Pooled Nucleic Acid Amplification Test for Screening of Stool Specimens for Shiga Toxin-Producing Escherichia coli

Stool specimens are routinely submitted to the British Columbia Centre for Disease Control (BCCDC) Public Health Laboratory (PHL) for stx detection by quantitative PCR (qPCR). Specimens that were positive by qPCR were subsequently cultured on sorbitol MacConkey agar (SMAC); suspect O157 (colorless) or non-O157 (pink) colonies were confirmed to be Shiga toxin producing by qPCR and serotyped using the Kaufmann classification scheme. Specimens were selected for the retrospective study on the basis of previous qPCR cycle threshold (C_T) values and recovered isolate serogroups. Specifically, samples were chosen to represent a wide range of C_T and serogroup findings. Previously stx-positive specimens (n = 30) were stored at −80°C prior to testing and evenly represented stx₁, stx₂, and stx₁, and stx₂ genotypes from recovered isolates (n = 10 each) (Table 1). Previously stx-negative specimens (n = 270) were randomly chosen from fresh samples kept at 4°C for up to 3 weeks.

Overnight enrichments (200 μl or “pea” size) in 5 ml brain heart infusion (BHI) broth (Bacto, Mississauga, ON, Canada), DNA extraction, and duplex quantitative PCR (qPCR) methods for stx detection from stool were performed as previously described (20) with the following modifications: (i) PerfeCTa qPCR ToughMix (UNG, Low ROX; Quanta Biosciences, Gaithersburg, MD, USA) was used for qPCR with an ABI 7500 Fast System (Life Technologies, Grand Island, NY, USA) and an initial 5-min incubation at 45°C, and (ii) FAM/ZEN/IABkFQ-labeled stx₁ and MAX/IABkFQ-labeled stx₂ TaqMan probes (Integrated DNA Technologies, Coralville, IA, USA) were used. The limit of detection (LoD) was determined from serial dilutions of cultures of three stx₁ and three stx₂ clinical isolates spiked into an enriched stx-negative specimen performed in triplicate. Briefly, isolates grown overnight in BHI broth were serially diluted 10-fold to achieve dilutions of approximately 10⁹ to 10¹ CFU/ml. Overnight-enriched STEC-negative stool was spiked with each dilution at a 10:1 ratio, and subsequent DNA extraction and qPCR was performed as described above. In addition, 100-μl aliquots of 10⁴, 10³, and 10² dilutions were plated in duplicates for colony counts taken after 24 h at 37°C. These counts were used to calculate the average actual numbers of CFU/ml of the dilutions used. The LoD was defined as the CFU/ml concentration that was consistently detected by three technical repeats of qPCR.

Pooling of stool specimens was performed using two approaches, where pooling occurred either prior to enrichment in BHI broth or after enrichment (Fig. 1). Pea-sized samples of solid stool or 200 μl of liquid stool were used. For preenrichment pooling, nine stx-negative stools were mixed with one stx-positive stool in 5 ml BHI broth prior to overnight enrichment. Enrichment, DNA extraction, and qPCR were then performed as described above. For postenrichment pooling, 10 stool samples were enriched separately overnight in 5 ml BHI broth before 200-μl samples from each specimen were combined. Accuracy and sensitivity, as determined by the change in C_T (ΔC_T), were evaluated for the two pooling strategies relative to individual specimen testing that was performed in parallel on postenrichment specimens.

The cost of consumables was calculated on the basis of U.S. catalog pricing at the time of the study and is shown in dollars. Consumables included plastics required for extraction and qPCR procedures, primers, probes, buffers, and disinfecting agents. Cost calculations incorporated amounts of consumables required to run two positive controls and one negative (no-template) control. Labor costs were not included in these calculations because they vary greatly among institutions. Labor time for calculations included hands-on time for extraction and qPCR procedures, room preparation and sample retrieval to room cleanup, sample storage, and result analysis and reporting. Labor time excluded events with set times, such as incubations and instrument run times; these, combined with hands-on time, were captured in the reported turn-around-times.

Over 30 days during June, August, September, and October 2014, postenrichment pooling for stx detection by qPCR was performed in parallel with routine individual specimen stx qPCR at the BCCDC PHL using the same enrichment broth. Historical data indicate that peak STEC detection in British Columbia occurs from June to October. A total of 512 specimens were tested. Resolution of the positive pools was performed the following day because pooled NAATs were started in the afternoon. To accurately compare hands-on time between the individual and pooled NAAT approaches, each day, the technical staff recorded the start and end times associated with procedures that varied depending on the number of samples and/or the testing strategy used. Start time was defined as the time of room preparation for DNA extraction of overnight stool cultures. End time was defined as the time following room cleanup after the start of stx qPCR.

Differences between pooling strategies were determined by paired Student's t tests using GraphPad QuickCalcs Software (La Jolla, CA). The performances of the pooled qPCR strategies (sensitivity) were calculated in comparison to individual qPCR results.

