Development of a Pefloxacin Disk Diffusion Method for Detection of Fluoroquinolone-Resistant Salmonella enterica

Abstract

Fluoroquinolones (FQs) are among the drugs of choice for treatment of Salmonella infections. However, fluoroquinolone resistance is increasing in Salmonella due to chromosomal mutations in the quinolone resistance-determining regions (QRDRs) of the topoisomerase genes gyrA, gyrB, parC, and parE and/or plasmid-mediated quinolone resistance (PMQR) mechanisms including qnr variants, aac(6′)-Ib-cr, qepA, and oqxAB. Some of these mutations cause only subtle increases in the MIC, i.e., MICs ranging from 0.12 to 0.25 mg/liter for ciprofloxacin (just above the wild-type MIC of ≤0.06 mg/liter). These isolates are difficult to detect with standard ciprofloxacin disk diffusion, and plasmid-mediated resistance, such as qnr, is often not detected by the nalidixic acid screen test. We evaluated 16 quinolone/fluoroquinolone disks for their ability to detect low-level-resistant Salmonella enterica isolates that are not serotype Typhi. A total of 153 Salmonella isolates characterized for the presence (n = 104) or absence (n = 49) of gyrA and/or parC topoisomerase mutations, qnrA, qnrB, qnrD, qnrS, aac(6′)-Ib-cr, or qepA genes were investigated. All isolates were MIC tested by broth microdilution against ciprofloxacin, levofloxacin, and ofloxacin and by disk diffusion using EUCAST or CLSI methodology. MIC determination correctly categorized all isolates as either wild-type isolates (MIC of ≤0.06 mg/liter and absence of resistance genes) or non-wild-type isolates (MIC of >0.06 mg/liter and presence of a resistance gene). Disk diffusion using these antibiotics and nalidixic acid failed to detect some low-level-resistant isolates, whereas the 5-μg pefloxacin disk correctly identified all resistant isolates. However, pefloxacin will not detect isolates having aac(6′)-Ib-cr as the only resistance determinant. The pefloxacin disk assay was approved and implemented by EUCAST (in 2014) and CLSI (in 2015).

INTRODUCTION

Human infections caused by Salmonella enterica subsp. enterica represent a major burden worldwide (1). Typhoidal Salmonella serotypes (S. enterica serotypes Typhi and Paratyphi A) cause enteric fever, a severe systemic and febrile illness, whereas nontyphoidal serotypes primarily cause self-limiting diarrhea with occasional bacteremia. Timely treatment with antimicrobial agents is critical for optimal treatment of both enteric fever and invasive nontyphoidal Salmonella infections. Fluoroquinolones (FQs) (e.g., ciprofloxacin) are highly efficient against fully susceptible Salmonella bacteria (i.e., isolates without any resistance mechanisms), whereas their efficacy is in doubt as soon as any resistance can be detected, and there is concern about the rapidly increasing fluoroquinolone resistance in Salmonella (2–5).

Fluoroquinolone resistance in Salmonella is mainly caused by chromosomal mutations in the quinolone resistance-determining regions (QRDRs) of the topoisomerase genes gyrA, gyrB, parC, and parE (6, 7). These mutations usually confer stepwise resistance; a single mutation is associated with a ciprofloxacin MIC of 0.12 to 0.5 mg/liter, whereas two or more mutations result in higher MIC values. Topoisomerase mutations are associated with resistance to the quinolone nalidixic acid (MIC > 16 mg/liter) (6).

