QuantiFERON-TB performance enhanced by novel Mycobacterium tuberculosis-specific antigens

To the Editor:

One of the most significant developments in the diagnosis of tuberculosis (TB) infection has been the introduction of whole-blood based interferon-γ release assays (IGRAs) [1–3]. IGRAs, commercially available as QuantiFERON-TB Gold In-Tube test (QFT) (QIAGEN, Germantown, MD, USA) and T-SPOT.TB (Oxford Immunotec Ltd, Abingdon, UK), are based on the detection of a T-cell immune response towards RD1 antigens (ESAT-6 and CFP-10), with the addition of the TB7.7 antigen for the QFT only [1]. IGRAs are endowed with great specificity, as the antigens used are almost exclusively expressed by the Mycobacterium tuberculosis (MTB) complex, but not Mycobacterium bovis bacille Calmette–Guerin (BCG) [1–3]. However, the diagnostic sensitivity of IGRAs can be improved (75–85% in HIV-negative active TB patients) especially in countries with a high TB burden [3].

To improve the performance of the IGRAs, addition of MTB antigens TB7.7, Rv3425 and EsxV to the current antigen cocktail has already been attempted [4, 5]. Following a similar strategy, we evaluated the addition of six novel peptides to the QFT antigen cocktail to improve the diagnostic performance in terms of accuracy and dynamic range of interferon (IFN)-γ level in active TB patients.

To this end, a multicentre study was performed under different clinical settings. Study group characteristics are reported in table 1. Active TB patients (n=85) were enrolled in three different clinical centres at the time of admission for being suspected of having TB and diagnosed according to Centers for Disease Control and Prevention/American Thoracic Society criteria [6]. Control subjects (n=290) without any risk factors for latent TB infection were enrolled in two TB low-risk countries (Australia and Italy) and in a TB medium-risk country (Bulgaria). The study was approved by the institutional review boards of all participating clinical institutions. Informed consent was obtained from each study subject before blood sample collection.

The novel peptides were derived from MTB-specific genes overexpressed in an in vitro model of macrophage infection and validated ex vivo in the lung of active TB patients [7–9]. Four MTB proteins (Rv0724A, Rv1251c, Rv1478 and Rv3479) were selected for further investigation upon expression profiles, antigenicity and cross-reactivity. Multi-epitope, human leukocyte antigen (HLA) class II promiscuous peptides were identified on the protein sequences by peptide binding motif analysis [8]. Six peptides (Rv0724A: GEIIFISGRLNGAA; Rv1251c: ELMARAAVLGSAH and AVIVRSELLTQYL; Rv1478: TAWITAVVPGLMV; Rv3479: RPVRRVLLFVVPSSGPAP and GSVRQLPSVLKPPLITLRTLTLSG) were designed, synthesised by Fmoc chemistry (American Peptide Company, Inc., Sunnyvale, CA, USA) and obtained >90% purity.

The QFT assay was performed according to the manufacturer's instructions. In parallel, a QFT plus the six novel peptides (hereafter referred to as QFT-Six) was performed by adding to one QFT antigen tube the pooled novel peptides at a final concentration of 1 μg·mL⁻¹ each. The QFT-Six tube was then treated the same as any of the QFT tube assays. IFN-γ levels were determined using the QFT ELISA and results were interpreted according to the manufacturer's instructions.

The intracellular expression of IFN-γ in CD4⁺ and CD8⁺ T-lymphocyte subsets upon stimulation with the pool of the six novel peptides was evaluated using flow cytometry as previously described [10]. Data are expressed as median (interquartile range) or percentages. Comparison between groups was made using a paired or unpaired nonparametric test or Fisher's test, as appropriate. Variation in the QFT MTB antigen test after addition of the novel peptides was calculated by comparison with the normal QFT MTB antigen test. Deviation from the normal coefficient of variation of the QFT was determined according to the QFT reproducibility data [11]. All analyses were performed using the GraphPad Prism 5.0 software (Graphpad, San Diego, CA, USA).

