Inhaled therapies, azithromycin and Mycobacterium abscessus in cystic fibrosis patients

RESULTS

We included 30 MABSC-positive cases and 60 NTM-negative controls in this study. The cases and controls had similar baseline characteristics. The controls tended to have better forced expiratory volume in 1 s values than the cases, but this difference was not statistically significant (table 1). The use of inhaled therapies and low-dose azithromycin by cases and controls in the year preceding MABSC isolation is shown in table 2. No significant association was found between these treatments and infection. Inhaled steroid treatment tended to be more frequent among the controls (OR 0.4, 95% CI 0.1–1.2), although this difference was not significant (p=0.08). A similar analysis was performed with data collected in years 2, 3 and 4. Again, no significant association was found with either inhaled therapies or low-dose azithromycin.

The use of inhaled therapies and low-dose azithromycin at the 34 French CF centres that participated in the 2004 NTM prevalence study is shown in table 3. The treatments administered differed considerably between centres, even within the paediatric, mixed and adult subgroups. However, the centres with the highest prevalence of MABSC infection in each subgroup (centres A, B and C, respectively) did not seem to use inhaled therapies or low-dose azithromycin with a higher frequency than expected, except for a slightly higher frequency of inhaled steroids at the paediatric centre (centre A) than at other paediatric centres (47%, 95% CI 24–42.6%). The frequency of inhaled rhDnase use at the paediatric centre and the mixed centre with the highest prevalence of MABSC infection was also particularly low (table 3).

DISCUSSION

Unlike non-CF bronchiectasis patients, CF patients are particularly susceptible to MABSC infection. This susceptibility may result either from CF-related treatments or from CF itself. The genetic defect underlying CF was identified in 1989 (mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR)), but the mechanisms by which CFTR defects cause lung disease remain unclear. Various mechanisms have been proposed, such as defective submucosal gland secretion in the airways, abnormal airway surface liquid composition and volume and intrinsic inflammation [29]. In addition, CFTR was found to be involved in phagosomal pH control in alveolar macrophages in one study [30]. Indeed, lysosomes from CFTR-null macrophages displayed acidification defects and were unable to kill internalised bacteria. However, CFTR-dependent lysosomal acidification in alveolar macrophages was not confirmed by another group [31]. Therefore, there is no clear evidence of an intrinsic immune defect of bacterial killing in CF patients.

In a previous whole-genome analysis of M. abscessus (sensu stricto), we were surprised to discover that this mycobacterium had a large series of specific genes in common with the other two major CF pathogens Pseudomonas aeruginosa and Burkholderia cepacia complex. Among the “nonmycobacterial” factors relevant to the infection of CF patients is the Dnd phenotype conferred by a large number of cysteine desulfurases [22, 23]. We speculated that rhDnase treatment permitted the use of sputum DNA as a nutrient by Dnd-positive MABSC strains in CF patients. However, we did not find any such association. Whether an underlying host mechanism explains the striking susceptibility of CF patients to MABSC, P. aeruginosa and B. cepacia complex infection remains to be determined.

Azithromycin has been shown to improve lung function of CF patients, particularly those colonised with P. aeruginosa [32]. These benefits are believed to result from the attenuation of inflammation and the prevention of bacterial biofilm formation [33]. As MABSC isolates are usually resistant to azithromycin, MABSC may be selected by long-term azithromycin treatment in CF patients. In addition, a recent study also suggested that azithromycin may increase susceptibility to mycobacterial disease by inhibiting the autophagy-dependent killing of mycobacteria [26]. Prior treatment with azithromycin was shown to result in significant alkalinisation of the mycobacterial phagosome, to pH levels inhibiting most lysosomal proteases in RAW 264.7 cells, and to prevent basal intracellular killing and the interferon (IFN)-γ- and tumour necrosis factor (TNF)-dependent killing of M. abscessus in primary human macrophages. The authors also demonstrated that azithromycin treatment was associated with persistent lung infection and lower levels of interleukin-12, IFN-γ and TNF production in their mouse model of M. abscessus lung infection. Thus, the potential favouring role of azithromycin on MABSC infection is a matter of concern.

