Current and emerging polymyxin resistance diagnostics: a systematic review of established and novel detection methods

Evidence before this review

Methods used to detect polymyxin resistance have been reviewed ^{1, 12, 23}. Bardet and Rolain (2018) narratively described methods used to detect colistin resistance, focusing mainly on their efficiency to detect all mechanisms of colistin resistance. They also analysed methods specifically used to detect plasmid-mediated colistin resistance. Osei Sekyere (2019) provided a comprehensive description of polymyxin resistance and mcr-detecting diagnostic methods up to 2018. The review included the composition of culture media, primers, and cycling conditions of PCR methods. Osei Sekyere (2019) summarized the sensitivities, specificities, turnaround time (TAT), skill, relative cost, essential agreement (EA), categorical agreement (CA), major error (ME) and very major error (VME) of polymyxin-resistance detection methods. Since 2018, new evaluation studies have been reported, broadening our understanding and conclusions of the best colistin resistance diagnostics/methods. Therefore, we provide a comprehensive update and an expanded review, based on broader evaluation studies, of all the current diagnostic methods designed to detect colistin resistance.

Literature search strategy

A comprehensive literature search was performed using Pubmed, Web of Science, and ScienceDirect. Articles published in English, from January 2016 to September 2020, were retrieved and screened using the following keywords: “colistin AND resistan*”, “polymyxin AND resistan*” in permutation and combination with “detection” and “diagnostics”, in a factorial order. The search was based on articles that were evaluating methods that are currently used for the detection of colistin resistance and mcr-genes. Studies based on epidemiology, risk factors, surveillance, non-English language articles, other reviews, case reports, or case studies were excluded. The inclusion and exclusion methods used in this review are demonstrated in Figure 1. The following data was extracted from the included articles and summarized in Table 1: Diagnostic methods used, types and sample size (in numbers) of bacterial species used in the evaluation, sensitivity, specificity, EA, CA, ME, VME, relative cost and TAT.

• Download figure
• Open in new tab

Figure 1. Literature search strategy and sorting techniques used to obtain articles for inclusion in the systematic review.

Broth Microdilution (BMD)

The Clinical and Laboratory Standard Institute (CLSI) and European Committee on Antimicrobial Susceptibility Testing (EUCAST) have jointly recommended that the MIC (minimum inhibitory concentration) testing of colistin be performed according to the ISO-20776 standard BMD method ^{25, 26}. Methods such as agar dilution, disc diffusion and gradient diffusion have been ruled out as it was shown that the large molecular size of colistin makes it poorly diffusible through agar ^{1, 26, 27}.

It has been recommended that BMD be used with plain polystyrene trays and colistin sulfate salt without the addition of any surfactants ^{26, 28}. The CLSI had initially recommended the addition of polysorbate-80 (P-80) to alleviate the binding of polymyxin to negatively charged polystyrene surfaces (binding of colistin to polystyrene reduces the concentration of colistin)¹. However, there were concerns that the surfactant would act in a synergistic manner with colistin, and is therefore not recommended at this time ²⁸. Moreover, several studies have suggested that the loss of colistin concentration could be reduced by minimizing contact with unexposed pipette tips and by storing colistin solution in glass tubes ²⁸.

The CLSI and EUCAST have established a colistin susceptible breakpoint of ≤ 2 mg/L and a resistant breakpoint of >2 mg/L for Acinetobacter spp. and Pseudomonas aeruginosa ²⁵. However, only the EUCAST has the same breakpoints for Enterobacterales whilst the CLSI has an epidemiological cut-off value of 2 mg/L that defines Escherichia coli, Klebsiella pneumoniae, Raoultella ornithiolytica, Enterobacter aerogenes and Enterobacter cloacae as either wild type or non-wildtype ^{1, 25}.

BMD performed by the reference ISO-20776 method is currently the only gold standard for determining colistin MIC and evaluating CA, EA, ME and VME; yet it is laborious and rarely performed in routine clinical microbiology laboratories ¹. Instead, diffusion methods and automated AST systems are more commonly used ¹. Chew et al (2017) evaluated the detection of mcr-1 positive Enterobacterales by BMD in comparison with commercial automated AST systems viz., Sensititre, Microscan and Vitek 2, and a gradient diffusion test i.e., E-test. This study found that none of the commercial testing methods meet the CLSI recommendation standard for commercial antimicrobial susceptibility testing systems: EA ≥90%, CA ≥90%, VME ≤1.5%, and ME ≤ 3.0%. Even so, the Sensititre and MicroScan were shown to detect 100% of the mcr-1 positive isolates (Refer to Automated AST systems below) ²⁵. The BMD’s overall sensitivity could be improved by reducing the susceptible breakpoint to ≤1mg/L and using microtitre plates that were manufactured to reduce adsorption ^{25, 28}.

