Plasma biomarkers in the prediction and diagnosis of tuberculosis-associated immune reconstitution inflammatory syndrome

Abstract

Tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) is the most common form of immune restoration disease (IRD) complicating early antiretroviral therapy (ART) in patients from resource-limited countries with HIV infection [1]. Increasing attention is being paid to the definition of plasma biomarkers that might assist in the prediction and diagnosis of TB-IRIS, as well as other forms of IRD [2,3]. The findings of the study by Haddow et al.[4] are, therefore, an important contribution to the literature on IRD but might have been more informative if considered in the context of data from other studies of plasma biomarkers. We described a prospective study of the immunopathogenesis of TB-IRIS and ART-associated TB (ART-TB; equivalent to ‘unmasking’ TB-IRIS [1] reported by Haddow et al.) undertaken using plasma from whole blood interferon (IFN)-gamma release assays (IGRAs) [5,6]. The advantage of using whole blood IGRAs in this context is that plasma from antigen-stimulated and antigen-unstimulated tubes can be analyzed simultaneously. However, this advantage must be balanced against the uncertainty of knowing whether biomarkers in plasma from unstimulated tubes reflect the in-vivo state or the production or consumption of biomarkers during the 24-h culture. The demonstration by Haddow et al.[4] that plasma CCL2 (monocyte chemotactic protein-1) levels are low in patients with ‘paradoxical’ TB-IRIS is supported by our finding of low CCL2 levels in unstimulated IGRA plasma from patients with TB-IRIS [6]. In addition, we demonstrated low levels of CCL2 before ART. In contrast to the findings of Haddow et al.[4], we found similar levels of interleukin (IL)-10 in unstimulated IGRA plasma from TB-IRIS cases and controls [6]. However, low production of IL-10 by antigen-stimulated peripheral blood mononuclear cells from patients with IRD associated with infection by nontuberculous mycobacteria has been demonstrated previously [7,8]. In the study by Seddiki et al.[8], low IL-10 production was associated with increased numbers of circulating regulatory T (Treg) cells, so the authors suggested that low IL-10 production might reflect dysfunctional Treg cells. As CCL2 and IL-10 are both produced by monocytes and macrophages, an alternative explanation is that CCL2 and IL-10 deficiency reflect monocyte/macrophage dysfunction. We agree with Haddow et al.[4] that there may be differences in the immunopathogenesis of TB-IRIS and ART-TB [5,6]. They reported that pre-ART plasma IFN-gamma levels were approximately 10-fold higher in patients who subsequently developed ‘unmasking’ TB IRIS compared with controls, but that there was no difference in pre-ART IFN-gamma levels when ‘paradoxical’ TB-IRIS cases were compared with controls [4]. We reported that pre-ART IFN-gamma responses to both region of difference 1 and purified protein derivative antigens of Mycobacterium tuberculosis in whole blood IGRAs, corrected for CD4+ T cell counts, were increased in patients who developed ART-TB and had strong performance characteristics in the prediction of ART-TB by receiver operator curve (ROC) analysis [5]. We also reported that levels of CXCL10 (IP-10) in unstimulated IGRA plasmas were higher in TB-IRIS patients than ART-TB patients during 24 weeks of ART and, furthermore, that pre-ART levels of CXCL10 in combination with CCL2 and IL-18 levels predicted the occurrence of TB-IRIS by ROC analysis. In contrast, Haddow et al.[4] did not demonstrate a difference in plasma levels of CXCL10 levels between ‘paradoxical’ TB-IRIS or ‘unmasking’ TB-IRIS cases and their controls [4], perhaps suggesting that in our study, these biomarkers were produced in cell culture. Unlike previous studies of various forms of IRD [2], Haddow et al.[4] did not demonstrate increased plasma levels of IL-6 in patients with ‘paradoxical’ TB-IRIS. A possible explanation for this comprises differences in the assay method. In most previous studies of IRD, plasma IL-6 levels have been assayed using enzyme-linked immunosorbent assays (ELISAs), whereas Haddow et al.[4] used a multiplex bead array assay (MBAA). Plasma IL-6 levels in HIV patients may differ significantly when assayed by ELISA or MBAA [9]. It is now clear that assays of selected biomarkers can assist in both the prediction and diagnosis of TB-IRIS and ART-TB. Large-scale studies are now needed to delineate the value of these assays as diagnostic tests. In doing this, consideration should be given to the use of both plasma and IGRA plasma from antigen-stimulated and antigen-unstimulated tubes and also to the method used to assay cytokines and chemokines. Acknowledgements Conflicts of interest There are no conflicts of interest.