770394421034J Food ProtJ. Food Prot.Journal of food protection0362-028X1944-909729985068619375210.4315/0362-028X.JFP-18-052HHSPA991738ArticleInterlaboratory Evaluation of the U.S. Food and Drug Administration Escherichia coli Identification Microarray for Profiling Shiga Toxin–Producing Escherichia coliPATELISHA R.1*GANGIREDLAJAYANTHI1LACHERDAVID W.1MAMMELMARK K.1BAGILORI2BARANZONIGIAN MARCO2FRATAMICOPINA M.2ROBERTSELIZABETH L.3DebROYCHITRITA3LINDSEYREBECCA L.4STONEBURGDEVON V.4MARTINHALEY4SMITHPEYTON4STROCKBINENANCY A.4ELKINSCHRISTOPHER A.4SCHEUTZFLEMMING5FENGPETER C. H.6U.S. Food and Drug Administration, Division of Molecular Biology, 8301 Muirkirk Road, Laurel, Maryland 20708, USAU.S. Department of Agriculture, Agricultural Research Service, Eastern Regional Research Center, 600 East Mermaid Lane, Wyndmoor, Pennsylvania 19038, USAEscherichia coli Reference Center, Pennsylvania State University, University Park, Pennsylvania 16802, USACenters for Disease Control and Prevention, 1600 Clifton Road, Atlanta, Georgia 30329, USAWHO Collaborating Centre for Reference and Research on Escherichia and Klebsiella, Artillerivej 5, DK-2300, Copenhagen, DenmarkU.S. Food and Drug Administration, Division of Microbiology, 5001 Campus Drive, College Park, Maryland 20740, USAAuthor for correspondence. Tel: 240-402-3594; Fax: 301-210-7976; isha.patel@fda.hhs.gov.10102018820181810201881812751282

The U.S. Food and Drug Administration Escherichia coli Identification (FDA-ECID) microarray provides rapid molecular characterization of E. coli. The effectiveness of the FDA-ECID for characterizing Shiga toxin–producing E. coli (STEC) was evaluated by three federal laboratories and one reference laboratory with a panel of 54 reference E. coli strains from the External Quality Assurance program. Strains were tested by FDA-ECID for molecular serotyping (O and H antigens), Shiga toxin subtyping, and the presence of the ehxA and eae genes for enterohemolysin and intimin, respectively. The FDA-ECID O typing was 96% reproducible among the four laboratories and 94% accurate compared with the reference External Quality Assurance data. Discrepancies were due to the absence of O41 target loci on the array and to two pairs of O types with identical target sequences. H typing was 96% reproducible and 100% accurate, with discrepancies due to two strains from one laboratory that were identified as mixed by FDA-ECID. Shiga toxin (Stx) type 1 subtyping was 100% reproducible and accurate, and Stx2 subtyping was 100% reproducible but only 64% accurate. FDA-ECID identified most Stx2 subtypes but had difficulty distinguishing among stx2a, stx2c, and stx2d genes because of close similarities of these sequences. FDA-ECID was 100% effective for detecting ehxA and eae and accurately subtyped the eae alleles. This interlaboratory study revealed that FDA-ECID for STEC characterization was highly reproducible for molecular serotyping, stx and eae subtyping, and ehxA detection. However, the array was less useful for distinguishing among the highly homologous O antigen genes and the stx2a, stx2c, and stx2d subtypes.