The methodological challenges surrounding the standard BMD have led to interest in finding alternative means for detecting polymyxin resistance in Gram-negative bacteria^{1, 28}. Broth macrodilution method has been shown to be an obvious alternative, as it employs the use of glass tubes instead of polystyrene ²⁸. Notwithstanding, the broth macrodilution method is not a popular alternative because it requires the same preparation and TAT as the standard BMD. Hence, commercially available selective media and rapid colorimetric assays have become popular for screening (Table 1) ²⁸.

Diffusion and Agar dilution methods

Diffusion methods for colistin susceptibility testing are still commonly used despite being disapproved by the CLSI and EUCAST ^{1, 27, 29, 30}. Two of the commercial gradient tests, E-test and MIC Test Strip, performed poorly for colistin-resistant isolates in one study ³⁰. Disc diffusion is not an MIC determiner, although it perform similarly to gradient diffusion tests, generating high levels of VMEs (false-susceptible) whilst agar dilution has a tendency of generating higher MICs (which results in high MEs) than the reference method ^{1, 25, 27, 28, 30-32}. These findings further support the conclusion that polymyxins are poorly diffused in agar and therefore corroborates the CLSI and EUCAST recommendations to abandon diffusion and agar dilution methods ^{27, 30, 32}

Manual commercial MIC-determiners: UMIC, MMS, and ComASP Colistin

UMIC (Biocentric, Bandol, France), MICRONAUT MIC-Strip (MMS) (MERLIN Diagnostika GmbH, Bornheim, Germany), ComASP Colistin (formerly SensiTest™, Liofilchem, Italy) are non-automated broth microdilution-based tests ³⁰. UMIC and MMS both consist of a plastic device with 12 wells that allow for different colistin concentrations to be tested for a single isolate without the need for any additional equipment ^{26, 30, 33}. Whilst the ComASP (SensiTest) Colistin, consists of a compact panel for four isolates with freeze-dried antibiotics in seven two-fold dilutions ^{5, 34}. Matuschek et al (2018) evaluated all three tests, where ComASP had a poor EA for Acinetobacter spp., and UMIC was poor for Acinetobacter spp. and P. aeruginosa. The overall performance of ComASP was improved (as shown in Table 1) when certain species were removed and K. pneumoniae and E. coli were tested, suggesting that ComASP is not suitable for testing all species ⁵.

UMIC was generally found to lower the MIC of some colistin-resistant isolates, which may result in failure to detect colistin-resistant isolates with low MICs (≤8 mg/L) ^{26, 33}. More so, UMIC failed to detect four Stenotrophomonas maltophilia isolates with MICs ranging from 8mg/L to 32mg/L in one study, although in another study all S. maltophilia isolates were detected accordingly ^{26, 33}. The MMS had the best correlation to the BMD amongst the three tests; however, there aren’t sufficient studies evaluating this test (Table 1)³⁰.

Automated AST systems: BD Phoenix, MICRONAUT-S, MicroScan, Sensititre and Vitek 2

Automated antimicrobial susceptibility testing (AST) systems are of particular interest for colistin susceptibility testing due to their ease of use than the reference BMD ^{26, 30}. Several studies have shown that some automated AST systems can achieve results that are relatively similar to those of the reference BMD ^{25, 26, 30, 32, 35}. The Sensititre recorded a high rate of agreement with the reference BMD in several studies, particularly demonstrating the highest potential for detecting mcr-1-producing Enterobacterales together with MicroScan in one study ^{25, 26, 30}. However, MicroScan has a tendency of overestimating MICs (which may result in false-resistant isolates) of E. cloacae, Salmonella spp. and non-fermenters (A. baumannii, P. aeruginosa and S. maltophilia) ^{5, 26}. Vitek 2 had the highest rate of VMEs at 36% for mcr-1-producing isolates and failed to detect colistin resistance in eight mcr-negative E. cloacae complex isolates ²⁵. However, another study found that Vitek 2 had VMEs of 2.04% for K. pneumoniae isolates that were mcr-negative ³⁶. MICRONAUT-S (MERLIN Diagnostika GmbH, Bornheim, Germany) performed similarly to the Sensititre although it had two VMEs whereas the Sensititre had none ³⁰. The Sensititre test is unique as it can be performed manually or semi-automated unlike MicroScan, MICRONAUT-S, Vitek 2 and BD Phoenix that require an automated inoculation delivery system ^{26, 30}.

