Systematic application of multiplex PCR enhances the detection of bacteria, parasites, and viruses in stool samples

Transcription

1 Journal of Infection (2013) 67, 122e129 Systematic application of multiplex PCR enhances the detection of bacteria, parasites, and viruses in stool samples Gary N. McAuliffe a, *, Trevor P. Anderson a, Mary Stevens a, Jacqui Adams a, Robyn Coleman a, Patalee Mahagamasekera a, Sheryl Young a, Tom Henderson b, Maria Hofmann d, Lance C. Jennings a,c, David R. Murdoch a,c a Canterbury Health Laboratories, P O Box 151, Christchurch 8140, New Zealand b Medlab South, C/O P O Box 151, Christchurch 8140, New Zealand c University of Otago, Christchurch, New Zealand d Fast-Track Diagnostics, 38 Rue Hiehl Z.I Langweis, Junglinster 6131, Luxembourg Accepted 11 April 2013 Available online 18 April 2013 KEYWORDS Multiplex PCR; Faeces; Bacteria; Parasites; Viruses Summary Objectives: To determine whether systematic testing of faecal samples with a broad range multiplex PCR increases the diagnostic yield in patients with diarrhoea compared with conventional methods and a clinician initiated testing strategy. Methods: 1758 faecal samples from 1516 patients with diarrhoea submitted to two diagnostic laboratories were tested for viral, bacterial, and parasitic pathogens by Fast-Track Diagnostics multiplex real-time PCR kits and conventional diagnostic tests. Results: Multiplex PCR detected pathogens in 530 samples (30%): adenovirus (51, 3%), astrovirus (95, 5%), norovirus (172, 10%), rotavirus (3, 0.2%), Campylobacter jejuni/coli (85, 5%), Salmonella spp. (22, 1%), Clostridium difficile (72, 4%), entero-haemorrhagic Escherichia coli (21, 1%), Cryptosporidium spp. (3, 0.2%), Entamoeba histolytica (1, 0.1%), and Giardia lamblia (59, 3%). In contrast, conventional testing detected a pathogen in 324 (18%) samples. Conclusions: Using a systematic approach to the diagnosis of gastroenteritis improved diagnostic yield. This enhanced detection with PCR was achieved by a combination of improved detection of individual pathogens and detection of pathogens not requested or unable to be tested by conventional tests. This approach also allowed earlier identification for most pathogens and created a workflow which is likely to adapt well for many diagnostic laboratories. ª 2013 The British Infection Association. Published by Elsevier Ltd. All rights reserved. * Corresponding author. Microbiology Department, LabPlus, P O Box , Auckland City Hospital, Auckland 1148, New Zealand. Tel.: þ ; fax: þ addresses: GMcAuliffe@adhb.govt.nz, mcauliffegary@hotmail.com (G.N. McAuliffe) /$36 ª 2013 The British Infection Association. Published by Elsevier Ltd. All rights reserved.

2 Systematic multiplex PCR for gastroenteric pathogens 123 Introduction Diarrhoeal disease is a leading cause of mortality and morbidity worldwide, 1 and can also place considerable financial burden on health care systems through hospitalisations and complications such as GuillaineBarré syndrome. 2 The laboratory detection of gastroenteric pathogens still relies on microscopy, culture, and immunoassays. 3 Turnaround times for these tests are variable, but it can take several days to exclude organisms such as Campylobacter spp. and Salmonella spp. 4 PCR is an established tool for the diagnosis of norovirus infection, 5 and several studies have examined the use of multiplex PCR for the diagnosis of bacterial, 6,7 viral, 8,3 or parasitic 9,10 causes of diarrhoea. These have demonstrated enhanced detection of most pathogens with more rapid turn-around times, and the ability to test for a greater range of targets. The presence of important pathogens such as Giardia lamblia has also been demonstrated in samples requesting examination only for bacteria. 11 We describe the prospective application of combined viral, bacterial and parasite multiplex PCR panels from Fast-Track Diagnostics (FTD), (Fast-track Diagnostics, Junglinster, Luxembourg) on consecutive faeces specimens from patients with diarrhoea. The main aim of the study was to assess whether the systematic application of multiplex PCR would result in an increased diagnostic yield over existing diagnostic testing based on tests routinely performed at our laboratories, and a clinician initiated testing strategy. Materials and methods Specimens Sequential faecal specimens submitted from community or hospital sources to the microbiology laboratories of Canterbury Health Laboratories (CHL) or Medlab South (MLS), Christchurch, New Zealand between April and June 2011 were included in the study. At the time, both CHL and MLS were operating from