CharacterizationMicroarrayShiga toxin–producing Escherichia coli

Shiga toxin–producing Escherichia coli (STEC) strains are a large, complex group of bacteria with various serotypes and phenotypes. The sole characteristic for classification as STEC is the presence of the stx genes or the production of Shiga toxin (Stx), with the two main types designated Stx1 and Stx2. Stx1 has three known subtypes (1a, 1c, and 1d), and Stx2 has at least seven known subtypes (2a through 2g) (17). Some Stx subtypes are produced mostly by STEC of environmental or animal origin and do not seem to adversely affect humans (8, 12), but subtypes Stx1a, Stx2a, Stx2c, and Stx2d are most often associated with severe illnesses such as hemorrhagic colitis and hemolytic uremic syndrome (5, 16). Researchers have identified more than 470 STEC serotypes (13), and these can produce any one of the Stx subtypes alone or in combination with other subtypes. Therefore, determining the Stx subtype produced by STEC strains can be a complicated and extensive process (17). The mechanism for STEC pathogenesis is complex, and the production of Stx alone without an adherence factor is deemed insufficient to cause hemolytic uremic syndrome. The most notable STEC adherence factor is the eae-encoded intimin protein. Intimin is also a virulence factor of enteropathogenic E. coli, and more than 40 eae alleles have been reported (14). However, not all STEC possess eae. STEC strains that have caused hemolytic uremic syndrome but are eae negative, such as STEC serotypes O91:H21 and O113:H21, likely have other means of adherence that have yet to be identified (15). Many STEC strains also possess a large virulence plasmid that contains the enterohemolysin-encoding gene ehxA (2). Although ehxA can be present in non-STEC from environmental sources (1), it is commonly found in many STEC strains that have caused infections. However, the role of enterohemolysin in STEC pathogenesis remains undetermined; therefore, enterohemolysin is regarded as a putative virulence factor.

The STEC pathotype is highly complex, both phenotypically and genotypically. Characterization of an isolate often entails the use of a large panel of PCR assays to identify the various virulence genes and their alleles or subtypes. E. coli has at least 183 somatic (O) and 53 flagellar (H) antigens, so serological typing of STEC is extremely time-consuming and labor intensive. Very few laboratories in the United States can perform complete E. coli serotyping analyses, which often can take a few weeks. Because over 50% of the STEC isolates from fresh produce yield only partial serotypes or are not typeable (4), traditional serology is not a reliable technique for STEC characterization.

The U.S. Food and Drug Administration E. coli Identification (FDA-ECID) microarray, which tests for 41,932 E. coli gene targets with a single assay, has been evaluated for its potential to quickly characterize STEC isolates. The FDA-ECID microarray assay was more effective and much faster than traditional serology for identifying serotypes of STEC isolates from fresh produce (10) and was an effective tool for strain characterization because it could be used to detect the virulence genes and alleles carried by STEC isolates in foods (11). A single-laboratory validation was performed by an FDA laboratory to verify the effectiveness of the FDA-ECID assay for use in molecular serotyping (3); however, the assay has not been evaluated by other laboratories outside of the FDA. In this study, we performed an interlaboratory evaluation of the FDA-ECID microarray assay. An identical panel of 54 E. coli reference strains was sent to laboratories at the Agricultural Research Service of the U.S. Department of Agriculture (USDA), the Centers for Disease Control and Prevention (CDC), and the E. coli Reference Center at the Pennsylvania State University. The results obtained by these three laboratories were compared with our results and with the known data for the reference strains to assess the performance and reproducibility of FDA-ECID assay for characterizing STEC strains.

MATERIALS AND METHODSBacteria.

The reference strains used in this study were obtained from the WHO Collaborating Centre for Reference and Research on Escherichia and Klebsiella (Copenhagen, Denmark). The Centre conducts a yearly international external quality assurance (EQA) evaluation funded by the European Centre for Disease Prevention and Control (Solna, Sweden), where panels of 10 to 15 reference strains are sent to participants worldwide to evaluate the performance characteristics of the assays used by the respective laboratories to identify and characterize STEC strains (18). For this interlaboratory study, we used the bacterial reference panels from the 2011 to 2014 EQA programs, including nine strains used as representatives for each of the known Stx subtypes, for a total of 54 strains. The strains sent were identified by EQA numbers, but the EQA data for each strain were not provided to the participating laboratories until the end of the study.

FDA-ECID.

Each participating laboratory used the FDA-ECID microarray assay as described previously (14). The R-Bioconductor software packages affy and made4 were used to extract robust multiarray summarized probe set intensities and MAS5 calls (presence and divergence) for each scanned array-generated cel file (6, 7). The custom R-script (R Foundation for Statistical Computing, Vienna, Austria) used by all the laboratories is available upon request.