The detection of colistin resistance by BD Phoenix highly agreed with that of the reference BMD method ^{32, 35}. Nevertheless, the BD Phoenix 100 system failed to detect colistin-resistant isolates that may have hetero-resistance; the BD Phoenix M50 had 5.7% MEs, all of which were within twofold dilutions, as compared to the reference BMD ^{35, 37, 38}. Notably, the BD Phoenix M50 has not been evaluated on strains that exhibit heteroresistance; hence, its evaluation is limited compared to BD Phoenix 100 (Table 1) ³⁵.

Chelator-based and non-chelator-based tests

Lateral Flow-Immunoassay

Monoclonal antibodies (MA) were used to develop lateral flow assays to detect MCR-1-producing bacterial isolates ⁶². In this study, 177 and 121 Enterobacterales isolates from human and animal samples were obtained, respectively ⁶². Bacterial colonies were isolated from agar plates, suspended in extraction buffer and dispensed on the MA-containing cassette where they were allowed to migrate for 15 minutes ⁶². All MCR-1-producing isolates were detected accordingly as shown by a pink band on the test line and control line of the assays. Furthermore, this test was able to detect MCR-2-producing isolates ⁶². The same assay is currently marketed as the NG-Test MCR-1 by NG Biotech in France ⁵⁸. Initially, the evaluation of the NG-Test MCR-1 resulted in eight false-positive results that were ultimately resolved to negatives apart from one isolate that was found to be an MCR-2 producer ⁵⁸. Whilst the detection of the MCR-2 product by the NG Test MCR-1 further confirms the results of the developers, cross-reactivity with MCR-2 limits the accuracy of the assay for the MCR-1 producers. ⁵⁸. Nonetheless, the lateral flow immunoassays were found to be highly sensitive, easy to use and cost-effective for detecting MCR-1/-2 ^{58, 62}.

Agar-based screening medium

Amplidiag Carba-R+ MCR assay

The Amplidiag Carba-R+ MCR assay is a multiplex nucleic acid-based test developed for detecting carbapenemase and mcr-1/-2 genes from rectal swabs and bacterial culture ⁷³. The assay was performed retrospectively and prospectively on 215 Gram-negative bacilli and 51 Enterobacterales isolates, respectively ⁷³. The Amplidiag Carba-R+MCR assay did not detect one GES carbapenemase-producing P. aeruginosa. However, all mcr-1 and mcr-2-producing isolates were accurately detected (Table 1)⁷³.

Additionally, this assay was performed on DNA extracted from 100 rectal swabs, including 40 carbapenemase-positive samples: the sensitivity was 92.5% ⁷³. Two NDM and one OXA-48 producers were not detected due to a low concentration of bacteria; therefore, the samples were subjected to an overnight enrichment in brain heart infusion with 0.5 µg/mL ertapenem ⁷³. The enrichment step allowed for the detection of two of the three samples that were not previously detected. One of the samples was an OXA-48 producing E. cloacae with an AcOXA (Acinetobacter oxacillinases with carbapenemase activity) gene ⁷³. The overall performance of this assay was acceptable, demonstrating sensitivity and specificity ranging from 92-100% and 86-100% respectively ⁷³. Moreover, the assay can be performed on cultured bacteria as well as DNA extracted from rectal swabs in no more than three hours (Table 1)⁷³.

AusDiagnostic MT CRE EU assay

AusDiagnostics MT CRE EU assay is a two-step nested multiplex-tandem PCR (MT-PCR) assay by AusDiagnostics (Chesham, UK) ⁷⁴. One study evaluated the AusDiagnostics MT CRE EU assay for detecting carbapenemase, mcr-1, and mcr-2 genes ⁷⁴. A collection of Enterobacterales, Pseudomonas spp. and Acinetobacter spp., including carbapenemase or mcr-1/-2 producers, were used to evaluate the performance of this assay ⁷⁴. The assay was performed by suspending two to three bacterial colonies grown overnight on Columbia blood or cystine lactose electrolyte agar in tubes with a sample buffer ⁷⁴. The tubes were loaded onto the AusDiagnostics MT processor platform for template extraction and the first round of PCR ⁷⁴. Lastly, the nested RT-PCR was performed by loading a 384-PCR plate containing the reaction mix onto the AusDiagnostic MT analyser ⁷⁴. The results were automatically read using the AusDiagnostics MT assay software. The assay failed to detect four out of the 22 mcr genes; however, the mcr genes were also not detected by the reference PCR ⁷⁴.