the same facility and were using similar laboratory protocols. Formed specimens and those with insufficient volume for PCR testing and storage were excluded. All specimens were routinely cultured for bacterial pathogens and, at the request of referring clinicians, were tested for other enteric pathogens (Table 1). In addition, approximately 200 mg of faeces was transferred into 1 ml of Stool Transport and Recovery buffer (STAR, Roche Diagnostics, Auckland, New Zealand) and tested as soon as possible by multiplex PCR. An additional aliquot of unprocessed faeces and the STAR buffer suspension were stored at 80 C in case further testing was required. The study was approved by the Upper South B Ethics Committee. Nucleic acid extraction Specimens in STAR buffer were pre-processed by centrifugation at 13,000 rpm for 2 min to remove faecal matter. 60 ml of supernatant was added to 240 ml of NucliSENS lysis buffer (BioMerieux, Sydney, Australia) with 2 ml of internal control (FTD) and extracted using the NucliSENS easymag (BioMerieux) generic 2.0A protocol. Nucleic acid was eluted into a final volume of 75 ml. Specimens unable to be pipetted due to the presence of mucous were first diluted 1:2 and centrifuged. If still unable to be pipetted, mucous was withdrawn from the specimen. Specimens found to be inhibited after PCR amplification and analysis were diluted 1:10 in STAR buffer and re-extracted as above. PCR amplification PCR was performed according to the manufacturer s instructions using FTD Bacterial gastroenteritis (2 pools); FTD Viral gastroenteritis (2 pools) FTD Parasite gastroenteritis (1 pool) kits and AgPath-IDä One-Step enzyme (Applied Biosystems, Life technologies, Melbourne, Australia) with Table 1 Pathogen detection by conventional methods according to clinician request, multiplex PCR (M-PCR) detection in samples where both methods were applied. Pathogen Conventional positive % Multiplex PCR positive % Total number tested Campylobacter spp C. difficile Entero-haemorrhagic Escherichia coli (EHEC) Salmonella spp Shigella spp Yersinia spp Adenovirus Norovirus GII Rotavirus Cryptosporidium E. histolytica G. lamblia

3 124 G.N. McAuliffe et al. 10 ml of nucleic acid per pool on the ABI 7500 real-time PCR thermocycler (Applied Biosystems) platform. Each run included 12 patient samples plus positive and no template controls on a MicroAmp 96 well optical reaction plate (Applied Biosystems). The multiplex PCR targets are listed in Table 2. Results were interpreted as positive if an exponential fluorescence trace was demonstrated, as recommended by the manufacturer. All PCR curves and the database were subsequently double-read to prevent interpretation and data-entry errors. Other laboratory methods Faecal culture for Salmonella, Shigella, Yersinia, Plesiomonas, Aeromonas and Campylobacter species was performed using the following selective agars: xylose lysine desoxycholate (XLD), cefsulodin irgasan novobiocin (CIN), and Campylobacter blood free agar (Fort Richard Laboratories, Auckland, New Zealand). The plates were incubated at 36 C in ambient air (XLD), 30 C in ambient air (CIN), and 42 C in a microaerophilic environment (Campylobacter agar). Faeces specimens were also inoculated into Selenite F broth, and Yersinia selective enrichment broth (Fort Richard Laboratories) and incubated at 36 C and 30 Cin ambient air for 24 h prior to inoculation onto XLD and CIN plates respectively. In addition, macroscopically bloody specimens, specimens from patients with suspected haemolytic-uraemic syndrome (HUS), bloody diarrhoea, or specimens requesting testing for entero-haemorrhagic Escherichia coli (EHEC) were inoculated on cefixime tellurite sorbitol MacConkey agar (CTSMAC, Fort Richard Laboratories) and incubated at 36 C in ambient air for 24 h. At MLS, testing for EHEC also extended to specimens from patients who lived in rural locations or were aged less than 5 years. Suspicious colonies were identified using standard bacteriological phenotypic methods, including hippurate hydrolysis for the differentiation of Campylobacter jejuni and Campylobacter coli. Specimens that were sent for Clostridium difficile testing were tested using Premier Toxins A & B enzyme immunoassay (Meridian Bioscience Inc., Cincinnati, USA) according to manufacturer s instructions. Specimens that were sent for adenovirus or rotavirus testing were tested using Adenoscreen or Rotascreen latex agglutination (LA) test (Microgen Bioproducts, UK) respectively using a modified version of the manufacturer s instructions (20 ml of stool and 20 ml of test reagent). In addition, all