RESULTS

The FDA-ECID data obtained by the four laboratories were examined for reproducibility for O and H typing, stx gene detection and subtyping, and detection of eae and ehxA. The results were also compared with the EQA data to determine their accuracy for identifying these gene targets.

O typing.

With the FDA-ECID assay, an O type could be identified in 94% (51 of 54) of the EQA strains (Table 1). According to the EQA data, three nontypeable strains (II9, 481, and 3879) all belong to type O41, an O type not represented on the array. O typing results obtained by the four laboratories were in agreement for 96% (52 of 54) of the strains. The first discrepancy, strain 2745, was positive for both O121 and O166 according to the results obtained by the USDA but was identified by the other laboratories to be O166, consistent with the EQA data. The second discrepancy, strain D4159, was characterized as O128 by the CDC, but the other laboratories identified it as O78, consistent with the EQA data. The FDA-ECID assay also could not precisely determine the O types for strains 1258 (O124 or O164) and D3428 (O118 or O151), which the EQA data indicated were O124 and O118, respectively (Table 1).

H typing.

Three of the 54 strains were classified as H negative according to the EQA data. For these strains (481, 3393, and D3522), the FDA-ECID results from all four laboratories were in agreement and identified these types as H26, H8, and H19, respectively (Table 1). A comparison of the results obtained by the four laboratories with the EQA data revealed an agreement of 96% (49 of 51 strains). Two strains were typed by the USDA laboratory as having two H types: strain 2745 (H15 and H19) and strain 3336 (H11 and H26) (Table 1). However, the other laboratories identified strains 2745 and 3336 as H15 and H11, respectively, consistent with the EQA data. Excluding these two discrepancies, the FDA-ECID H typing results obtained by the four laboratories were the same as the EQA data for 100% (51 of 51) of the isolates tested.

<italic>stx</italic><sub>1</sub> subtyping.

The FDA-ECID data from all four laboratories identified 31 strains as negative for the Stx1 gene (stx1), in agreement with the EQA data (Table 2). The 23 stx1-positive strains possessed various known allelic subtypes (stx1a, stx1c, and stx1d), all of which were correctly identified by all four laboratories (Table 2). The USDA laboratory identified one additional strain, 3336, as carrying stx1d, but the other laboratories determined that this strain was stx1 negative, consistent with the EQA data (Table 2). Excluding the USDA result for strain 3336, which appears to be a mixed culture, the overall FDA-ECID stx1 subtyping results obtained by the four laboratories were in 100% agreement and consistent with the EQA data.

<italic>stx</italic><sub>2</sub> subtyping.

The FDA-ECID subtyping data for the Stx2 gene (stx2) obtained by all four laboratories were in agreement for 98% (53 of 54) of the strains, including 21 strains characterized as stx2 negative. The lone discrepant result was for strain D4159, which the CDC laboratory identified as stx2f positive but the other three laboratories characterized as stx2 negative, consistent with the EQA data. The FDA-ECID assay consistently made it difficult to distinguish among the stx2a, stx2c, and stx2d subtypes. These alleles have a high degree of sequence homology, especially stx2c and stx2d, which can differ by as few as two polymorphic sites (17). For strains with stx2a alone, the FDA-ECID results from all four laboratories correctly identified the subtype as stx2a (Table 2). However, for strains with stx2c and/or stx2d or with stx2a in combination with either stx2c or stx2d, the array results from all four laboratories indicated all three stx2 subtypes in these strains (Table 2). The FDA-ECID stx2 subtyping data matched that of the EQA data for 64% (21 of 33) of the isolates, with all 12 of the discrepancies due to strains with stx2c and/or stx2d.

<italic>eae</italic> and <italic>ehxA</italic>.

Analysis of the FDA-ECID eae results indicated that all four laboratories were in complete agreement on the presence or absence of this gene, with the exception of the USDA result for strain 2745 and the CDC result for strain D4159 (Table 3). The combined FDA-ECID results from the four laboratories matched the EQA data for 53 (98%) of the 54 strains. According to the EQA data, strain D3522 is expected to be eae positive, but the FDA-ECID results from all four laboratories indicated that this strain is eae negative. The EQA data indicated only the presence or absence of eae, but the FDA-ECID assay is capable of allele determination. The eae alleles identified with the FDA-ECID assay among the EQA strains are α2, β1, γ1, γ2, ε1, and ι1 (Table 3). Analysis for the presence of ehxA revealed that the FDA-ECID results from all four laboratories were in 100% agreement (Table 3). Of the 54 strains tested, 26 strains had ehxA, and this result is consistent with the EQA data.

