Performance of modified colistin broth disc elution vis-a-vis broth microdilution method for susceptibility testing of Enterobacterales

INTRODUCTION

Carbapenem-resistant Gram-negative bacteria have become a global threat, and polymyxins are the most relied option for them.^[1] In the year 2017, WHO declared colistin as one of the drugs in the “RESERVE” category, which can only be used in the most severe circumstances when all other alternatives have failed.^[2] However, resistance to polymyxin is also increasing. Only broth microdilution (BMD) is recommended by the Clinical and Laboratory Standards Institute (CLSI) and European Committee on Antimicrobial Susceptibility Testing (EUCAST) for polymyxin susceptibility testing, which is impractical for most clinical microbiology laboratories.^[3,4] Most clinical microbiology laboratories rely on disk diffusion or gradient diffusion susceptibility testing methods, but unfortunately, the polymyxins, being large cationic molecules, diffuse poorly, and there are high rates of very major errors (VMEs) as an isolate is that is read susceptible when it is resistant by the recommended BMD method. Most clinical microbiology laboratories are unable to provide an accurate colistin susceptibility report to clinicians due to the lack of a simple and easy-to-perform colistin susceptibility test method. Nordmann et al. have developed a simple and rapid polymyxin NP test as an alternative to BMD for screening of colistin-resistant Enterobacterales (CoRE) isolates.^[5] Although rapid, this test requires fresh preparation of colistin-containing solution that has a maximum shelf-life of 72 h and appropriate storage. This also fails to detect the minimum inhibitory concentration (MIC) of colistin, and only detect colistin resistance qualitatively. Simner et al. had reported an easy and practical method to perform colistin MIC testing by colistin broth disk elution (CBDE) test for colistin susceptibility testing of Gram-negative bacteria and got an encouraging result of 98% categorical agreement (CA), 99% essential agreement (EA), and no major errors (MEs) when compared to reference BMD method.^[6] However, the limitations of this method are use of 0, 1, 2, and 4 numbers of colistin (10 µg) disks in four different 10-mL cation-adjusted Mueller–Hinton broth (CAMHB) tubes, to get a presumed colistin concentrations in the tubes to be 0, 1, 2, and 4 µg/mL/mL, respectively. Thus, for each isolate, a minimum of 7 colistin discs and five tubes of 10 mL capacity containing CAMHB medium are exhausted. To achieve the same drug concentration in the lower volumes, the present study was planned to use a modified CBDE (mCBDE) test in a single set of CBDE test materials by performing in a 96-well microtiter plate so that multiple isolates can be tested at once for colistin susceptibility using a single set. Thus, there shall be less labor and use of a smaller number of tubes, colistin discs, and CAMHB medium and will be compared simultaneously with reference BMD. The results will be helpful for the adoption of a simpler and cost-effective method for colistin susceptibility testing in routine practice.

MATERIALS AND METHODS

The proposed work was a retrospective evaluation of CBDE test in a microtiter plate for Colistin Susceptibility Testing of clinical isolates of multidrug resistant (MDR) Enterobacterales.

RESULTS

In the present study, we have included a total of 160 non-duplicate Carbapenem resistant Enterobacterales (CRE) isolates obtained from various clinical samples received for routine culture and susceptibility test. As the study was conducted during the period of the ongoing COVID pandemic, of the 160 CRE isolates, the majority were recovered from inpatients (159/160, 99.3%), and 46.2% of them (74/160) were obtained from various intensive care units. Among the 160 CRE isolates, the majority were obtained from respiratory samples (70/160, 43.7%), followed by urine (37/160, 23.1%), blood (26/160, 16.2%), pus (21/160, 13.1%), and sterile body fluids (6/160, 3.7%). All the isolates were simultaneously subjected to BMD and mCBDE tests for colistin susceptibility. The results for positive and negative controls were satisfactory. Among these 160 isolates, 88.75% (142/160) were intermediate susceptible (MICs ≤2 µg/mL), and 11.2% (18/160) were resistant (MIC from >4 µg/mL) to colistin according to the reference BMD test. The susceptible isolates included 59.2% (84/142) Klebsiella spp., 39.4% (56/142) E. coli, and 1.4% (2/142) Enterobacter spp. and the resistant isolates were 55.5% (10/18) Klebsiella spp., 33.3% (6/18) E. coli, and 11.1% (2/18) were Enterobacter spp. Similarly, 87.5% (140/160) isolates were found to be intermediate susceptible including 57.8% (81/140) Klebsiella spp., 40.7% (57/140) E. coli, and 1.4% (02/140) Enterobacter spp. and 12.5% (20/160) were resistant to colistin including 65% (13/20) Klebsiella spp., 25% (5/20) E. coli, and 10% (2/20) Enterobacter spp. by mCBDE test. Seventeen isolates were resistant, and 139 were intermediate susceptible by both methods, showing the same profile.