samples from children aged <5 years submitted to MLS were tested for rotavirus. All specimens requesting parasites submitted to CHL were examined by light microscopy using a trichrome and modified ZiehleNeelsen stain. First specimens from a given patient submitted to MLS requesting parasites were tested by giardia and cryptosporidium EIA (Thermo Fisher Scientific, Remel, USA) unless clinical details suggested chronic diarrhoea, when a modified iron-haematoxylin stain and EIA were performed. Subsequent samples from a given patient were tested by modified iron-haematoxylin stain only. For specimens requesting norovirus testing approximately 50 mg of stool was inoculated into 1 ml of viral transport medium, centrifuged at 13,000 rpm for 10 min. 200 ml of supernatant was extracted using the SPR-ITEä Viral NA extraction kit (BeckmaneCoulter, Auckland, New Table 2 M-PCR targets. Pathogen Target Serotypes/subtypes/species detected Bacterial panel Campylobacter jejuni mapa gene C. jejuni Campylobacter coli glya gene C. coli, other Campylobacter spp. Clostridium difficile tcdb gene C. difficile, C. sordelii EHEC vtx 1A, 1B, vtx 2A,2B gene EHEC, S. dysenteriae Salmonella spp. ttr gene S. enterica serotypes Shigella spp. ipah gene S. boydii, S. dysenteriae, S. flexneri, and S. sonnei, enteroinvasive E. coli (EIEC) Yersinia enterocolitica ail gene Y. enterocolitica Viral panel Adenovirus Hexon gene D,10,13,15,17,19,20,22e30,32,33,36e49, 51,53,54, all 51 subtypes Astrovirus Outer capsid protein gene 1 Norovirus Genogroup 1 ORF 1e2 junction G1-2, G1-4 Norovirus Genogroup 2 ORF 1e2 junction G2-4 Rotavirus Segment 7, NSP3 gene A Parasite panel Cryptosporidium spp. DNA-J like protein gene C. parvum, C. hominis, C. meleagridis Entamoeba histolytica SSU rrna gene sequences for E. histolytica E. histolytica & E. dispar, Specific E. histolytica probe Giardia lamblia SSU rrna gene sequences for G. lamblia G. lamblia

4 Systematic multiplex PCR for gastroenteric pathogens 125 Zealand) and eluted to 60 ml of nucleic acid. PCR amplification for the detection of norovirus genogroup I (GI) and genogroup II (GII) was performed on the ABI 7500 platform. 12 Statistical analysis The proportion of specimens testing positive in various groups of patients were compared using the chi-square test. Results During the study period 1758 faeces specimens were tested from 1516 patients. Of these patients, the median age was 49 years (range 2 weekse98 years), 238 (14%) were 5 years old or younger, and 918 (52%) were female. Of the specimens, 1190 (68%) were sent from community primary care clinics, and the remainder were from hospital inpatients or outpatient clinics. When tested by conventional methods according to original requests, gastroenteric pathogens were detected in 324 (18%) specimens, with 45 (3%) specimens having two or more pathogens (Table 1). Conventional positives also included Aeromonas spp. (43), Plesiomonas spp. (2) Blastocystis hominis (82), and Dientamoeba fragilis (15). In comparison, using the systematic approach on all samples, multiplex PCR detected pathogens in 530 (30%), with 49 (3%) specimens having two or more pathogens. Table 1 illustrates multiplex PCR results in the group of specimens where both conventional and multiplex PCR were initially applied, and Table 3 demonstrates the multiplex PCR yield in samples where additional testing was not requested by the referring clinician. A pathogen was detected in 642 (37%) of samples by combining conventional and multiplex PCR results. Thirty specimens (2%) demonstrated inhibition of internal control amplification, but gave adequate internal control amplification on dilution and repeat testing. Of the 84 Campylobacter isolates detected by culture, 81 were identified by biochemical profiles, molecular sequencing, and/or PCR 13 as C. jejuni, two as C. coli, and one as C. lari. Three culture-positive specimens were negative by the multiplex PCR; one of these specimens isolated C. lari, which would not have been expected to have been detected by this assay. Four culture-negative specimens were positive by the multiplex PCR; three of these subsequently had positive results for C. jejuni (2) and C. coli (1) by another PCR assay. 