DISCUSSION

The efficiency of the FDA-ECID microarray assay for characterizing STEC strains was evaluated by public health laboratories from three federal government agencies and the E. coli Reference Center at the Pennsylvania State University. The FDA-ECID results obtained for the analysis of 54 EQA E. coli reference strains were examined for reproducibility (data compared among the four laboratories) and accuracy (compared with the EQA data) for identifying O and H types and Shiga toxin gene subtypes and determining the presence or absence of eae and ehxA (Table 4).

Three discrepancies were found when comparing the data from the four laboratories. First, O166:H15 strain 2745 was characterized by the USDA laboratory as positive for both O166 and O121. The USDA H type results for this strain were also mixed, listing the strain as positive for H15 and H19. The observation of two O types and two H types obtained for the same sample suggests that the strain 2745 analyzed by the USDA was most likely a mixed culture with the contaminant being an O121:H19 strain, possibly strain 2266 from the same EQA reference set. For strains that have a combination of stx2a, stx2c and/or stx2d, the probe sets on the array cannot differentiate among the alleles. Because strain 2266 contains stx2a and strain 2745 has stx2d, no contamination is readily apparent. However, further evidence for contamination of the USDA strain 2745 with strain 2266 in indicated by the ehxA and eae data. For ehxA, the hybridization signal intensity for the USDA strain 2745 run was marginal such that the stringent normalization criteria used resulted in a negative result but could also be interpreted as a weak positive result under less stringent conditions. Strain 2745 is supposed to be eae negative, but in the USDA analysis, the strain was positive for ε1, which is the allele found in strain 2266. The second discrepancy, strain 3336, was analyzed by the USDA and is likely a mixture of O26:H11 (the real strain 3336) and O41:H26 (the contaminant, possibly strain 3879 from the same EQA set). As mentioned previously, O41 is not represented on the array, so a mixed culture of O26:H11 and O41:H26 would give a single O type but two H types on the array, the situation observed with the USDA results for strain 3336. Third, strain D4159 was identified by the CDC as O128:H2 with stx2f and β1 eae, but the other three laboratories identified this strain as O78:H2 and negative for both stx2 and eae, which is consistent with the EQA data. These results suggest that the results attributable to strain D4159 by the CDC analysis are actually for a different strain, most likely either strain CC3 or D3546, both of which are O128:H2 with stx2f and β1 eae.

The FDA-ECID O typing results were 100% reproducible (51 of 51 strains) when the contamination and strain mix-up issues were not included in the calculation. Compared with the EQA data, the FDA-ECID O typing results were in complete agreement for 51 of the 54 strains. Three strains in the O41 serogroup were not typeable because targets for this O type are not represented on the array. The FDA-ECID array was designed when sequence data were available for only 152 of the 183 known O types. Another apparent O typing discrepancy involved strains 1258 and D3428, which were typed by the array as O124-O164 and O118-O151, respectively. The EQA data lists strain 1258 as O124 and strain D3428 as O118. Although the array results seem to be mixed, they were not treated as discrepant. The discriminatory ability of the FDA-ECID probe sets is limited to those O antigens for which the degree of homology is below 98%. Several O types are known to possess highly similar O antigen gene clusters (9), making it difficult if not impossible to distinguish among them with this array. This is the case for both the O124-O164 and O118-O151 pairs of O types. Within each pair, the sequences of the wzx and wzy loci used as targets on the array are identical or nearly identical.