Retrospective and prospective evaluation of the assay resulted in eight and 18 false-positive results, respectively ⁷⁴. An overall sensitivity and specificity of 95.5% and 99.8%, respectively, were obtained, which improved to 100% following repeats of the assay ⁷⁴. The AusDagnostic MT CRE EU assay detected mcr-1/-2 genes as well as carbapenemase genes with minimal hands-on time (Table 1)⁷⁴.

Loop-mediated isothermal amplification (LAMP)

Loop-mediated isothermal amplification (LAMP) is a nucleic acid amplification method that allows autocycling strand displacement DNA synthesis at constant temperature using Bst DNA polymerase ⁷⁵. The use of LAMP for detecting mcr-1 gene was first described by two studies ^{76, 77}. Zou et al (2017) established a LAMP assay for to detect mcr-1 gene from cultured bacteria and spiked human stools. In this study, the LAMP assay was performed in 25 µL reaction mixtures that contained 20mM Tris-HCl, 10mM KCl, 10mM (NH4)₂ SO₄, 8mM MgSO₄, 0.8 M betaine, 0.1% Tween-20, 1.4 mM of each DNTP and 8U Bst DNA polymerase ⁷⁷. Each reaction mixture had specified amounts of forward and backward inner primers, outer forward and backward primers, loop primers as well as the appropriate amount of DNA template ⁷⁷. The mixture was covered with 25µL wax and incubated in a dry bath incubator for 60 minutes at constant temperature, the amplification products were read visually and by turbidimetry ⁷⁷. Visual detection was by colour change of a fluorescent metal indicator, a positive reaction was demonstrated by the formation of a magnesium pyrophosphate precipitate, which changed the reaction mixture from orange to green during amplification ⁷⁷.

Imirzalioglu et al (2017) evaluated the eazyplex SuperBug mcr-1 kit (Amplex Biosystems GmbH, Giessen, Germany) for the rapid detection of mcr-1 gene. Colistin-resistant isolates were grown on LB media containing colistin sulfate to prevent the loss of mcr-1 encoding plasmid, whilst colistin-susceptible isolates were grown in the absence of colistin ⁷⁶. The mcr-1-detecting LAMP assays were found to detect mcr-1 genes accurately, and the in-house LAMP assay was stated to be more sensitive than conventional PCR assays (Table 1) ^{76, 77}.

However, mcr-1 detecting LAMP assays cannot detect other potential target genes ^75-77. Therefore, Zhong et al (2019) developed a restriction endonuclease-based multi-LAMP for the detection of multiple mcr-genes. Two separate LAMP systems were established, a double-LAMP (mcr-2 and mcr-5) and triple-LAMP system (mcr-1, mcr-3 and mcr-4) which were performed in 25 µL reaction mixtures ⁷⁵. The 25 µL reaction mixture consisted of 12.5 µL LAMP-reaction mix, 1 µL Bst 2.0 polymerase, 1.25 µL primer mix, 8.25 µL nuclease-free water, and 2 µL DNA lysate ⁷⁵. Amplification products were detected visually by change in colour of SYBR Green I, which changed from yellow to orange for a positive reaction ⁷⁵. Amplification products were also stained with GoldView TM and analysed by electrophoresis on 2% agarose gel ⁷⁷. Multiplex detection of mcr-1 to mcr-5 genes was established through restriction digestion of the LAMP products based on band numbers and fragment lengths using Hind restriction enzyme ⁷⁵.

LAMP assays have advantages over conventional PCR in that LAMP is more sensitive, has a shorter processing time of <60 minutes, is relatively easier to run, and multiplex detections can be conducted in the same detection system (Table 1) ^75-77.

Matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF MS)

MALDI-TOF MS is a technique that is based on the production of mass spectra from whole cells and their contrast to a reference spectrum ⁷⁸. This method is widely used for species identification of pathogens in clinical microbiology laboratories. Giordano and Barnini were the first to evaluate the possibility of detecting colistin resistance using MADI-TOF MS ⁷⁸. In this study, 139 K. pneumoniae isolates from clinical samples were collected, from which protein was extracted for identification using MALDI-TOF MS (Bruker Daltonik GmbH, Bremen, Germany)⁷⁸. The MICs of the isolates were determined by Sensititre using a range of antibiotics including colistin ⁷⁸. First, a training set for mass peak analysis was established using 50/139 of the K. pneumoniae isolates from which 400 spectra were obtained and used for database entry (Main Spectrum Profile) as well as to classify algorithm models ⁷⁸. Finally, the remaining 89/139 isolates were used to conduct the test; 712 spectra were collected from this set ⁷⁸. However, from the 712 spectra, 158 were excluded as they constituted flat-line spectra or outliers, demonstrating identification score below 2.3 ⁷⁸. Based on the mass signals and intensities of the bacterial protein samples, two-dimensional peak distribution classified the training set spectra into two main groups viz., colR-KPn (colistin-resistant K. pneumoniae) and colS-KPn (colistin-susceptible K. pneumoniae) isolates ⁷⁸.

The newly created database entry consisted of using MALDI Biotyper RTC and MALDI Biotyper v3.0 to identify K. pneumoniae isolates and for the automatic detection of colistin resistance, respectively ⁷⁸. The automatic classification of the test set resulted in the correct classification of 71% colR-KPn and 40% colS-KPn ⁷⁸. Furthermore, different algorithm models were tested using ClinProTools v3.0 (Bruker Daltonics). The three algorithms tested included the Genetic Algorithm (GA), Supervised Neural Network (SNN) and Quick Classifier (QC)⁷⁸. However, the tested algorithms either had good recognition capability and cross validation but poor classification of colistin resistance or poor recognition capability and acceptable classification of colistin resistance ⁷⁸. The GA seemed more promising as it was better suited for biological samples ⁷⁸.Therefore, a reliable classification model was created by combining the most relevant peaks detected from the GA algorithm⁷⁸. The resulting peak combination of 4507.28/5142.84 Da from GA demonstrated a sensitivity and specificity of 78% and 89%, respectively ⁷⁸.

Three studies have evaluated MALDIxin, a MALDI-TOF-based assay, ^{7, 22, 79}. The MALDIxin test was developed to detect pETN modification in lipid A directly from bacterial colonies in less than 15 min ^{22, 79}. Dortet et al (2018) evaluated MALDIxin on A. baumannii isolates, where the mass spectrum in colistin-susceptible isolates was characterised by two sets of peaks at the centre of m/z 1728.1 and m/z 1910.3. The peaks were assigned to bis-phosphorylated hexa-acyl and bis-phosphorylated hepta-acyl lipid A that had 12 to 14 carbons making up the acyl chain, respectively ⁷⁹. The mass spectrum in colistin-resistant isolates was observed by two sets of peaks at the centre of m/z 1935.3 and m/z 2033.3, showing m/z+25 and m/z+123 shifts of mass unit of the bis-phosphorylated hepta-acyl lipid A at m/z 1910.3 ⁷⁹. The peaks observed at m/z 2033.3 and m/z 1935.3 were assigned to pETN-modified-bis-phosphorylated hepta-acyl and pETN-modified-mono-phosphorylated hepta-acyl lipid A, respectively with an acyl chain of 12 carbons in length ⁷⁹. The peaks (m/z 2033.3 and m/z 1935.3) associated with pETN-modified lipid A were observed in all colistin-resistant isolates and were not observed in any of the colistin-susceptible isolates ⁷⁹.

Furniss et al (2019) and Dortet et al (2020) described the optimization of the MALDIxin test for detecting colistin resistance in clinical E. coli and K. pneumoniae isolates, respectively. Furniss et al (2019) optimized the MALDIxin test by adopting the low-resolution linear mode used by the MALDI Biotyper Sirius system. The optimization was achieved by adding a mild-acid hydrolysis step, which is required for analysis of clinical isolates in negative ion mode ⁷. The mild-acid hydrolysis step was performed by resuspending a single bacterial colony grown on MH agar for 18-24h in 200µL distilled water ^{7, 22}, after which 50-100 µL of 2% acetic acid was added to double-distilled water containing bacterial suspension and heated for 5-15 min at 98-100□^{7, 22}. For MALDI-TOF analysis, Furniss et al (2019) used a MALDI Biotyper Sirius system, whereas Dortet et al (2020) used a 4800 Proteonic Analyzer. The optimization of the MALDIxin allowed for the identification of L-Ara4N-and pETN-modified lipid A in E. coli and K. pneumoniae isolates. Moreover, the optimized methods were able to distinguish between chromosome-encoded and MCR-mediated colistin resistance ^{7, 22}.

