INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Fluoroquinolones have a good in vitro and clinical activity against isolates of the Salmonella species (1). Yet, during the last years several treatment failures with ciprofloxacin and other fluoroquinolones have been reported both in immunocompromised patients and in those with normal host defense (2, 10, 11, 13, 18, 19, 21, 24, 26, 27). According to the National Committee for Clinical Laboratory Standards (NCCLS) guidelines, which use the MICs of 1 and 4 µg/ml as respective breakpoints for susceptibility and resistance (15), these infections have been caused by ciprofloxacin-susceptible isolates. The majority of cases have involved Salmonella enterica serotype Typhi (hereafter Salmonella serovar Typhi) infections (10, 11, 24) and invasive infections with nontyphoidal salmonellas caused by strains which initially have been fully susceptible to ciprofloxacin with MICs of 0.064 µg/ml (2, 18, 19, 21, 26). After fluoroquinolone treatment failure, however, the ciprofloxacin MICs for these strains were 0.125 µg/ml.

Single point mutation in the quinolone resistance-determining region (QRDR) of the topoisomerase gene gyrA (amino acids 67 to 122) in salmonellas usually leads simultaneously to resistance against nalidixic acid, a nonfluorinated narrow-spectrum quinolone, and to decreased ciprofloxacin susceptibility (18, 22). It has been suggested that resistance to nalidixic acid may be an indicator of low-level resistance to ciprofloxacin (10, 14, 26). During our previous study focusing on the susceptibility of salmonellas to fluoroquinolones (7), we observed that the strain collection defined as susceptible according to the NCCLS guidelines could be divided into two subpopulations, one fully susceptible and the other with reduced susceptibility. The purpose of the present study was to confirm this finding in a large collection of Salmonella isolates and to illustrate the phenomenon by using scattergram analysis. In addition, we aimed at assessing whether resistance to nalidixic acid could be used to screen for decreased fluoroquinolone susceptibility in salmonellas. In doing so, we compared the susceptibilities of 1,010 epidemiologically unrelated Salmonella isolates to nalidixic acid and three fluoroquinolones.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Salmonella isolates and susceptibility testing. We included in this study a total of 1,010 clinical Salmonella isolates collected in Finland between 1995 and 1998. Of these isolates, 810 were collected in three different phases yearly from 1995 to 1997 (7) and 200 were collected in one phase in 1998. All isolates were considered to be epidemiologically unrelated based on their recovery from distinct sources. For each Salmonella outbreak recognized, only one isolate representing the epidemic strain was included. The Salmonella collection consisted of 83 different serotypes. The most prevalent serotypes were Salmonella serovar Enteritidis and Salmonella serovar Typhimurium, accounting for 27 and 25% of the isolates, respectively. The study collection contained three Salmonella serovar Paratyphi B and no Salmonella serovar Typhi isolates.

PCR and sequencing. The QRDR of the gyrA gene was sequenced from all isolates found to be nalidixic acid resistant and from a number of nalidixic acid-susceptible isolates selected based on variable disk zone diameters and ciprofloxacin MICs. Chromosomal DNA was prepared from each strain by boiling for 10 min and proteinase K digestion. The four oligonucleotide primers used in the PCR amplification and DNA sequencing of the gyrA gene fragments were as described by Ouabdesselam et al. (18). Oligonucleotides were synthesized by Eurogentec (Herstal, Belgique). The final volume of each PCR mixture was 50 µl, and the mixtures contained 2 U of DyNAzyme DNA polymerase, 1× PCR buffer (Finnzymes, Espoo, Finland), 0.01 µmol of each deoxynucleoside triphosphate, 20 pmol of each primer, and 5 µl of the template. PCR amplifications were performed in an automated thermal cycler, the DNA Engine PTC-200 (MJ Research, Inc., Watertown, Mass.), with 30 cycles of denaturation at 95°C for 1 min, annealing at 54°C for 1 min, and extension at 72°C for 2 min, followed by a final step with extension at 72°C for 5 min. Primers and free nucleotides were removed with the High Pure PCR Product Purification Kit (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer's instructions. Sequencing was performed with the ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit with AmpliTaq DNA Polymerase, FS (PE Applied Biosystems, Foster City, Calif.), and analyzed in an automatic DNA sequencer, ABI PRISM 377 DNA Sequencer (PE Applied Biosystems). The gyrA fragments of all mutated and five wild-type strains were sequenced in both directions.