The reproducibility of FDA-ECID H typing results was determined to be 100% (52 of 52 strains), with 100% accuracy (51 of 51 strains) compared with the EQA data. The calculations for reproducibility excluded the USDA data on the mixed-culture strains 2745 and 3336. An advantage of the FDA-ECID assay over traditional serology is the ability to characterize nonmotile (H negative) strains. For example, 3 of the 54 EQA strains are classified as H negative, but the FDA-ECID assay identified the specific H type for all three strains. By omitting these outliers, both the reproducibility and accuracy of FDA-ECID assay for H typing was 100%. The FDA-ECID array contains probe sets for all 53 known E. coli H type genes, and others have reported 96% efficiency for H typing using the FDA-ECID microarray (10, 11).

The reproducibility and accuracy of FDA-ECID assay for stx1 detection were 98 and 100%, respectively. Strain 3336 was listed by the USDA laboratory as having stx1d, but this appears to be a false-positive result due to contamination because the other three laboratories and the EQA data indicated that this strain is stx1 negative. The stx1 subtyping data obtained by all four laboratories for the 23 stx1-positive strains were in 100% agreement and consistent with the EQA data, indicating that the FDA-ECID assay effectively identified all the known stx1 subtypes. These results are in agreement with that of a previous study (11), in which the FDA-ECID stx1 subtyping results were comparable to those obtained with the standard stx subtyping protocol utilizing conventional PCR (17).

The FDA-ECID stx2 subtyping results obtained by the four laboratories were in 98% agreement and were correlated 100% with the EQA data for identifying stx2b, stx2e, stx2f, and stx2g. The FDA-ECID assay was correctly and consistently identified stx2a in strains with stx2a alone. However, the array had some difficulty discriminating between stx2c and stx2d in strains that had these subtypes alone or in combination with stx2a. The inability of FDA-ECID to differentiate between stx2c and stx2d has been reported previously (11). The sequences of stx2a, stx2c, and stx2d are very similar, and stx2c and stx2d are nearly identical (17). Although the FDA-ECID assay cannot accurately resolve members of the stx2a, stx2c, and stx2d complex, it does not misclassify them as another stx2 subtype. Compared with the EQA data, the accuracy of the FDA-ECID stx2 subtyping results was 64% (21 of 33 strains), and all the discrepant results were attributable to the 12 strains with stx2c, stx2d, or combinations including these subtypes.

The FDA-ECID assay was very effective for detecting eae and ehxA, with 100% reproducibility and 98% accuracy for eae and 100% reproducibility and 100% accuracy for ehxA. This assay also identified the various eae alleles with 100% reproducibility. The accuracy of eae subtyping could not be determined because the EQA database did not include allelic data; however, the eae allele data from all four laboratories were consistent. The 98% eae accuracy was due to the results for strain D3522. According to the EQA data, D3522 is eae positive, but the FDA-ECID results from all four laboratories indicated that this strain is eae negative. This discrepancy could reflect an error in the EQA data, and the accuracy of the eae status of strain D3522 in the EQA database remains to be determined.

In conclusion, the results of this interlaboratory study indicate that the FDA-ECID microarray assay is a simple and fast alternative for characterizing selected traits of STEC strains with a high degree of reproducibility. Compared with the EQA database, the FDA-ECID results were accurate for ehxA detection, eae detection and subtyping, and stx1 detection and subtyping. The assay was also able to reliably identify the more distinct stx2 subtypes but had some difficulty with stx2a, stx2c, and stx2d. The FDA-ECID assay was highly accurate for O and H serotyping, and the pangenomic platform of the array enabled the detection of mixed strain cultures. The FDA-ECID microarray is a useful tool for stratifying strain identification and assessing the pathogenic potential of STEC and other E. coli pathotypes in food and clinical diagnostic laboratories.

