Development and Validation of a Novel Anaerobic Carbapenem Inactivation Method (Ana-CIM) for the Detection of Carbapenemase Production in Bacteroides fragilis

ABSTRACT

Antibiotic resistance, particularly to carbapenems, is of increasing concern in Bacteroides fragilis. Carbapenem resistance in B. fragilis is most often mediated by the activation of chromosomally encoded metallo-β-lactamase cfiA by the presence of an upstream insertion sequence (IS). While traditional phenotypic susceptibility methods and molecular tests to detect carbapenem resistance in B. fragilis exist, they are not available in most clinical microbiology laboratory settings. Here, we describe the development of the anaerobic carbapenem inactivation method (Ana-CIM) for predicting carbapenemase production in B. fragilis based off the principles of the well-established modified carbapenem inactivation method (mCIM) for Enterobacterales and Pseudomonas aeruginosa. We also present the clinical validation and reproducibility of the Ana-CIM at three clinical laboratory sites (with 60 clinical isolates, 45% ertapenem resistant). Compared to ertapenem susceptibility by Etest interpreted by CLSI M100 Ed30, the Ana-CIM accurately detected carbapenem resistance in B. fragilis with categorical agreement (CA) of 87% (52/60) and 0% (0/21) very major error (VME), 11% (4/36) major error (ME), and 7% (4/60) minor error (mE) rates across all sites. Additionally, the Ana-CIM demonstrated high reproducibility with 5 clinical and 3 quality control (QC) isolates tested in triplicate with 3 commercial Mueller-Hinton media across all sites, with 93% (604/648) of replicates within a 2-mm zone size of the mode for each isolate. We conclude that the Ana-CIM can be readily deployed in clinical laboratories at a low cost for detection of carbapenemase-mediated resistance in B. fragilis.

INTRODUCTION

Anaerobic infections are associated with significant morbidity and mortality. Bacteroides species are among the most common and virulent agents of anaerobic infections, largely attributed to high levels of antimicrobial resistance that have been shown to contribute to poor outcomes (1). Most Bacteroides species are resistant to penicillin and ampicillin and have variable susceptibility to cephalosporins due to the activity of chromosomally encoded β-lactamase cepA, which hydrolyzes cephalosporins and aminopenicillins, and cfxA, which hydrolyzes cefoxitin and other β-lactams (2).

Carbapenem antibiotics remain an effective therapeutic option for multidrug-resistant (MDR) Bacteroides species, with most isolates testing susceptible (3). However, recent national and international surveys have found increasing rates of carbapenem resistance, nearly doubling from 0.5 to 1.6% carbapenem intermediate/resistant in 2006 to 2007 to 1.1 to 2.4% in 2010 to 2012 in the United States (3, 4) and up to 3.4% and 2.4% in Europe and South America, respectively (5, 6). In Bacteroides fragilis, most of the carbapenem resistance is mediated by chromosomally encoded metallo-β-lactamase cfiA, which normally exhibits low-level expression that is increased in the presence of insertion sequences (IS), upstream of the cfiA promoter (7). Conversion of cfiA-positive B. fragilis isolates from carbapenem susceptible to carbapenem resistant has been previously demonstrated during carbapenem therapy (7). Identification and differentiation of carbapenemase-producing B. fragilis isolates from nonenzymatic mechanisms of carbapenem resistance are important for evaluating the therapeutic potential of carbapenem therapy, as carbapenemase-producing isolates will likely be resistant to all carbapenems (6). Due to diversity in sequence and location of IS elements, PCR-based tests have limited sensitivity for detection of cfiA-mediated carbapenem resistance and are not practical to implement in the workflow in most clinical microbiology laboratories. Phenotype-based tests including double-ended Etest strip with or without EDTA, double disk diffusion, and detection of enzymatic activity via colorimetric assay (e.g., Carba-NP), as well as the molecular biology-based matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-ToF MS), have been approved for aerobes but have not been widely evaluated or implemented by clinical laboratories for anaerobic microorganisms (8–10). Phenotypic susceptibility test methods for anaerobes such as agar dilution and gradient diffusion strips require ≥24 h after assay setup to yield results; however, the availability of this testing is often delayed as most clinical laboratories do not perform this testing and must send out isolates to a reference laboratory.

