Epidemiology of Mycobacterium abscessus in England: an observational study

Abstract

Background: Mycobacterium abscessus has emerged as a significant clinical concern following reports that it is readily transmissible in health-care settings between patients with cystic fibrosis. We linked routinely collected whole-genome sequencing and health-care usage data with the aim of investigating the extent to which such transmission explains acquisition in patients with and without cystic fibrosis in England.

Methods: In this retrospective observational study, we analysed consecutive M abscessus whole-genome sequencing data from England (beginning of February, 2015, to Nov 14, 2019) to identify genomically similar isolates. Linkage to a national health-care usage database was used to investigate possible contacts between patients. Multivariable regression analysis was done to investigate factors associated with acquisition of a genomically clustered strain (genomic distance <25 single nucleotide polymorphisms [SNPs]).

Findings: 2297 isolates from 906 patients underwent whole-genome sequencing as part of the routine Public Health England diagnostic service. Of 14 genomic clusters containing isolates from ten or more patients, all but one contained patients with cystic fibrosis and patients without cystic fibrosis. Patients with cystic fibrosis were equally likely to have clustered isolates (258 [60%] of 431 patients) as those without cystic fibrosis (322 [63%] of 513 patients; p=0·38). High-density phylogenetic clusters were randomly distributed over a wide geographical area. Most isolates with a closest genetic neighbour consistent with potential transmission had no identifiable relevant epidemiological contacts. Having a clustered isolate was independently associated with increasing age (adjusted odds ratio 1·14 per 10 years, 95% CI 1·04-1·26), but not time spent as an hospital inpatient or outpatient. We identified two sibling pairs with cystic fibrosis with genetically highly divergent isolates and one pair with closely related isolates, and 25 uninfected presumed household contacts with cystic fibrosis.

Interpretation: Previously identified widely disseminated dominant clones of M abscessus are not restricted to patients with cystic fibrosis and occur in other chronic respiratory diseases. Although our analysis showed a small number of cases where person-to-person transmission could not be excluded, it did not support this being a major mechanism for M abscessus dissemination at a national level in England. Overall, these data should reassure patients and clinicians that the risk of acquisition from other patients in health-care settings is relatively low and motivate future research efforts to focus on identifying routes of acquisition outside of the cystic fibrosis health-care-associated niche.

Funding: The National Institute for Health Research, Health Data Research UK, The Wellcome Trust, The Medical Research Council, and Public Health England.

Figure 1

Phylogeny of 944 Mycobacterium abscessus isolates The phylogenetic tree includes one unique isolate per patient per cluster (36 were excluded due to missing data). The inner ring shows the presence or absence of a diagnosis of cystic fibrosis. The outer ring shows the geographical region of England in which the patient lives.

Figure 2

Distribution of Mycobacterium abscessus isolate cluster sizes by sample type and patient diagnosis Distribution of cluster sizes for all clusters identified using a genomic distance threshold of fewer than 25 SNPs, by diagnosis (A) and by sample types (B). The algorithm used to assign respiratory diagnoses to patients is in the appendix (p 7).

Figure 3

Timeline of genomic clusters identified in this study Isolates (deduplicated per cluster) are shown for patients with and without cystic fibrosis. Clusters of isolates defined using a genomic distance threshold of fewer than 25 SNPs are connected by dark grey lines. Ligh grey lines show the time to the earliest non-sequenced isolate belonging to a member of a given cluster. The orange bars show the 95% highest posterior density interval for the inferred date of the root for the time-scaled phylogenetic trees for larger clusters (n≥10); the mean point estimate of these dates is shown as a black dot.

Figure 4

Dated phylogenies and geographical distribution of Mycobacterium abscessus isolates in the two largest clusters in this study Clusters were identified using a genomic distance threshold of fewer than 25 SNPs. The side panel shows the region of England in which the patient lived at the time of isolate collection. Dated phylogenies for all clusters are shown in the appendix (p 13).