Individual performance characteristics of mCBDE compared to BMD

The agreement analysis of mCBDE to BMD is depicted in Figure 1a. The number of isolates with the same MIC was shown in yellow box, whereas ± 1 log₂ and ± 2 log₂ MIC values were shown in light brown box and light blue box respectively. These values are used for extrapolating the CA and EA. In our study, mCBDE showed a CA and EA of 97.5% and 98.7%, respectively. Of the 160 isolates tested, 152 had exactly the same MIC as that of BMD. Among the rest, eight isolates, six had MIC within ±1 log₂ dilutions and two had higher MIC (8 µg/mL) compared to BMD (1 µg/mL). Out of 142 colistin intermediate susceptible isolates, three showed ME (03/142, 2.1%) or false resistance. Similarly, out of 18 colistin-resistant isolates, one showed VME or false susceptibility in mCBDE (01/18, 5.5%). The sensitivity and specificity of mCBDE test as compared to reference BMD were found to be 97.8% and 94.4%, respectively [Figure 1b].

Close

Figure 1:

Performance of modified colistin broth disc elution (mCBDE) compared to broth microdilution (BMD)- (n = 160). (a) Categorical agreement analysis of minimum inhibitory concentration values of colistin in broth microdilution (BMD) versus mCBDE and (b) performance of mCBDE compared to BMD (n = 160). Circle denotes isolates showing major error (ME), where as Square denotes very major error (VME). TP: True positive, FN: False negative.

Export to PPT

DISCUSSION

Colistin, a five-decade-long antimicrobial, regained a cult status in the 21^st century to address the growing menace of infection due to CRE. Although the pharmacokinetics and pharmacodynamics of polymyxins are still in debatable status, there is an increase in usage of colistin due to the lack of availability of alternative antimicrobials, leading to emergence of colistin-resistant pathogens.^[9] A rapid colistin susceptibility test is highly needed in diagnostic clinical microbiology laboratories, because the joint CLSIEUCAST recommended method (BMD) is time-consuming (>48 h), expensive, and demands expertise, thus difficult to implement in routine practices.^[10] Over the years, many studies have evaluated colistin susceptibility by automated system, agar dilution, E-test, and disk diffusion, considering BMD as the gold standard. However, most of them have reported high error rates in all these methods. Khurana et al. had reported the CA to be 88% (798/910) with 10% (95/910) VME and 1% (9/910) ME for Gram-negative bacilli by the VITEK 2 system.^[11] Kar et al. also reported high VME rates for both agar dilution (11%) and E-test (37%) in Enterobacterales.^[12] A rapid colorimetric assay for polymyxin susceptibility (NP test) by Nordmann et al. gained popularity as one of the screening tests with a sensitivity and specificity of >95%.^[5] However, the disadvantage of NP test was, its performance is limited to the Enterobacterales only.^[13] Kar et al. could accurately detect 26/31 CoRE by polymyxin NP test with an overall CA of 83.8%, but failed to detect five Enterobacter spp. 5/31 (16.1%) with overall VME to be 16%,which decreased to <3% when those Enterobacter spp. (n = 5) were excluded from the study.^[14]

Recently, the CLSI antimicrobial susceptibility testing subcommittee endorsed the CBDE and colistin agar test (CAT-10) methods for colistin susceptibility testing of Enterobacterales and Pseudomonas aeruginosa based on the results of CBDE (97.9% CA and 94.4% EA) and CAT-1/CAT-10 (99.4% CA and 99.7% EA) as compared to reference BMD.^[9] Unlike polymyxin NP test, CBDE can also be used for glucose-non-fermenting organisms. Although both CBDE and CAT are showing excellent agreement with BMD, CBDE is gaining popularity compared to CAT as the former is simpler and easy to perform.^[9]

In our study, among the 160 isolates, 88.75% (142/160) were intermediate susceptible (MICs ≤2 µg/mL), and 11.2% (18/160) were resistant (MIC >4µg/mL) to colistin according to the reference BMD test.

