- 作者：Dr David W Eyre ; BM BCha ; b ; david.eyre@ndm.ox.ac.uk ; Madeleine L Cule ; PhDa ; c ; d ; A Sarah Walker ; PhDa ; e ; Prof Derrick W Crook ; MB BCha ; b ; Prof Mark H Wilcox ; MDf ; Prof Tim E A Peto ; DPhila ; b
- 刊名：The Lancet
- 出版年：2012
- 出版时间：23 November, 2012
- 年：2012
- 卷：380
- 期：supp_S3
- 页码：S12
- 全文大小：1K
Background
Clostridium difficile infection (CDI) has traditionally been classed as a health-care-associated pathogen, predominantly transmitted within hospitals. Bacterial genotyping is used for investigation of CDI outbreaks, but its endemic nature hampers identification of the source of most infections. We aimed to determine whether the finer genetic resolution of whole-genome sequencing (WGS) could illuminate routes of C difficile transmission, as an exemplar of endemic disease transmission.Methods
All CDIs in Oxfordshire, UK, from Sept 1, 2007, to March 31, 2011, underwent Illumina WGS. Toxin enzyme-immunoassay-positive, culture-positive samples were obtained from Oxford University Hospitals' microbiology laboratory, which provides all C difficile diagnostic testing for the county, including other hospitals and primary care. Single nucleotide variants (SNVs) between cases were estimated from maximum likelihood trees and compared with SNV differences in 145 serially sampled patients. The estimated rate of evolution of 0¡¤86 SNVs per genome per year (95 % CI 0¡¤35-1¡¤46) equated to expecting 0 to 2 SNVs between transmitted isolates sampled less than 146 days apart, and 0 to 3 SNVs 146-350 days apart (95 % prediction intervals). Plausible epidemiological links between genetically related cases were identified from hospital admission and community location (postcode district and general practice postcode) data.
Findings
1195 of 1250 (96 % ) CDIs were successfully sequenced. For 938 CDIs from April 1, 2008, to March 31, 2011, 381 (41 % ) were within 2 SNVs of one or more previous cases since Sept 1, 2007; 345 (37 % ) were within more than ten SNVs from all previous cases (Sept, 2007, to March, 2008, cases included as potential sources for later CDIs). Of 381 patients with a previous case within two SNVs (consistent with transmission), only 111 (29 % ) spent time on the same hospital ward within plausible limits on infectious periods (1 week before to 8 weeks either side of a positive test) and incubation periods (12 weeks). A further 60 of 381 (16 % ) were linked by either shared time in the same hospital (but never on the same ward) or by a shared ward (not at the same time, up to 28 days apart). Of 210 cases without any hospital link, only 35 (17 % ) shared a general practice or home postcode district. Overall, 175 of 381 (46 % ) CDIs with a previous case within 2 SNVs had no record of any previous hospital or community contact. Even when all restrictions on durations of infectious, incubation, and ward contamination periods were removed, 104 of 381 (27 % ) remained without any known epidemiological link. The numbers of intermediate patients in common between epidemiologically unlinked CDI cases who were closely (two or less SNVs) or distantly (more than ten SNVs) genetically related were similar (p=0¡¤68), suggesting that unidentified cases in hospitals were unlikely to be a major cause of infection.
Interpretation
Substantial transmission arises from outside the symptomatic population, from currently unsampled reservoirs. The main limitation of the study was that we included only idenfied enzyme-immunoassay-positive CDI, and not asymptomatic and environmental sampling. Routine access to WGS could provide insights into novel routes of transmission for endemic as well as epidemic microbial pathogens and identify new targets for public health interventions.
Funding
NIHR Oxford Biomedical Research Centre, UKCRC Modernising Medical Microbiology Consortium, UKCRC Translational Infection Research Initiative, Medical Research Council, Biotechnology and Biological Sciences Research Council, National Institute for Health Research, Department of Health, Wellcome Trust.