Assessment of Imprecision in Gamma Interferon Release Assays for the Detection of Exposure to Mycobacterium tuberculosis

New gamma interferon (IFN-γ) release assays (IGRAs) to detect an exposure to Mycobacterium tuberculosis have recently been launched. The majority of the studies in temperate-climate countries agree that these methods have superior specificity and equal or even superior sensitivity over tuberculin skin tests (TSTs) in the diagnosis of latent tuberculosis (TB) infection (LTBI). However, reproducibility data of IGRAs are virtually missing. We assessed within-run, between-run, and total imprecision of two commercial IGRAs by testing samples from subjects with a stable state of TB infection or treated pulmonary TB, a sample from a healthy volunteer, and internal quality control samples. We calculated coefficients of variance (CV%s) to describe assays variability and compared the obtained results to the reported CV%s for other commercial immunodiagnostic methods. We illustrate an example of assay variability near the cutoff zone to demonstrate the necessity of a gray zone. Due to the strict adherence to the standard operation procedures (SOP) adopted in our laboratory, the total imprecision of enzyme-linked immunospot (ELISPOT)- and enzyme immunoassay (EIA)-based IGRAs was at a maximum CV% of 37.8% for the samples with moderate and high reactivities. Imprecision of testing samples with very low reactivity levels or nonreactive samples may, however, exceed 100%. In conclusion, despite multiple steps of the method performance, the analytical imprecision of IGRAs, which in our study design included also between-lot variability and had a component of normal biological variation, was well in accordance with the reported imprecisions of other manual immunodiagnostic tests. The recognition of the variability around the cutoff point advocates the use of a gray zone to avoid ambiguous result interpretations.Evaluation of immunometric tests for infectious diseases abides by the same rules as methods for clinical chemistry and is based on the same general principles (12). Evaluation of analytical performance includes, among other parameters, reproducibility data. As a prerequisite for CE mark registration to get a license to market the products for in vitro diagnostic use, manufacturers should provide reproducibility characteristics as a part of the overall performance data.Two new kits, namely, the T-SPOT.TB (Oxford Immunotec, Oxford, United Kingdom) (8) and QuantiFERON-TB Gold In-Tube (Cellestis Limited, Carnegie, Victoria, Australia) (www.cellestis.com/IRM/content/aust/qtfproducts_tbgoldintube_techinfo-perfparameters.html) kits, have been recently launched. The kits utilize the ability of sensitized CD8⁺ and CD4⁺ T lymphocytes to release gamma interferon (IFN-γ) when stimulated with synthetic peptides specific to Mycobacterium tuberculosis and detect exposure to Mycobacterium tuberculosis. While the first method measures the frequency of reactive lymphocytes in the peripheral blood mononuclear cell (PBMC) fraction, the latter measures the concentration of released IFN-γ into supernatants. These methods are collectively called IFN-γ release assays (IGRA). Although these were launched for diagnostics, clinically relevant information on assay reproducibility was available from only one test series (www.cellestis.com/IRM/content/aust/qtfproducts_tbgoldintube_techinfo-perfparameters.html) as of May 2009. From the literature search, we have found only a few publications that are related to this topic (4, 9, 14), whereas test sensitivity and specificity have been extensively tested and reviewed in recent meta-analyses (7, 10). The importance of the reproducibility parameter is emphasized by the demand to assess immunological conversions and reversions, in other words, a significant decrease in immunological responses that exceeds the total imprecision of the method. The clinical phenomenon of immunological reversion was reported to associate with, e.g., successful chemotherapy (2). A spontaneous reversion, which is a phenomenon that is not yet well understood, may be, indeed, a very important observation meaning pathogen clearance. However, the decrease in the response should be well documented and should clearly exceed the method''s total imprecision. Immunological conversion may mean a rise in the reactivity that is above the variation of technical and biological noise. Because the data on reproducibility are scarce, we assumed that ethical considerations may have constituted the major obstacle. Indeed, it may be difficult to obtain an ethical permission to collect blood samples consecutively from tuberculosis (TB) patients who may need urgent treatment.Both ethical considerations and unstable sample material make imprecision study of IGRAs difficult. In fact, IGRAs were the first diagnostic methods to exploit cell-mediated immunity and utilize the ex vivo activity of vital lymphocytes. For example, another recently introduced CE mark-registered test to evaluate the effect of immunosuppression on the function of lymphocytes (ImmuKnow; Cylex, Columbia, MD) provides only repeatability data (results obtained in different laboratories from the same venipuncture). The information on the between-run imprecision was not yet available in the kit instructions as of January 2009. The shortage of information reflects obvious practical problems in obtaining samples from critically ill subjects to study between-run imprecision as required.Imprecision data represent a very important parameter when the cutoff point and the width of the gray zone should be considered. Surprisingly, the interpretation guides for the results in both IGRA kit inserts do not discuss the topic of analytical uncertainty, i.e., the variation of a positive response around the cutoff point. Based on our pilot results, we have suggested earlier the use of a gray zone (11). This concept has been introduced recently in only one method (8), albeit without reference to the method imprecision.The aims of this study were (i) to provide an assessment of the total imprecisions of both IGRAs, (ii) to assess between-run imprecisions of both methods by using internal quality control (QC) samples, and (iii) to demonstrate an example from our daily routine for the need of a cautious interpretation of the result falling on a single cutoff point (per manufacturers'' instructions).