The qPCR limits of detection for specimens tested individually were 7.23 × 10² for stx₁ and 1.03 × 10³ CFU/ml for stx₂. A total of 300 patient stool specimens (30 stx positive, 270 stx negative) were used to evaluate two different pooling strategies simultaneously as approaches for high-throughput stx detection. Pool resolution, which would normally follow positive pool test results, was not performed at this time. The overall sensitivity for positive samples was 83% for the preenrichment pooling strategy and 100% for the postenrichment pooling strategy relative to individual qPCR results (Table 2). Two stx₁, one stx₂, and two stx₁ and stx₂ samples positive by individual NAAT (10 of each were tested in total) were not detected by the preenrichment pooled NAAT. Prepooling C_T values of samples undetected in preenrichment pools ranged from 17.52 to 33.75, suggesting that factors other than the concentration of STEC present in the specimen (as reflected by the relative C_T values) were at play. Therefore, the loss of detection in the preenrichment pooling evaluation were not predicted using the stx genotype or C_T value of individually tested specimens. Furthermore, for both stx genes, the ΔC_T between individual and pooled NAAT results for specimens was significantly lower and more consistent for postenrichment pooling (overall means ± standard deviation [SD]: stx₁ = 3.90 ± 1.08, stx₂ = 4.28 ± 1.36) than those from preenrichment pooling (overall mean ± SD excluding specimens that were undetected in pools: stx₁ = 9.34 ± 5.18, stx₂ = 8.96 ± 5.83) (P ≤ 0.0013) (Table 2). On the basis of average pooling ΔC_T results, the theoretical qPCR LoDs for preenrichment pooling were 6.96 × 10⁵ for stx₁ and 8.67 × 10⁵ CFU/ml for stx₂, while the qPCR LoDs for postenrichment pooling were 1.27 × 10⁴ for stx₁ and 2.57 × 10⁴ CFU/ml for stx₂.

To assess the operational requirements of the promising postenrichment pooled NAAT, cost and workflow analyses were performed to evaluate the costs of consumables and technician hands-on time (labor). Cost of consumables and labor time for the testing of 90 specimens simultaneously were estimated to be $223.47 ($76.50 for extraction procedures and $146.97 for qPCR procedures) and 4.75 h (3.47 h for extraction procedures and 1.28 h for qPCR procedures), respectively. Major consumables cost and labor savings (48 to 81% and 6 to 66%, respectively, based on an expected 1 to 2% positivity rate) were estimated with the postenrichment pooled NAAT for the same number of specimens, which corresponds to the maximum of 9 pools of 10 samples that would fit on a 96-well plate, although savings were impacted by the expected frequency of pool resolution (Table 3). Consumables cost savings were equally associated with extraction and qPCR procedures (80 and 81%, respectively, with no pool resolution), while extraction procedures gave greater labor savings over those of qPCR procedures (71 versus 53%, respectively, with no pool resolution). The estimated cost of consumables and labor time savings were highest with the greatest number of pools tested (9) and were decreased when fewer pools were tested given the incremental loss in efficiency (examples shown in Table 3). Moreover, while resolving positive pools necessitates additional testing, the turnaround time of the pooled postenrichment NAAT remains the same (within 1 day) as that of individual testing, given the time savings associated with the pooled strategy.

Evaluation of the postenrichment pooled NAAT involved a 30-day direct comparison between the pooling strategy and the routine individual specimen testing currently performed at the BCCDC PHL. A total of 512 specimens were tested, with 9 to 27 specimens tested per day and consequently 1 (n = 1), 2 (n = 24), or 3 (n = 5) pools of 10 specimens or less per day (Table 4). An overall sensitivity of 100% was achieved once again for the postenrichment pooling strategy. Specimens positive for stx genes were identified on 7/30 days (23.3%), during which only 1 pool was positive on each day. Resolved pools contained 1 (n = 5), 2 (n = 1), or 4 (n = 1) positive specimens, for a total of 11 specimens (7 positive for stx₁, 4 positive for stx₁ and stx₂)from 8 unique patients and, therefore, a positivity rate of 3% (2% excluding repeat patient samples). Isolates were recovered from 10 of the 11 specimens and included stx₁- and stx₂-positive O157:H7 (n = 4, two from one patient) and stx₁-positive O103:H2 (n = 1), O118:H untypeable (n = 1), O126:H8 (n = 3, two from one patient), and O186:H2 (n = 1) serotypes. Consistent with retrospective evaluation findings, for both stx genes, the mean and range ΔC_T values between the routine (individual) and pooled NAAT results for specimens in pools containing one positive specimen were 4.99 (2.00 to 7.66) for stx₁ (n = 5) and 4.65 (4.41 to 4.88) for stx₂ (n = 2). For the pool with two positive specimens, the pooled NAAT stx₁ C_T was 26.96, while individual NAAT stx₁ C_T values were 19.85 and 33.36. For the pool with four positive specimens, the pooled NAAT stx₁ C_T was 23.29, and the stx₂ C_T was 21.93, while individual NAAT C_T values ranged from 19.78 to 32.05 (n = 4) and from 18.39 to 18.72 (n = 2), respectively. Interestingly, the C_T values of the individual postresolution tests were generally higher than those of the individual routine tests that were conducted prior to pooling in this study (overall means: stx₁, 27.02 versus 24.66, respectively; stx₂, 19.63 versus 18.24, respectively), resulting in ΔC_T values of only 2.35 and 4.21 between postresolution and pooled NAAT results for specimens in pools containing one positive specimen. As expected, overall labor time savings were associated with routine performance of the postenrichment pooled NAAT compared with those of the individual NAAT, even in a scenario with only 2 pools per test (Table 4).