Amplified fragment length polymorphism and whole genome sequencing: a comparison of methods in the investigation of a nosocomial outbreak with vancomycin resistant enterococci

Abstract

Background: Recognition of nosocomial outbreaks with antimicrobial resistant (AMR) pathogens and appropriate infection prevention measures are essential to limit the consequences of AMR pathogens to patients in hospitals. Because unrelated, but genetically similar AMR pathogens may circulate simultaneously, rapid high-resolution molecular typing methods are needed for outbreak management. We compared amplified fragment length polymorphism (AFLP) and whole genome sequencing (WGS) during a nosocomial outbreak of vancomycin-resistant Enterococcus faecium (VRE) that spanned 5 months.

Methods: Hierarchical clustering of AFLP profiles was performed using unweighted pair-grouping and similarity coefficients were calculated with Pearson correlation. For WGS-analysis, core single nucleotide polymorphisms (SNPs) were used to calculate the pairwise distance between isolates, construct a maximum likelihood phylogeny and establish a cut-off for relatedness of epidemiologically linked VRE isolates. SNP-variations in the vanB gene cluster were compared to increase the comparative resolution. Technical replicates of 2 isolates were sequenced to determine the number of core-SNPs derived from random sequencing errors.

Results: Of the 721 patients screened for VRE carriage, AFLP assigned isolates of 22 patients to the outbreak cluster. According to WGS, all 22 isolates belonged to ST117 but only 21 grouped in a tight phylogenetic cluster and carried vanB resistance gene clusters. Sequencing of technical replicates showed that 4-5 core-SNPs were derived by random sequencing errors. The cut-off for relatedness of epidemiologically linked VRE isolates was established at ≤7 core-SNPs. The discrepant isolate was separated from the index isolate by 61 core-SNPs and the vanB gene cluster was absent. In AFLP analysis this discrepant isolate was indistinguishable from the other outbreak isolates, forming a cluster with 92% similarity (cut-off for identical isolates ≥90%). The inclusion of the discrepant isolate in the outbreak resulted in the screening of 250 patients and quarantining of an entire ward.

Conclusion: AFLP was a rapid and affordable screening tool for characterising hospital VRE outbreaks. For in-depth understanding of the outbreak WGS was needed. Compared to AFLP, WGS provided higher resolution typing of VRE isolates with implications for outbreak management.

Keywords: AFLP; Molecular typing; Nosocomial outbreak; VRE; WGS.

Fig. 1

Timeline showing the number of cases and contacts during the VRE outbreak. Bar chart: number of patients that tested positive for a VRE outbreak-type isolate by AFLP (Y-axis, left); solid line: cumulative number of ward and room contacts traced (Y-axis, right); dotted line: room and ward contacts that tested negative for the VRE outbreak-type (Y-axis, right). One hundred thirty-nine patients were traced but lost to follow up. X-axis: week numbers in 2017. The 1st outbreak wave was from week 16–21, the 2nd outbreak wave from week 29–35

Fig. 2

Dendrogram based on AFLP patterns of all VRE strains identified during the outbreak. Unrelated E. faecium and E. faecalis strains were included in the analysis as outgroup. Hierarchical clustering of AFLP types was performed using unweighted pair-grouping. The cut-off value for identical strains was 90% relative similarity (Bionumerics, Applied Maths, Belgium). For the outbreak strain (in green), the relative similarity was 92%. The relative similarity for S15, S17, and S18 was 89% compared to the outbreak strain. However, this value was deemed to be below 90% due to issues with signal to noise ratios and these three strains were therefore also considered part of the outbreak, based on their AFLP patterns. Unrelated isolates U02–1 and U03 (blue) clustered together with 97% similarity to each other

Fig. 3

Maximum likelihood tree based on alignment of core-SNPs, mid-point rooted. Included are all outbreak isolates, plus previous isolates collected across the Netherlands (indicated by the letter “E” as the first letter of their name) and unrelated VRE isolates (downloaded from European Nucleotide Association database; indicated by the letters SRR or VRE). SNP count differences with the index isolate (S01), reference free core-SNPs only. Outbreak isolates are numbered and lettered. The letters “S”, “H” and “U” indicate Surgery, Haematology and Unrelated. S, H and U isolates are numbered in chronological order