Our goal was to develop and evaluate a novel assay, the Anaerobic Carbapenem Inactivation Method (Ana-CIM), for the detection of carbapenemase production in B. fragilis. The Ana-CIM assay is a simple methodology derived from the principles of the modified carbapenem inactivation method (mCIM) for Enterobacterales and Pseudomonas aeruginosa and utilizes materials readily available in clinical laboratories capable of isolating anaerobic bacteria (11). This study was established with B. fragilis sensu stricto isolates only, as carbapenemase production has not been characterized in other non-fragilis Bacteroides species. After initial method development and optimization, we performed a pilot study at three study sites, followed by a multicenter validation study and reproducibility study (Fig. 1). We report that the Ana-CIM reproducibly detects carbapenemase production in B. fragilis isolates with high agreement to phenotypic and predicted genomic carbapenem susceptibility. The Ana-CIM can be readily implemented and can provide information about carbapenemase production results about a day faster than conventional susceptibility results for B. fragilis isolates from sterile body sites or when carbapenem resistance is suspected.

FIG 1.

Overview of study workflow. After initial assay development at WU, a pilot study collection containing 20 isolates of B. fragilis from WU was tested by three clinical laboratories to evaluate the assay protocol. Quality control (QC) isolates were tested at all sites at least daily. Next, for the clinical validation study, 35 clinical isolates originating from all three sites were tested by the Ana-CIM at all sites. Finally, a reproducibility study was performed on a total of eight (five test and three QC) isolates. These eight isolates were tested in triplicate on three different days with three different media by all three sites for a total of 81 replicates per isolate. Abbreviations: Ana-CIM, anaerobic carbapenem inactivation method; MC, Mayo Clinic; NCH, Nationwide Children’s Hospital; QC, quality control; WU, Washington University School of Medicine.

MATERIALS AND METHODS

Clinical isolates.

Sixty deidentified clinical isolates of B. fragilis were obtained from Washington University School of Medicine (WU), St. Louis, MO; Mayo Clinic (MC), Rochester, MN; and Nationwide Children’s Hospital (NCH), Columbus, OH (n = 27, 22, and 11 isolates, respectively). Isolates were recovered from various specimen types including abscesses, blood, body fluids, and wounds and were identified by MALDI-ToF MS (see Table S1 in the supplemental material). All isolates were transferred to WU, where antimicrobial susceptibility testing (AST) for ertapenem and meropenem was performed by Etest (bioMérieux, Durham, NC) before isolates were coded and returned to each site for Ana-CIM testing.

Molecular detection of carbapenemase genetic determinants.

All isolates were subjected to endpoint PCR to detect cfiA and upstream IS elements as previously described (12). Prediction of CfiA by Bruker Biotyper was evaluated as previously described (12).

Anaerobic susceptibility testing.

All isolates were tested for susceptibility to ertapenem and meropenem initially by Etest at WU and subsequently by agar dilution at MC. For the Etest, a 1 McFarland standard suspension of 24- to 48-h growth on prereduced anaerobically sterile (PRAS) Brucella blood agar with hemin and vitamin K (H+K) (Hardy Diagnostics, Santa Maria, CA) in Brucella broth (BD BBL, Sparks, MD) was used to inoculate PRAS Brucella blood H+K agar plates. Etest strips were placed on lawn-struck plates and incubated at 35°C in AnaeroGen 2.5-L atmosphere generating systems (Oxoid, Ltd.) for up to 72 h. MICs were read and recorded at 72 h. Agar dilution testing of ertapenem and meropenem was performed as described in the Clinical and Laboratory Standards Institute (CLSI) document M11-A9 at MC (13). Interpretive criteria were applied according to the CLSI M100 Ed30 document (14) and EUCAST 2020 guidelines (15).

Anaerobic carbapenem inactivation method (Ana-CIM).

Please refer to supplementary file 1 in the supplemental material for the detailed development of Ana-CIM methods. B. fragilis isolates were subcultured for isolation to prereduced Brucella blood agar with hemin and vitamin K (Hardy Diagnostics, Santa Maria, CA). To maintain selective pressure, a 10-μg meropenem disk was placed in the first quadrant of each plate. Plates were incubated at 35°C in an anaerobic atmosphere for 24 to 48 h. Following incubation, a new 10-μg meropenem disk was placed into a 5-mL tube of Brucella broth (BD BBL, Sparks, MD) and incubated at room temperature for 15 min to distribute antibiotic (without introducing excess oxygen by vortexing or inverting the tubes). Growth of each isolate from solid medium closest to the meropenem disk was taken and suspended in the Brucella broth containing the meropenem disk to obtain a 1 McFarland standard suspension. The final suspensions were incubated anaerobically at 35°C for 6 h.

After incubation, using a sterile 10-μL loop, meropenem disks were removed from Brucella broth tubes and placed on a 15- by 150-mm Mueller-Hinton (MH) plate (Hardy Diagnostics, Santa Maria, CA) inoculated with a confluent lawn of a 0.5 McFarland standard suspension of Escherichia coli ATCC 25922. A maximum of 8 disks was placed on one plate (5 test isolates and 3 quality control [QC] organisms). MH plates were incubated for 18 h at 35°C in an air incubator. Zone sizes around each meropenem disk were measured using reflected light and recorded using a metric ruler. Ana-CIM results were interpreted according to both CLSI M100 Ed30 mCIM interpretive criteria for Enterobacterales and interpretive criteria developed in our pilot study (Table 1). In keeping with the CLSI guidance on mCIMs, microcolonies or colonies inside the zone of inhibition were read and measured as growth (Fig. 2).