Data analysis. The susceptibility data were analyzed by using the WHONET4 computer program, available from J. Stelling (WHO/EMC, Geneva, Switzerland). The nucleotide sequence data were assembled and edited by using SeqEd software version 1.0.3 (PE Applied Biosystems) and GCG program package version 10.0 (Genetics Computer Group, Madison, Wis.), and the sequences were compared with the published section of the sequence of the Salmonella serovar Typhimurium gyrA gene (6). Statistical analysis was done with the independent sample t test and the Mann-Whitney U test. P values of <0.05 were considered significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Correlation between MICs of fluoroquinolones and nalidixic acid. All 1,010 Salmonella isolates were classified as ciprofloxacin susceptible (MIC, 1 µg/ml) according to the NCCLS recommendations. A total of 62 isolates were resistant to nalidixic acid (MIC, 32 µg/ml). The MIC histogram of ciprofloxacin for all Salmonella isolates showed a bimodal distribution with a range of 0.008 to 0.5 µg/ml (Fig. 1). The isolates were divided into two populations based on nalidixic acid susceptibility, with the ciprofloxacin MICs ranging from 0.008 to 0.064 µg/ml for the nalidixic acid-susceptible population and from 0.032 to 0.5 µg/ml for the nalidixic acid-resistant population.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Ciprofloxacin (CIP) MIC histogram for 1,010 epidemiologically unrelated Salmonella isolates collected in Finland from 1995 to 1998. The MICs of ciprofloxacin are plotted on the x axis, and the number of isolates are plotted on the y axis. White columns indicate nalidixic acid-susceptible (n = 948) isolates, and black columns indicate nalidixic acid-resistant (n = 62) isolates.

View larger version (11K): [in this window] [in a new window]   FIG. 2.   Scattergrams for 1,010 Salmonella isolates correlating the MICs of nalidixic acid to those of ciprofloxacin (a), ofloxacin (b), and norfloxacin (c). The vertical dashed lines of panels a to c indicate the NCCLS breakpoint recommendations for susceptibility and resistance, respectively, to ciprofloxacin (MIC, 1 and 4 µg/ml), ofloxacin (MIC, 2 and 8 µg/ml), and norfloxacin (MIC, 4 and 16 µg/ml). The horizontal solid lines indicate the respective NCCLS breakpoint recommendations for nalidixic acid (MIC, 16 and 32 µg/ml). The numbers within the graph indicate the numbers of Salmonella isolates. CIP, ciprofloxacin; OFL, ofloxacin; NOR, norfloxacin; NAL, nalidixic acid.

Disk diffusion test results. All 129 Salmonella (the 62 nalidixic acid-resistant and 67 randomly selected nalidixic acid-susceptible) isolates evaluated by means of disk diffusion tests were classified as ciprofloxacin susceptible according to the NCCLS recommendations (inhibition zone diameter, 21 mm). The mean inhibition zone diameters around nalidixic acid and five fluoroquinolone disks (three different fluoroquinolones) for these isolates are shown in Table 1. The differences in the inhibition zone diameters between the two study groups were statistically significant regarding every fluoroquinolone tested, supporting the association between nalidixic acid resistance and decreased fluoroquinolone susceptibility.

Ciprofloxacin MICs versus disk diffusion tests. The correlations between the ciprofloxacin MICs and zone diameters around 5-µg ciprofloxacin disks for the 129 Salmonella isolates described above are shown in Fig. 3. All 31 isolates for which the ciprofloxacin MICs were 0.25 µg/ml had inhibition zone diameters of 33 mm. The inhibition zone diameter also was 33 mm for 16 isolates for which the MIC of ciprofloxacin was 0.125 µg/ml and 7 isolates for which the MICs of ciprofloxacin were between 0.064 and 0.016 µg/ml. All 50 isolates for which the ciprofloxacin MICs were 0.125 µg/ml had inhibition zone diameters of 37 mm. Also 38 isolates for which the MICs of ciprofloxacin were between 0.064 and 0.016 µg/ml were included in this category.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   Scattergram plotting the MICs of ciprofloxacin (x axis) and the inhibition zone diameters around a 5-µg ciprofloxacin disk (y axis) for 62 nalidixic acid-resistant and 67 nalidixic acid-susceptible Salmonella isolates. Nalidixic acid-resistant isolates (MIC, 32 µg/ml) are indicated with an asterisk, and the resistant isolates exhibiting a mutation in the gyrA gene are indicated with a double asterisk. The dashed lines are discussed in the text.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Scattergram plotting the MICs of ciprofloxacin (x axis) and the inhibition zone diameters around a 30-µg nalidixic acid disk (y axis) for 62 nalidixic acid-resistant and 67 nalidixic acid-susceptible Salmonella isolates. The horizontal dashed lines indicate the NCCLS recommendations for susceptibility (inhibition zone diameter, 19 mm) and resistance (inhibition zone diameter, 13 mm) to nalidixic acid. Isolates designated as resistant to nalidixic acid based on MIC (32 µg/ml) determinations are indicated with an asterisk, and the resistant isolates exhibiting a mutation in the gyrA gene are indicated with a double asterisk.

