates.

The QFT sensitivity of all of the TB patients, including inpatients and outpatients, was 70.7%, which is within the range of QFT sensitivities of prior reports, including those obtained by meta-analysis [14]. Among prior reports, Mazurek et al. [15] reported QFT-G sensitivity of 69.6% in patients with culture-confirmed pulmonary TB, which is very similar to the sensitivity in the same population in our study. However, our data showed a significant difference of QFT sensitivity between inpatients and outpatients with culture-confirmed pulmonary TB. In general, inpatients with TB admitted to the ED are expected to have a more severe form of TB than outpatients with TB. Thus, our data suggests that the sensitivity of the QFT test may be significantly different depending on disease severity in the study population.

Previous studies have shown that host factors that are significantly associated with indeterminate QFT results are immunosuppressive therapies, extremes of age, lymphocytopenia, hypoalbuminemia, and HIV-infection with lower CD4 lymphocyte counts [6, 7, 9, 10, 16]. Other causes of indeterminate QFT results include pre-analytical errors such as the storage of tubes outside the recommended temperature range, the over- or under-filling of tubes with blood, insufficient mixing, and incomplete washing of the enzyme-linked immunosorbent assay plate [13]. However, while recent technical modifications of the QFT, such as vortexing or immediate incubation, have reduced the rate of indeterminate QFT-GIT results [17, 18], they are unlikely to be able to overcome the effect of host factors that lead to low cell-mediated immune responses. Indeed, the high rate of indeterminate results observed in our ED setting did not seem due to errors in the testing process, as the manufacturer's instructions were followed strictly and the rate of indeterminate QFT results in outpatients with pulmonary TB (whose samples were handled in an identical manner to those of inpatients) was very low. Therefore, we believe that patient-specific factors rather than technical problems are the main reasons for the high rate of indeterminate QFT results in our inpatients. Further analysis revealed that those factors are lymphocytopenia, hypoalbuminemia, and increased CRP levels. This list differs somewhat from the list of host factors mentioned above that were identified by previous studies that were based on routine hospital use [6, 7, 10].

The association of lymphocytopenia with indeterminate QFT results is not surprising as it is well known that lymphocytopenia can decrease the production of IFN-γ. Moreover, as suggested by a previous study [10], the association of indeterminate QFT results with hypoalbuminemia may reflect a poor nutritional status that compromises the immune system. However, our study is the first to show an association between higher CRP levels and indeterminate QFT results. Since CRP is a major positive acute phase reactant [19], higher CRP levels are likely to reflect more intensive inflammation, which in turn suggests that patients with indeterminate QFT results have higher levels of inflammation than those with determinate QFT results. Notably, since albumin is a major negative acute phase reactant [19], a higher inflammatory status result may explain the association between hypoalbuminemia and poor QFT results better than the malnutrition hypothesis. Moreover, lymphocytopenia is an extremely common feature of the acute response to stress in acutely ill patients, regardless of the nature of the underlying illness [20]. Although this may reflect the small numbers of patients with immunosuppressive therapies or old age in our study, it is also possible that these underlying conditions play less important roles in acute status than in chronic status. In addition, since the effects of underlying conditions, which can vary between individuals and during the clinical course, would be ultimately reflected in lymphocyte counts and acute phase reactants, these parameters may be good objective markers that indicate the strength of cellular immunity and the risk of poor QFT performance.

Hospitalization-related stress is known to promote the systemic secretion of glucocorticoids and catecholamines, which, in turn, influence immune responses [21]. In particular, glucocorticoids are known to suppress the production of TNF-α, IFN-γ, and IL-2 in animals and humans both in vitro and in vivo [22, 23]. In addition, catecholamines inhibit the development of Th1-type cells, thereby inhibiting IFN-γ production [24]. Thus, the fact that inpatients with pulmonary TB have higher rates of indeterminate QFT result than outpatients may partly reflect the derangement of immune responses caused by hospitalization at the ED.

Lymphocytopenia and hypoalbuminemia were found to be risk factors of non-positive QFT results in inpatients with active TB. This is consistent with our observation that clinically significant inflammation associated with indeterminate QFT results in inpatients in general. However, acute systemic inflammation did not seem to be strongly associated with non-positive QFT results in OPD-based TB patients. Indeed, the outpatients with non-positive QFT results did not differ significantly from the other outpatients in terms of lymphocyte counts and the levels of major acute phase reactants such as albumin and CRP. However, the outpatients with non-positive QFT results did have significantly higher ESR values than the other outpatients. Since CRP concentrations change rapidly whereas the ESR changes relatively slowly [19], it may be that the OPD-based TB patients with non-positive QFT results were at a more chronic and stable stage of the inflammatory process (namely, a stage where CRP responses were no longer being elicited).

Radiographically more severe disease was also significantly associated with non-positive QFT results in the outpatients. This is consistent with the suggestion that the limited sensitivity of IGRAs in patients with active TB may relate to the immunological incapacity of the host to contain mycobacterial replication [25]. Moreover, IFN-γ secretion has been shown to correlate inversely with disease severity in patients with TB [25–27]. These observations may explain why outpatients with non-positive QFT results had extensive disease more frequently than the other outpatients. In a certain aspect, radiographically more extensive diseases may be in line with higher ESR values shown in TB outpatients with non-positive QFT results.

Taken together, these observations suggest that the level of immune competence varies among patients with mycobacterial diseases. The different clinical presentations of TB appear to be associated with the degree of alteration of the immune response against the mycobacteria. T cell dysfunction by alterations in the expression of several T cell signal-transduction proteins was reported in patients with pulmonary TB [28]. In addition, severe TB in a macaque model induces unbalanced up-regulation of immune gene networks, causing overexpression of genes encoding inflammatory cytokines and chemokines without proportional increases in antigen-specific cellular responses [29]. Thus, TB inpatients, who are expected to have more severe disease than OPD-based TB patients, are likely to have more intense inflammation and more severe T cell dysfunction, perhaps explaining why more intense inflammation is associated with indeterminate result of the QFT test.

The main limitation of the present study is the fact that the study population was tested by two different QFTs (QFT-G and QFT-GIT) because the QFT-G was replaced by the newer generation test QFT-GIT during the study period. However, a recent meta-analysis that also included the indeterminate result rates has found that QFT-G and QFT-GIT may be similarly sensitive [14]. Thus, this factor is not likely to have significantly affected our results. Another potential limitation is the fact that only a small number of patients received immunosuppressive treatment. Thus, the power of our study to detect the impact of immunosuppressive therapy on QFT performance was probably rather weak. In addition, the numbers of inpatients with TB and outpatients after they were subgrouped according to their QFT results were small. As well, our data could not directly determine the mechanism by which more intense inflammation, as evidenced by higher CRP levels and lower albumin concentrations, is associated with indeterminate results in the QFT test. These relationships will require further study. Finally, our observations cannot be generalized to all populations with pulmonary infiltrates because most of non-TB population were consisted of patients with pneumonia (91%).

Our findings indicate that, unlike QFT of outpatients with pulmonary TB, QFT performed in the ED for inpatients with pulmonary infiltrate frequently return indeterminate results. In addition, our study showed that inpatients and outpatients with pulmonary TB may differ in terms of the risk factors that are associated with non-positive QFT results.