TABLE 1.

CLSI mCIM interpretive criteria and lab-developed Ana-CIM interpretive criteria

Result	CLSI mCIMa (zone size, mm)	Ana-CIM (zone size, mm)
Positive	≤15	≤8
Negative	≥19	≥15
Indeterminate	16–18	9–14

^aPer CLSI M100 Ed30 for Enterobacterales.

FIG 2.

Ana-CIM zone size reading guide. Representative carbapenemase positive (A, C, and D) and negative (B) clinical isolates are shown with confluent growth up to the disk (A) or microcolonies within the zone of inhibition (C and D). Consistent with the Clinical and Laboratory Standards Institute (CLSI) M100 Ed30, colonies within the zone are read as growth (6 mm). Solid black lines represent measurement, and measured zone size is listed at the bottom of each panel in millimeters.

QC was performed every day of Ana-CIM testing, such that out of the 8 disks on the test plate, 3 were QC isolates. Two positive-control isolates were utilized to control for carbapenem inactivation and adequate anaerobiosis, respectively: (i) Klebsiella pneumoniae ATCC BAA-1705, a K. pneumoniae carbapenemase (KPC) producer that is utilized as the mCIM positive control recommended by CLSI, and (ii) B. fragilis WIS-ImiR-001, a carbapenem-resistant (ertapenem, imipenem, and meropenem MICs of ≥32 μg/mL) isolate that tested positive for cfiA and upstream IS by previously described PCR (16). Both positive-control isolates have mCIM/Ana-CIM results of ≤8 mm. B. fragilis ATCC 25285 served as the negative QC strain, with an expected Ana-CIM result of ≥15 mm.

Anaerobic incubation.

For anaerobic conditions, WU and NCH used the AnaeroGen 2.5-L atmosphere generating systems (Oxoid, Ltd.) for all studies. MC utilized the Coy Laboratory Products Inc. anaerobic chamber with a gas mixture of 90% nitrogen, 5% carbon dioxide, and 5% hydrogen (Praxair Inc.). Anaerobic indicators were included with all testing to ensure an anaerobic environment had been achieved.

Pilot study.

Twenty isolates originating from WU were tested at each site for proof of concept and to refine lab-developed Ana-CIM interpretive criteria. At this time, each site performed testing at the respective institution using the described Ana-CIM procedure. Each institution used the same brands of solid media (Brucella blood H+K and MH agar; Hardy Diagnostics), broth (Brucella broth; BD BBL), and meropenem disks (BD BBL) for testing.

Clinical isolate validation study.

Thirty-five isolates, representing strains recovered in clinical specimens from all three institutions (WU, n = 6; MC, n = 20; and NCH n = 9), were tested using the Ana-CIM procedure using the same conditions described for the pilot study testing.

Reproducibility study.

Reproducibility testing consisted of five clinical isolates (WU, n = 1; MC, n = 2; and NCH, n = 2) and the three QC isolates previously described. Each isolate was tested in triplicate, on three different days, using three different brands of MH plates (Hardy Diagnostics [Santa Maria, CA], BD Biosciences [San Jose, CA], and Remel [Lenexa, KS]). All sites used the same lot number and expiration date of all three MH brands. Due to supply shortages during the COVID-19 pandemic, only WU was able to perform testing using commercially prepared Brucella broth as described in the Ana-CIM procedure and used for the pilot and clinical validation study. NCH and MC used the same lot of dehydrated Brucella broth (BD BBL, Sparks, MD) resuspended per manufacturer’s instructions and dispensed into 5-mL tubes prior to sterilization for testing.

All pilot, clinical, and reproducibility study isolates tested are listed with isolate source, site of origin, phenotypic carbapenem susceptibility, and genotypic carbapenemase results (see Table S1 in the supplemental material).

Data analysis.

