Abstract

Background: Owing to its low incidence, the management of Mycobacterium xenopi pulmonary infections is not clearly defined. A multicentre retrospective study was performed to describe the features of the disease and to evaluate its prognosis.

Methods: All patients with M xenopi satisfying the 1997 ATS/IDSA criteria from 13 hospitals in north-east France (1983–2003) were included in the study. Clinical, radiological and bacteriological characteristics and data on the management and outcome were collected.

Results: 136 patients were included in the analysis, only 12 of whom presented with no co-morbidity. Three types of the disease were identified: (1) a classical cavitary form in patients with pre-existing pulmonary disease (n = 39, 31%); (2) a solitary nodular form in immunocompetent patients (n = 41, 33%) and (3) an acute infiltrate form in immunosuppressed patients (n = 45, 36%). 56 patients did not receive any treatment; the other 80 patients received first-line treatment containing rifamycin (87.5%), ethambutol (75%), isoniazid (66.2%), clarithromycin (30%) or fluoroquinolones (21%). After a follow-up of 36 months, 80 patients (69.1%) had died; the median survival was 16 months (range 10–22). Two independent prognostic factors were found: the acute infiltrate form was associated with a bad prognosis (hazard ratio 2.6, p = 0.001) and rifamycin-containing regimens provided protection (hazard ratio 0.325, p = 0.006). Clarithromycin-containing regimens did not improve the prognosis.

Conclusions: In contrast to recent guidelines, this study showed three different types of the disease (cavitary, nodular or diffuse infiltrate forms) with a different prognosis. In order to improve survival, all patients with M xenopi infection should be treated with a rifamycin-containing regimen. The usefulness of clarithromycin remains to be evaluated.