The authors have declared that no competing interests exist.
Conceived and designed the experiments: LB RE CJ TD JW KG GA. Performed the experiments: NP CJ. Analyzed the data: LB RE NP CJ TD PA. Wrote the paper: LB CJ RE. Critically reviewed the manuscript and approved the final version for submission: TD NP PA JW GA KG.
Multilocus variable number tandem repeat analysis (MLVA) provides microbiological support for investigations of clusters of cases of infection with Shiga toxin-producing
Shiga toxin-producing
The main reservoir of STEC in England is cattle although it is carried by other animals, mainly ruminants.
Each year, over 1000 isolates of presumptive
From the 1st May 2012, all confirmed STEC O157 isolates submitted to GBRU were typed using MLVA, with the aim of evaluating this typing approach for prospective outbreak identification and assessing how it would impact on epidemiological investigations. Data and findings from the first six months of this strategy are presented herein.
Detection and confirmation of STEC at GBRU included biochemical identification and serotyping of bacterial isolates. Real time polymerase chain reaction (PCR)was performed to determine the presence of Shiga toxin genes
From 1st May 2012, all STEC O157 strains submitted to GBRU were typed using MLVA. Amplification of eight MLVA loci was performed in 20 ul reactions volumes on all isolates in two quadruplex PCR reactions.
Local laboratories report presumptive isolates of STEC directly to PHE Centres, who arrange for an ESQ to be administered to patients either by local health protection practitioners or environmental health professionals. The ESQ collects demographic details; risk status; clinical condition (including progression to HUS); household or other close contact details; laboratory results; exposures including travel, food and water consumption, contact with animals and environmental factors; epidemiological case classification; and outbreak/cluster status. Completed questionnaires are forwarded for inclusion in the National Enhanced Surveillance System for STEC in England (NESSS) which is managed by the PHE GEZI department.
Domestic case: An STEC case with no travel outside of the UK reported in the seven days prior to onset of illness.
Travel case: An STEC case with travel outside of the UK reported in the seven days prior to onset of illness.
Household case: An STEC case epidemiologically and microbiologically linked to one or more cases in the same household.
Cluster case: An STEC case epidemiologically and microbiologically linked to one or more case outside of the same household.
Sporadic case: An STEC case with no known links to any other STEC cases.
Case definitions were applied first on the basis of ESQ data and results of standard microbiological methods (i.e. phage typing/
Clusters were grouped into three categories:
Household clusters
Known Clusters- detected through routine public health follow-up of STEC cases identifying linked cases
MLVA identified clusters – newly identified through MLVA typing
For clusters newly identified through MLVA, the ESQ's for cases were retrospectively reviewed for epidemiological links including:
Temporal linkage: Cases with onset of illness occurring within seven days of each other.
Exposure linkage: Cases reporting shared exposures, including geographical links (all cases resided or visited a location within one PHE region), as reported on the ESQ.
Between 1st May and 31st October 2012, 556 confirmed cases of STEC O157 in England were reported and 16 different PT's were identified amongst cases; the most frequently reported PT overall was PT 8 (n = 189, 33.9%), followed by PT 21/28 (n = 160, 28.8%). STEC O157 isolates from 539 cases (96.9%) were further typed by MLVA and a total of 341 unique (>2 SLV's) MLVA profiles were identified. Over three-quarters (n = 258) of profiles were unique to one case. Of the remaining profiles (24%), half (n = 41) occurred more than once due to known household transmission and half (n = 42) occurred more than once outside of households (i.e. community clusters). The concurrence between MLVA profiles and PT was high, with just three instances where the same MLVA profile covered more than one PT (
1. Includes 539 confirmed cases of STEC O157 in England. The size of nodes is proportional to the number of instances of that unique profile. Join lines represent locus variants: Related SLV's and DLV's are represented through a solid line while >two locus variants are denoted through a dotted join.In three clusters MLVA profiles spanned two phage types, these are highlighted in the figure. Inset presents two of these instances where the same MLVA profile was reported in two different phage types.
ESQ data indicated a history of foreign travel for 150 cases (27.8%), including 20 cases comprising eight household clusters and 42 cases linked to 13 travel associated clusters (
1.Includes 539/556 confirmed cases who had isolated typed by MLVA. 2.Travel cases reported travel outside of the UK in the seven days prior to onset of illness
A fifth (n = 106) of all cases comprised 42 different household clusters, eight of which were travel-related. Household clusters ranged between two and seven cases in size, with a median of 2 cases (IQR: 2, 3). Of the 42 clusters, in 24 households all cases had the same MLVA profile, 11 exhibited Single locus variants (SLVs), five double locus variants (DLVs) and two had >2 locus variants.