Ana-CIM testing was compared to ertapenem Etest as the reference method and interpreted using CLSI M100 Ed30 (Table 2). Ertapenem was used as the reference for phenotypic susceptibility based on previous reports of higher mean MICs for ertapenem than for meropenem in B. fragilis with the rationale that this would result in a more conservative assessment of Ana-CIM performance (3, 17). Categorical agreement (CA) was defined as the percentage of total Ana-CIM test results in agreement with expected results from traditional phenotypic susceptibility testing. For the pilot, clinical, and reproducibility studies utilizing the Ana-CIM, the minor error (mE) rate was defined as the percentage of total isolates for which the ertapenem Etest with CLSI interpretation was resistant or susceptible but the Ana-CIM result was indeterminate or when the Etest result was intermediate and the Ana-CIM was either positive or negative. The major error (ME) was defined as the percentage of isolates susceptible by Etest but interpreted as positive (i.e., resistant) by the Ana-CIM. The very major error (VME) was defined as the percentage of isolates testing ertapenem resistant by Etest that tested Ana-CIM negative (i.e. susceptible). For the reproducibility studies, the range and mode of millimeter zone sizes were calculated for all replicates per site and MH agar brand, as well as across all replicates (see Table 4 and 5).

TABLE 2.

Overall performance of Ana-CIM compared to ertapenem Etest susceptibility using CLSI and EUCAST breakpoints^a

Study	Comparator	CLSI (14)				EUCAST (15)
		CA	VME	ME	mE	CA	VME	ME	mE
Pilot	mCIM	95% (19/20)	0% (0/8)	0% (0/12)	5% (1/20)	95% (19/20)	0% (0/8)	0% (0/12)	5% (1/20)
	Ana-CIM	95% (19/20)	0% (0/8)	0% (0/12)	5% (1/20)	95% (19/20)	0% (0/8)	0% (0/12)	5% (1/20)
Clinical	Ana-CIM	86% (30/35)	0% (0/12)	10% (2/20)	9% (3/35)	94% (33/35)	17% (2/12)	0% (0/20)	0% (0/35)
Repro	Ana-CIM	60% (3/5)	0% (0/1)	50% (2/4)	0% (0/5)	100% (5/5)	0% (0/3)	0% (0/2)	0% (0/5)
Overall	Ana-CIM	87% (52/60)	0% (0/21)	11% (4/36)	7% (4/60)	95% (57/60)	7% (2/29)	0% (0/31)	2% (1/60)

^aCA, categorical agreement; mE, minor error; ME, major error; VME, very major error; Repro, reproducibility.

TABLE 4.

Ana-CIM reproducibility testing by medium manufacturer and testing site^a

Isolatec	Mode (range) by manufacturer and testing site:
	Hardy			Remel			BD
	WU	MC	NCH	WU	MC	NCH	WU	MC	NCH
R1	19 (18–20)	18 (17–19)	19 (18–20)	18 (18–21)	18 (17–19)	18 (18–19)	18 (17–20)	19 (17–19)	19 (19–20)
R2	6 (6)	6 (6)	6b (6–16)	6 (6)	6 (6)	6b (6)	6 (6)	6 (6)	6b (6)
R3	19 (18–20)	18 (17–19)	19 (18–20)	19 (18–20)	18 (17–19)	19 (18–20)	18 (18–20)	19 (18–19)	19 (19–20)
R4	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)
R5	15 (7–16)	6b (6–14)	16b (6–17)	6b (6–16)	6b (6–14)	17b (6–18)	13 (13–15)	6b (6)	6b (6–17)
Kp pos control	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)
Bf pos control	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)	6 (6)
Bf neg control	19 (18–20)	18 (17–19)	20 (18–20)	18 (18–20)	18 (17–18)	18 (18–20)	19 (19–20)	19 (17–20)	19 (19–20)

^aKp pos control, K. pneumoniae ATCC BAA-1705; Bf pos control, B. fragilis WIS-ImiR-001; Bf neg control, B. fragilis ATCC 25285.

^bMicrocolonies within zone.

^cEach isolate was tested a total of 81 times (3 replicates per day on 3 different testing days at 3 different sites).

TABLE 5.

Ana-CIM reproducibility testing by number of tests with zone size off the mode^a

Isolateb	Mode for all tests (mm)	No. of tests with zone size (mm) off the mode
		<−5	−5	−4	−3	−2	−1	0	+1	+2	+3	+4	+5	>+5
R1	19					4	29	37	10	1
R2	6							78						3
R3	19					2	27	41	11
R4	6							81
R5	6							39	1					41
Kp pos control	6							81
Bf pos control	6							81
Bf neg control	19					3	29	30	19
Total		0	0	0	0	9	85	468	41	1	0	0	0	44

^aKp pos control, K. pneumoniae ATCC BAA-1705; Bf pos control, B. fragilis WIS-ImiR-001; Bf neg control, B. fragilis ATCC 25285.

^bEach isolate was tested a total of 81 times (3 replicates per day on 3 different testing days at 3 different sites).

