Severe Paradoxical Reaction During Treatment of Disseminated Tuberculosis in a Patient With Neutralizing Anti-IFNγ Autoantibodies

Abstract

Interferon-gamma (IFNγ) neutralizing autoantibodies are associated with disseminated nontuberculous mycobacterial infections. We report a previously healthy Thai woman with disseminated tuberculosis and high-titer IFNγ-neutralizing autoantibodies, who developed a severe inflammatory reaction during anti-tuberculosis treatment. IFNγ contributes to host control of tuberculosis but appears inessential for tuberculosis paradoxical reactions.

The interferon-gamma (IFNγ)-interleukin (IL)-12 axis is critical for control of mycobacteria as evidenced by Mendelian defects, which confer susceptibility to these organisms. Anti-IFNγ autoantibodies are associated with an adult-onset immunodeficiency largely (although not exclusively) in Southeast Asians who develop disseminated nontuberculous mycobacterial (NTM) disease and other intracellular opportunists [1, 2]. However, anti-IFNγ autoantibodies have not previously been associated with disseminated Mycobacterium tuberculosis infection in absence of other opportunistic infections, even in M. tuberculosis endemic regions.

We report a Thai woman with disseminated tuberculosis and a dramatic paradoxical inflammatory response after treatment initiation, who was found to have high-titer neutralizing anti-IFNγ autoantibodies.

CASE SUMMARY

A 55 year-old previously healthy Thai woman living in the United States since 2000 was admitted to a community hospital in November 2012 with respiratory failure and sepsis. Chest computerized tomography (CT) showed bilateral pulmonary infiltrates with right lung opacification, but no organisms were identified on bronchoscopy, including a negative acid-fast bacilli (AFB) smear and culture. She improved with vancomycin, piperacillin-tazobactam, and levofloxacin, and after resolution of sepsis and respiratory failure, was discharged home.

In January 2013, she presented with 4 weeks of neck swelling, intermittent fevers, and chills. Biopsy of cervical and axillary lymphadenopathy yielded caseating granulomata and grew pan-sensitive M. tuberculosis. Two weeks after initiation of rifampin, isoniazid, pyrazinamide, and ethambutol, she developed pleuritic chest pain and malaise with a new left lung cavity (Figure 1).

Figure 1.

A, Computed tomography series. Top left: Air-filled, thin-walled left upper lobe cavity. Top right: Bilateral cervical soft tissue densities contiguous with the subcutaneous surface. Bottom left: Numerous, predominantly sclerotic, osseous lesions within the lumbar spine and right iliac bone. Bottom right: Soft tissue abscesses in left axilla. B, Multiplex screening of plasmas from patient and 8 normal control plasmas for anticytokine autoantibodies. C, Evaluation of anti-interferon (IFN) autoantibody immunoglobulin G (IgG) subclass in patient plasma. D, Normal or patient peripheral blood mononuclear cells (PBMC) were incubated in the presence of normal or patient plasma and left unstimulated or stimulated for 15 minutes with IFNγ. Cells fixed and stained for intracellular phosphorylated STAT-1 were measured by flow cytometry, gating on CD14+ monocytes.

Eleven weeks after starting therapy, a new 4.8 × 3 cm suppurating anterior cervical lymph node appeared, followed by development of draining abscesses in the neck, axilla, and shoulder, all AFB smear and culture-negative.

During the 14th week of 4-drug therapy, new left shoulder pain led to the discovery of lytic lesions in the left humeral head with extensive soft tissue extension, as well throughout the spine, pelvis, and sacrum (Figure 1). T2 hyperintense liver lesions were presumed to be additional foci of disseminated M. tuberculosis. AFB smear and culture from a left shoulder aspirate were negative; echocardiogram showed no valvular lesions. She continued to experience night sweats and fevers along with increasing erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP).

