Natural Killer cell activation distinguishes M. tuberculosis-mediated Immune reconstitution syndrome (IRIS) from chronic HIV and HIV-MTB co-infection

Introduction

Approaches to expand treatment of HIV infection in sub-Saharan Africa have been complicated by the high prevalence of Mycobacterium tuberculosis (MTB) infection ^{1, 2} According to WHO, 1.37 million people worldwide have HIV/tuberculosis, accounting for 456,000 deaths every year, almost 25% of all HIV-related deaths. In addition to the challenge of combined drug toxicity, simultaneous treatment of HIV and MTB co-infection often results in immune reconstitution syndrome (IRIS), an unanticipated aggravation of tuberculosis symptoms in a previously controlled subject (paradoxical IRIS) or in unmasking of a previously latent infection (unmasking IRIS or ART-associated TB) ^{3, 4}. Recent studies support initiating HIV treatment in TB-infected persons 2 weeks before initiation of anti-TB treatment (CAMELIA study ⁵), and in November 2009, WHO recommended treating patients with antiretrovirals “as soon as possible” within the first 2 months following the start of tuberculosis treatment ¹. As the risk of IRIS has been shown to be increased when ART is initiated together or shortly after anti-MTB treatment, these recommendations are likely to result in an even higher incidence of TB-IRIS in sub-Saharan Africa.

The immunopathogenesis of IRIS remains unclear; in addition to the wide variety of conditions associated with IRIS (reviewed in ⁶), growing evidence indicates that even in M. tuberculosis infection the pathogenesis of paradoxical and unmasking IRIS may differ significantly: Elliot et al. ⁷ showed that IFN-γ responses to RD1 antigen were greater in unmasking IRIS, whereas skin test response to PPD was greater in paradoxical IRIS. Haddow et al. ⁸ recently reported that, while CRP serum levels are typically elevated in all forms of IRIS, levels of MCP-1 and IL-10 are reduced in paradoxical TB-IRIS, whereas cases of unmasking TB-IRIS have higher pre-ART IFN-γ, as well as higher TNF-α at the time of event. Finally, ex-vivo stimulation with mycobacterial antigens was shown to induce higher levels of IL-18 and CXCL10 and lower CCL2 in subjects with paradoxical TB-IRIS, whereas in unmasking TB-IRIS only IL-18 was markedly elevated ⁹, suggesting a prevalence of adaptive responses in unmasking TB-IRIS. Subjects with active MTB infection have been shown to have increased T-cell activation (CD3⁺/HLA-DR⁺ cells), as well as decreased production of IL-2, IL-4, IL-5 and IL-10 IFN-γ and TNF-α ¹⁰ and higher serum IL-6 and IP-10 ¹¹. Importantly, serum TNF-α is increased in HIV/MTB co-infection ¹², which also results in higher serum levels of IL-12 and neopterin as compared to MTB alone ¹³. Based on this knowledge, a number of studies have focused on the detection of serum factors associates with IRIS presentation, demonstrating increased serum levels of IFNγ IL-6 and TNFα in association with IRIS (^{14, 15}, also reviewed in ¹⁶).

A number of studies has assessed the contribution various cellular subsets to the pathogenesis of IRIS: increased antigen-specific responses in CD4⁺ Th1 cells, as well as KIR⁻ TCRγδ⁺ T cells ¹⁷ and production of type 1 and pro-inflammatory cytokines ¹⁸ have been reported in paradoxical TB-IRIS. Evidence for a pathogenetic role of regulatory T cells remains inconclusive ^{19, 20}. In addition to triggering adaptive responses via TCR αβ or γδ²¹, mycobacteria interact with cell surface receptors (e.g. in macrophages TLR2 ^{22, 23}, mannose receptor and complement receptor²⁴), increasing production of expression of nitric oxide ²⁵ and TNF-α ²⁶. Markers associated with monocyte/macrophage function have been associated with IRIS presentation (e.g., IL-6 and related CRP production ²⁷) indicating that acute activation of innate effectors may be a distinct feature of IRIS ²⁸.

In this manuscript we present immune correlated of unmasking TB-IRIS, identifying innate immune parameters uniquely associated which may help diagnose and differentiate unmasking IRIS in HIV-TB co-infected control subjects.

Methods

Results

Discussion

By comparing subjects with acute, unmasking TB-related IRIS to both HIV-infected and MTB/HIV-co-infected control subjects, we identified for the first time a number of innate response parameters that are uniquely associated with uTB-IRIS.

NK cell activation, as determined by surface expression of CD69 alone or with HLA-DR is significantly higher in uTB-IRIS. CD56⁺ NK cells have been reported to be involved in the host response to MTB, as indicated by the activation (HLA-DR and CD69 expression) and expression of chemokine receptors (including CXCR2, CCR1, 2 and 7 in CD56^bright and CCR5 on CD56^dim) on pleural NK cells from TB subjects, as well as their ability to migrate in response to IL-8, IP-10, MCP-1 and SLC ³⁵. It is possible that the observed NK cell activation may result in part from their interaction with mycobacterial products (e.g. via TLR2, ³⁶), as well as from enhanced chemokines production by infected/bystander macrophages at the sites of infection. Our observation that IL-8 is significantly increased is uTB-IRIS further supports this model, suggesting that IL-8 might be involved in the activation and chemotaxis of NK cells expressing CXCR1 ³⁷ or CXCR2 ³⁵.