Using CLSI M100 Ed30 interpretive guidelines (14), Ana-CIM results were also compared to ertapenem agar dilution, meropenem Etest, and meropenem agar dilution susceptibility (see Table S5). The CLSI breakpoints for meropenem and ertapenem at the time of this study were ≤4 μg/mL, susceptible; 8 μg/mL, intermediate; and ≥16 μg/mL, resistant. Additionally, in a separate, standalone analysis, Ana-CIM testing was also compared to meropenem and ertapenem susceptibility evaluated with Etest and agar dilution methods interpreted using the European Committee on Antimicrobial Susceptibility Testing (EUCAST) 2020 breakpoints (15) (Table 2 and Table S5). The EUCAST breakpoints for ertapenem and meropenem at the time of this study were ≤0.5 μg/mL, susceptible, and >0.5 μg/mL, resistant, and ≤2 μg/mL, susceptible, and >8 μg/mL, resistant, respectively.

Discrepant analysis.

Isolates with discrepant results (difference in Ana-CIM zone sizes of ≥5 mm between sites) were subject to discrepant analysis. New isolate stocks were prepared at WU and resent to MC and NCH. All sites retested new and original isolate stocks. Only 1 isolate was subjected to discrepant analysis among the sites; this resolved with repeat testing. Thus, results of repeat testing were used for final analyses.

RESULTS

Pilot study.

To verify that the proposed Ana-CIM was robust and reproducible in different clinical laboratories, a set of 20 B. fragilis isolates with variable carbapenem susceptibility and 3 QC isolates (positive controls, Klebsiella pneumoniae ATCC BAA-1705 and B. fragilis WIS-ImiR-001; negative control, B. fragilis ATCC 25285) were selected by WU and sent to MC and NCH sites for Ana-CIM. MC and NCH test sites were blind to the expected results. Among pilot study isolates (WU01 to WU20), 8 were resistant to ertapenem (≥16 μg/mL) by Etest by both CLSI and EUCAST. Across all sites, QC isolates tested within the expected range per the CLSI M100 Ed30 Enterobacterales mCIM (see File 1 in the supplemental material). Eighteen (90%) pilot study test isolates had zone sizes within 2 mm across all sites (Table S2). When interpreted using CLSI M100 Ed30 mCIM Enterobacterales interpretive criteria (Table 1), categorical agreement (CA) of Ana-CIM for the pilot study isolates was 95% at WU, 95% at MC, and 75% at NCH. The difference among sites was due to 6 isolates, WU02, WU05, WU07, WU10, WU16, and WU20, which were susceptible to ertapenem but were interpreted as indeterminate by mCIM criteria at 1 or more sites (Table S2). However, review of discrepant isolates revealed that all zone sizes were within 3 mm (range, 18 to 21 mm) across all sites (Table S2). Interpretive criteria for mCIM were therefore not suitable for this assay as B. fragilis isolates that were phenotypically carbapenem susceptible were interpreted as indeterminate when mCIM criteria were applied to Ana-CIM testing. As such, we proposed laboratory-developed Ana-CIM interpretative criteria (Table 1). When lab-developed Ana-CIM interpretive criteria were applied to pilot study testing, 95% (19/20) concordance in interpretation was observed across the 3 sites, with one indeterminate result at MC. CA compared to ertapenem Etest phenotypic susceptibility was 95% (19/20 isolates) with 1 mE (1/20, 5%). There were no VMEs (0/8) or MEs (0/12) observed at any sites (Table 2 and Table S2).

Clinical validation study.

Having demonstrated that the Ana-CIM could be performed at multiple sites with greater than 90% concordant results in the pilot study, we undertook a multicenter clinical validation study. Thirty-five B. fragilis isolates (C1 to C35) were tested by Ana-CIM at all three sites. Twelve isolates tested resistant to ertapenem by Etest interpreted by CLSI criteria, while 18 isolates tested resistant using EUCAST breakpoints (Table S1). MC and NCH sites were blind to phenotypic carbapenem susceptibility status of all isolates. When the proposed Ana-CIM interpretive criteria were applied, concordance was observed in 94% (33/35 isolates) at all 3 sites (Table 3). MC observed indeterminate results for C19 and C25, while WU and NCH observed positive Ana-CIM results for both of these isolates. Similar to the pilot study testing, the majority (86%; 30/35) of clinical validation study isolates had Ana-CIM zones sizes within 2 mm across all sites (Table 3 and Fig. 3). CA of the Ana-CIM results compared to ertapenem Etest susceptibility result was 86% (30/35) with no VMEs (0/12) and 10% (2/20) ME and 9% (3/35) mE rates (Table 2). Three isolates (C12, C19, and C26) tested intermediate by ertapenem Etest when applying CLSI interpretive criteria, but resistant by EUCAST criteria, which accounted for the 3 mEs. Isolates C24 and C25 accounted for the ME, as both tested susceptible by ertapenem Etest (resistant by agar dilution, resistant to meropenem by Etest and agar dilution) by CLSI guidelines (Tables 2 and 3 and Table S1).