She was referred to the National Institutes of Health (NIH) for progressive clinical disease despite pan-sensitive M. tuberculosis. Human immunodeficiency virus (HIV) infection was excluded by serology and antigen studies. Multiple therapeutic drainages of her axillary abscess remained AFB culture-negative, supporting a paradoxical post-treatment inflammatory response. The patient received an appropriate weight-based regimen by direct observation, and serum levels of isoniazid and rifampin were confirmed to be in the expected therapeutic range, suggesting that under-dosing or nonadherence did not explain her persistent illness. Prednisone 1 mg/kg per day was started with rapid improvement in appetite and energy. By month 3 of her 4-month prednisone taper, neck and axillary lesions were closed and dry, range of motion in her neck and shoulder had significantly improved, and she eventually returned to work. Pyrazinamide was discontinued in the 7th month of chemotherapy, and she remained on isoniazid, rifampicin, and moxifloxacin for a total of 16 months. Since then, she has had no relapse and has remained off corticosteroids.

METHODS

The patient was consented under National Institute of Allergy and Infectious Diseases (NIAID) institutional review board (IRB)-approved protocols 10-I-0195 (NCT01212003) and 93-I-0119 (NCT00001355). Blood from healthy controls was obtained from the NIH Blood Bank under IRB-approved protocols. Plasma was separated and stored at −80°C until testing. Peripheral blood mononuclear cells (PBMC) were obtained by density-gradient centrifugation [3].

Anti-cytokine Autoantibody Screening

Patient and control plasmas were screened for anticytokine autoantibodies using a multiplexed, particle-based approach [4]. Immunoglobulin isotype and IgG subclass of the detected anti-IFNγ autoantibodies were determined similarly.

Detection of IFNγ-induced Phosphorylated STAT-1 Production

Patient and normal PBMC (1 × 10⁶) were cultured in complete RPMI 1640 media (2 mM glutamine, 20 mM Hepes, 100 U/mL penicillin, 100 µg/mL streptomycin) with 10% patient or control plasma. Cultures were unstimulated or stimulated with IFNγ (1000 U/mL, Actimmune) for 15 minutes at 37°C. Monocytes were identified by CD14 (BD Pharmingen) surface staining before being fixed and permeabilized for intracellular staining with antiphosphorylated Signal Transducer and Activator of Transcription (STAT)-1 (Y701) antibody (BD Biosciences) [5]. Data were collected using FACSCalibur (BD Biosciences), analyzed using FlowJo (Treestar) and graphed with Prism6 (Graphpad).

RESULTS

The patient had a normal number of lymphocytes at NIH admission [CD3 581/µL (47%), CD4 304/µL (24%), CD8 225/µL (18%), NK 227/µL (37%)]. We detected high-titer anti-IFNγ autoantibodies primarily of IgG₃ subclass (Figure 1).

Patient but not control plasma inhibited IFNγ-induced STAT-1 phosphorylation in normal PBMC, whereas patient PBMC washed free of autologous plasma demonstrated normal IFNγ-induced STAT-1 phosphorylation, consistent with her normal expression of IFNγR1 (Figure 1).

Although net plasma interferon gamma levels were suppressed (see Supplementary Table), hyper-production of IFNγ and tumor necrosis factor α (TNFα) were observed when CD4 T cells were stimulated with tuberculin antigen in vitro (see Supplementary Figure). Systemic steroids were initiated on June 4, 2013.

DISCUSSION

IFNγ is produced largely by CD4⁺ Th1 and NK cells, among others, and is critical for macrophage activation and to control mycobacteria and intracellular fungi, as shown by IL-12/IFNγ axis disorders [6]. Autoantibodies to IFNγ have been identified predominantly in patients of Southeast origin who have disseminated nontuberculous microbacteria (NTM) infections or other intracellular pathogens, confirming the critical role of this pathway in their control [7, 8]. Rituximab has reduced anti-IFNγ titers in refractory patients, enabling clearance of mycobacterial infections [9].