Twenty-one of the 41 community clusters defined through MLVA typing were known clusters, previously identified as epidemiologically linked through surveillance data from ESQ's and phage typing. This included 13 clusters of cases associated with travel to Malta (3), Turkey (3), Cape Verde (2), Tunisia (1), Morocco (1), Egypt (1), Spain (1) and Israel (1). Eight non-travel community clusters comprising 37 cases (
Cluster | PT/ |
Total cases | Cases identified through MLVA |
1. minced beef | PT 21/28 |
9 | 6 |
2. wedding reception | PT 8 |
2 | 0 |
3. nursery/school A |
PT 4/47 |
2 | 0 |
4. nursery/school B | PT 21/28 |
6 | 0 |
5. nursery/school C | PT 8 |
4 | 0 |
6. nursery/school D | PT 32 |
5 | 0 |
7. Petting farm |
PT 4 |
2 | 0 |
8. Country park | PT 54 |
8 | 0 |
Total | 38 | 6 |
1 Cases not linked with recognised clusters through phage typing and VESQ data.
2 One case from the farm seeded the outbreak in nursery A.
MLVA did not identify additional cases in the six other clusters. However, in Cluster#3 (Nursery cluster A), MLVA indicated that despite different PT's (
An additional 20 community clusters not detected through ESQ data and phage typing were identified through MLVA. These clusters ranged in size between two and 12 cases, with a median of three cases (IQR: 3, 7). Eighteen of the 20 clusters (92 cases) had associated strains which were either PT 21/28
Evidence of linkage between cases | No. clusters | Total No. cases | Min. cases/cluster | Max. cases/cluster |
MLVA, temporal |
4 | 29 | 2 | 12 |
MLVA & temporally related cases |
8 | 45 | 2 | 11 |
MLVA linked only | 8 | 27 | 2 | 7 |
All clusters | 20 | 101 | 2 | 13 |
1 Cases with onset dates between zero and seven days apart.
2 Includes residing in or travel within the UK to the same area, shared direct or indirect contact with animals and/or their environment.
Four MLVA clusters had cases both occurring close together in time and additional evidence of linkage between cases. Definitive sources of infection were not identified but cases were linked geographically and/or through exposures: Three of the clusters consisted of cases from different English regions who had travelled to the same areas on holiday and some had visited the same attractions or undertaken the same activities. The fourth cluster was of two cases who lived in the same town. One cluster of two in a household comprised cases with an identical MLVA proflle but one case had a PT54 strain and the other PT8.
One MLVA cluster of seven cases comprised cases from different regions in England. Reviewing the ESQ data revealed that two cases had visited the same village, where another case was resident. Four cases were re-interviewed and all reported visiting the same public house. There were no common consumption patterns between the cases, and two cases had visited the premises outside of the seven day exposure period. The potential source could have been a person or the environment or it may that contamination was taken into the public house. Three other cases did not report travel to the same location and no obvious shared exposures were reported on the ESQ. Because of the time that had elapsed between notification and onset no further follow-up of these cases was undertaken.
Eight MLVA clusters consisted of 45 cases with onset dates close together in time (maximum seven days between cases) but no common geography or exposures were reported on the ESQ's. Five clusters each had two to three cases, but one larger cluster comprised eight cases (one with a different PT,
No additional evidence for a cluster could be found from the ESQ for eight MLVA clusters, totalling 27 cases. The size of these clusters varied from two to seven cases. Within these clusters, cases did not appear to be linked geographically, were dispersed over time, spanning several months, and no obvious shared exposures were reported on the ESQ's.
A number of studies have described the ease of use and timeliness of MLVA and demonstrated the discriminatory power of this approach over phage typing and pulse-field gel electrophoresis.