TABLE 3.

Ana-CIM clinical study testing^a

Isolate	Erta Etest	WU		MC		NCH
		Zone size (mm)	Interp Ana-CIM	Zone size (mm)	Interp Ana-CIM	Zone size (mm)	Interp Ana-CIM
C1	S	18	Neg	19	Neg	20	Neg
C2	S	18	Neg	18	Neg	21	Neg
C3	R	6	Pos	6	Pos	6	Pos
C4	R	6	Pos	6	Pos	6	Pos
C5	R	6	Pos	6	Pos	6	Pos
C6	S	17	Neg	19	Neg	21	Neg
C7	S	18	Neg	19	Neg	20	Neg
C8	S	19	Neg	19	Neg	20	Neg
C9	R	6	Pos	6	Pos	6	Pos
C10	S	18	Neg	18	Neg	20	Neg
C11	S	18	Neg	19	Neg	20	Neg
C12	I	17	Neg	18	Neg	21	Neg
C13	R	6	Pos	6	Pos	6	Pos
C14	R	6	Pos	6	Pos	6	Pos
C15	S	18	Neg	18	Neg	20	Neg
C16	S	19	Neg	19	Neg	20	Neg
C17	S	19	Neg	18	Neg	20	Neg
C18	R	6	Pos	6	Pos	6	Pos
C19	I	6	Pos	9	Ind	6	Pos
C20	R	6	Pos	6	Pos	6	Pos
C21	S	17	Neg	18	Neg	19	Neg
C22	S	19	Neg	18	Neg	20	Neg
C23	S	18	Neg	18	Neg	20	Neg
C24	S	6	Pos	6	Pos	6b	Pos
C25	S	6b	Pos	10	Ind	6	Pos
C26	I	6b	Pos	6	Pos	6b	Pos
C27	S	18	Neg	18	Neg	19	Neg
C28	S	19	Neg	18	Neg	19	Neg
C29	S	19	Neg	18	Neg	18	Neg
C30	R	6	Pos	6	Pos	6	Pos
C31	S	19	Neg	18	Neg	19	Neg
C32	R	6	Pos	6	Pos	6b	Pos
C33	R	6	Pos	6	Pos	6	Pos
C34	S	19	Neg	17	Neg	18	Neg
C35	R	6	Pos	6	Pos	6	Pos

^aErta, ertapenem; Ind, indeterminate; Interp, interpretation; MC, Mayo Clinic; NCH, Nationwide Children’s Hospital; Neg, negative; Pos, positive; WU, Washington University; S, susceptible; R, resistant; I, intermediate.

^bMicrocolonies within zone.

FIG 3.

Ana-CIM zone size distribution of clinical isolates tested at all sites. Zone sizes for Ana-CIM testing for each clinical validation study isolate are shown. Results for Nationwide Children’s Hospital (NCH) (orange triangles), Mayo Clinic (MC) (green squares), and Washington University School of Medicine (WU) (blue circles) are shown. The isolates labeled in red along the x axis tested resistant by the ertapenem Etest, while the isolates in black tested susceptible. Lab-developed Ana-CIM interpretive criteria are shown as red solid lines, and CLSI mCIM interpretive criteria for Enterobacterales are shown in dashed gray lines.

Reproducibility testing.

To assess the reproducibility of the Ana-CIM, 5 isolates (R1 to R5; Table S1) as well as the 3 QC isolates were tested using 3 different commercially available MH agar plates from Hardy, BD, and Remel. Each isolate was tested in triplicate on each MH agar on three different days at each site (Fig. 4 and Tables 4 and 5). The mode and range were reported for each MH agar brand, at each test site, and across all replicates (Tables 4 and 5). All QC isolates were within the expected range (≤8 mm for K. pneumoniae ATCC BAA-1705 and B. fragilis WIS-ImiR-001, ≥15 mm for B. fragilis ATCC 25285) with 100% CA. All measured zone sizes were within 2 mm of the mode for each control isolate (Fig. 4 and Tables 4 and 5). There was 100% categorical agreement across all sites for isolates R1, R3, and R4 with all observed values within 2 mm between replicates (Fig. 4A, C, and D and Tables 4 and 5). However, categorical agreement for isolates R2 and R5, which both tested susceptible by ertapenem Etest CLSI guidelines, was 4% (range, 0 to 4% between sites) and 37% (range, 0 to 59% between sites), respectively. For isolate R2, the majority of readings resulted in an Ana-CIM zone size of 6 mm with only 3.7% (3/81) of replicates with zone sizes measuring greater than 5 mm from the mode (6 mm). However, for isolate R5, 50% (41/81) of replicates had zone sizes measuring greater than 5 mm from the mode (mode, 6 mm; range, 6 to 18 mm. Isolate R5 zone sizes that were off the mode hovered around the indeterminate and negative interpretations using the proposed Ana-CIM interpretive criteria (Fig. 4E). While colonies within the zone of inhibition were variably observed at all sites for isolate R5, isolate R2 consistently displayed microcolonies at NCH (22/27 replicates) only; this was rarely observed at other sites (WU, 1/27; MC, 0/27). There was no apparent association of microcolonies during Ana-CIM testing of R2 and R5 with specific MH medium bands (Table 4).