Table 1 shows the results obtained in the different study populations with the QFT and the QFT-Six tests. As expected, the QFT showed high sensitivity (range: HIV-negative 75–87%; HIV-infected 67–100%) and specificity (range: 96–100%) in BCG-vaccinated and nonvaccinated control populations. The addition of the six novel peptide antigens to the QFT did not alter the specificity of the test in either BCG-vaccinated or normal control populations (table 1), although the antigens were derived from MTB proteins present in the BCG strain genomes. The addition of the six novel peptide antigens to the QFT did not determine a significant increase in test sensitivity, probably due to the already high performance of QFT in this study population [3]. However, with the limitation of the small numbers, the addition of the six novel peptides demonstrated a conversion to a positive result in three (30%) out of 10 HIV-negative and one (25%) out of four HIV-infected active TB patients with a negative result in QFT alone.

The addition of the six novel antigen peptides to QFT determined overall a significant increase of IFN-γ release with respect to the QFT alone, both in HIV-negative (QFT: 3.88 (0.88–10) IFN-γ IU·mL⁻¹; QFT-Six: 5.01 (1.20–10) IFN-γ IU·mL⁻¹; p<0.01 Wilcoxon paired test) and HIV-infected (QFT: 0.82 (0.22–2.10) IFN-γ IU·mL⁻¹; QFT-Six: 0.99 (0.39–2.26) IFN-γ IU·mL⁻¹; p<0.03 Wilcoxon paired test) active TB patients.

This observation was confirmed by the analysis of the coefficient of variation of the IFN-γ release after addition of the six novel peptide antigens to QFT. Considering the IFN-γ release inter-test reproducibility of QFT (14% [11]), a significant number of both HIV-negative (24 (51%) out of 47) and HIV-infected (five (56%) out of nine) active TB patients presented an increase of IFN-γ release greater than the normal range of variability of the IFN-γ release for QFT (p<0.001, all comparisons).

To corroborate this data, the contribution of the CD4⁺ and CD8⁺ antigen-specific responses to the six novel peptides in the pool were determined by flow cytometry in a subgroup of 12 active TB patients and nine QFT-negative controls. Active TB patients present, compared with controls, a significantly higher frequency of peptide-specific IFN-γ⁺/CD69⁺/CD4⁺ (active TB: 0.08% (0.02–0.14%); controls: 0.01% (0–0.02%); p<0.02 Mann–Whitney test) and IFN-γ⁺/CD69⁺/CD8⁺ (active TB: 0.07% (0.02–0.12%); controls: 0.01% (0–0.02%); p<0.02) T-lymphocytes. Therefore, although designed to present HLA class II binding capabilities, the novel peptides also contain HLA class I epitopes, as assessed by Immune Epitope Database analysis [12], supporting both CD4⁺ and CD8⁺ responses.

The development of rapid and sensitive diagnostic methods for active TB, latent TB infection and recent MTB infection is a key aspect of the TB control strategy [1–3]. The addition of the novel peptides to the QFT resulted in a significant increase in the IFN-γ release in active TB, without altering specificity, and also in the BCG-vaccinated control population. This is consistent with previous data in the literature indicating that the addition of other non-RD1 proteins to QFT can result in an increase of the test performance [4, 5], as also observed in the changes implemented in the QFT test versions [1].

The significant increase of IFN-γ release after adding the novel peptide antigens to QFT in HIV-infected TB patients is of particular importance as poorer performance of IGRAs in immunocompromised patients has been reported [2]. This increase is likely to be due to the contribution of the CD8⁺ T-cell response directed against the novel peptides, better preserved in HIV-infected subjects than CD4⁺ responses [13, 14].

Furthermore, the significant increase in IFN-γ release in active TB, and thus dynamic range, after addition of the novel peptide antigens to the QFT assay might help in interpreting mild fluctuations in IFN-γ responses observed during serial testing with the QFT [15] and in reducing the variability of IGRA response often obtained under different settings and/or host backgrounds. This is consistent with the homogeneous performance obtained across the different settings in this study with the use of QFT and the addition of the novel peptide antigens.

The addition of other antigens to QFT may be perceived to have a potential negative impact on assay specificity, particularly when these antigens are present in the genome of BCG strains. However, the use of antigens preferentially expressed by MTB during its active replication phase [7–9], as in the present study, did not alter the specificity of the QFT assay, including in BCG-vaccinated controls, supporting the utility of this antigen selection strategy.

In conclusion, although there is a need for larger confirmatory studies, in particular in HIV-infected subjects, this study suggests that the addition of novel antigens to the QFT assay can strongly help in the development of a new generation of IGRAs, circumventing some of the challenges of the current assay in certain populations and maintaining the high specificity of the current QFT assay compared with the tuberculin skin test.