Results

Five different species of mycobacteria were isolated in 14 patients, with M fortuitum, M avium complex, and M chelonae occurring most commonly (table 1). The 14 cases comprised nine adults and five children, 4% and 3.5% of the respective clinic populations. The age range was 9–25 years with a median of 16 years. The median age of the paediatric clinic population is 11 years, and for the adult clinic 24 years. Eight of the nine adult cases were male compared with an adult clinic sex ratio of 52% female to 48% male.

Each case with its two controls constitutes a single set. Data on each of the variables assessed were complete for all sets apart from the following categories: steroid therapy (13 sets complete); Northern score (12 sets); respiratory function, nutritional status and S-K score (11 sets). Only those sets with complete data were subject to analysis. Follow up data on cases 1, 3, 6, and 13 were limited to 18 months.

Overall, no significant effect of non-tuberculous mycobacteria was seen on respiratory function, nutritional status, or S-K score (table 2). Cases tended to have worse Northern chest radiograph scores, although this did not achieve statistical significance (p = 0.06). There was no significant difference between cases and controls in the incidence of diabetes mellitus, genotype, frequency of other bacteria or fungi isolated, or in the use of nebulised antibiotic therapy (table 3). Cases had received significantly more intravenous antibiotic courses before the first isolate was identified (p = 0.002) and, although they were more likely to have received steroid therapy of any kind (p = 0.08), particularly if they were adults (p = 0.06), neither of these changes was statistically significant. There was no observable attributable effect in those individuals who received treatment for non-tuberculous mycobacteria (cases 9, 10, and 14). A subgroup analysis excluding sets 7, 11, 12, and 13 (which contain cases with a single mycobacterial isolate only) resulted in the same conclusions; the difference between cases and controls with respect to the number of intravenous antibiotic courses received before the isolate was identified was reduced but still significant (p = 0.015).

Discussion

Although initially discovered shortly after Koch identified the tubercle bacillus in 1882, non-tuberculous mycobacteria were not generally recognised as potentially pathogenic until the 1950s.1 16 More recently there have been reports of fatal non-tuberculous mycobacteria infection in cystic fibrosis3 17 and subsequent studies have cited prevalence rates for non-tuberculous mycobacteria ranging from 1.8% to 20% in cystic fibrosis populations.3 5-9 18 Differences in inclusion and diagnostic criteria, as well as the varying geographical distribution of environmental mycobacteria, lead to difficulties in making direct comparisons between these reports. Some authors have seen patients improve with treatment for non-tuberculous mycobacteria,5 while others have failed to note any benefit.9 The issue is further complicated by the difficulties in distinguishing simple colonisation from infection, although the presence of at least two positive smears or cultures is recommended by the American Thoracic Society to establish a diagnosis of the latter.16 In a recent necropsy study, Tomashefskiet al found evidence of active mycobacterial disease in only two of six patients with cystic fibrosis who had had multiple positive sputum cultures, although interestingly only these same two cases had multiple positive smears for acid-fast bacilli.19 Twelve other patients with a single positive sputum culture had no evidence of mycobacterial pathology at necropsy.

Risk factors for pulmonary infection with non-tuberculous mycobacteria in the general population include chronic lung disease and immunosuppression.1 16 There is no evidence of any systemic immune defect in patients with cystic fibrosis20and interest has focused on determining other predisposing factors in this group. Aitken et al reported eight cases (from 64 adult patients with cystic fibrosis) who tended to be older and have lower clinical scores, but found no association with radiographic findings, diabetes mellitus, respiratory function, or genotype.8 Kilby et al found that adult patients with non-tuberculous mycobacteria were similar to those without, although cases again tended to be older. In addition, they noted no association with the presence of Pseudomonas aeruginosa, Staphylococcus aureus, orAspergillus species in the sputum.7 Hjelteet al reported five cases (aged 11–25 years) from a cystic fibrosis population of 54 and did not observe any association with corticosteroid usage, diabetes, or poor nutritional state.5 All five patients had progressive radiographic shadowing and deteriorating lung function at the time of positive sputum cultures and all five improved with specific chemotherapy. However, one of these cases had persistently negative smears and one positive culture yielding M gordonae, which is rarely pathogenic. More recently, Fauroux et al reported two teenage patients with severe lung disease from a cystic fibrosis population of 106 who had positive smears and sputum cultures forM chelonae.18 Both were treated with clarithromycin with subsequent clearance of the non-tuberculous mycobacteria from the sputum. However, clinical and functional status continued to decline in one, and the other saw only an initial improvement which may have been temporary. Apart from the possibility that the non-tuberculous mycobacteria were not related to the clinical status, these cases highlight the fact that clearance of non-tuberculous mycobacteria from sputum in those with suspected infection does not necessarily indicate cure, and that claims of successful treatment may be premature. In this regard, Heurlin and Petrini have reported 25 cases of non-tuberculous mycobacteria infection in patients without AIDS, 22 of whom received chemotherapy.21 After more than two years follow up, following the completion of chemotherapy, only seven appeared to be free of disease.

In our study there was only a slight increase in the prevalence of non-tuberculous mycobacteria isolated from the adult cystic fibrosis population when compared with the children, but the child cases were all aged nine years or older. There was a male preponderance in the adult cases that was not reflected in the background population. Overall, cases tended to have worse Northern chest radiographic scores and were more likely to have received corticosteroids than controls, but neither of these changes reached statistical significance. This may reflect the small sample size and resultant low study power. Cases had also received significantly more courses of intravenous antibiotics. This latter association may simply be a surrogate marker for progressive lung damage, predisposing to infection with non-tuberculous mycobacteria, although we did not observe a parallel significant deterioration in respiratory function or S-K score. Alternatively, it may be a true effect by way of repeatedly altering the bacterial flora of the airways, permitting colonisation with non-tuberculous mycobacteria and subsequent infection. A third possibility is that a delay in recognition of non-tuberculous mycobacteria infection leads to prolonged courses of intravenous antibiotics for presumed bacterial infection. We did not note any association with diabetes mellitus, genotype, nutritional status, or the frequency of other bacteria or fungi isolated. The three cases who received treatment for non-tuberculous mycobacteria (cases 9, 10, and 14) did so because of clinical deterioration despite antibacterial and antifungal treatment. Case 10 showed some initial improvement, but this was not uniformly sustained over the treatment course. Cases 9 and 14 did not appear to benefit from non-tuberculous mycobacterial treatment, although there was poor compliance with therapy in case 14. There has been no significant attributable deterioration in those cases with multiple positive smears and sputum cultures who did not receive treatment for non-tuberculous mycobacteria (cases 3, 5, 6, and 8). There were not enough data on skin testing to warrant analysis. Others have found its predictive value to be low in patients with cystic fibrosis because of the high prevalence of cross reactivity between mycobacterial species and the problem of anergy in patients with advanced disease.11

This study is limited by its small size and retrospective nature. However, it does indicate that the overall prevalence of non-tuberculous mycobacteria in patients with cystic fibrosis is low and that the clinical impact is likely to be minimal in most individuals over a two year period. The possible link with frequent intravenous antibiotic usage and steroid therapy warrants further investigation and can only be clarified by a large, multicentre, prospective trial. Such a trial would not only help to define significant risk factors for non-tuberculous mycobacteria infection in cystic fibrosis but would also help in producing guidelines for diagnosis and identification of those patients who would benefit most from chemotherapy.