FIG 4.

Distribution of Ana-CIM zone sizes for reproducibility study isolates. Histograms include all 81 Ana-CIM zone-of-inhibition measurements for each of the 5 test isolates (A to E) and the quality control (QC) isolates (F to H) in the reproducibility study. Each isolate was tested in triplicate on 3 different commercially available Mueller-Hinton agar plates on 3 different days at the 3 clinical laboratories (27 zone measurements from each site for each of the isolates for a total of 81 replicates for each isolate).

Agreement of Etest and agar dilution phenotypic susceptibility methods.

All B. fragilis isolates in this study were initially tested by Etest for phenotypic susceptibility to ertapenem and meropenem. Subsequently, phenotypic susceptibility testing was performed by the gold standard method of agar dilution for both ertapenem and meropenem. The methods were largely concordant for both ertapenem (92% CA) and meropenem (98% CA) by CLSI standards (Table S3). Furthermore, Ana-CIM performance compared to Etest and agar dilution susceptibility by CLSI breakpoints was highly concordant for both ertapenem (87% CA Etest and 93% CA agar dilution) and meropenem (98% CA Etest and 97% CA agar dilution) (Table S4). As such, ertapenem Etest was maintained as the reference method for Ana-CIM performance assessment. Overall, Ana-CIM had 87% or greater CA compared to either meropenem or ertapenem AST tested by either Etest or agar dilution and interpreted by CLSI or EUCAST breakpoints (Table 2 and Tables S4 and S5).

We reviewed isolates with carbapenem categorical results that were discordant between CLSI and EUCAST guidelines. In this study, 8 isolates (13%) by ertapenem Etest exhibited MICs in this range of >0.5 μg/mL (EUCAST ertapenem-resistant breakpoint) and <16 μg/mL (CLSI ertapenem-resistant breakpoint) (Table S6). Six of these 8 isolates were tested in the clinical study, where 2 (C12 and C23) tested negative for carbapenemase production by Ana-CIM at all sites and C19 tested positive at 2/3 sites and indeterminate at one site. The remaining 3 isolates (C24, C25, and C26) tested positive across all 3 sites with 2 noted to have microcolonies. Isolates R2 and R5 tested positive by Ana-CIM, both with a mode of 6 mm; microcolonies were noted for both isolates. None of the 8 isolates had an IS element detected by PCR, though 6 of the 8 isolates had a CfiA callout by MALDI-ToF MS (Table S1).

DISCUSSION

With carbapenem resistance on the rise in B. fragilis, rapid methods readily adaptable to clinical laboratories will be needed to detect resistance and decrease the likelihood of carbapenem treatment failure. PCR-based methods that detect the carbapenemase gene cfiA and activating upstream insertion sequences are not practical in most clinical microbiology laboratory settings. Double-ended Etest strips of meropenem or imipenem with or without EDTA have been proposed; however, preliminary analyses indicate that sensitivity is highly variable based on the carbapenem utilized and the resistance level of the isolates tested. For meropenem ± EDTA and imipenem ± EDTA Etests, sensitivity of 87% and 27%, respectively, has been reported for detection of cfiA-positive B. fragilis isolates (18).

More recently, MALDI-ToF-based methods for carbapenemase detection and confirmation have been proposed using MALDI subtyping and MBT-STAR-Carba (Bruker Daltonik, Germany) (19). Using the MBT-STAR-Carba kit that measures imipenem hydrolysis following organism plus antibiotic incubation for up to 60 min, Cordovana and colleagues reported 100% sensitivity for detection of cfiA-positive B. fragilis isolates among a collection of 70 cfiA-positive and 33 cfiA-negative isolates (19). In contrast, they reported that off-label anaerobic adaptations of Carba NP and disk diffusion synergy tests had sensitivity of only 20.7% and 79.3%, respectively, and preferentially detected carbapenemase production in isolates with high-level carbapenem resistance (19). While the MBT-STAR-Carba appears promising, this test requires additional reagents and specific instrumentation (Microflex LT mass spectrometer) and software and is not widely utilized in clinical microbiology laboratories despite excellent performance characteristics.

