Background
The incidence of non-tuberculous mycobacterial (NTM) infection is rising [
1], and this appears to be driven largely by an increase in pulmonary
Mycobacterium avium-intracellulare complex (MAI) infection [
2]. Pulmonary NTM disease is characterized by respiratory and constitutional symptoms, with significant impact on quality of life [
3,
4]. Treatment requires long-courses of antibiotics, which often do not eradicate the mycobacteria [
5]. Furthermore, many of the antimicrobials used to treat NTM disease have significant adverse-effects and drug-drug interactions [
6]. As a result, often the benefit of attempting to treat localized infection is outweighed by its complications - leaving patients with persistent symptoms. Investigating the aetiology of pulmonary NTM infections may therefore enable us to devise more effective treatments and so manage the increasing number of affected individuals.
Whilst disseminated NTM infections are associated with severe immunodeficiency states such as HIV infection or defects in interferon gamma (IFNγ) and STAT3 pathways [
7], the pathophysiology of localised pulmonary NTM infection is unclear [
8]. Pulmonary NTM disease is most often seen in patients with chronic respiratory illness but this is perhaps confounded by doctors most often testing for NTM in such populations. Research in patients with cystic fibrosis demonstrates a specific association between NTM infection and allergic-bronchopulmonary aspergillosis (ABPA) [
9] - suggesting that the increased predilection for NTM infections in patients with chronic lung disease is more complex than simply mycobacteria thriving in damaged lungs [
10,
11].
In parallel with the increasing incidence of pulmonary NTM disease over the last decades, there is a globally rising prevalence of allergic diseases and the “atopic march” by which patients sequentially develop eczema, food allergies, allergic rhinitis and eosinophilic asthma [
12,
13]. Dysregulated Th2 (T-helper lymphocyte type 2) immune responses are thought to underlie these states [
14,
15]. Asthma itself is associated with increased susceptibility to, and severity of, certain infections, and this appears to be independent of the effect of corticosteroids [
16]. For example, Kloepfer and colleagues have shown asthmatic children to be at greater risk of influenza infection [
17], whilst Talbot and colleagues have demonstrated an association between asthma and invasive pneumococcal disease [
18]. Furthermore, Th2-type inflammation is known to be associated with dissemination and impaired clearance of
Cryptococcus [
19,
20] and
Histoplasma infections [
21].
Following the observation in our NTM clinic of several patients with eosinophilia, we hypothesised that the increasing incidence of MAI disease reflects an increase in allergic disease, with a direct association between Th2-type cytokine predominant immune responses and MAI infection. Therefore, we reviewed biomarkers of Th2-type inflammation – peripheral blood eosinophil counts and serum Immunoglobulin E (IgE) levels [
22] – in pulmonary NTM patients and compared them, where available, to a pulmonary tuberculosis cohort attending over the same period.
Methods
Patient Groups
A complete list of all patients with positive mycobacterial cultures from pulmonary samples at the Royal Free Hospital, London, UK, over the five years August 2010 – August 2015 was generated from the microbiology information system. Patients with known HIV, current haematological malignancy or significant primary immunodeficiency condition were excluded. Our NTM cohort was defined as those with at least two positive mycobacterial cultures, at least one identified NTM cultured and no positive cultures of Mycobacterium tuberculosis complex organisms. A cohort of patients with culture-positive pulmonary tuberculosis (TB) was similarly identified.
Patients’ pathology records were interrogated for the preceding eosinophil count (as an absolute value and as a percentage of the total white cell count) and total IgE before and closest to the date of the first positive mycobacterial culture. If these were not available then the first succeeding counts were recorded. A raised peripheral blood eosinophil count was defined as >0.4x10
9/L and a raised serum IgE level as >150kU/L. An eosinophil count cut-off of 0.27x10
9/L was also evaluated, given research showing this has a good sensitivity and specificity for detection of active eosinophilic airway inflammation in asthmatic patients [
23].