In addition to the QRDR topoisomerase mutations, a number of plasmid-mediated quinolone resistance (PMQR) mechanisms have been described; the PMQR mechanisms include qnr variants, aac(6′)-Ib-cr, qepA, and oqxAB, with qnr genes being the predominant PMQR mechanism among Salmonella bacteria (7, 8). The PMQR mechanisms result in reduced susceptibility to ciprofloxacin (MIC of 0.125 to 1.0 mg/liter) but only a modest or no increase in susceptibility to nalidixic acid (MIC of 8 to 32 mg/liter) (8). Although the PMQR mechanisms confer only a moderate increase in fluoroquinolone MICs, they are clinically relevant; patients infected with both Salmonella Typhi and nontyphoidal Salmonella isolates with ciprofloxacin MICs of 0.125 to 1.0 mg/liter have more treatment failures and longer times to fever clearance than patients with isolates fully susceptible to ciprofloxacin (MICs ≤ 0.06 mg/liter) (2, 9–11).

Since Salmonella isolates with low-level fluoroquinolone resistance may be associated with failed response to fluoroquinolone treatment, it is important that these isolates are detected during routine antimicrobial susceptibility testing. This has been recognized by the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST), the two international bodies providing antimicrobial susceptibility testing guidelines. Initially, both EUCAST and CLSI included warnings for using fluoroquinolones for treating infections caused by Salmonella spp. with ciprofloxacin MICs of >0.06 mg/liter, and later, both organizations introduced species-specific MIC breakpoints for Salmonella and fluoroquinolones to aid in the detection of isolates with acquired resistance (12–14).

For laboratories using disk diffusion, detection of low-level-resistant isolates is challenging. With ciprofloxacin, an overlap in inhibition zone diameters between wild-type isolates and isolates with low-level resistance has been reported for the 5-μg disk (15–17). For many years, both CLSI and EUCAST therefore recommended the use of a nalidixic acid disk as a screening test. However, this does not adequately detect isolates with PMQR, which are often susceptible to nalidixic acid (MICs ≤ 16 mg/liter) (8).

The aim of this study was to investigate whether or not one of 16 evaluated fluoroquinolone disks could identify all known fluoroquinolone resistance mechanisms in Salmonella isolates.

MATERIALS AND METHODS

Bacterial isolates.

A total of 153 nontyphoidal Salmonella isolates were included in this study: 98 from Statens Serum Institut (SSI), Copenhagen, Denmark, and 55 through the National Antimicrobial Resistance Monitoring System at the Centers for Disease Control and Prevention (NARMS-CDC), Atlanta, GA, USA (see Table S1 in the supplemental material). To ensure a high proportion of isolates exhibiting difficult-to-detect low-level fluoroquinolone resistance, the collection consisted of 104 isolates with ciprofloxacin MICs in the range of 0.125 to 0.5 mg/liter and 49 isolates fully susceptible to nalidixic acid and ciprofloxacin (MIC ≤ 0.064 mg/liter). Serotype distribution was very similar in the two sets of isolates.

DNA isolation, PCR amplification, and sequencing.

All 153 isolates were investigated for the presence of gyrA and/or parC topoisomerase mutations, qnrA, qnrB, qnrD, qnrS, aac(6′)-Ib-cr, and qepA genes. For each isolate, crude genomic DNA was prepared by lysing the bacteria at 95°C for 10 min and collecting the supernatant after a brief centrifugation. The presence of these genes and mutations were investigated by PCR and Sanger sequencing using the following primer pairs (5′ to 3′) for the indicated genes and mutations: CTATGCGATGTCAGAGCTGG and TAACAGCAGCTCGGCGTATT for parC (18), ATGAGCGACCTTGCGAGAGAAATTACACCG and TTCCATCAGCCCTTCAATGCTGATGTCTTC for gyrA (19), GGATGCCAGTTTCGAGGA and TGCCAGGCACAGATCTTG for qnrA (20), GGMATHGAAATTCGCCACTG and TTTGCYGYYCGCCAGTCGAA for qnrB (21), CGAGATCAATTTACGGGGAATA and AACAAGCTGAAGCGCCTG for qnrD (22), TCGACGTGCTAACTTGCG and GATCTAAACCGTCGAGTTCGG for qnrS (20), TTGCGATGCTCTATGAGTGGCTA and CTCGAATGCCTGGCGTGTTTfor aac(6′)-Ib-cr (23), andTGGTCTACGCCATGGACCTCA and TGAATTCGGACACCGTCTCCG for qepA (24). Each PCR was performed in a 100-μl volume using AmpliTaq (PerkinElmer, Waltham, MA, USA) on a GeneAmp PCR system 2400 (PerkinElmer). Amplicons were purified using a QIAquick PCR purification kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions and subsequently sequenced on an ABI PRISM 373 DNA sequencer (PE Applied Biosystems, Foster City, CA, USA).