REFERENCESBoczekLA, JohnsonCH, RiceEW, and KinkleBK. 2006 The widespread occurrence of the enterohemolysin gene ehlyA among environmental strains of Escherichia coli. FEMS Microbiol. Lett 254:281–284.16445757BrunderW, SchmidtH, FroschM, and KarchH. 1999 The large plasmids of Shiga-toxin–producing Escherichia coli (STEC) are highly variable genetic elements. Microbiology 145:1005–1014.10376815ChouK, PatelI, GangiredlaJ, TranN, Williams-HillD, RichterR, FengP, LampelK, and ElkinsCA. 2017 Utility of FDA ECID microarray for comprehensive molecular serotyping of Escherichia coli, P2-130 Abstr. 2017 Annu. Meet IAFP. International Association for Food Protection, Des Moines, IA.FengPCH, and ReddyS. 2013 Prevalences of Shiga toxin subtypes and selected other virulence factors among Shiga-toxigenic Escherichia coli strains isolated from fresh produce. Appl. Environ. Microbiol 79:6917–6923.23995936FriedrichAW, BorellJ, BielaszewskaM, FruthA, TschapeH, and KarchH. 2003 Shiga toxin 1c–producing Escherichia coli strains: phenotypic and genetic characterization and association with human disease. J. Clin. Microbiol 41:2448–2453.12791863GautierL, CopeL, BolstadBM, and IrizarryRA. 2004 affy—Analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20:307–315.14960456GentlemanRC, CareyVJ, BatesDM, BolstadB, DettlingM, DudoitS, EllisB, GautierL, GeY, GentryJ, HornikK, HothornT, HuberW, IacusS, IrizarryR, LeischF, LiC, MaechlerM, RossiniAJ, SawitzkiG, SmithC, SmythG, TierneyL, YangJY, and ZhangJ. 2004 Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5:R80.15461798HoferE, CernelaN, and StephanR. 2012 Shiga toxin subtypes associated with Shiga toxin–producing Escherichia coli strains isolated from red deer, roe deer, chamois, and ibex. Foodborne Pathog. Dis 9:792–795.22891940IguchiA, IyodaS, KikuchiT, OguraY, KatsuraK, OhnishiM, HayashiT, and ThomsonNR. 2015 A complete view of the genetic diversity of the Escherichia coli O-antigen biosynthesis gene cluster. DNARes 22:101–107.LacherDW, GangiredlaJ, JacksonSA, ElkinsCA, and FengPCH. 2014 Novel microarray design for molecular serotyping of Shiga toxin–producing Escherichia coli strains isolated from fresh produce. Appl. Environ. Microbiol 80:4677–682.24837388LacherDW, GangiredlaJ, PatelI, ElkinsCA, and FengPCH. 2016 Use of the Escherichia coli identification microarray for characterizing the health risks of Shiga toxin–producing Escherichia coli isolated from foods. J. Food Prot 79:1656–1662.28221838MartinA, and BeutinL. 2011 Characteristics of Shiga toxin–producing Escherichia coli from meat and milk products of different origins and association with food producing animals as main contamination sources. Int. J. Food Microbiol 146:99–104.21371769MoraA, HerrreraA, LopezC, DahbiG, MamaniR, PitaJM, AlonsoMP, LlovoJ, BernardezMI, BlancoJE, BlancoM, and BlancoJ. 2011 Characteristics of the Shiga-toxin–producing enter oaggregative Escherichia coli O104:H4 German outbreak strain and of STEC strains isolated in Spain. Int. Microbiol 14:121–141.22101411PatelIR, GangiredlaJ, LacherDW, MammelMK, JacksonSA, LampelKA, and ElkinsCA. 2016 FDA Escherichia coli identification (FDA-ECID) microarray: a pangenome molecular toolbox for serotyping, virulence profiling, molecular epidemiology, and phylogeny. Appl. Environ. Microbiol 82:3384–3394.27037122PatonAW, WoodrowMC, DoyleRM, LanserJA, and PatonJC. 1999 Molecular characterization of a Shiga toxigenic Escherichia coli O113:H21 strain lacking eae responsible for a cluster of cases of hemolytic-uremic syndrome. J. Clin. Microbiol 37:3357–3361.10488206PerssonS, OlsenKE, EthelbergS, and ScheutzF. 2007 Subtyping method for Escherichia coli Shiga toxin (verocytotoxin) 2 variants and correlations to clinical manifestations. J. Clin. Microbiol 45:2020–2024.17446326ScheutzF, TeelLD, BeutinL, PierardD, BuvensG, KarchH, MellmannA, CaprioliA, TozzoliR, MorabitoS, StrockbineNA, Melton-CelsaAR, SanchezM, PerssonS, and O’BrienAD. 2012 Multicenter evaluation of a sequence-based protocol for subtyping Shiga toxins and standardizing Stx nomenclature. J. Clin. Microbiol 50:2951–2963.22760050SchjorringS, NiskanenT, TorpdahlM, BjorkmanJT, and NielsenEM. 2016 Evaluation of molecular typing of foodborne pathogens in European reference laboratories from 2012 to 2013. Euro Surveill 21(50):pii–30429.

