Evaluation of the Colistin HiMIC Plate Kit for Colistin Susceptibility Testing of Klebsiella pneumoniae

Introduction

Colistin, also known as polymyxin E, is a broad-spectrum antibiotic with activity against most Enterobacterales, Acinetobacter spp., and Pseudomonas aeruginosa. ¹ It is often the only available active antimicrobial against multidrug-resistant and extensively drug-resistant pathogens, including carbapenemase producers.^2,3 The reference method for colistin antimicrobial susceptibility testing is minimal inhibitory concentration (MIC) determination with broth microdilution (BMD).^4,5 Using disk diffusion or gradient diffusion is not recommended due to poor diffusion of large cationic colistin molecules.^6,7 Reference BMD is not a practical method for most clinical microbiology laboratories, therefore several commercial BMD tests have been developed and are now available on the market.^8–10

Materials and Methods

This study aimed to evaluate the Colistin HiMIC Plate Kit (HiMedia Laboratories, Mumbai, India), a new commercial BMD test for colistin susceptibility testing. A total of 73 isolates of carbapenemase-producing K. pneumoniae with different level of colistin susceptibility and originating from various clinical samples such as urine (n = 32), rectal swab (n = 19), wound swab (n = 5), throat swab (n = 5), blood culture (n = 4), tracheal aspirate (n = 2), sputum (n = 1), bronchoalveolar lavage (n = 1), intraoperative specimen (n = 1), stool (n = 1) and groin swab (n = 1) were used. Isolates were identified with MALDI-TOF MS (MALDI Biotyper Microflex LT/SH, Bruker Daltonics GmbH, Bremen, Germany), stored at 4°C, and subcultured before testing. The carbapenemases produced by the study isolates were OXA-48 (n = 67), NDM (n = 3), KPC (n = 2), and OXA-48/NDM (n = 1) and were detected by immunochromatographic assay (RESIST-5 O.K.N.V.I., Gembloux, Belgium) as well as by detecting genes conferring resistance to β-lactams including class A KPC, class B metallo-β-Lactamases (blaVIM and blaNDM) and carbapenem hydrolyzing oxacillinases (blaOXA-48-like) using multiplex PCR as previously described. ¹¹ BMD according to ISO standard 20776-1 with two-fold dilutions from 128 to 0.125 mg/L was used as a reference method. ⁴ Pure colistin powder (Colistin sulfate salt C4461-1G, Sigma–Aldrich, St. Louis, Missouri, SAD) was used for the preparation of stock solution which was diluted in sterile Mueller-Hinton (MH) broth (BBL™ Mueller Hinton II Broth, Becton Dickinson and Company, Sparks, USA) to get the concentration of 256 μg/mL. Two-fold dilution series ranging from 128 to 0.125 mg/L were diluted with MH broth in the 96-well plates using an eight-channel micropipette. The first column of the microtiter plate contained pure broth, acting as a growth control for each strain tested. The bacterial suspensions were prepared by mixing overnight K. pneumoniae colonies from an agar medium with saline solution and adjusting them to a turbidity equivalent to 0.5 McFarland standard using a densitometer (DensiCHEK Plus, BioMérieux, Marcy l’Etoile, France). The inoculum was then diluted in MH broth to achieve a final 5 × 10⁵ cfu/mL density. A 50 μL of bacterial suspension was added to each well with 50 μL diluted antimicrobial solution for testing. Colistin HiMIC Plate Kit containing dehydrated colistin in the concentration range of colistin in two-fold dilutions from 16 to 0.25 mg/L was evaluated with 0.5 McFarland bacterial suspensions prepared according to the manufacturer’s recommendations. The results were read after incubation in an aerobic atmosphere at 35 ± 1°C for 18 hours and were interpreted according to EUCAST breakpoints (susceptible, ≤ 2 mg/L; resistant, > 2 mg/L). The colistin-susceptible P. aeruginosa ATCC 27853 and E. coli ATCC 25922, and mcr-1 positive E. coli NCTC 13846 were included as quality control (QC) strains and analyzed according to EUCAST QC Tables version 14.0.^4,12,13

Performance of the commercial kit was assessed by calculating the rate of essential agreement (EA, i.e., MICs within ± 1 dilution of reference MICs adapted to the dilution range of the corresponding commercial product), categorial agreement (CA, i.e., test results with correct susceptibility categorization), very major error (VME) and major error (ME) according to Clinical and Laboratory Standards Institute guideline [14]. To calculate 95% CIs, we used Jeffreys’ method using Epitools Epidemiological Calculators available at https://epitools.ausvet.com.au.