Importantly, we identified a profile (CD69⁺ NK and lymphocyte count) that emphasizes the contribution of innate variables in classifying uTB-IRIS subjects in our cohort: further validation in an independent cohort, is needed to elucidate the generalizability of this finding.

We only observed a moderate raise in IL-6 in uTB-IRIS, which was significant only before adjustment for multiple testing. this is in contrast with prior studies in paradoxical TB-IRIS ¹⁵ showing increased IL-6: this discrepancy may depend on the experimental design (cell culture supernatants of MTB-stimulated PBMC rather than serum/plasma) and IRIS presentation, as the pathogenesis of paradoxical TB IRIS is thought to be at least in part separate from that of unmasking TB-IRIS ^3,7. Further studies assessing the cellular subsets source of the cytokines associated with uTB-IRIS may help further define these differences.

While we observed an increased T-cell activation (HLA-DR/CD69 expression, plasma IL-2 receptor) in MTB co-infected subjects, uTB-IRIS subjects had levels of T-cell activation similar to HIV/MTB co-infected subjects, suggesting that T cell activation is driven by MTB infection, irrespective of clinical presentation.

In combination with the increased activation of NK cells in uTB-IRIS, these observations suggest that, in addition to the proposed role for adaptive immune responses ⁷, which have also been shown to underlie paradoxical TB-IRIS ^{17, 18}, innate immune effectors play a role in the pathogenesis of unmasking TB-related IRIS, by contributing to the Th1 response to M. Tuberculosis (NK cells) or producing pro-inflammatory mediators, as suggested by prior in vitro observations (reviewed in ³⁸). Prior demonstration that NK cell rapidly achieve partial functional recovery upon ART initiation ³⁹ further support their potential contribution to the pathogenesis of unmasking TB-IRIS.

While we could confirm the association of increased acute phase reactants (APRs) with uTB-IRIS in our cohort by assessing between-group differences, our results indicate that APRs are not the most useful variable to classify unmasking TB-IRIS, as they are outranked by NK cell activation and other hematological parameters (lymphocyte and neutrophil count) in our ctree analysis. It is possible that the chronic activation underlying both advanced HIV and MTB infection may result in some degree of elevation APRs, thus obfuscating their association with IRIS. Given the relatively small size of our cohort, larger studies will be required to confirm this result, particularly in reference to the cut-off values identified here, and to assess the usefulness of these markers in diagnosing other forms of IRIS, particularly in reference to different pathogens.

We have identified a number of immunological parameters associated with MTB infection, independent of IRIS presentation; among these are decreased lymphocyte and increased neutrophil counts. While prior studies in MTB-infected subjects in the same geographical area did not show alteration of these subsets ⁴⁰, lymphopenia is a relatively common feature of MTB infection ^{41, 42} it is likely that the advanced HIV co-infection, as well as ongoing anti-MTB treatment, may have contributed to the profile of our subjects. CD4⁺ T cell counts were also lower in MTB co-infected subjects; while MTB-related CD4⁺ T cell depletion has been reported^{41, 43, 44}.

The study has a number of limitations, the first being the relatively small size of the cohort: while our confidence in the significance of the tests conducted is based on a conservative approach to multiple testing adjustment, it is possible that other parameters (e.g. IL-6 and IP-10 levels) would have been found to be significantly different using a larger cohort.

Secondly, our work focused exclusively on subjects with the unmasking form of TB-IRIS: further studies will be required to assess whether or not our findings apply also to other presentations (i.e. paradoxical IRIS) or pathogens.

It is also important to consider that, due to the fact that anti-TB treatment is initiated before ART in the vast majority of HIV-infected subjects, in order to match time on ART our MTB control group included only subjects undergoing anti-mycobacterial treatment, which limits the interpretation of the results due to the regressing MTB infection and the potential effects of anti-MTB drugs.

Another limitation is that our study assessed uTB-IRIS subjects at presentation, and lacked pre-treatment samples and extensive follow-up data, due to the fact that a) most uTB-IRIS subject were referred as hospital inpatients, and b) most subjects in the MTB-HIV co-infection control group were referred from TB clinics in the geographical catchment of the Themba Lethu clinic Partly due to acuteness of the clinical presentation, blood drawing was limited in 5 of uTB-IRIS group subjects, limiting the amount of flow cytometry data available for analysis.

Finally, the cross-sectional nature of the cohort did not address whether pre--treatment or early “on-treatment” elevation of cell activation may be predictive of IRIS onset; further prospective studies will be required to determine this point. A prospective study enrolling subject at the time of HIV diagnosis and with extended follow-up will therefore be required to confirm and extend these observations.

Conclusions

We sought to identify immune parameters that uniquely characterize MTB-related IRIS by studying a cohort of subjects with unmasking IRIS presentation. Our results indicate that elevated NK cell activation, together with increased plasma levels of acute phase reactants and IL-8, is uniquely associated with unmasking TB-related IRIS, supporting the hypothesis that innate immune effectors participate in the pathogenesis of IRIS. Based on our conditional inference tree classification we propose that NK activation and lymphocyte counts may be clinically useful to support a diagnosis of MTB-related IRIS.

Supplementary Material