Antimicrobial susceptibility testing.

MICs for ciprofloxacin, levofloxacin, nalidixic acid, and ofloxacin were determined using broth microdilution (BMD) according to the ISO standard 20776-1 (25). Testing was performed using customized frozen (TREK Diagnostics/Thermo Fisher Scientific, Oakwood, OH) or lyophilized (TREK Diagnostics/Thermo Fisher Scientific, Basingstoke, United Kingdom) panels.

Disk diffusion (DD) was performed according to EUCAST and CLSI methodology, i.e., Mueller-Hinton (MH) agar medium, inoculum of 0.5 McFarland standard taken from fresh overnight cultures and incubation for 16 to 20 h at 35°C in ambient air (26, 27).

Escherichia coli ATCC 25922 was used as a quality control (QC) strain in all assays.

Identification of candidate disks suitable for screening for fluoroquinolone resistance.

Sixteen fluoroquinolone and quinolone disks (Table 1) were evaluated for their ability to detect low-level fluoroquinolone resistance in Salmonella enterica. This work was performed at the EUCAST Development Laboratory (EDL), Växjö, Sweden, and at the Reference Laboratory for Antibiotic Resistance at Statens Serum Institut (SSI), Copenhagen, Denmark. For this evaluation, 87 of the 153 isolates were included and tested on two brands of Mueller-Hinton media (BBL [Becton Dickinson, Baltimore, MD, USA] and Oxoid [Thermo Fisher Scientific, Basingstoke, United Kingdom]). All tests were read by two persons, giving four readings per isolate, i.e., a total of 87 isolates × 2 media × 2 persons = 348 readings per disk. Four disks were included in a second round of testing (ciprofloxacin [1-μg], enoxacin [10-μg], norfloxacin [2-μg], and pefloxacin [5-μg] disks) along with ciprofloxacin (5-μg), nalidixic acid (30-μg), levofloxacin (5-μg), and ofloxacin (5-μg) disks. This second evaluation was performed at three laboratories (EDL, SSI, and NARMS-CDC) and included 126 isolates (Table 2). All three sites used the same lots of disks and performed testing on two different types of MH media using commercial BBL Mueller-Hinton agar plates (Becton Dickinson) as common media. In addition, NARMS-CDC used commercial plates from Remel (Thermo Fisher Scientific, Waltham, MA, USA), EDL used in-house-prepared plates based on Mueller-Hinton powder from Oxoid, and SSI used in-house-prepared plates based on BBL MH powder from Becton Dickinson. Altogether, this resulted in 126 isolates × 3 laboratories × 2 media = 756 readings for each disk. Different brands of media and different readers were used to investigate the robustness of the method by simulating the everyday situation where media from different manufacturers would be used in many laboratories and where results would depend on the acuity of many readers. The results were therefore interpreted collectively rather than by individual media and reader.

TABLE 1.

