Evaluation of the NG-Test MCR-1 Lateral Flow Assay and EDTA-Colistin Broth Disk Elution Methods To Detect Plasmid-Mediated Colistin Resistance among Gram-Negative Bacterial Isolates

DISCUSSION

The chronic and systematic use and misuse of antimicrobials have made multidrug-resistant bacterial infections an increasing reality for both clinicians and public health authorities. As resistance continues to become increasingly prevalent, clinical microbiology laboratories will be required to provide reliable and rapid information about a particular isolate’s susceptibility to new, rarely used, and/or last-line antimicrobial agents (23). Information about antimicrobial resistance phenotypes and their respective mechanism(s) also has significant value for public health and hospital infection control epidemiologists as they track the local, regional, and global spread of resistance, especially that conferred by plasmid-borne resistance genes, which are readily transferrable. The NG-Test MCR-1 LFA and the EDTA-CBDE methods are two recently described assays that can be used to screen Gram-negative bacterial isolates for PMCR.

In this study, we demonstrate that the NG-Test MCR-1 LFA provides rapid (15 min for the time to a result), sensitive (100%), and specific (99%) screening for the presence of MCR-1 in American bacterial isolates directly from culture. These findings are largely consistent with those described by Volland et al. (14), who helped develop the assay and who first described the performance characteristics of the assay using European isolates. They observed a sensitivity of 100% and a specificity of 98%. In contrast, we experienced a relatively large number of faintly positive T bands (i.e., false-positive results; 3.4%), resulting in an initially decreased specificity that was resolved upon repeat testing. However, it is important to note that the package insert specifically instructs the reader to consider any T band to be a positive result, regardless of intensity (19). While Volland et al. (14) did note some faintly positive results, they specifically noted that these, in fact, either were truly positive for MCR-1 or were due to cross-reactivity with MCR-2.

One possible explanation for this difference may be rooted in the fact that our inoculum volume was larger than that of Volland et al. (14). The decision to err on the side of a larger inoculum was made because we observed a faintly positive T band when testing our MCR-1-positive control isolate during the initial preexperiment walkthrough using the recommended inoculum volume of a single colony (19). We therefore opted to use a heaped 1-μl loopful as the inoculum to help ensure strongly positive T bands, which may partially explain the increased numbers of faintly positive T bands that we observed, although this would not explain why these discrepant results resolved upon repeat testing with the same inoculum volume. Another, more likely explanation for our increased number of initial false-positive T bands is that there is an inherent degree of subjectivity in interpreting the test bands, particularly in cases with faintly positive T bands.

The LFA’s cross-reactivity with MCR-2 (14, 19) is a dual-edged sword, providing additional (albeit off-label) detection coverage of additional MCR enzymes while simultaneously reducing the assay’s analytical specificity, as defined by the manufacturer. The possibility of cross-reactivity with some MCR-2 variants is noted in the NG-Test MCR-1 LFA package insert (19). No cross-reactivity with MCR-3 and MCR-4 producers was observed in this study. One major disadvantage of the LFA is that it is (nominally) MCR-1 specific, failing to detect other members of the MCR family. This potentially limits the utility of negative results in the absence of additional testing for phenotypic resistance, particularly in low-incidence settings. The necessity of taking epidemiological factors into account is illustrated by the fact that we were able to find only a single MCR-1-positive clinical isolate for our study, despite specifically evaluating many colistin-resistant isolates from 3 major U.S. academic medical centers.

The EDTA-CBDE method also provided a sensitive (100%) and specific (94.3%) screening for the presence of PMCR among Enterobacterales and P. aeruginosa isolates directly from culture, albeit with a longer time to a result than the LFA (16 to 20 h versus 15 to 20 min). However, its turnaround time (TAT) is comparable to that of other MIC-based methods that simultaneously determine in vitro antimicrobial susceptibility results and that detect PMCR by determining polymyxin MICs in the presence and absence of a chelator (24, 25). As previously noted, EDTA is not a specific inhibitor of mcr, with its inclusion resulting in a slightly decreased specificity of the assay, which is the reason why EDTA-CBDE is a screening test for which all positive results must be confirmed by a secondary method (e.g., a molecular method) (15). Furthermore, the EDTA-CBDE method is relatively easy to perform using materials that are readily available in most clinical laboratories and is capable of detecting multiple mcr variants (e.g., mcr-1 to mcr-4 were detected in this study).

Unfortunately, the CBDE and EDTA-CBDE methods are not recommended for use with Acinetobacter species due to high error rates, and large numbers of isolates are excluded from analysis due to the absence of a reference MIC (16). Recently, a resazurin-based rapid test to detect colistin resistance in Acinetobacter isolates was described and had a 93.3% sensitivity and specificity (26). This method does not distinguish between chromosome- or plasmid-mediated colistin resistance but provides an alternative to BMD to detect colistin resistance among Acinetobacter species.

There are some limitations to the EDTA-CBDE method. First, the materials may be costly relative to the cost of the materials used for other phenotypic methods, but the costs are comparable to those of the NG-Test MCR-1 LFA (estimated cost, $15 to $20). Clinical laboratories offering this testing need to have available stores of perishable, premade 10-ml CA-MHB tubes, which can cost anywhere from $1 to $5 per tube, with each tested isolate requiring 8 tubes for the combined CBDE and EDTA-CBDE methods. This excludes the need for 25-ml CA-MHB tubes, as the newly CLSI-approved colistin breakpoints were set at 2 μg/ml for intermediate and 4 μg/ml for resistant, thereby eliminating the need for testing at a concentration of 0.4 μg/ml. Alternatively, labs may choose to prepare and dispense their own 10-ml CA-MHB tubes to reduce costs. Furthermore, interpretation of broth turbidity can be subjective and subject to variability both between technologists and between bacterial isolates.

While both tests have their own individual advantages and disadvantages, there is some overlap between the NG-Test MCR-1 LFA and EDTA-CBDE in both categories (Table 4). For example, both methods are sensitive for detecting MCR-1-mediated colistin resistance. On the other hand, because EDTA-CBDE is a screening test, all positive results require subsequent confirmatory testing by a second method (e.g., mcr PCR). Additionally, neither the NG-Test MCR-1 LFA nor EDTA-CBDE provides clinically actionable results as a stand-alone test. However, the concurrent CBDE testing required for EDTA-CBDE interpretation would ultimately result in an actionable colistin MIC value, as was previously described (17). Importantly, this means that for clinical microbiology laboratories, the utility of both the NG-Test MCR-1 LFA and the EDTA-CBDE test is largely centered around epidemiological, surveillance, hospital infection control, and research purposes, as different methods (e.g., CBDE, CAT, or rBMD) will be used to determine colistin in vitro susceptibility results.

Limitations of this study include the fact (i) testing was performed at a single center, (ii) only a single manufacturer of CA-MHB and colistin disks was used to evaluate the EDTA-CBDE method, and (iii) few mcr-positive isolates were evaluated, in spite of assessing as many colistin-resistant isolates as possible from 3 U.S. academic medical centers.

Despite the rarity of PMCR among clinical isolates in the United States, these methods are valuable tools that may be implemented in public health and clinical microbiology laboratories to further discern the mechanism of resistance among colistin-resistant Gram-negative isolates and to screen for PMCR for infection prevention and control purposes.