FDA-ECID molecular serotyping results obtained by various laboratoriesa

O type
H type
SetEQA no.EQACDCECRCFDAUSDAEQACDCECRCFDAUSDA
EQA-3AA117417417417417488888
BB2555555555577777
CC312812812812812822222
DD41771771771771772525252525
EE511111111111111188888
FF611311311311311344444
GG710310310310310322222
HH826262626261111111111
II941NTNTNTNT2626262626
JJ1015715715715715777777
EQA-44261461461461461462121212121
48141NTNTNTNT–26262626
94915715715715715777777
111111311311311311344444
121415715715715715777777
1258124124/164124/164124/164124/1643030303030
155978787878781111111111
18711771771771771772525252525
194111111111111111188888
211212812812812812822222
216026262626261111111111
22661211211211211211919191919
271010410410410410444444
2745166166166166121 & 1661515151515 & 19
296810310310310310322222
EQA-5D41411451451451451453434343434
D41421211211211211211919191919
D41491661661661661661515151515
D415191919191911414141414
D4152555555555577777
D415510310310310310322222
D415626262626261111111111
D415815715715715715777777
D41597812878787822222
D416011111111111111188888
EQA-6333626262626261111111111 & 26
3393111111111111111–8888
34311661661661661661515151515
3570636363636366666
374410410410410410477777
384415715715715715777777
387941NTNTNTNT2626262626
398715715715715715777777
39941741741741741742121212121
41311211211211211211919191919
STX referenceD265315715715715715777777
D3428118118/151118/151118/151118/1511212121212
D343115715715715715777777
D3509222222525252525
D352288888−19191919
D354612812812812812822222
D360217417417417417488888
D364813913913913913911111
D41341661661661661661515151515

EQA, External Quality Assurance program; CDC, Centers for Disease Control and Prevention; ECRC, E. coli Reference Center; FDA, U.S. Food and Drug Administration; USDA, U.S. Department of Agriculture. Differences from EQA data are bold. NT, nontypeable; −, negative.

FDA-ECID stx subtyping results obtained by various laboratoriesa

stx1
stx2
SetEQA no.EQACDCECRCFDAUSDAEQACDCECRCFDAUSDA
EQA-3AA1cccccbbbbb
BB2aaaaa−−−−−
CC3−−−−−fffff
DD4−−−−−c, dacdacdacdacd
EE5aaaaaaaaaa
FF6cccccbbbbb
GG7aaaaa−−−−−
HH8aaaaa−−−−−
II9ddddd−−−−−
JJ10−−−−−cacdacdacdacd
EQA-4426−−−−−dacdacdacdacd
481ddddd−−−−−
949−−−−−aaaaa
1111cccccbbbbb
1214−−−−−a, cacdacdacdacd
1258−−−−−−−−−−
1559−−−−−−−−−−
1871−−−−−−−−−−
1941aaaaa−−−−−
2112ccccc−−−−−
2160−−−−−aaaaa
2266−−−−−aaaaa
2710−−−−−−−−−−
2745−−−−−dacdacdacdacd
2968aaaaa−−−−−
EQA-5D4141−−−−−fffff
D4142−−−−−aaaaa
D4149−−−−−dacdacdacdacd
D4151−−−−−bbbbb
D4152aaaaa−−−−−
D4155aaaaa−−−−−
D4156aaaaa−−−−−
D4158−−−−−a, cacdacdacdacd
D4159−−−−−−f−−−
D4160aaaaa−−−−−
EQA-63336−−−−daaaaa
3393aaaaa−−−−−
3431−−−−−dacdacdacdacd
3570−−−−−fffff
3744ccccc−−−−−
3844−−−−−a, cacdacdacdacd
3879ddddd−−−−−
3987aaaaaaaaaa
3994−−−−−b, cb, acdb, acdb, acdb, acd
4131−−−−−aaaaa
STX referenceD2653aaaaaaaaaa
D3428−−−−−bbbbb
D3431−−−−−cacdacdacdacd
D3509−−−−−ggggg
D3522ddddd−−−−−
D3546−−−−−fffff
D3602cccccbbdbb
D3648−−−−−eeeee
D4134−−−−−dacdacdacdacd