13 Of the specimens tested for C. difficile by both EIA and PCR, 13 were positive by both methods, two were negative by PCR but positive by EIA, and 11 samples were positive by PCR but negative by EIA. In the subgroup of 1488 specimens for which C. difficile testing was not requested by the attending clinicians, 48 (3%) specimens were positive by PCR. This group was notable in that 28 (58%) of these specimens were from children <5 years old, that 15 (31%) had co-pathogens (Campylobacter spp., (3) astrovirus, (3) and norovirus (9)) and that 38 (79%) came from community sources. Of the 21 specimens positive for EHEC by PCR, 16 (66%) were confirmed by isolation of E. coli O157 and/or detection of toxin by PCR by a reference laboratory (Institute of Environmental Science & Research, New Zealand (ESR)). Two of these specimens came from patients with bloody diarrhoea, but none were associated with haemolytic-uraemic syndrome. The 31 Salmonella strains isolated were: S. enteritidis, (3) S. infantis, (3) S. stanley, (6) S. saintpaul, (3) S. typhimurium, (11) S. weltevreden, (1) S. birkenhead, (1) and S. larochelle. (3) Of these, nine specimens were negative by PCR but positive by culture: S. stanley, (1) S. saintpaul, (3) S. typhimurium, (3) S. larochelle. (2) Pure culture isolates of these organisms all tested positive by the bacterial panel of the multiplex PCR. FTD have subsequently updated this panel in order to improve sensitivity. When the original specimens were retested, five of the nine specimens that tested negative by the original PCR panel tested positive by the updated version: S. saintpaul, (2) S. typhimurium, (1) S. larochelle. (2) This improved the sensitivity of PCR versus culture from 71% to 87%. The Yersinia isolates were Yersinia enterocolitica biotype 1A, (1) Y. enterocolitica biotype 4 (1) and Yersinia frederiksenii. (7) None were positive by PCR in the original specimens. When the pure culture isolates were tested by the bacterial PCR panel, only the Y. enterocolitica biotype 4 tested positive. This panel has subsequently been updated by FTD to include a target for the ystb gene. FTD were able to detect both the biotype 1A and biotype 4 Table 3 Additional pathogen detection using systematic M-PCR approach. Pathogen Additional Total M-PCR Percentage of all study samples positive (n Z 1758) M-PCR positive positive M-PCR approach Conventional methods C. difficile EHEC Adenovirus Astrovirus e Norovirus GI Norovirus GII Rotavirus Cryptosporidium E. histolytica G. lamblia

5 126 G.N. McAuliffe et al. isolates from the pure culture isolates with their revised assay, however no positives were detected when the original specimens in STAR buffer at CHL were retested. Adenovirus was not detected by LA in any of the 45 specimens requested for testing. PCR was positive in one of these samples, and in 50 additional specimens not requesting LA. With retrospective testing of the PCR positives, 6/46 (13%) were also positive by LA from the archived specimens. Astrovirus PCR was positive in 95 specimens. An in-house real-time PCR assay 14 confirmed a subset of 20 of 24 archived samples. Norovirus was detected by multiplex PCR in all specimens that tested positive by the in-house PCR. One additional specimen that tested negative by in-house PCR was positive by multiplex PCR. When retrospectively tested using the in-house method norovirus GII and norovirus GI were detected in 119 and 16 specimens, respectively, in samples for which norovirus testing had not been initially requested. Rotavirus was detected in 25 of 373 (7%) specimens by LA testing. One of these specimens was also PCR positive. An additional two PCR positive specimens had not been tested by LA. These were subsequently tested by LA from stool stored at 80 C and were both positive. Twenty-three specimens positive by LA were retested from archived specimens using the Rotaclone enzyme immunoassay (Premier, Meridian Bioscience Inc., Cincinnati, USA). Twentytwo of these tested negative by EIA. The one positive EIA specimen had tested positive by PCR. Of the specimens sent for parasitology, only one specimen was positive for Cryptosporidium spp. by both microscopy and EIA. This specimen was negative by PCR. In addition three specimens for which parasitology was not requested, tested positive by PCR. Of these, two were subsequently found to be positive by Immunocard Stat (Meridian Bioscience Inc., Cincinnati, USA) antigen test and EIA for Cryptosporidium spp. from archived specimens. E. histolytica/dispar was detected in one specimen by microscopy, and tested negative by PCR. One specimen for which parasitology was not requested was PCR positive. The aliquot of this specimen stored at 80 C was examined using a modified D Antoni s iodine stain, no cysts or trophozoites were seen. Giardia lamblia was detected by PCR in all specimens testing positive by conventional means and in an additional 19 specimens negative by conventional methods. 