Microarray

A commercial CT103XL microarray system that allows for the simultaneous detection of mcr-1/-2 and β-lactamase genes was evaluated ⁸⁰. The study was conducted on 106 Enterobacterales isolates including mcr-1 and mcr-2 positive strains, as well as carbapenemase and extended-spectrum β-lactamase (ESBL)-producing strains ⁸⁰. The CT103XL microarray, which uses a multiplex ligation detection reaction, was performed following bacterial DNA extraction from bacterial cultures ⁸⁰. The commercial CT103XL microarray was confirmed to simultaneously detect mcr-1/-2 and β-lactamase genes with accuracy, although it failed to detect mcr-3, which shares 45% and 47% identity to mcr-1 and mcr-2, respectively ⁸⁰.

Multiple Cross Displacement Amplification coupled with Gold Nanoparticles-Based lateral Flow Biosensor

A multiple cross displacement amplification (MCDA) method, coupled with gold nanoparticles-based lateral flow biosensor (LFB) assay for detecting mcr-1 gene was developed ⁸¹. The MCDA reaction was performed on extracted DNA from 59 bacterial isolates, where each 25 µL reaction consisted of 12.5 µL reaction buffer, 1µL Bst DNA polymerase 2.0, 1µL colorimetric indicator, 1.6 µM of each cross primers, 0.4 µM of each displacement primers, 0.4 µM amplification primers and 1 µL DNA template ⁸¹. The MCDA reaction systems were then subjected to isothermal temperature (63□) for 40min, after which the amplification products were analysed using 2% agarose gel electrophoresis, colorimetric indicator and LFB ⁸¹. For mcr-1 detection by LFB, 0.2 µL of the amplicons was added to the well of the sample pad, followed by the addition of three drops of running buffer (1% Tween 20 and 0.01 mol/L phosphate-buffered saline) ⁸¹. The results were visually read after 1-2 min; a positive result was demonstrated by two red bands, one at the test-line and the other at the control line ⁸¹. The results were positive for all mcr-1 positive isolates and negative for all non-mcr-1 isolates ⁸¹.

The MCDA-LFB assay was further applied to stool samples spiked with 100 µL dilutions of bacterial strains ⁸¹. The resulting detection limit was 600fg of mcr-1 plasmid DNA per microliter in bacterial culture and 4.5 10³ CFU/mL in the spiked faecal samples ⁸¹. The MCDA-LFB has demonstrated the same sensitivity as the mcr-1 LAMP, which is more sensitive than the conventional PCR. Further, the MCDA-LFB has demonstrated a shorter reaction time ^{77, 81}.

Recombinase polymerase amplification (RPA)

The rapid detection of mcr-1, using a RPA, has been described ⁸². RPA is a novel isothermal amplification method, which can be performed in no more than 30 minutes at body temperature without the need for thermal cycling instruments ⁸². This study used basic RPA (B-RPA) and RPA with lateral flow (LF-RPA) on 23 genomic DNA extracted from 20 mcr-1 positive and three mcr-1-negative Enterobacterales ⁸². The primers for the B-RPA assay were designed by Clone Manager 8 and validated on three mcr-1 positive and one mcr-1 negative DNA samples ⁸². The B-RPA was based on the TwistAmp Basic kit reaction system, which was incubated at room temperature for 30 min, after which the amplicons were extracted by phenol/chloroform solution or purified using an amplicon purification kit ⁸². The LF-RPA reaction required primers and a probe, which were labelled with biotin and fluorescence ⁸². The LF-RPA was based on TwistAmp Nfo kit reaction system, which was incubated as described for the B-RPA ⁸². The amplification products for the LF-RPA were diluted at 1:50 with running buffer, after which a downstream operation was carried out ⁸².

The results for the B-RPA assay were read by agarose gel electrophoresis, whereas the results for the LF-RPA assay were visually read using Hybridetect 2T dipsticks ⁸². A positive mcr-1 detection by LF-RPA was demonstrated by two purple bands at the test line and the quality control line ⁸². Both assays detected the mcr-1 positive and mcr-1 negative DNA samples accordingly; therefore, both assays are equally suitable for detecting mcr-1 genes ⁸².

