Methods
GEL bag test, the limit is estimated to be 1.0 CFU/100 ml. For each method, a subset of presumptive ESBL-positive E. coli samples was isolated for further characterization. Overall, one to five presumptive ESBL E. coli colonies were selected from each membrane filtration plate, ESBL CBT, or GEL ESBL E. coli bag. The col- onies were re-streaked for isolation on TBX medium with 4 mg/L CTX as an initial ESBL E. coli confirmation step. Colonies were picked at random from these plates using a sterile loop. The exte- rior of positive compartments of CBTs were swabbed with 70% ethanol. Next, the com- partments were pierced with a sterile syringe and needle, and a drop of medium was spot- ted onto TBX plates with CTX and streaked as described to obtain individual colonies after incubation. For GEL ESBL E. coli positives, the exterior of the GEL bag was swabbed with 70% ethanol, and a large-gauge sterile syringe was used to pierce the GEL medium of the bag. In some instances, it was necessary to use the sterile syringe in combination with a sterile isolation needle to spread the positive colony from the GEL bag. Colonies isolated from re-streaked TBX plates were picked with a sterile loop, cul- tured initially on tryptic soy agar (TSA), cultured again overnight in tryptic soy broth (TSB), diluted 1:1 with sterile glycerol, and stored at -80 °C in sterile 2-ml cryovials. Stored isolates were thawed and further char- acterized by biochemical testing, specifically the indole test, according to the manufac- turer’s instructions. Isolates were further con- firmed as ESBL by Kirby–Bauer susceptibility testing using the criteria defined in the Tri- cycle protocol (WHO, 2021) for CTX, ceftazi- dime (CAZ), CTX + clavulanic acid (CLA), and CAZ + CLA as paper discs. The distributions of presumptive and con- firmed ESBL E. coli CFU and MPN concen- trations were characterized for each method (membrane filtration, ESBL CBT, and GEL ESBL). E. coli concentrations were subjected to Shapiro–Wilk normality tests, and the geo- metric mean, arithmetic standard deviation, and range (minimum and maximum) were calculated. A 0.5 minimum limit of detec- tion was used to reduce bias for nondetects, such that a nondetect for a 100-ml undiluted surface water sample would be calculated as 0.5/100 ml—rather than 0/100 ml—to mini-
mize bias and enable log 10 -transformation of count data where needed. The confirmed proportion of ESBL E. coli was calculated as the ratio of confirmed ESBL E. coli isolates to total isolates tested, adjusted for the number of total isolates collected from each sample type. Dierences in log 10 -trans- formed concentrations between each test method were evaluated using nonparamet- ric methods. All analyses were conducted in GraphPad Prism version 10. Results Concentrations of ESBL E. coli in surface water samples were relatively low (<100 CFU or MPN per 100 ml) throughout the study period and the total percentage of presump- tively resistant E. coli to nonresistant E. coli varied between 1.5% and 15.2%. Table 1 pres- ents the occurrence of presumptively positive ESBL E. coli by assay method, and Figure 1 displays a box and whisker plot of the pre- sumptively positive concentrations. To further evaluate the three methods used to detect ESBL E. coli , a Friedman test was used because a normal distribution did not adequately represent this data set. At an α level of .05, there was no statistically significant dif- ference between the median detected concen- tration of the Tricycle protocol membrane fil- tration method compared with the ESBL CBT ( p = .57), the membrane filtration method compared with the GEL ESBL method ( p > .99), and the ESBL CBT method compared with the GEL ESBL method ( p > .99). Isolate Analysis for ESBL E. coli Analysis was performed on 306 presump- tively positive ESBL E. coli isolates detected in the 100 surface water samples (Table 2). There were 117 ESBL E. coli isolates ana- lyzed from the membrane filtration method, as well as 91 and 98 isolates analyzed from the ESBL CBT and GEL ESBL methods, respectively. The isolates were initially confirmed by streak plating on TBX agar medium containing CTX. Of the samples first identified as presumptive ESBL E. coli , 92.8% were confirmed on the CTX agar plates. By method, 94.0%, 96.7%, and 87.8% of isolates were confirmed using this tech- nique from the membrane filtration, ESBL CBT, and GEL ESBL methods, respectively. Next, isolates were confirmed as E. coli using an indole test. Overall, 85.6% of the
Sample Collection A total of 100 samples were collected from May–September 2023 from a variety of sur- face water sources in central North Carolina. Sample sites included two reservoirs, two lakes (Sunset and Jordan), and three rivers (Cape Fear, Neuse, and Eno). Grab samples of surface waters were collected using sterile polypropylene bottles. Samples were trans- ported and stored on ice at 4 ° C and analyzed within 24–48 hr after collection. Sample Processing and Data Analysis To address anticipated low concentrations of ESBL E. coli , surface water samples were mixed well and analyzed as 100-ml volumes without dilution by all three test methods. Membrane filtration was performed by fil- tering samples through 0.45-µm cellulose nitrate filters followed by incubation at 44 ° C for 24 hr on 100-mm diameter Tryptone Bile X-glucuronide (TBX) agar plates with or without 4 mg/L added cefotaxime (CTX) per the Tricycle protocol (WHO, 2021). We tested duplicate plates for each water sample ( n = 200). Parallel 100-ml samples were analyzed using the ESBL CBT for MPN concentra- tions/100 ml and the GEL ESBL E. coli colony test for CFU concentrations/100 ml. All CBT and GEL samples were incubated at 35 °C for 24 hr. If a CBT compartment exhibited a blue-green color after incubation or if a GEL bag had a blue-green colony, that compart- ment or colony was counted as positive. Positive and negative control plates (for membrane filtration method) and bags (for CBT and GEL methods) were tested one time per week. A positive control ESBL E. coli , a non-ESBL-negative control bacteria, and a negative dilution control (phosphate buered saline [PBS]) were used for each set of experi- ments. Additionally, at the beginning of our experiment, all positive control bacteria were compared in a clean matrix (PBS) using each of the methods (Appling et al., 2023) to deter- mine if the methods were comparable to no outside water interactions. The limit of detec- tion for each method is described in the manu- facturer’s instructions (www.aquagenx.com/). For the ESBL CBT method, the limit is estimated to be 0.0 MPN, with an upper 95% confidence limit of 2.87 MPN/100 ml. For the
9
November 2024 • Journal of Environmental Health
Powered by FlippingBook