Screening for isolates with decreased ciprofloxacin susceptibility. The relevance of using the resistance to nalidixic acid as a marker for decreased ciprofloxacin susceptibility in salmonellas was evaluated by comparing the MICs of ciprofloxacin and nalidixic acid for the 1,010 Salmonella isolates included (Fig. 2a). When an MIC of ciprofloxacin of 0.125 µg/ml was adopted as a breakpoint, screening for nalidixic acid resistance (MIC, 32 µg/ml) led to the detection of all 50 isolates with decreased ciprofloxacin susceptibility (MIC, 0.125 µg/ml) and, in addition, of 12 of the 960 susceptible isolates. Thus, the sensitivity of the approach was 100%, and the specificity was 98.8%. When an MIC of ciprofloxacin of 0.25 µg/ml was selected as a breakpoint, screening for nalidixic acid resistance led to the detection of all 31 isolates with decreased ciprofloxacin susceptibility (MIC, 0.25 µg/ml) and, in addition, of 31 of the 979 susceptible isolates, resulting in a sensitivity of 100% and a specificity of 96.8% of the approach.

Nucleotide sequence analysis. The QRDR of the gyrA gene was sequenced from all 62 nalidixic acid-resistant isolates (50 for which the ciprofloxacin MICs were 0.125 µg/ml) and 23 nalidixic acid-susceptible isolates. All isolates for which the ciprofloxacin MICs were 0.125 µg/ml had a mutation in the QRDR of gyrA. Mutated isolates related to ciprofloxacin MICs and 5-µg ciprofloxacin disk zone diameters are shown in Fig. 3, and those related to ciprofloxacin MICs and 30-µg nalidixic acid disk zone diameters are shown in Fig. 4. Altogether, 58 (94%) nalidixic acid-resistant isolates had a mutation in the QRDR of gyrA, whereas none of the nalidixic acid-susceptible isolates analyzed had a mutation in that region. All of the mutations detected were localized at codon 83 or 87. Five different kinds of point mutations were found: two at codon 83 and three at codon 87 (Table 2).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have shown in this study that a collection of 1,010 Salmonella isolates classified as fluoroquinolone susceptible according to the NCCLS guidelines contained two distinct subpopulations, with the great majority of the isolates being entirely ciprofloxacin susceptible and a minority exhibiting reduced ciprofloxacin susceptibility (MICs ranging between 0.125 and 0.5 µg/ml). Although at the present time the MICs of 1 and 4 µg/ml are still widely accepted as breakpoints of ciprofloxacin susceptibility and resistance, respectively, many authors have focused attention on Salmonella isolates with decreased susceptibility (2, 4, 7, 10, 11, 13, 18, 19, 21, 24-27). Indeed, recognition of these strains is of concern due to the increasing number of treatment failures in invasive salmonellosis reported in association with reduced fluoroquinolone susceptibility. We believe that, in addition to the categories of resistant and fully susceptible, a category of decreased fluoroquinolone susceptibility could be useful in the susceptibility determinations of salmonellas. A corresponding classification is accepted for the Neisseria gonorrhoeae species (5, 8, 9, 17).

In most laboratories, the proportion of reported Salmonella strains with decreased fluoroquinolone susceptibility is still small compared to the fully susceptible population. Therefore, alertness is required to recognize these strains. It has been previously recommended, based mainly on solitary cases of fluoroquinolone treatment failures, that all isolates of the Salmonella species should be tested for nalidixic acid resistance in order to avoid reporting false susceptibility to fluoroquinolones (10, 26). In our collection of clinical Salmonella isolates, nalidixic acid susceptibility testing proved both sensitive and specific in screening for isolates with decreased ciprofloxacin susceptibility. Identification of nalidixic acid resistance by the disk diffusion method led to the detection of all isolates for which the MICs of ciprofloxacin were 0.125 µg/ml, providing a sensitivity of 100% and a specificity of 87.3% as a screening approach. Alternatively, nalidixic acid MICs can be employed as a basis for screening. Calculated by the MIC determinations for our Salmonella isolates, the sensitivity of finding isolates for which the MICs of ciprofloxacin were 0.125 µg/ml was 100% and the specificity was 98.8%, when an MIC of nalidixic acid of 32 µg/ml was used as a selection criterion. Technically, screening could be accomplished by using a selective breakpoint plate containing nalidixic acid in a concentration of 16 µg/ml, which allows the growth of Salmonella isolates for which the MICs of nalidixic acid are 32 µg/ml.