Electronic records of clinical correspondence contemporaneous to the first positive NTM sample were also reviewed for evidence of patient or doctor reported co-morbid diagnoses such as asthma.
Sample Analysis
Samples for mycobacterial analysis were received in sterile containers, homogenized and centrifuged as appropriate and decontaminated using NALC-NaOH method. All specimens were inoculated into growth indicators tubes (BD MGIT; BD, US), using 0.5 ml of processed specimen. Specimens that flagged positive on the BD MGIT system were examined for acid-fast bacilli and sub-cultured onto pH neutral pyruvate-based Löwenstein Jensen slope. For each patient with NTM the first two positive specimens were sent to the Public Health England (PHE) Mycobacterial Reference Laboratory for identification; thereafter new positive samples were sent monthly. The reference laboratory performed a Genotype Mycobacterium CM VER 2.0 analysis (Hain Lifescience, Germany).
Complete blood counts, including total white cell counts and eosinophil counts, were assayed using Sysmex XN-9000 analyzers (Sysmex, Japan). Total IgE measurement was performed on an ImmunoCAP 250 analyzer (ThermoScientific, US) using a fluorescence enzyme immunoassay technique (FEIA).
Statistics
GraphPad Prism 6.0 (GraphPad Software, USA) was used for all parametric and non-parametric statistical tests. Unpaired t-tests were used to compare mean absolute eosinophil counts and mean percentage of total leukocyte counts between patient groups, and to compare IgE concentrations after logarithmic transformation of the variable. Chi-squared tests were used to compare distributions between patient groups of eosinophil counts in the ranges <0.27x109/L vs 0.27-0.4x109/L vs >0.4x109/L. For comparisons of subject demographics between groups Chi-squared tests were employed (except for age in which an unpaired t-test was used). Figures show mean and standard deviation for all data except logarithmically transformed IgE concentrations for which geometric mean and 95% confidence interval are shown.
Discussion
Over recent decades there has been a rising incidence of both NTM infection, driven by increasing MAI infections, and atopic/allergic diseases, with the latter underpinned by Th2-type immune responses [
1,
2,
12,
13]. We therefore hypothesized a possible association between MAI infection and Th2-type immune responses and investigated our cohort of patients with mycobacterial infection to evaluate this, using blood eosinophil count and serum IgE biomarkers of Th2-type responses. In the absence of a hematological malignancy, the Th2-type cytokine interleukin-5 (IL-5) is critical for the development of a peripheral eosinophilia [
24] whilst B lymphocyte class-switching to IgE synthesis is dependent on the Th2-type cytokines IL-4/IL-13 [
25]. We found significantly higher peripheral blood eosinophil counts in patients with NTM infection compared to pulmonary TB; and significantly higher eosinophil counts and serum IgE levels in patients culturing MAI complex NTM compared to those culturing only NTM other than MAI. Given the sample size in this study, the statistical significance of these differences is notable. We did not find asthma to be a more frequent co-morbidity in patients culturing MAI; however, not all cases of asthma are necessarily underpinned by allergic mechanisms or Th2-type inflammation [
26]. Bronchiectasis was more common in patients culturing MAI NTM, and is of some interest given recent research regarding the association between NTM disease and bronchiectasis [
27], and the frequency in asthma of airway changes along the bronchiectatic spectrum [
28].
The association between MAI infection and Th2-type inflammation, as indicated by raised peripheral blood eosinophil levels and serum IgE levels, could have several explanations. One possibility is that inhaled corticosteroids used to treat Th2-associated airway inflammation (eg asthma, ABPA and eosinophilic COPD) may underlie the association. Previous studies have suggested steroids are a risk factor for NTM disease [
9,
10]. If this were the explanation then it is of concern, as peripheral blood eosinophilia is increasingly regarded as a biomarker of under-treated steroid-responsive airways inflammation that requires increased steroid doses [
23,
29] - which might paradoxically further increase the risk of NTM disease and impair anti-mycobacterial immune responses.