In England, the most prevalent STEC O157 PT's are PT21/28 and PT8. Five of the eight known clusters identified through public health investigation and supplemented by phage typing, were of a PT other than PT 21/28 or PT 8. Detection of rarer PT's facilitates identification of clusters. In addition, all but one of the known clusters were linked to a precise location, and all cluster cases were temporally linked. In contrast, only a fifth of the MLVA identified clusters were linked geographically, and not to precise locations. In addition, most were of PT 21/28 or PT8, as per all STEC O157 cases, making it extremely difficult to detect such clusters with phage typing alone. There may be under-reporting of PT 8 cases as symptoms appear less severe which may increase the likelihood of PT8 clusters occurring undetected. In contrast, PT 21/28 is associated with more severe disease and progression to HUS than other PT's.
MLVA was useful in informing investigation of known clusters. The additional data can be used to elucidate previously undisclosed links, such as in the farm cluster seeding the nursery cluster, thus highlighting gaps in control measures. MLVA data had a substantial impact on investigations of Cluster#1 (associated with minced beef outbreak), where it captured additional cases and widened the focus of the outbreak control team. Through increased ascertainment of linked cases, both the likelihood and accuracy of determining the source of infection was improved.
Although MLVA identified additional clusters, analyses of data did not lead to an increased chance of a public health intervention. For most of the 20 clusters newly detected through MLVA, reviewing the ESQ data revealed temporal clustering of cases, but additional evidence of shared exposures was indicated in just four clusters. In such clusters where cases are linked to a relatively broad geographic area and the potential source of infection is environmental, it is difficult to pinpoint the source of infection from ESQ data. For eight clusters, no temporal, geographic or exposure linkage was detected. The ability to establish epidemiological links between cases is in part diminished due to the small sample size of the clusters, making it difficult to confidently establish commonality of exposures. ESQ data were retrospectively reviewed when some time had passed since most cases were ill, the clusters were small in size and were temporally contained. Therefore the ability to undertake further analytical study was limited.
There were three larger clusters comprising nationally distributed cases with apparent temporal links suggestive of a food-borne source of infection. While, the ESQ, introduced in England in 2009, has made a major contribution to STEC surveillance, it was not designed to capture the highly detailed information often required to definitively link cases to a shared exposure. The ESQ is not a trawling questionnaire so the vehicle of infection, for example, could be a food-stuff not included in the questionnaire or included but not in enough detail to identify it. Additionally, similarities in consumption between cases may reflective of common population behaviours, such as in the cluster of cases reporting consumption of pre-packed salad.
While this study demonstrated the added value of MLVA in increasing recognition of clusters and providing insight into the microbiology and epidemiology of STEC O157 in England, it raises questions in terms of how to respond in terms of public health. Most STEC clusters are small so the chance of proof of source and hence intervention is reduced, while analytical studies are not possible. The resource involved in investigating clusters is not inconsiderable; input from a wide range of multi-disciplinary professionals is required and in nationally dispersed clusters, co-ordinating local investigations is challenging. In real-time surveillance the eventual size of a cluster is unknown, and difficulties lie in deciding at what point hypothesis generation exercises should be initiated. The delay in obtaining MLVA data (although MLVA is more rapid than PFGE, the MLVA data is often available more than eight days after the patient becomes symptomatic), and thereby identifying linked cases, may also impede investigation. The increased detection of clusters through the additional sensitivity of MLVA typing provides a forerunner for the advent of next generation sequencing of clinical isolates and highlights both the advantages to, and the challenges in public health practice, in utilising molecular methods as routine, which need to be addressed.
PFGE remains the method of choice for outbreak investigation in many countries, although the use of MLVA is increasing as the contribution of the MLVA approach to overall surveillance of STEC O157, in addition to outbreak investigation, is recognised. Despite the challenges described above, the universal MLVA approach has revealed a more accurate picture of the extent of linked cases in England and demonstrated that while STEC O157 clusters in England are small, they contribute a significant proportion of cases. Continuation of MLVA typing on all confirmed STEC O157 isolates submitted to the GBRU will generate more data on the spatial and temporal dispersion of cases of STEC O157. Furthermore, in conjunction with enhanced surveillance data and timely public health investigation of clusters, MLVA offers the potential to provide a more accurate measure of clusters, insight into routes of this infection in England and evidence to inform health protection responses at local and national levels.
We would like to thank Kirsten Glen, Natalie Adams, Naomi Launders and Radha Patel for their contributions to the National Enhanced Surveillance System for STEC in England. Thanks to all public health practitioners who report to the scheme. We would also like to acknowledge Marie Chattaway, Vivienne do Nascimento and Michael Wright in GBRU for their laboratory expertise.