Among the 17 CoRE detected by both BMD and mCBDE, Klebsiella pneumonia was the major isolates (58.5%, 10/17) with higher MIC value (≥8 µg/mL). In our study, the result of mCBDE was satisfactory with a CA and EA of 97.5% and 98.7%, respectively. The ME of mCBDE was within the acceptable range (2.1%), whereas VME was high (5.5%). Dalmolin et al.^[15] had also performed CBDE in diminished volumes of the reagents (colistin broth microdilution) with a final volume of 1 mL and the microelution-plates test – final volume of 200 µL on 85 g-negative rods obtained from two different research centers in Brazil. They had evaluated one more modified method – colistin susceptibility test tube to overcome the issue of adhesion to the microelution-plates. They performed the test in the same manner in one glass tube (5 mL of CA-MHB + 10 µg colistin disk which gave concentration of 2 µg/mL and to that 25 µL of test inoculum [10⁸ CFU/mL] added) and interpreted resistant when growth (turbidity) was observed (MIC >2 µg/mL). They reported satisfactory results for Enterobacterales with fair agreement, sensitivity, and specificity by all three methods, but the major and VME were less satisfactory. However, their results for non-fermentative isolates were not satisfactory, the agreement, sensitivity, and specificity were low and error rates were high.^[15] The possible reason for high error rates was hypothesized by various authors that the assumed colistin concentrations eluted from the discs in the broth medium may not be as accurate as desired, thus leading to the high VME.^[15] Further studies are needed to measure the exact concentration of polymyxin in the broth after elution.

Ngudsuntia et al.^[16] evaluated the performance of CBDE and rapid colistin disk elution (RCDE) test for Gram-negative bacilli as compared to the reference BMD test. They performed RCDE test using a 10-µg colistin disk in 2.7 mL volume (final colistin concentration-3.7 µg/mL) of CAMHB or phenol red broth base, added 1-µL loop of bacterial inoculums, incubated 1–4 h. for Enterobacterales and 16–20 h. for Acinetobacter baumannii. For Enterobacterales, the CA and VME rate were 98.3% and 5.4%, respectively, in the RCDE test and 97.9% and 7.1%, respectively, in the CBDE test with a ME of 0.6% in both the tests. However, for the A. baumannii isolates, the RCDE and CBDE tests showed high VME rates (8.3% and 16.7%, respectively). As per their observation, the performance of RCDE was good and comparable with CBDE; moreover, it was cheaper, more rapid (3 h), and convenient, thus suggesting RCDE to be an alternative for detecting colistin resistance among Enterobacterales in low-income countries.^[16]

The ad hoc working group reported low performance of QC strain (E. coli ATCC 25,922) used in CBDE, thus CLSI recommended E. coli AR Bank no. 0349 (mcr1 positive) as the QC strain.^[9] We had used NCTC E. coli 13846 (mcr1) as a positive control and got satisfactory results.

The mCBDE conducted at our setup had solved the issues raised by Simner et al. regarding the use of numerous discs, tubes, and occupying more space in incubator for CBDE.^[6]

CONCLUSIONS

This is the first report from Eastern India to perform modified CBDE in reduced volume. Although there is no specific recommended brand of colistin disc/CAMHB, we used only one brand (HiMedia) and got excellent sensitivity (97.8%) and specificity (94.4%). Thus, mCBDE can be used as an alternative to CBDE to save a lot of resources.

The limitations of our study were small sample size, and non-inclusion of medically important non-fermenters (Pseudomonas spp., Acinetobacter spp.). Moreover, the mCBDE could have been compared with that of CBDE for agreement analysis. Further, it is suggested to perform the tests by two different persons to ascertain its repeatability and reproducibility.