Alternatively, Th2-type inflammation itself could predispose to MAI infection. Successful host defense against mycobacteria with clearance/control of mycobacterial infection requires an effective Th1, and to lesser extent functioning Th17, immunological response rather than a Th2-type response [
30,
31]. These different facets of the adaptive immune system are capable of cross-regulation – which allows an acute immune response to be specifically optimized to the character of any infection. However (chronic) disequilibrium between the different divisions of the adaptive immune system may lead to pathology and susceptibility to infection [
32].
Mediators of Th2-type inflammation in allergic disease may suppress anti-mycobacterial responses in patients leading to persisting NTM infection. For example, the Th2-type cytokine IL-4 is known to inhibit Th1 and Th17 responses [
33,
34]. Such mechanisms are thought to underlie the diminished IFN responses and more severe pathology seen in respiratory viral infections in asthmatics [
35]. IL-5 also has the capacity to impair IFN responses [
36]. If this were the case then suppressing Th2-type inflammation may be beneficial in MAI infection. For example, it is possible that anti-Th2 cytokine monoclonal antibody therapy could be a useful component of treatment for MAI, optimizing the body’s own anti-mycobacterial immune responses. It is notable that patients with common variable immune deficiency (CVID) and X-linked agammaglobulinaemia (XLA), who are unable to generate IgE, only rarely seem to develop NTM infections despite their increased susceptibility to other organisms and a very high incidence of bronchiectasis ([
37,
38] and unpublished observations).
Finally, it is possible that MAI complex mycobacteria have developed defensive mechanisms to skew immune responses towards a Th2-type bias that decreases the ability of the immune system to clear the mycobacteria. For example,
Francisella tularensis has previously been shown to promote macrophage differentiation into an ‘alternatively activated’ phenotype, more associated with allergic disease and tissue repair/remodeling than killing of intracellular pathogens [
39].
There is a growing research field investigating the pulmonary microbiome in airway diseases: a prevailing theory is that loss of microbial diversity in patients with asthma may be responsible for dysregulated pulmonary immune tolerance and onset of disease pathology [
40]. However, there is now increasing interest in the mechanisms by which respiratory microbes may perturb pulmonary immune responses and in particular how epithelial stimulation from microbial proteases may elicit Th2-type inflammation, with production of IL-4, IL-5 and IL-13 (for example) by other cells of the immune system such as innate lymphoid cells [
41,
42].
Hence, there are several possible mechanisms by which MAI infection could directly promote aberrant pulmonary Th2-type inflammation. Interestingly, Fritscher and colleagues have reported a case-series of difficult-to-control asthma patients found to have NTM infection, the majority of whom had
Mycobacterium avium complex infection [
43]. In most of the patients treated with anti-mycobacterial therapy there was an improvement in their symptomatology.
A limitation of our work is that patient data were collected retrospectively. Therefore information on prevalence of eczema, food allergies and allergic rhinitis (the other diseases of the atopic march) was not available for most of these patients. Furthermore, the basis of asthma diagnoses in this cohort, whether patients had diagnostic lung function studies, and fractional exhaled nitric oxide (FeNO) measures of Th2-associated airway inflammation were not included in the available patient data. Additionally our eosinophilic patients were not universally screened for possible helminth infections that could potentially underlie the peripheral blood eosinophilia seen in these individuals (which is important given the current research interest into impaired anti-mycobacterial immune responses in patients with helminth and TB co-infections [
44]). However the prevalence of helminth infection in London, UK, is very low and allergic/atopic disease is the cause of most peripheral blood eosinophilia in developed countries [
45]. Further prospective research is needed to better investigate these observations, and on the basis of these results we would therefore advise that future studies of risk-factors for NTM disease should record whether patients have allergic/atopic co-morbidities, including aero-allergen testing, and include markers of Th2-type inflammation.