Background
Prevalence and incidence of nontuberculous mycobacterial lung disease (NTM-LD) have been increasing worldwide [
1‐
3]. In Japan, the incidence rate of NTM-LD in 2014 was estimated to be 14.7 per 100,000 people [
4]. The most common pathogen for NTM-LD is the
Mycobacterium avium complex (MAC), which comprises of
M. avium and
M. intracellulare [
1,
3,
5]. MAC lung disease (MAC-LD) is mainly classified into two forms: the nodular bronchiectatic type (NB) and the fibrocavitary type (FC) [
5]. Although the 2007 American Thoracic Society and Infectious Diseases Society of America (ATS/IDSA) guidelines recommend a combination treatment of multiple drugs [
5], antibiotic treatment has limited efficacy [
6]. Because there is no consensus on the optimal timing to start antibiotic therapy and the treatment period, it would be beneficial to establish useful biomarkers to monitor therapeutic effectiveness and evaluate prognosis.
Although the prognosis for MAC-LD is generally good, disease progression varies between patients, with some experiencing rapid, fatal deterioration. A few reports have evaluated the severity of MAC-LD and its prognosis. We reported that profiling data of variable number tandem repeats of
M. avium are associated with disease progression in MAC-LD [
7]. Other studies have suggested that anti-glycopeptidolipid (GPL)-core antibodies reflect disease activity; for example, Kitada et al. reported that serum levels of anti-GPL-core antibodies in patients with MAC-LD were related to the severity of chest computed tomography (CT) findings and were decreased after chemotherapy [
8,
9]. In addition, there have been several surveys of clinical prognostic factors, which indicated that radiographic findings, undernutrition, anemia and high inflammation test values may be negative prognostic factors for MAC-LD [
10,
11].
Limited information is available about the cytokine networks involved in the pathogenesis of MAC-LD. Activation of T helper type 1 (Th1) cells and cytokine production play important roles against intracellular parasites such as mycobacteria, but their roles in MAC infection remain poorly understood. It has been reported that peripheral blood mononuclear cells from NTM-LD patients produced less Th1 cytokines [interferon-γ (IFN-γ), interleukin-12 (IL-12) and tumor necrosis factor-α (TNF-α)] than those from healthy controls [
12,
13] and that Th17 immunity may be associated with susceptibility to NTM-LD [
14]. In addition, a previous report suggested that lower CXC motif ligand 10 (CXCL10) levels were associated with therapeutic response [
15]. Thus, differences in cytokine response may be implicated in the differences between patients in the progression of MAC-LD.
The aim of the present study was to elucidate the pathophysiology of MAC-LD and develop a new biomarker that reflects the immune response to the disease. To do this, we simultaneously measured the protein concentrations of 38 serum cytokines and growth factors (GFs) in patients with MAC-LD. We also investigated the association between these protein concentrations and the poor prognostic factors described above.
Discussion
The natural history of MAC-LD depends on the type of clinical disease: FC or NB [
5]. A radiographic finding of FC is one of the negative prognostic factors for both all-cause and MAC-specific mortality. We measured 38 serum cytokine/GFs protein concentrations in 27 patients with MAC-LD and compared them between those with the FC and the NB forms. We found that CXCL10 concentrations were significantly higher in those with the FC form and that CXCL10 concentration correlated with other poor prognostic factors, the HRCT scores and disease activity.
MAC-LD typically presents as an apical fibrocavitary lung disease, FC form. If left untreated, this form of disease is gradually progressive, resulting in extensive cavitary lung destruction with increased mortality [
10,
18,
19]. On the other hand, the NB form tends to have a much slower progression than the FC form, but it too may progress, leading to death [
20]. Using a multivariate Cox proportional hazard model, Hayashi et al. found that a radiographic finding of FC or FC combined with NB was a negative prognostic factor for MAC-specific mortality, as were low BMI, presence of anemia and high CRP [
10]. We investigated the relationship between these poor prognostic factors and serum cytokine concentrations. It is plausible that levels of many cytokines were low because they act locally and transiently rather than systemically and chronically. Two proinflammatory cytokines, G-CSF and sCD40L, and one chemokine, CXCL10, were significantly elevated in the FC group, whereas anti-inflammatory cytokines, IL-1RA and IL-10, were significantly elevated in the FC group. This finding might indicate that a strong immune response to MAC infection and accompanying inhibitory response are variable, particularly in the FC form.