EQA, External Quality Assurance program; CDC, Centers for Disease Control and Prevention; ECRC, E. coli Reference Center; FDA, U.S. Food and Drug Administration; USDA, U.S. Department of Agriculture. Differences from EQA data are bold. −, negative.

FDA-ECID eae subtyping and ehxA results obtained by various laboratoriesa

eae
ehxA
SetEQA noEQACDCECRCFDAUSDAEQACDCECRCFDAUSDA
EQA-3AA1−−−−−−−−−−
BB2+γ1γ1γ1γ1−−−−−
CC3+β1β1β1β1−−−−−
DD4+β1β1β1β1+++++
EE5+γ2γ2γ2γ2+++++
FF6−−−−−+++++
GG7+ε1ε1ε1ε1+++++
HH8+β1β1β1β1−−−−−
II9−−−−−−−−−−
JJ10+γ1γ1γ1γ1+++++
EQA-4426−−−−−−−−−−
481−−−−−−−−−−
949+γ1γ1γ1γ1+++++
1111−−−−−+++++
1214+γ1γ1γ1γ1+++++
1258−−−−−−−−−−
1559−−−−−−−−−−
1871+β1β1β1β1+++++
1941+γ2γ2γ2γ2+++++
2112−−−−−−−−−−
2160+β1β1β1β1+++++
2266+ε1ε1ε1ε1+++++
2710−−−−−−−−−−
2745−−−−ε1−−−−−
2968+ε1ε1ε1ε1+++++
EQA-5D4141+ι1ι1ι1ι1−−−−−
D4142+ε1ε1ε1ε1+++++
D4149−−−−−−−−−−
D4151−−−−−+++++
D4152+γ1γ1γ1γ1−−−−−
D4155+ε1ε1ε1ε1+++++
D4156+β1β1β1β1−−−−−
D4158+γ1γ1γ1γ1+++++
D4159−β1−−−−−−−−
D4160+γ2γ2γ2γ2+++++
EQA-63336+β1β1β1β1+++++
3393+γ2γ2γ2γ2+++++
3431−−−−−−−−−−
3570+α2α2α2α2−−−−−
3744−−−−−−−−−−
3844+γ1γ1γ1γ1+++++
3879−−−−−−−−−−
3987+γ1γ1γ1γ1+++++
3994−−−−−−−−−−
4131+ε1ε1ε1ε1+++++
STX referenceD2653+γ1γ1γ1γ1+++++
D3428−−−−−−−−−−
D3431+γ1γ1γ1γ1+++++
D3509−−−−−+++++
D3522+−−−−−−−−−
D3546+β1β1β1β1−−−−−
D3602−−−−−−−−−−
D3648−−−−−−−−−−
D4134−−−−−−−−−−

EQA, External Quality Assurance program; CDC, Centers for Disease Control and Prevention; ECRC, E. coli Reference Center; FDA, U.S. Food and Drug Administration; USDA, U.S. Department of Agriculture; −, negative; +, positive.

Summary of FDA-ECID assay reproducibility among laboratories and accuracy versus EQA data

LocusAssay% reproducibility (no. of strains/total strains)% accuracy (no. of strains/total strains)
OSerotyping96 (52/54)94 (51/54)
HSerotyping96 (52/54)100 (51/51)
stx1Detection98 (53/54)100 (54/54)
Subtyping100 (23/23)100 (23/23)
stx2Detection98 (53/54)100 (54/54)
Subtyping100 (33/33)64 (21/33)
eaeDetection98 (53/54)98 (53/54)
Subtyping100 (30/30)NDa
ehxADetection100 (54/54)100 (54/54)

ND, not determined; the EQA database did not contain eae allelic information, so accuracy could not be determined.