Results

Reference BMD showed 50.7% (37/73) of colistin-susceptible and 49.3% (36/73) of colistin-resistant K. pneumoniae isolates. The colistin reference MICs ranged from 0,25 to 128 mg/L (Table 1.) with 15 (20.5%; 15/73) isolates having colistin reference MICs close to the current EUCAST breakpoint (MICs of 2, 4, and 8 mg/L).

The MIC distribution of the included isolates as determined by the Colistin HiMIC Plate Kit and broth microdilution reference standard are presented in Figure 1. In more than half of the isolates (82.2%; 60/73), the MICs determined with Colistin HiMIC^TM Plate Kit were either the same (27.4%; 20/73) or within ± 1 dilution (54.8%;40/73) in comparison to referral BMD method. On 12 occasions (16.4%;12/73) there was a difference of ± 2 and in one isolate (1.4%; 1/73) of ± 3 dilutions. Differences of ± 2 dilutions and ± 3 dilutions were observed more often in colistin-susceptible (18.9%; 7/37) than colistin-resistant isolates (16.7%; 6/36).

OPEN IN VIEWER

FIG. 1.

Comparison between MIC values of reference broth microdilution method (BMD) and Colistin HiMIC^TM Plate Kit for 73 carbapenemase-producing Klebsiella pneumoniae isolates. Perfect matching values and MIC values within the EA (within ± one dilution of reference MIC) are reported in dark gray and light gray, respectively. EUCAST breakpoints (susceptible ≤ 2, resistant > 2 mg/L) are shown as lines.

Compared with the reference method, Colistin HiMIC^TM Plate Kit showed CA and EA of 100% (73/73; 95% CI: 0.97–1.00) and 82.2% (60/73; 95% CI: 0.72–0.90), respectively. No ME (false-resistant results) and VME (false-susceptible results) were detected. Results for the quality control strains were within acceptable ranges.

Discussion

Our study evaluates a recently developed commercial BMD kit for colistin susceptibility testing, for which to our knowledge there aren’t any published data yet. When the following CLSI guidelines criteria are met: (a) CA and EA ≥ 90% when compared to the referral method, (b) VME rate < 3% of the total resistant isolates, and (c) ME rate < 3% of the total susceptible isolates, the antimicrobial susceptibility system should be considered verified for the particular drug/microorganism combination. ¹⁴ Our results showed that the Colistin HiMIC Plate Kit has acceptable CA, ME, and VME, whereas the EA (82.2%) did not meet the ≥ 90% threshold.

Three different guidelines followed in our study (EUCAST, CLSI, ISO) are the result of the complexity of colistin susceptibility testing and different guidelines authorities making joint efforts to standardize this method. Regarding BMD method, a joint CLSI-EUCAST working group investigated colistin BMD testing and decided that the recommendations in the ISO standard should be adhered to.^4,12 Susceptibility was interpreted according to EUCAST because this is the guideline we use in our routine laboratory work and performance of the commercial kit was evaluated according to CLSI guidelines because neither EUCAST nor ISO has an official document for verification of commercial antimicrobial susceptibility testing systems as CLSI does. ¹⁴

Our analysis may have been limited by a small number of isolates with reference MICs near the current EUCAST breakpoint of > 2 mg/L. Previous research has demonstrated that when a subgroup of more difficult isolates is employed, the rates of VME are rising and the rates of CA and EA are falling. ¹⁵ This argument’s significance is diminished, nevertheless, because two comparable approaches—the commercial BMD test and the referral BMD test—were compared. We were also limited by the current colistin resistance rate in our hospital; as colistin resistance is luckily not present in other Gram-negative bacteria, only K. pneumoniae and no other carbapenem-resistant Gram-negative bacteria such as A. baumannii and P. aeruginosa could be included. Although the testing was performed with a skilled staff, it was not logistically possible to test all isolates with both the new and the reference method on the same day, which would have been optimal. Furthermore, the variation among carbapenemases is minimal.

Conclusions

When reporting data related to colistin, accurate results, and good essential agreement are more crucial than when reporting data related to other antimicrobial agents. Laboratories must be aware of and check for possible limitations of commercial kits before they can be used for colistin susceptibility testing.