According to the NCCLS recommendations, all Salmonella isolates should be routinely tested for susceptibility to ampicillin, a quinolone, and trimethoprim-sulfamethoxazole, and the extraintestinal isolates should be tested, in addition, for susceptibility to chloramphenicol and an expanded-spectrum cephalosporin (17). In this context, most laboratories evidently use one of the fluoroquinolones, since nalidixic acid is not administered for the treatment of Salmonella infections. Therefore, an approach of including an additional nalidixic acid disk in all susceptibility testing may not be cost-effective in field laboratories. The results of the present study also suggest that the ciprofloxacin disk diffusion test can be applied to detect decreased ciprofloxacin susceptibility: a zone diameter of 37 mm around a 5-µg ciprofloxacin disk was appropriate as a selection criterion to find all Salmonella isolates for which the MICs were 0.125 µg/ml. Admittedly, a considerable number of isolates for which the MICs of ciprofloxacin were 0.064 µg/ml were also included in this category. Thus, a further screening with a 30-µg nalidixic acid disk (inhibition zone diameter, 13 mm) could be used to decrease the number of isolates referred to full-range ciprofloxacin MIC determinations.

In addition to a mutation in the QRDR of the gyrA gene, decreased permeability and active efflux have been reported to confer low-level resistance to fluoroquinolones (12, 20). In the present study, a mutation of the gyrA gene could be demonstrated in all isolates with decreased ciprofloxacin susceptibility and in 94% of the nalidixic acid-resistant isolates. All of the five different mutations detected have been previously described in the literature in association with fluoroquinolone resistance in salmonellas (2, 6, 18, 22, 23). No mutations in the QRDR of the gyrA gene were observed in four (6%) of our nalidixic acid-resistant isolates, indicating other mechanisms of resistance. At the present time, it is important to keep in mind that a mutation may not always translate into clinical resistance. Moreover, the fluoroquinolone treatment failures reported in salmonellosis have, so far, been anecdotal. The final proof of the clinical importance of these less fluoroquinolone susceptible Salmonella strains can be obtained only after careful, randomized, controlled studies. Such studies are urgently needed.

In our Salmonella population, all isolates manifesting decreased fluoroquinolone susceptibility were nalidixic acid resistant. Consistently, previous studies have shown that all Salmonella strains recovered from patients who failed fluoroquinolone therapy have been resistant to nalidixic acid, suggesting an alteration in DNA gyrase as a consequence of mutation (2, 10, 11, 18, 19, 21, 26, 27). This finding is interesting and could explain the emergence of clinical fluoroquinolone resistance in this particular group of isolates, since it is conceivable that the bacterial strains which have undergone one mutation are easily prone to a second mutation, resulting in high-level fluoroquinolone resistance (3). Unexpectedly, Salmonella strains isolated from patients after unsuccessful fluoroquinolone treatment have commonly expressed only low-level resistance to this antimicrobial group. However, there are data showing that the presence of nalidixic acid resistance, in and of itself, may occasionally influence the therapeutic outcome of salmonellosis. According to a recent report from Vietnam, treatment failures with short-course ofloxacin for uncomplicated typhoid were significantly more common in patients infected with nalidixic acid-resistant strains than in those infected with susceptible strains (27).

In recent years, many papers have focused on fluoroquinolone resistance in Salmonella isolates with special reference to strains for which the ciprofloxacin MICs were 0.125 µg/ml (2, 10, 11, 21, 24, 26, 27). In England and Wales, ciprofloxacin resistance (MIC, 0.25 µg/ml) in salmonellas increased from 0.3 to 2.1% between 1991 and 1994 (4). In that country, low-level resistance (MIC, 0.125 to 1 µg/ml) was identified in 1996 in 7% of Salmonella serovar Typhi, 4% of Salmonella serovar Paratyphi A, and 4% of nontyphoidal salmonellas (25). We have shown that in Finland ciprofloxacin resistance (MIC, 0.25 µg/ml) emerged during the years 1995 to 1997 among the domestic Salmonella isolates, with the proportion of resistant strains increasing from 0.0 to 2.2% (7). Among the foreign salmonellas isolated in our country, the simultaneous increase (from 2.0 to 8.4%) was statistically significant (P = 0.037). It is of note that similar strains would have been classified as susceptible in those microbiological laboratories which follow the current breakpoint recommendations. Thus, the lack of universally observed guidelines for breakpoints of susceptibility and resistance severely impedes the worldwide surveillance of the emergence and spread of fluoroquinolone-resistant Salmonella strains, as well as of those with decreased fluoroquinolone susceptibility.

In conclusion, we have demonstrated here the presence of a defined subpopulation with decreased fluoroquinolone susceptibility in our Salmonella collection. These isolates differed from the fully fluoroquinolone-susceptible population by being uniformly resistant to nalidixic acid and exhibiting a mutation in the QRDR of the gyrA gene. We regard it as important that microbiology laboratories endeavor to recognize these less susceptible Salmonella strains, in order to reveal their clinical importance and to survey their epidemic spread. In our large collection of clinical Salmonella isolates, detection of nalidixic acid resistance by the disk diffusion method proved useful as a tool to screen for decreased ciprofloxacin susceptibility.