Whole-genome sequencing for investigating hospital outbreaks

A couple of papers have come out this month demonstrating the use of whole genome sequencing (using the Illumina MiSeq) for investigating hospital outbreaks in the UK.

In the Cambridge corner, we have a paper in NEJM looking at 14 ST22 MRSA genomes from a local hospital, with the help of the Sanger Institute team, using 150 bp paired end sequencing. Using SNP-based phylogenies, they showed that a cluster of isolates from an outbreak in the neonatal intensive care unit belonged to a very tightly-linked genetic cluster, quite distinct from the other MRSA genomes isolated from other wards in the hospital during the same period. They also showed complete concordance between phenotypic antimicrobial resistance profiles and genetic mechanisms for resistance.

Rapid whole-genome sequencing for investigation of a neonatal MRSA outbreak.

Köser CU, Holden MT, Ellington MJ, Cartwright EJ, Brown NM, Ogilvy-Stuart AL, Hsu LY, Chewapreecha C, Croucher NJ, Harris SR, Sanders M, Enright MC, Dougan G, Bentley SD, Parkhill J, Fraser LJ, Betley JR, Schulz-Trieglaff OB, Smith GP, Peacock SJ.

N Engl J Med. 2012 Jun 14;366(24):2267-75.


The authors conclude that they would therefore be able to correctly rule-in or rule-out MRSA isolates from an outbreak. I.e. they use the set of isolates defined by the infection control team as an outbreak as the gold standard outbreak definition; then by showing that sequencing can detect clustering of these strains, conclude that sequencing is just as good at identifying an outbreak.

Personally (probably because I am a genomicist and think sequencing is the bee’s knees) I think of it the other way around, that whole genome sequencing by definition provides the ‘correct’ answer against which we can test a hypothesis that an outbreak is occurring. I.e. the sequencing provides accurate data against which we can properly decide whether a set of cases really does constitute a transmission chain, or merely reflects a coincidental cluster of infections with distinct causative agents. If you want to define an outbreak on the basis of case numbers alone then there is no need for sequencing. However if you consider the key issue to be determining whether an increase in cases is the result of transmission within the hospital (presumably because you want to do something about it), then sequencing can give you the answer. I think the authors would agree, since they also conclude that sequencing revealed a transmission event in another ward, which they had not missed previously.

Anyway, the most important thing is that they showed it can be done, and for a cost ($150 per isolate) and time frame (<36 hours) that is highly competitive with current microbiological approaches. I didn’t notice any discussion of analysis time (i.e. data-to-conclusions time), only isolate-to-sequence data time, which will be a very important consideration for the rollout of this approach into clinical practice.

In the Oxford corner we have a paper in BMJ looking at five disease clusters, including MRSA and Clostridium difficile, in three different UK hospitals:

A pilot study of rapid benchtop sequencing of Staphylococcus aureus and Clostridium difficile for outbreak detection and surveillance.

Eyre DW, Golubchik T, Gordon NC, Bowden R, Piazza P, Batty EM, Ip CL, Wilson DJ, Didelot X, O’Connor L, Lay R, Buck D, Kearns AM, Shaw A, Paul J, Wilcox MH, Donnelly PJ, Peto TE, Walker AS, Crook DW.

BMJ Open. 2012 Jun 6;2(3). pii: e001124


The perspective presented in this paper is more in line with my own, i.e. they framed the study as an investigation of whether sequencing could “provide high precision support and rejection of transmission”. So here the assumption is that very close genomes provides evidence of transmission. The authors show that, on this assumption, the sequencing supports the existence of an outbreak (transmission chain) on the basis of 0-SNP differences in the core genome, even though the antimicrobial resistance phenotypes differed between isolates (note that resistance genotypes were concordant with these phenotypes, as in the other paper). So this is an example where sequencing can change our mind, providing strong evidence of an outbreak (i.e. rule-in) where other typing methods would lead you to the opposite conclusion.

The authors also give some examples where sequencing of C difficile clusters with suspected transmission links, based on MLST and PCR ribotyping, actually showed that the isolates from ‘linked’ cases were quite distinct and therefore did not constitute transmission events. So this is an example where sequencing can change your mind in the other direction, providing convincing evidence against an outbreak (i.e. rule-out) where other typing methods would lead you to think transmission was occurring.

In this paper, the authors quote turnaround times of “under a  week”. I am assuming this time this does include analysis and interpretation, as they also say that they obtained results quickly enough to influence cluster investigations and inform hospital responses. This is really the key.

The authors mention the Ion Torrent as an alternative platform, and quote the possibility of the Oxford Nanopore being used at a cost of $10… a little premature still, but hopefully this is not too long away as it would be the real game-changer that’s needed to propel sequencing into routine clinical and public health use. Can’t wait.


One thought on “Whole-genome sequencing for investigating hospital outbreaks

  1. Pingback: Links 6/23/12 | Mike the Mad Biologist

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s