However, the role of anti-IFNγ autoantibodies in tuberculosis is less clear. The 14 Southeast Asian individuals in Browne, et al found to have anti-IFNγ autoantibodies with pulmonary or disseminated tuberculosis also had other opportunistic infections, such as disseminated NTM, salmonella, fungi, and/or varicella zoster infections [1]. Increased levels of anti-IFNγ autoantibodies were previously identified in patients with moderate to severe pulmonary tuberculosis compared to levels in healthy controls, but they were in low titer and were not examined for neutralization or biological activity [10]. Even among other genetic forms of mycobacterial susceptibility, including IFNγ receptor and IL-12 receptor mutations, tuberculosis is uncommon. Although the microbial etiology of our patient's initial presentation with pneumonia and respiratory failure remains unclear, this severe event may have been a consequence of her immunodeficiency or, alternatively, precipitated production of the anti-IFNγ autoantibodies that left her susceptible to disseminated tuberculosis.

In addition to being the first described case with disseminated tuberculosis as a sole opportunistic infection in the presence of high-titer neutralizing anti-IFNγ autoantibodies, this patient offers novel insights into the paradoxical inflammatory reaction that occurs during tuberculosis treatment. Mycobacterium tuberculosis infection suppresses host immunity through down-modulation of responsiveness to activation, depressed production of inflammatory cytokines such as IFNγ and IL-2, and induction of T-cell apoptosis [11, 12]. Upon initiation of anti-mycobacterial therapy, paradoxical reactions occur in up to 20% of tuberculosis monoinfected patients, often in association with high baseline M. tuberculosis bacillary burden [13].

The mechanism of this exuberant immune reconstitution, which can emerge with the waning of the immunosuppressive features of M. tuberculosis infection, may parallel that of TB-HIV coinfected patients who start antiretroviral therapy (ART). In HIV-associated immune reconstitution inflammatory syndrome (IRIS), ART permits the rapid expansion of M. tuberculosis-specific CD4 T-cells, characterized by high levels of IFNγ and IFNγ- producing CD4 T-cells [14]. Among longitudinal studies, comparable antigen-specific IFNγ-responses have been variably found in non-IRIS patients and the true contribution of IFNγ in IRIS pathology remains under question [14, 15]. In a murine model of Mycobacterium avium-associated immune reconstitution disease, injection of wild type compared with IFNγ-/- CD4 T-cells into chronically infected T-cell deficient mice resulted in more rapid wasting and death [16]. However, even in this murine model, IRIS still occurs in the absence of IFNγ signaling, albeit with delayed kinetics. Therefore, although IFNγ appears to contribute to IRIS pathology, it is not essential for IRIS to occur in mice.

Looking beyond the IFNγ response, high antigen burden is a feature of both paradoxical reactions and IRIS and is thought to lead to excessive priming of the innate immune response, with significant contributions to IRIS pathology [17]. Our patient's cytokine levels prior to systemic steroid therapy (see Supplementary Table) suggest active innate and myeloid responses, which may have been affected by her autoantibody levels [18]. Low IL-10 levels, seen in our patient and variably seen in TB-HIV associated IRIS, suggest transiently impaired regulatory function [19]. Additionally, our patient's elevated presteroid levels of IL-6 (see Supplementary Table) were comparable to those in TB-HIV IRIS patients [20]. Both TNF and IL-6 directly contribute to elevated CRP, severe wasting, and reduced survival in murine IRIS. Moreover, IL-6 production and its role in IRIS are completely independent of IFNγ in the murine model [16, 21]. Likewise, the high bacterial loads allowed by IFNγ deficiency and elevated IL-6 and TNF may have also driven the inflammatory pathology in our patient.

This case of disseminated tuberculosis in the setting of neutralizing anti-IFNγ antibodies complicated by immune-mediated deterioration during anti-tuberculous treatment demonstrates that IFNγ is important for control of M. tuberculosis infection but is not necessary for the development of paradoxical treatment reactions.

Supplementary Data

Supplementary materials are available at http://cid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.