Ranges of inhibition zone diameters for various quinolone disks versus ciprofloxacin susceptibility and fluoroquinolone resistance mechanisms for 87 isolates tested at two laboratories (348 readings per disk)

Antimicrobial agent	Disk strength (μg)	Range of inhibition zone diam (mm) for isolates with the following characteristic:					Reading overlap (%)b
		Ciprofloxacin MIC (mg/liter)		Fluoroquinolone resistance mechanisma
		≤0.06	>0.06	None (n = 37)	qnr (n = 31)	QRDR (n = 19)
Ciprofloxacin	1	26–36	15–28	26–36	15–26	16–28	6
	5	31–40	21–34	31–40	21–31	21–34	24
Enoxacin	10	26–34	14–25	26–34	16–25	14–24	0
Enrofloxacin	5	28–37	14–29	28–37	14–29	15–28	3
Gatifloxacin	2	25–37	13–27	25–37	13–23	15–27	13
	5	27–38	16–29	27–38	16–28	18–29	18
Levofloxacin	1	23–33	9–24	23–33	11–23	9–24	3
	5	28–39	18–29	28–39	20–29	18–29	11
Lomefloxacin	10	27–37	17–29	27–37	18–27	17–29	9
Nalidixic acid	5	6–21	6–14	6–21	6–14	6	97
	10	9–25	6–17	9–25	6–17	6	39
	30	19–30	6–23	19–30	6–26	6	42
Norfloxacin	2	26–36	14–24	26–36	14–23	14–24	0
Ofloxacin	5	27–37	16–28	27–37	18–27	16–28	4
Pefloxacin	5	26–35	6–26	26–35	12–26	6–24	1
Sparfloxacin	5	27–38	14–31	27–38	14–27	19–31	31

^aAmong the 87 isolates included, none exhibited the aac(6′)-Ib-cr resistance mechanism.

^bOverlap in zone diameter between isolates without and with resistance mechanisms.

TABLE 2.

Ranges of inhibition zone diameters for various quinolone disks versus ciprofloxacin susceptibility and fluoroquinolone resistance mechanisms for 126 isolates tested at three laboratories (756 readings per disk)

Antimicrobial agent	Disk strength (μg)	Range of inhibition zone diam (mm) for isolates with the following characteristic:						Reading overlap (%)a
		Ciprofloxacin MIC (mg/liter)		Fluoroquinolone resistance mechanism
		≤0.06	>0.06	None (n = 43)	qnr (n = 37)	QRDR (n = 45)	aac(6′)-Ib-cr (n = 1)
Ciprofloxacin	5	29–39	20–32	29–39	20–30	22–32	20–21	21
Enoxacin	10	24–33	12–25	24–33	12–25	13–24	16–18	3
Levofloxacin	5	27–36	18–29	27–36	19–27	18–29	22–23	16
Nalidixic acid	30	18–29	6–25	18–29	6–25	6	13–16	34
Norfloxacin	2	24b–31	11–24	24b–31	13–24	13–24	11–14	1b
Ofloxacin	5	24–35	15–27	24–35	16–25	15–27	19–21	21
Pefloxacin	5	24–34	6–24	24–34	11–24	6–23	14–16	0.3

^aOverlap in zone diameter between isolates without and with resistance mechanisms.

^bOne isolate with no identified FQ resistance mechanism and a ciprofloxacin MIC of ≤0.064 mg/liter had a 2-μg norfloxacin inhibition zone diameter of 21 mm. Including this isolate would result in a 14% overlap.

Validation of the 5-μg pefloxacin disk.

Once we had chosen to develop the method on the pefloxacin disk, two steps were taken to validate the disk diffusion assay. As a first step, differences in pefloxacin disk potency among disks from different manufacturers was assessed by testing 24 selected isolates against 5-μg pefloxacin disks from Becton Dickinson, Bio-Rad (Marnes-la-Coquette, France), Mast Diagnostic (Bootle, Merseyside, United Kingdom), and Oxoid. The disk potency was investigated independently at two sites using a bioassay (28).

Second, 5-μg pefloxacin disks (Oxoid) were evaluated on consecutive clinical isolates as part of the routine disk diffusion testing at the Department of Clinical Microbiology, Kronoberg and Blekinge counties, Sweden, alternating between Mueller-Hinton agar from BD, Bio-Rad, and Oxoid and 10 different readers.