17 specimens not requesting parasitology also tested PCR positive. Analysing the multiplex PCR results, pathogen detection frequency also varied by sample origin and age group (Table 4). Discussion We have systematically applied a commercial multiplex PCR assay to detect bacterial, viral and parasitic pathogens in faecal specimens from patients with diarrhoea. Using this combined multiplex PCR panel approach has allowed us to increase the frequency of detection of gastroenteric pathogens from 18% to 30%. Previous studies have primarily focussed on the evaluation of either viral, bacterial or parasite multiplex PCR panels. Consistent with the findings from these studies, we have demonstrated equivalent or improved detection of most pathogens by PCR compared with conventional techniques used in our laboratories. By applying these combined panels to all specimens submitted for bacterial culture regardless of specific clinician requests, we have also enhanced detection of pathogens which would not normally have been tested. Additional targets in the multiplex PCR panels have allowed us to determine a burden of astrovirus and non-o157 EHEC disease for which we have not had routine testing available. Consistent with previous studies, Campylobacter PCR performed well compared with culture. 15,6 The PCR targets the glya and mapa genes found in the two most common Campylobacter faecal isolates, C. jejuni and C. coli. However some uncommon isolates, such as C. lari, may be missed by this assay. 13 Detection of Salmonella spp. by PCR was less sensitive than culture. This has also been noted in other studies, 4,6 although the reason for this is unclear. de Boer et al. hypothesised that their lower sensitivity may be due to the positive effect of selenite enrichment on culture, however in our study Salmonella was isolated from the primary culture plate in eight of the nine PCR negative samples. All Table 4 Pathogen detection with M-PCR by patient location and age. Pathogen No. (%) positive by patient location P value No. (%) by age group (years) P value Hospital (n Z 568) Community (n Z 1190) 0e15 (n Z 351) 16e60 (n Z 746) >60 (n Z 661) Campylobacter 13 (2.3) 72 (6.1) (8) 35 (5) 22 (3) C. difficile 28 (4.9) 44 (3.7) (9) 15 (2) 27 (4) <0.001 EHEC 6 (1.1) 15 (1.3) (3) 5(0.7) 7 (1.0) 0.02 Salmonella 4 (0.9) 18 (1.5) (1) 11 (1) 7 (1.0) 0.77 Adenovirus 1 (0.2) 50 (4.2) < (13) 6 (0.8) 0 (0) <0.001 Astrovirus 15 (2.6) 80 (6.7) < (12) 43 (6) 10 (2) <0.001 Norovirus GI 1 (0.2) 15 (1.2) (1) 8 (1) 3 (0.5) 0.25 Norovirus GII 72 (12.6) 84 (7) < (11) 42 (6) 74 (11) <0.001 Rotavirus 0 (0) 3 (0.3) (0.3) 2 (0.3) 0 (0) 0.40 Cryptosporidium 1 (0.2) 2 (0.2) (0.6) 1 (0.1) 0 (0) 0.11 E. histolytica 1 (0.2) 0 (0) (0) 0 (0) 1 (0.2) 0.44 G. lamblia 8 (1.4) 51 (4.2) (3) 35 (5) 12 (2) 0.01

6 Systematic multiplex PCR for gastroenteric pathogens 127 culture isolates were subsequently detected by the PCR, demonstrating that the assay was able to detect these strains and implying that the assay may have suboptimal analytical sensitivity. FTD have since modified the bacterial gastroenteritis panel and this has improved detection of Salmonella, however testing on new clinical samples is warranted to further assess the sensitivity of the modified assay. Only two samples were culture-positive for Yersinia enterocolitica. Both were missed by direct specimen PCR, and only the biotype 4 was detected by testing the culture isolate. The absence of the ail gene in most subtypes of Y. enterocolitica biotype 1A, a biotype of uncertain pathogenicity, 16 is likely to explain the negative PCR results in the faecal specimen and culture isolate for this organism. Yersinia frederiksenii is also of uncertain clinical significance and is not targeted by the Yersinia PCR. Failure to detect the biotype 4 strain suggests that the assay may have suboptimal sensitivity, but additional testing on a greater number of Yersinia-positive specimens with the revised assay is needed to examine this further. We detected an increased number of C. difficile PCR positives which were not positive by toxin EIA. These most likely represent the suboptimal sensitivity of toxin EIA for the detection of toxigenic C. difficile infection. 17 There is also controversy regarding the use of PCR as a stand-alone test for the diagnosis of C. difficile infection due to concerns regarding its specificity compared to toxigenic culture. This may be partly due to the ability of PCR to detect toxin producing strains in specimens following antibiotic treatment. 18,19 We observed that 4% of community samples tested positive by PCR for C. difficile toxin. Although this may represent enhanced detection of C. difficile disease, it may also represent asymptomatic colonisation with toxin producing strains in adults. 