Conventional & real-time PCR and Whole genome sequencing

The presence or absence of mcr-genes is determinable by PCR assays and whole-genome sequencing (WGS) as standard ⁷⁶. Whilst WGS is able to characterize the mechanism of resistance and determine the molecular evolutionary trajectory of colistin-resistant isolates, PCR is only able to characterize resistance genes ^{83, 84} However, WGS technology is limited in settings that lack adequate resources and therefore PCR assays are widely adopted for detecting mcr-genes ^{75, 85}.

Nijhuis et al (2016) were the first to design a real-time PCR assay for detecting mcr-1 from clinical isolates using self-designed primers and probes. The assay was validated on 26 mcr-1 positive E. coli isolates, where the presence of mcr-1 was detected in all 26 isolates ⁸⁶. Additionally, the assay was evaluated on spiked stool samples and the efficiency of the PCR was 102.6% and the LOD was 3-30 cfu/reaction ⁸⁶. However, mcr-1 genes were not detected in other colistin-resistant strains i.e., Klebsiella, Enterobacter, Pseudomonas, Acinetobacter etc.

A multiplex PCR (M-PCR) assay for the simultaneous detection of mcr-1 and carbapenem-resistant genes, bla_KPC, bla_NDM, bla_IMP, bla_{OXA−48−like}, was described ⁸⁷. The assay was validated on reference strains including E. coli A434-59, which contains mcr-1 and bla_NDM−1⁸⁷. Evaluation of the M-PCR on 127 carbapenem-resistant, eight mcr-1-positive and 62 carbapenem-susceptible Enterobacterales found the assay to be 100% sensitive and specific ⁸⁷.

Additionally, three studies designed M-PCR assays to detect mcr-1 to mcr-5 genes ^{4, 85, 88}. The assay designed by Rebelo et al (2018) allowed for the simultaneous detection of mcr genes and their variants in bovine and porcine isolates ⁸⁵. This study did not use internal amplification controls as they were incompatible with DreamTaq Green PCR Master Mix ⁸⁵. The master mix contains DNA polymerase synthesized in E. coli and thus would produce amplicons if 16S rRNA primers are used ⁸⁵.

However, Lescat et al (2018) designed a more rapid (<2 hours) M-PCR assay that was compatible with internal controls. Recently, Joussett et al (2019) designed and evaluated an M-PCR assay on 50 E. coli, 41 K. pneumoniae and 12 Salmonella enterica isolates (from which a total of 40 were MCR-producers), which was 100% accurate in detecting mcr-positive isolates. The assay was additionally performed on 82 Aeromonas spp. and 10 Shewanella spp. that were previously described as potential originators of mcr-3 and mcr-4, respectively ⁴. None of the Aeromonas spp. were mcr-positive, although two Shewanella spp., S. bicestrii JAB-1 strain and S. woody S539 with MICs of 0.25mg/L and <0.12 mg/L respectively, were mcr-4 positive ⁴. However, cloning S. bicestrii JAB-1 genes into E. coli TOP10 resulted in an mcr-4 positive outcome by the PCR assay with a colistin MIC of 4 mg/L ⁴.

Borowiak et al (2020) described the detection of mcr-1 to mcr-9 in colistin resistant Salmonella enterica isolates using an M-PCR (mcr-1 to mcr-5) designed by Rebello et al (2018) and a newly designed M-PCR assay (mcr-6 to mcr-9). The assay was performed on 407 colistin-resistant S. enterica isolates from animals, animal feed, food and the environment ⁸⁹. Two hundred and fifty-four of the isolates had mcr genes. Moreover, the assay detected mcr-9 in isolates carrying mcr-1 ⁸⁹. However, two separate frameshift mutations of mcr-9 were shown to have occurred in the respective isolates as demonstrated by WGS analysis; the mutations are believed to have contributed to non-functional MCR-9 proteins ⁸⁹.