Among the cytokines analyzed, CXCL10 levels were associated with low BMI, low albumin and high HRCT scores. CXCL10 expressed by antigen-presenting cells induces chemotaxis, apoptosis, cytostasis and angiostasis [
21‐
23]. As well as in several autoimmune diseases [
24] and in infections with various viruses [
25‐
27], CXCL10 has been reported to be a useful biomarker for tuberculosis (TB) [
28,
29]. Another report indicated that plasma CXCL10 levels in patients with MAC-LD were higher than in those healthy participants [
15]. Our results show that CXCL10 is associated with CT findings and need for treatment of MAC-LD. Therefore, it increases against MAC infection and may be involved in the pathogenesis of the disease.
Although skin test reactivity to MAC is highly prevalent during young adulthood [
30], pulmonary MAC disease is relatively rare, suggesting the host immune response plays an important part in eliminating the infecting microbes. Inhaled mycobacteria are taken up by macrophages, but they survive and proliferate within these vacuoles as an intracellular pathogen. Macrophages containing MAC produce cytokines to recruit lymphocytes and other macrophages. Of these cytokines, IL-12, TNF-α, and IFN-γ are of particular importance in the anti-mycobacterial immune response. IFN-γ and IL-12 form a positive feedback loop that is pivotal in the immune response to mycobacteria [
31,
32]. The production of CXCL10 is driven by many signals, including IFN-γ, IFN-α/β, IL-2, and autocrine antigen-presenting cell-derived TNF-α and IL-1β in an autocrine manner [
33‐
35]. We found a significant increase in the concentration of CXCL10 between the FC and NB subtypes, but not in IFN-α, IL-2, TNF-α, or IL-1β. Perhaps the poor increase of these inflammatory cytokines may contribute to the pathology of MAC disease [
12,
13].
With MAC-LD, one of the major problems in clinical practice is when to start antibiotics treatment. If untreated, approximately half of patients with MAC-LD progress over the subsequent 2–10 years of follow-up [
11,
18,
36,
37]. However, there is still no clear standard for the initiation of treatment, especially for patients with NB form. Risk factors for disease progression include low BMI, cavitary disease on chest CT, number of lung segments involved, older age, male sex, and the presence of comorbidities, as well as anemia, hypoalbuminemia and elevated CRP and/or ESR levels [
11,
18,
36]. In the present study, we found statistically significant associations between CXCL10 levels and some of these risk factors. In addition, CXCL10 was significantly higher in the patients requiring treatment, as in a previous report [
15]. Thus, CXCL10 could be a promising biomarker for determining the treatment strategy for the disease.
Anti-GPL-core IgA antibody is reported to be a convincing diagnostic marker for MAC-LD [
38]. Some studies have suggested that this antibody may reflect the progression of MAC-LD to some extent [
8,
9]. However, in the present study, we did not find a statistical difference in anti-GPL-core IgA antibody levels that supported this hypothesis. Instead, we found that CXCL10 concentration was strongly correlated with the poor prognostic factors and the HRCT scores, suggesting that the cytokine may reflect the disease activity with greater sensitivity than anti-GPL-core IgA antibody levels.
The current study had some limitations. First, it was a cross-sectional, single-center study. The possibility of unintentional selection bias could not be ruled out. In addition, because our institution is a university hospital, there may have been other biases in patient recruitment. Second, the number of participants was small, including only six cases with the FC form of MAC-LD, which may have resulted in underestimates of the differences in cytokine concentrations. The number of patients was insufficient to analyze the association between CXCL10 and the poor prognosis factors for each of the two disease form groups separately. For the same reason, we could not efficiently exclude several confounding factors. In addition, the standard of treatment introduction was unclear. Our findings need to be confirmed in a larger cohort from multiple institutions, preferably in a prospective study for a longer study period to establish CXCL10 as a biomarker for prediction of the prognosis of MAC-LD.