20,21 This has been well documented in infants and young children and is likely to explain the majority of the positive PCRs in children <3 years. 22 This group formed a substantial proportion of the additional C. difficile positives detected by multiplex PCR 29/48 (60%), and laboratories using the multiplex approach may wish to either add a comment or restrict reporting of results to children considered to be at risk for disease and/or where C. difficile testing is requested, e.g. paediatric oncology/haematology patients. Diagnostic laboratories are currently faced with the dilemma of deciding which faeces specimens should be tested for EHEC, as EHEC disease is difficult to clinically differentiate from other diarrhoeal diseases, and whilst O157 serotypes can be suspected by the growth of nonfermenting colonies of E. coli on sorbitol MacConkey agar, sorbitol fermenting non-o157 EHEC serotypes are emerging as important causes of diarrhoea worldwide. 23,24 EIAs are available, but they suffer from low sensitivity unless broth enrichment is used. 25 Good correlation of the FTD PCR with the confirmatory PCR assay suggests that routine use as part of a multiplex panel would be a useful tool to identify this important pathogen. In addition this approach has the advantage of being able to detect non-viable organisms if patients have had prior antibiotic treatment. 25 Our results show that norovirus was frequently detected in patients from whom testing for this virus was not specifically requested. The finding of relatively large numbers of cases of astrovirus infection was particularly important given that we did not normally test for this virus. Notably 41 (43%) specimens were from children 15 years old, 80 (84%) came from community sources and copathogens were present in 15 (16%) (norovirus GII, (10) C. difficile, (4) G. lamblia (3) and Campylobacter (1)). The increased number of positives detected by adenovirus PCR compared with LA may be due to improved sensitivity, but may also be due to detection of non-group F adenovirus serotypes. 14 Most non-group F serotypes are not recognised causes of gastroenteritis. 26 PCR has been shown to enhance detection of rotavirus compared with LA or electron microscopy. 27 In contrast, we detected rotavirus in more samples by LA than PCR. The LApositive but PCR negative samples could not be confirmed by EIA. Sixteen of 25 (64%) rotavirus LA-positive specimens had additional pathogens detected, raising the possibility of false positive results due to cross reactivity to these pathogens. 28,3 The small number of PCR positives is likely to be explained by the study period falling outside the annual peak season of JulyeSeptember. 29 G. lamblia PCR performed well compared with conventional methods, and detected additional positives as previously reported 6,9 including demonstrating undetected Giardia infections in stool specimens requesting only bacterial pathogens. 11 There were co-infections detected by both conventional methods and multiplex PCR. Co-infections may occur for a number of reasons: A common source of exposure or a new infection with ongoing shedding from a previous infection. This is a complex area, and we did not specifically address the relative importance of individual organisms within a coinfection. It may not be possible to directly attribute the current illness to a single pathogen, however, the presence of pathogens such as Shigella, Giardia and Entamoeba histolytica may warrant specific treatment, EHEC further investigations, and the presence of viruses may be useful for avoiding further unnecessary investigations. It is also possible that some patients who had enteric pathogens detected by PCR ultimately had a clinical diagnosis other than infectious gastroenteritis as noted in a previous study. 3 In our study, formed stools were excluded, which makes this less likely, nonetheless clinical judgement is required in the interpretation of all diagnostic tests, and guidance on interpretation of new tests should be available from clinical microbiologists. Our study has several limitations. Firstly, we were unable to prospectively test all samples using conventional tests other than bacterial culture unless they were requested by the referring clinician. This, as well as the small numbers of some pathogens such as Shigella, Yersinia, Entamoeba, and Cryptosporidia has limited our ability to assess the performance of multiplex PCR for all pathogens. By testing by PCR only those samples sent for bacterial culture, we may have given a biased account of clinicians requesting patterns. However, most faeces specimens sent to our laboratory are cultured. The lack of suitable comparator gold standards have limited the ability to estimate sensitivity and specificity compared to established testing methods. In addition, we did not have testing available for confirmation of all additional PCR positives, so it is possible that some of these may have represented false