Two studies have described methods for broth enrichment of colistin-resistant E. coli followed by real-time PCR to detect mcr-genes ^{90, 91}. Chalmers et al (2018) were the first to describe a SYBR Green-based real-time PCR method for mcr-1 and mcr-2 following enrichment with E. coli (EC) broth containing colistin (1µg/mL). All the porcine faecal and chicken caecal samples were screened by real-time PCR after 16h of culture in EC broth ⁹¹. However, none of the mcr-1 and mcr-2 genes were detected by PCR in any of the samples after 16h of enrichment ⁹¹. As well, the method described by Chandler et al (2020) for detecting mcr-1 included enrichment using EC broth containing colistin (1µg/mL) and vancomycin (8µg/mL). The method was evaluated on 100 feral swine faecal samples, which were inoculated with one of five different mcr-1 positive E. coli strains ⁹⁰. The bacteria was inoculated at concentrations ranging between 0.1-9.99 CFU/g, 10-49.99 CFU/g, 50-99 CFU/g, 100-149 CFU/g and 200-2,200 CFU/g from which mcr-1 was detected with 32%, 72%, 88%, 95%, and 98% accuracy by real-time PCR, respectively ⁹⁰.

Four SYBR Green-based real-time PCR assays have been developed for mcr-1, mcr-2, mcr-3, mcr-4, and mcr-5 detection ^92-95. Bontron et al (2016) designed a SYBR Green-based real-time PCR assay for detection of mcr-1 from cultured bacteria and stools. The assay was validated on 20 Enterobacterales, where it was found to accurately detect the presence or absence of mcr-1 at a LOD of 10² cultured bacteria ⁹³. Furthermore, Dona et al (2017) described a SYBR Green real-time PCR assay to also detect mcr-1 from human faecal samples. However, in this study, 20µg of the stool samples were enriched overnight in 10mL LB broth containing 2µg/mL colistin and plated on four selective agar plates prior to DNA extraction ⁹⁵. The real-time PCR accurately identified mcr-1 harbouring E. coli isolates with an LOD of 10¹ and PCR efficiency of ca. 106% ⁹⁵.

Li et al (2017) also designed a multiplex SYBR Green-based real-time PCR assay for mcr-1, mcr-2 and mcr-3 detection. The assay was validated on 25 isolates including mcr-1 positive and mcr-3 positive strains; the mcr-2 gene was synthesized in the study due to a lack of mcr-2-positive isolates ⁹⁴. Although the mcr genes were detected with 100% accuracy with a LOD of 10², mcr-2 was not validated on cultured bacteria ⁹⁴. However, in this study all three mcr genes could not be simultaneously detected in one reaction unlike when the Taqman probe was used ⁹⁴. A more recent study evaluated the SYBR Green-based real-time PCR method proposed by Li et al (2017) in detecting and quantifying mcr-1 to mcr-3 as well as newly designed assays for mcr-4 and mcr-5 ⁹². The optimized mcr-1 to mcr-5 PCR assays were validated on bacterial isolates ⁹². The study found that SYBR Green real-time PCR, followed by melting curve analysis, was more efficient in detecting and quantifying mcr-1 to mcr-5 genes in both bacterial isolates ⁹². The described assays detected all five mcr genes with a lower limit of 10². Moreover, the assays enabled screening of five individual samples in a single reaction ⁹².

The parallel detection of mcr-1, mcr-2 and mcr-8 by real-time PCR using Taqman® probes has been described ^96-98. Chabou et al (2016) designed two quantitative real-time PCR assays with TaqMan® probes for the rapid detection of mcr-1 gene. Primers and probes were designed to develop the two PCR assays, designated PE1 and PE2 ⁹⁶. The assays were evaluated on 100 bacterial isolates (18 of which were colistin resistant) and 833 broiler faecal samples ⁹⁶. The sensitivity and specificity of both assays were 100%, with a calibration curve that was linear from 10¹ to 10⁸. However, the PE1 assay was recommended for initial screening of mcr-1 followed by PE2 assay for confirming the results ⁹⁶.

Daniels et al (2017) developed a multiplexed real-time PCR with TaqMan® probes to detect mcr-1 and mcr-2. The assay was validated on 25 bacterial isolates, some of which were mcr-positive ⁹⁸. The sensitivity and specificity of the assay was 100%, being able to detect mcr-1 and mcr-2 from dilutions containing 8.5 10³ and 7.7 10³cfu/mL respectively ⁹⁸. A specific real-time PCR assay using TaqMan probes to identify mcr-8 was designed for the first time by Nabti et al (2020). The specificity and sensitivity of the assay were evaluated on 290 bacterial isolates from clinical samples and 250 metagenomic DNA from human stools ⁹⁷. The PCR assay accurately detected mcr-8 from the one positive K. pneumoniae isolate with an overall efficiency of 92.64% and a limit of detection of 55 CFU/mL ⁹⁷.