RESULTS

A fluoroquinolone resistance mechanism was identified in all 104 ciprofloxacin non-wild-type isolates (ciprofloxacin MIC > 0.06 mg/liter); 53 isolates harbored a topoisomerase mutation in gyrA, 50 isolates harbored a qnr gene, and a single isolate harbored the aac(6′)-Ib-cr gene. The corresponding MIC values for levofloxacin were >0.125 mg/liter, and for ofloxacin, the MICs were >0.125 mg/liter. For nalidixic acid, isolates with a topoisomerase mutation all displayed nalidixic MICs of >16 mg/liter, whereas isolates with a plasmid-mediated mechanism exhibited MIC values of 4 to >64 mg/liter (see Table S1 in the supplemental material).

In the 49 isolates with ciprofloxacin MICs of ≤0.06 mg/liter, there were no fluoroquinolone resistance mechanisms [gyrA or parC, qnr, qep, or aac(6′)-Ib-cr gene]. The corresponding MIC values were ≤0.125 mg/liter for levofloxacin and ofloxacin and ≤16 mg/liter for nalidixic acid. Thus, with standardized broth microdilution MIC determination, ciprofloxacin, levofloxacin, and ofloxacin, accurately and equally well, categorized the 49 isolates as belonging to the wild type.

For E. coli ATCC 25922, all MIC values for ciprofloxacin (n = 23), levofloxacin (n = 13), ofloxacin (n = 18), and nalidixic acid (n = 18) were within the established QC ranges (13, 29).

Identification of candidate disks suitable for screening for fluoroquinolone resistance.

The results for each of the 16 disks are shown in Table 1. Since there was a perfect correlation between MICs and resistance mechanisms for all three fluoroquinolones, disk diffusion results were correlated with ciprofloxacin MIC values and resistance mechanisms throughout. Our results confirmed that the 30-μg nalidixic acid disk is reliable only for the detection of isolates with topoisomerase mutations. A considerable overlap, and thus poor distinction, between wild-type and non-wild-type isolates was also observed for the 5-μg ciprofloxacin, 5-μg levofloxacin, and 5-μg ofloxacin disks. In contrast, the 1-μg ciprofloxacin, 10-μg enoxacin, 2-μg norfloxacin, and 5-μg pefloxacin disks were able to distinguish between wild-type and non-wild-type isolates and were for this reason considered for the next stage of the study where three laboratories (EDL, SSI, and CDC) were involved in the evaluation (Table 2).

Again, excellent results were obtained with the 5-μg pefloxacin disk—only 2 of 756 readings (0.3%) resulted in overlapping zone diameters between isolates with and without a resistance mechanism. When each of the MH agars was analyzed separately, clear separation between wild-type and non-wild-type isolates was achieved on all media (see Fig. S1a to S1f in the supplemental material). This was not the case with the 5-μg ciprofloxacin disk (Fig. S1g to S1l).

As shown in Tables 1 and 2, 1-μg ciprofloxacin, 10-μg enoxacin, and 2-μg norfloxacin disks also performed well. Neither of the disks performing well were part of the CLSI and EUCAST standard panel of disks. Pefloxacin was chosen for further development because of its slightly better performance on the individual brands of Mueller-Hinton media (data not shown) and because it was available from at least four manufacturers. Based on the initial disk diffusion results, a tentative screening breakpoint of an inhibition zone diameter of ≥24 mm for wild-type isolates and of <24 mm for non-wild-type isolates was set for pefloxacin.

For E. coli ATCC 25922, 5-μg pefloxacin inhibition zone diameters from repeated testing at the three test sites using MH from three different manufacturers ranged from 27 to 33 mm, with 96% of inhibition zone diameters within 27 to 32 mm.

Validation of the 5-μg pefloxacin disk.