7 128 G.N. McAuliffe et al. positive results. Subsequent testing of stored isolates for confirmation may have affected results if samples had been degraded by freeze-thawing. 30,31 The multiplex PCR also lacked targets for some causes of diarrhoea that might be regionally important, such as Aeromonas spp., Vibrio spp., sapovirus, Dientamoeba fragilis and Blastocystis hominis. Inclusion of additional targets such as these, and future modifications in assays to improve the sensitivity for targets such as Salmonella could further improve the diagnostic yield of this systematic approach. In addition to the advantages of simultaneous detection of all key gastroenteric pathogens, the workflow of the multiplex PCR approach is likely to adapt well to many diagnostic laboratories. It is more efficient to create a single faeces specimen processing pathway rather than using several laboratory sections to perform different components as commonly occurs at present. This approach also provides results faster than some conventional techniques allow. During the study period, samples were tested the day following collection, and results were available on the day of testing. In contrast 3e5 days were required for the detection of Campylobacter and Salmonella by culture. Cost has previously been a barrier for the implementation of PCR assays, but this is now less of an issue as PCR costs have decreased relative to other diagnostic tests. In our laboratory, the total price for the multiplex PCR (NZ$152) performed in batches is less than the combined price for bacterial, viral, and parasite examination (NZ$280). In conclusion, we found that systematic application of multiplex PCR enhances the detection of gastroenteric pathogens. Laboratory test requesting patterns for gastroenteritis, with testing based on clinician requests, may miss important pathogens in both community and hospital patients. Systematically applying a multiplex PCR that simultaneously detects bacterial, viral and parasitic pathogens to all specimens will help overcome this problem. Acknowledgements This study was supported by Fast-Track Diagnostics who provided reagents, and developed the revised bacterial PCR panel, and the Bowel and Liver Trust, Canterbury, who provided funding. We thank staff from the Institute of Environmental Science & Research, New Zealand (ESR) for specialist testing. References 1. Kosek M, Bern C, Guerrant RL. The global burden of diarrhoeal disease, as estimated from studies published between 1992 and Bull World Health Organ 2003;81:197e Buzby JC, Allos BM, Roberts T. The economic burden of Campylobacter-associated Guillain-Barré syndrome. J Infect Dis 1997;176:S192e7. 3. Wolffs PF, Bruggeman CA, van Well GT, van Loo IH. Replacing traditional diagnostics of fecal viral pathogens by a comprehensive panel of real-time PCRs. J Clin Microbiol 2011;49: 1926e Cunningham SA, Sloan LM, Nyre LM, Vetter FA, Mandrekar J, Patel R. Three-hour molecular detection of Campylobacter, Salmonella, Yersinia, and Shigella species in feces with accuracy as high as that of culture. J Clin Microbiol 2010;48: 2929e Koopmans M. Progress in understanding norovirus epidemiology. Curr Opin Infect Dis 2008;21:544e de Boer RF, Ott A, Keszty us B, Kooistra-Smid AM. Improved detection of five major gastrointestinal pathogens by use of a molecular screening approach. J Clin Microbiol 2010;48: 4140e6. 7. Liu J, Gratz J, Maro A, Kumburu H, Kibiki G, Taniuchi M, et al. Simultaneous detection of six diarrhea-causing bacterial pathogens with an in-house PCR-luminex assay. J Clin Microbiol 2012;50:98e Higgins RR, Beniprashad M, Cardona M, Masney S, Low DE, Gubbay JB. Evaluation and verification of the Seeplex Diarrhea-V ACE assay for simultaneous detection of adenovirus, rotavirus, and norovirus genogroups I and II in clinical stool specimens. J Clin Microbiol 2011;49:3154e Stark D, Al-Qassab SE, Barratt JL, Stanley K, Roberts T, Marriott D, et al. Evaluation of multiplex tandem real-time PCR for detection of Cryptosporidium spp., Dientamoeba fragilis, Entamoeba histolytica, and Giardia intestinalis in clinical stool samples. J