In this study, we have described the Ana-CIM, a modification of the widely utilized carbapenem inactivation method (CIM) adapted for detection of carbapenemase production in B. fragilis isolates. There are two main modifications of the mCIM assay for the Ana-CIM assay. The first change is the broth medium utilized for the incubation step. For the mCIM assay, tryptic soy broth (TSB) is used; however, the Ana-CIM assay specifically utilizes Brucella broth, which is formulated for recovery of fastidious and anaerobic organisms such as B. fragilis. The second modification is the anaerobic incubation conditions immediately following the isolate inoculation into the broth medium with the meropenem disk. The mCIM mixture is incubated for 4 h at 35°C in ambient air, which is sufficient for Enterobacterales that replicate quickly in a nutrient-rich broth. During the development of the Ana-CIM, both 6- and 24-h anaerobic incubations were tested and there was no difference in zone sizes between the two incubation times. After the anaerobic incubation step, the Ana-CIM and mCIM are identical.

The Ana-CIM exhibited 87% (52/60) CA with ertapenem Etest results across all sites for the 60 isolates tested in the pilot, clinical, and reproducibility studies with no VME, 11% (4/36) ME, and 7% (4/60) mE. In addition to high CA compared to phenotypic susceptibility results, we also report that the Ana-CIM is reproducible across test sites and with 3 different brands of MH medium. While 3/5 reproducibility study isolates demonstrated 100% CA across all sites, some variability was observed with isolates R2 and R5. Interestingly, both of these isolates tested susceptible to ertapenem and meropenem when CLSI breakpoints were applied but were interpreted as ertapenem resistant or meropenem nonsusceptible using EUCAST breakpoints (see Table S1 in the supplemental material). Ana-CIM testing was positive for carbapenemase production in 96% (78/81) of instances for R2 and 49% (40/81) of instances for R5. Microcolonies were observed at multiple testing sites and on multiple medium types for both R2 and R5, suggesting that they are harboring a carbapenemase and may display a resistant phenotype (Table 4).

The variability in Ana-CIM results observed with R2 and R5 may be due to elevated but relatively low carbapenem MICs that may be interpreted as susceptible or resistant depending on the interpretive guideline applied. In addition to R2 and R5, 6 additional clinical isolates had ertapenem Etest MICs with discordant interpretations by CLSI versus EUCAST, of which 5/6 had consensus Ana-CIM interpretations across all sites (Table S6). The significance of isolates with carbapenem MICs in this range is unclear, as they are not commonly reported (6, 20).

The presence of microcolonies within the zone of inhibition of select isolates may also confound Ana-CIM interpretation. Colonies of the indicator strain, E. coli ATCC 25922, within the zone were interpreted as growth consistent with the CLSI guidance on interpretation of Enterobacterales and Pseudomonas aeruginosa mCIM (Fig. 2). Overall, microcolonies within the zone were infrequent, occurring in 6/60 of the isolates evaluated, but were more often observed in isolates with carbapenem MICs ranging from 2 to >32 μg/mL. Additional studies interrogating the genetic mechanism of resistance in isolates with intermediate carbapenem MICs or microcolonies within zones of inhibition by the Ana-CIM are needed to corroborate these findings.

Interestingly, molecular cfiA and IS detection was largely concordant with phenotypic resistance. All 60 study isolates had cfiA detected by PCR. Nineteen isolates had cfiA and IS detected via PCR, all of which were also phenotypically carbapenem resistant and Ana-CIM positive. Forty-one were cfiA positive but IS negative; of which 10 (24%) were resistant to either meropenem or ertapenem when evaluated by either CLSI or EUCAST breakpoints (Table S1). Eighty percent (8/10) of these isolates were Ana-CIM positive, further indicating that cfiA and IS PCR alone are insufficient to identify carbapenemase production in B. fragilis. Of these 8 isolates, 6 had a callout for CfiA by the Bruker Biotyper, suggesting that these isolates may harbor an IS that is not detected by the PCR assay.

This study has several strengths and weaknesses. Notable strengths include the inclusion of clinical B. fragilis isolates from multiple study sites, with a large proportion of resistant strains for analysis. Additionally, three different brands of MH agar were tested in three clinical microbiology laboratories in order to assess reproducibility. However, the overall study sample size is limited. Another limitation of this study is that the assay has not been evaluated for other Bacteroides species and thus can be deployed only in laboratory settings where species-level identification is available.

We demonstrated that this method has favorable performance characteristics in the hands of multiple test sites and reagent manufacturers and can be readily adapted by clinical microbiology laboratories with no additional instrumentation or software. These initial findings indicate the Ana-CIM can be an effective method for detection of carbapenemase production in B. fragilis isolates. For laboratories capable of anaerobic isolation, the Ana-CIM is a simple and low-cost addition to the test menu that can be used to confirm carbapenemase production in B. fragilis.