The variability of 5-μg pefloxacin disks from different manufacturers was evaluated at EDL using 24 isolates with zone diameters close to the tentative screening breakpoint of 24 mm. Disks from four manufacturers (BD, Bio-Rad, Mast, and Oxoid) were tested on MH agar from three manufacturers (BD, Bio-Rad, and Oxoid). Disks from BD and Mast produced very similar inhibition zones; Oxoid disks resulted in slightly larger zone diameters, whereas the inhibition zones obtained with the Bio-Rad disks were substantially larger (Fig. 1). The same was observed for E. coli ATCC 25922, where inhibition zone diameters from repeated testing ranged from 26 to 31 mm for disks from BD, Mast, and Oxoid (n = 114), whereas inhibition zone diameters for Bio-Rad pefloxacin disks ranged from 32 to 33 mm (n = 12).

FIG 1.

(A to D) Inhibition zone diameters for 5-μg pefloxacin disks per disk manufacturer (Oxoid[A], Mast [B], BD [C], and Bio-Rad [D]) for 24 selected Salmonella enterica isolates on Mueller-Hinton agar from three manufacturers. The Salmonella enterica isolates had fluoroquinole resistance due to chromosomal mutations in the quinolone resistance-determining regions (QRDRs) of topoisomerase genes or plasmid-mediated quinolone resistance (PMQR) mechanisms, such as qnr.

To explore these differences, the active compound contents of the different disks were measured independently at two sites using a bioassay, which confirmed that the pefloxacin contents in three of the disks were very similar and close to the target of 5 μg, whereas the content of the Bio-Rad disks was more than twice as high (detailed data not shown). Based on these findings, Bio-Rad was informed, and the results for the Bio-Rad disk were excluded from further analysis.

The performance of the 5-μg pefloxacin disk was further evaluated by testing consecutive clinical isolates of Salmonella spp. as part of the routine in the Departments of Clinical Microbiology in Blekinge and Kronoberg counties, Sweden. The resulting inhibition zone diameter distribution for wild-type isolates ranged from 24 to 36 mm.

Validation of the tentative 5-μg pefloxacin disk screening breakpoint.

Each isolate was tested and/or read 4 to 22 times by different technicians, depending on whether or not the isolate was part of one, several, or all the evaluation steps. The aggregated 1,391 pefloxacin readings on the 153 investigational isolates were from MH agar from four manufacturers and disks from three manufacturers.

Although the low end of wild-type organisms and the high end of the non-wild-type isolates coincided at an inhibition zone diameter of 24 mm, there was very little overlap between isolates without and with FQ resistance mechanisms (Fig. 2A and B). This corresponded to our tentative screening breakpoint of an inhibition zone diameter of <24 mm for resistance. For isolates with fluoroquinolone resistance mechanisms, pefloxacin inhibition zone diameters were 24 mm (six readings on one isolate) or more (one reading of 25 mm and one reading of 26 mm) for eight readings (0.6%). These readings were all with Oxoid disks and belonged to two isolates, one with qnr (six readings) and one with a topoisomerase mutation (two readings), both isolates with ciprofloxacin MICs of 0.125 mg/liter.

FIG 2.

(A to D) Aggregated inhibition zone diameter distributions for Salmonella enterica. (A) Pefloxacin (5 μg) versus fluoroquinolone resistance mechanism; (B) pefloxacin (5 μg) versus ciprofloxacin MIC (milligram/liter); (C) ciprofloxacin (5 μg) versus fluoroquinolone resistance mechanism; (D) ciprofloxacin (5 μg) versus ciprofloxacin MIC (milligram/liter). The number of isolates with the indicated inhibition zone diameters are shown. Vertical lines indicate zone diameter breakpoints for susceptible isolates for pefloxacin (CLSI and EUCAST) (A and B) and for ciprofloxacin (CLSI) (C and D). An asterisk indicates a false susceptibility result by disk diffusion (readings are for two isolates, all on disks from Oxoid).

The correlation between 5-μg ciprofloxacin disk and fluoroquinolone resistance mechanisms and ciprofloxacin MIC values are shown in Fig. 2C and D.