Clin Microbiol 2011;49:257e Taniuchi M, Verweij JJ, Noor Z, Sobuz SU, Lieshout L, Petri Jr WA, et al. High throughput multiplex PCR and probebased detection with Luminex beads for seven intestinal parasites. Am J Trop Med Hyg 2011;84:332e ten Hove R, Schuurman T, Kooistra M, Möller L, van Lieshout L, Verweij JJ. Detection of diarrhoea-causing protozoa in general practice patients in the Netherlands by multiplex real-time PCR. Clin Microbiol Infect 2007;13:1001e Wolf S, Williamson WM, Hewitt J, Rivera-Aban M, Lin S, Ball A, et al. Sensitive multiplex real-time reverse transcription-pcr assay for detection of human and animal noroviruses in clinical and environmental samples. Appl Environ Microbiol 2007;73: 5464e Leblanc-Maridor M, Beaudeau F, Seegers H, Denis M, Belloc C. Rapid identification and quantification of Campylobacter coli and Campylobacter jejuni by real-time PCR in pure cultures and in complex samples. BMC Microbiol 2011;11: van Maarseveen NM, Wessels E, de Brouwer CS, Vossen AC, Claas EC. Diagnosis of viral gastroenteritis by simultaneous detection of Adenovirus group F, Astrovirus, Rotavirus group A, Norovirus genogroups I and II, and Sapovirus in two internally controlled multiplex real-time PCR assays. J Clin Virol 2010;49:205e Bessede E, Delcamp A, Sifré E, Buissonniere A, Mégraud F. New methods for detection of campylobacters in stool samples in comparison to culture. J Clin Microbiol 2011;49:941e Fredriksson-Ahomaa M, Cernela N, H achler H, Stephan R. Yersinia enterocolitica strains associated with human infections in Switzerland 2001e2010. Eur J Clin Microbiol Infect Dis 2012;31:1543e Ferguson JK, Cheng AC, Gilbert GL, Gottlieb T, Korman T, McGregor T, et al. Clostridium difficile laboratory testing in Australia and New Zealand: national survey results and Australasian Society for Infectious Diseases recommendations for best practice. Pathology 2011;43:482e Crobach MJ, Dekkers OM, Wilcox MH, Kuijper EJ. European Society of Clinical Microbiology and Infectious Diseases (ESCMID): data review and recommendations for diagnosing Clostridium difficile-infection (CDI). Clin Microbiol Infect 2009;15: 1053e Knetsch CW, Bakker D, de Boer RF, Sanders I, Hofs S, Kooistra- Smid AM, et al. Comparison of real-time PCR techniques to cytotoxigenic culture methods for diagnosing Clostridium difficile infection. J Clin Microbiol 2011;49:227e Ozaki E, Kato H, Kita H, Karasawa T, Maegawa T, Koino Y, et al. Clostridium difficile colonization in healthy adults: transient

8 Systematic multiplex PCR for gastroenteric pathogens 129 colonization and correlation with enterococcal colonization. J Med Microbiol 2004;53:167e van den Berg RJ, Kuijper EJ, van Coppenraet LE, Claas EC. Rapid diagnosis of toxinogenic Clostridium difficile in faecal samples with internally controlled real-time PCR. Clin Microbiol Infect 2006;12:184e Rousseau C, Lemée L, Le Monnier A, Poilane I, Pons JL, Collignon A. Prevalence and diversity of Clostridium difficile strains in infants. J Med Microbiol 2011;60:1112e Johnson KE, Thorpe CM, Sears CL. The emerging clinical importance of non-o157 Shiga toxin-producing Escherichia Coli. Clin Infect Dis 2006;43:1587e Bennett-Wood VR, Russell J, Bordun AM, Johnson PD, Robins- Browne RM. Detection of enterohaemorrhagic Escherichia coli in patients attending hospital in Melbourne, Australia. Pathology 2004;36:345e Grys TE, Sloan LM, Rosenblatt JE, Patel R. Rapid and sensitive detection of Shiga toxin-producing Escherichia coli from nonenriched stool specimens by real-time PCR in comparison to enzyme immunoassay and culture. J Clin Microbiol 2009;47:2008e Wilhelmi I, Roman E, Sanchez-Fauquier A. Viruses causing gastroenteritis. Clin Microbiol Infect 2003;9:247e Logan C, O Leary JJ, O Sullivan N. Real-time reverse transcription-pcr for detection of rotavirus and adenovirus as causative agents of acute viral gastroenteritis in children. J Clin Microbiol 2006;44:3189e Brink N, Rice S, Pickering C, Thurlbeck S. False positive reactions with rotavirus latex agglutination test. J Clin Pathol 1992;45: Grimwood K, Huang QS, Cohet C, Gosling IA, Hook SM, Teele DW, et al. Rotavirus hospitalisation in New Zealand children under 3 years of age. J Paediatr Child Health 2006;42:196e Coulson BS, Holmes IH. An improved enzyme-linked immunosorbent assay for the detection of rotavirus in faeces of neonates. J Virol Methods 1984;8:165e Erlandsen SL, Sherlock LA, Bemrick WJ. The detection of Giardia muris and Giardia lamblia cysts by immunofluorescence in animal tissues and fecal samples, subjected to cycles of freezing and thawing. J Parasitol 1990;76: 267e71.