Recent studies suggest a relationship between MABSC infection and long-term low-dose azithromycin treatment [4, 26]. However, we did not find a significant positive association between the presence of MABSC in sputum and the use of the other “new” therapeutic approaches to CF such as inhaled therapies or long-term azithromycin. These discrepancies may result from differences in the methods used in the other studies. Indeed, these studies are subject to several limitations, small study populations and unsatisfactory control groups, which probably biased the findings. We overcame these problems by using an appropriate control group established by matching cases with controls for age, sex and centre. Age and sex must be taken into account because they are significantly associated with NTM in CF patients [1, 34]. Moreover, “centre” has been overlooked as a factor in previous studies. However, our analysis of data from the French CF registry showed that the therapeutic management of patients differed considerably between participating centres. For example, the frequency of azithromycin use varied from 12% to 72.5% in adult centres and 6.5% to 50.5% in paediatric centres. Similarly, the frequency of inhaled steroid use varied from 8% to 55% in adult centres, and from 8% to 70% in paediatric centres. It is therefore important to control for any potential “centre effect”.

There was also no evidence that centres with the highest prevalence of MABSC infection differed substantially in their treatment practices from other centres of the same type (paediatric, mixed or adult). Indeed, the centres with the highest prevalence of MABSC infection did not use more low-dose azithromycin, inhaled antibiotics or inhaled rhDNAse than the others. We only found a slightly higher frequency of inhaled steroids at paediatric centre A than at other paediatric centres. However, the difference in inhaled steroid use among different paediatric centres was huge, ranging from 8.3% to 70.3%.

Thus, contrary to our expectations, we found no evidence of a significant association between MABSC infection and the administration of inhaled therapies or long-course, low-dose azithromycin, whether at the individual or CF centre level. However, we cannot formally exclude the existence of such a link. Indeed, our study was retrospective and we were able to trace the use of these treatments only for the 4 years preceding the first MABSC isolation. It therefore remains theoretically possible, given the frequently indolent or slowly progressive nature of MABSC infection, that the MABSC was acquired years earlier, at a time at which the patient was treated with inhaled therapies and/or long-term azithromycin. However, this is unlikely, because treatment escalations are much more frequent than treatment reductions during the progression of the disease in a given patient. It is unlikely, for example, that a patient not treated with an aerosol at the time of the study would have received such a treatment in any significant manner several years previously. However, this residual doubt should incite us to remain vigilant concerning the possible impact of these treatments on the bacterial populations circulating in patients. Of course, it is out of the question that we should “let our guard down” concerning inhaled therapies or call into question the absolute necessity to respect a certain number of rules aimed at preventing contamination (e.g. the use of sterile solutes and careful disinfection of nondisposable items).

The emergence of MABSC in CF patients is estimated to have occurred in the mid-1990s [35] and was probably multifactorial. This period saw major improvements in the microbiological methods for isolating MABSC and other NTM [13–15]. It also coincided not only with the widespread use of inhaled therapies and long-course macrolides as immune-modulators, but also with more systematic intravenous antibiotic treatment, greatly increasing antibiotic pressure and the selection of organisms naturally resistant to antibiotics commonly used to treat CF patients. Stenotrophomonas maltophilia, which belongs to an entirely different bacterial group, provides a good example of the selection of multiresistant organisms by broad-spectrum antibiotic treatments in CF patients in the last 20 years [36, 37]. However, this link between antibiotic selection pressure and MABSC has yet to be demonstrated formally.

Finally, another disturbing finding is the extremely heterogeneous clinical presentation of MABSC disease, extending from totally asymptomatic infection to fulminant forms in some patients, whether or not they had undergone transplantation [7–10]. This observation suggests a particular susceptibility to MABSC infection in some patients. It could be a key point for the control and the prevention of the disease. We are currently investigating this issue.