DISCUSSION

Detection of low-level fluoroquinolone resistance in Salmonella spp. was always a challenge. For many years, the use of nalidixic acid resistance as a marker for FQ resistance solved this problem. However, with the discovery of plasmid-mediated fluoroquinolone resistance which exhibits resistance to the fluoroquinolones but often not to nalidixic acid, this was no longer a perfect surrogate test. In many parts of the world, routine determination of MIC values is not a standard procedure, which is why the loss of the robust nalidixic acid disk test was problematic.

The aim of this study was to identify a disk that reliably detects fluoroquinolone resistance in Salmonella isolates irrespective of resistance mechanism. We identified 16 commercially available quinolone and fluoroquinolone disks. These disks were investigated against a large selection of Salmonella enterica isolates genetically confirmed to have or not have the known variants of fluoroquinolone resistance mechanisms. The highly diverse collection of strains, from both Europe and the United States, represented most of the phenotypes described in the literature. We had only one isolate with the aac(6′)-Ib-cr gene despite asking many international colleagues. It should be noted that this isolate was also resistant to levofloxacin and ofloxacin but had no mutations in the QRDR region or a PMQR and must have an unidentified resistance mechanism, i.e., a mutation outside the QRDR region, as aac(6′)-Ib-cr does not confer resistance to these two fluoroquinolones (30).

Our studies confirmed that none of the traditional fluoroquinolone disks nor the nalidixic acid disk were able to discriminate between isolates without (wild-type) and with (non-wild-type) resistance mechanisms. We also confirmed that MIC determination by broth microdilution with either ciprofloxacin, levofloxacin, or ofloxacin was able to reliably discriminate between wild-type and non-wild-type isolates.

We chose to develop pefloxacin because of its consistently good performance on several brands of MH media in distinguishing wild-type and non-wild-type isolates, because pefloxacin behaved slightly better than the other disks when looking at individual brands of MH agar, because we had to choose one disk for further development and because pefloxacin disks were available from four manufacturers.

However, we also identified drawbacks with the choice of pefloxacin. It does not detect resistance mediated by the aac(6′)-Ib-cr gene, as ciprofloxacin and norfloxacin are the only fluoroquinolones which possess the piperazynil amide side chain which is the target for the enzyme encoded by aac(6′)Ib-cr (30). This resistance type seems to be very rare, and the one isolate we were able to include had a second unrecognized resistance mechanism, as it was resistant to levofloxacin and ofloxacin. Furthermore, despite the fact that we identified at least four manufacturers, pefloxacin disks are currently not commercially available in some countries, including the United States.

Partly based on the work presented here, both CLSI and EUCAST decided to recommend that pefloxacin be used as a single surrogate marker for fluoroquinolone resistance (31, 32). Two QC studies have been performed for CLSI and EUCAST. Using Gavin statistics, CLSI has established a QC inhibition zone diameter range of 25 to 33 mm for E. coli ATCC 25922 and pefloxacin (5 μg) (31), whereas EUCAST recommends a range of 26 to 32 mm and a target value of 29 mm (32).

In conclusion, we confirmed that fluoroquinolone-resistant Salmonella isolates including isolates with low-level resistance can be reliably detected by BMD MIC determination of either ciprofloxacin, levofloxacin, or ofloxacin but not with disk diffusion using standard disks of these antimicrobials or nalidixic acid. Instead, we developed a disk diffusion method based on the 5-μg pefloxacin disk for detection of fluoroquinolone-resistant Salmonella. Pefloxacin proved to be an excellent surrogate marker for fluoroquinolone resistance for disk diffusion assays except for isolates having the aac(6′)-Ib-cr gene as the only resistance determinant, but since the beginning of these studies, we have been able to acquire only one such isolate. The pefloxacin disk diffusion assay represents a practical and economical method which may be especially valuable in resource-limited settings where disk diffusion might be the only susceptibility testing method available. The pefloxacin disk assay was approved and implemented by EUCAST in 2014 and CLSI in 2015.