Evaluation of Susceptibility Testing To Detect Fluoroquinolone Resistance Mechanisms in Streptococcus pneumoniae

Decreased susceptibility ofStreptococcus pneumoniae to fluoroquinolones has been documented (3, 8) and in one study was associated with increased fluoroquinolone use (3). Fluoroquinolones inhibit bacterial growth by trapping gyrase or topoisomerase IV on DNA as ternary drug-enzyme-DNA complexes (4).

Decreased susceptibility to fluoroquinolones may develop as a result of changes in the fluoroquinolone targets by mutations in the genes which encode ParC in topoisomerase IV and GyrA in DNA gyrase and/or by diminished drug accumulation by efflux (7, 9, 10, 17, 19, 20). Mutations in gyrB have not to date been associated with decreased susceptibility to the fluoroquinolones inS. pneumoniae, and whether mutations in parEalter the susceptibility is still controversial (1, 16, 17). Mutations in parC and gyrA occur in the quinolone resistance-determining region (QRDR) of the genes. InS. pneumoniae, the target to which the fluoroquinolone has the greatest binding affinity (primary target) depends on the fluoroquinolone (7, 9, 10, 17, 19). For levofloxacin and ciprofloxacin, the primary target is ParC. For gatifloxacin and moxifloxacin, the primary target is GyrA (18). Mutations in the primary target that result in a reduction in the affinity of the fluoroquinolone are associated with an increase in the MIC. A subsequent mutation in the original secondary target will result in a further increase in the MIC (7, 9, 10, 17, 19, 20). Laboratory identification of isolates with decreased susceptibility to fluoroquinolones is paramount for the effective treatment of patients with pneumococcal disease. Of equal importance may be the ability to detect isolates that have a fluoroquinolone resistance mechanism, even if the isolate is still susceptible, since this organism may more readily become resistant during therapy when a subsequent mutation occurs in the original secondary target. In this study, we compared the activities of ciprofloxacin and levofloxacin by broth microdilution, disk diffusion, and E-test susceptibility testing procedures against clinical isolates of S. pneumoniae for which the fluoroquinolone MICs varied in order to determine if the results could be used to reliably distinguish an isolate with one or more fluoroquinolone resistance mechanisms from one with no resistance mechanisms.

S. pneumoniae clinical isolates were from the Canadian Bacterial Surveillance Network; their collection and identification have been detailed elsewhere (1, 3). The initial in vitro broth microdilution testing and subsequent disk diffusion testing of ciprofloxacin (using a 5-μg disk; Oxoid, Nepean, Ontario, Canada) and levofloxacin (using a 5-μg disk; Becton Dickinson and Co., Cockeysville, Md.) were done according to National Committee for Clinical Laboratory Standards (NCCLS) (3, 12, 13). NCCLS disk diffusion testing breakpoints for levofloxacin are as follows: susceptible, ≥17 mm; intermediate >13 and <17 mm; and resistant, ≤13 mm. MIC breakpoints are as follows: susceptible, ≤2 μg/ml; intermediate, 4 μg/ml; and resistant, ≥8 μg/ml. The antimicrobials were provided by their respective manufacturers. The E-test was performed according to the manufacturer's instructions. Following storage at −70°C, isolates were thawed and subcultured onto blood agar (Oxoid) twice before susceptibility testing was performed. A total of 101 clinical isolates were selected for analysis; these included 75 isolates for which the ciprofloxacin MIC was ≥4 μg/ml (MICs of 64 μg/ml for 6 isolates, 32 μg/ml for 4, 16 μg/ml for 12, 8 μg/ml for 12, and 4 μg/ml for 41), 14 for which the MIC was 2 μg/ml, 3 for which the MIC was 1 μg/ml, and 9 for which the MIC was <1 μg/ml.

Crude cell lysates were used as DNA templates for PCR. Briefly, a loopful of overnight growth on Columbia nutrient agar supplemented with 5% sheep blood was suspended in 100 μl of lysis buffer (100 mM NaCl, 10 mM Tris-HCl [pH 8.3], 1 mM EDTA, 1% Triton X-100) and boiled for 10 min; 10 μl of the supernatant was used as a template in a 50-μl reaction volume. QRDR regions of parC and gyrA of all strains were amplified by PCR. Primers were based on published sequences (14, 16). Amplification products were purified using either the QIAquick PCR purification kit (Qiagen Inc., Mississauga, Ontario, Canada) or the Concert Rapid PCR Purification (Life Technologies, Burlington, Ontario, Canada).

DNA sequencing was performed by ABI Prism Big Dye terminator cycle sequencing with the ABI 377 automated sequencer (PE Applied Biosystems, Mississauga, Ontario, Canada). Nucleotide and protein sequence comparisons were performed using the multiple alignment sequence function of Vector NTI Suite software (InforMax Inc., Bethesda, Md.). The GenBank accession numbers for wild-type sequences used for comparison purposes were Z67739 for parC andparE (15), AB010387 for gyrA, andZ67740 for gyrB (11).

Isolates were examined for active efflux by agar dilution on Mueller-Hinton agar containing 5% sheep blood in the presence of ciprofloxacin with or without 10 mg of the alkaloid reserpine (Sigma Chemical Co., St. Louis, Mo.) per ml (2). Strains for which the ciprofloxacin MIC decreased at least fourfold in the presence of reserpine were considered positive for active efflux.

Using two-by-two tables, a wide range of MICs or disk diffusion diameters were tested to find the threshold with the highest sensitivity and specificity for detecting resistance mechanisms. For positive and negative predictive values, the estimated prevalence of these resistance mechanisms in the 7,551 isolates from the Canadian Bacterial Surveillance Network was used (3).

There were 27 isolates with mutations in the QRDR of bothparC and gyrA, 3 of which were also positive for efflux. Fifty isolates had an isolated parC mutation, and none had an isolated gyrA mutation. Of those with aparC mutation but no gyrA mutation, 14 were also positive for efflux. Finally, there were 24 isolates with neither aparC nor a gyrA mutation, 9 of which were positive for efflux, leaving 16 with no identified resistance mechanisms.

Figure 1A graphs the broth microdilution MICs versus the disk diffusion diameters for levofloxacin, with the breakpoints for each shown. There were no major, or very major, errors. Figure 1B graphs the broth microdilution MICs versus the disk diffusion diameters for ciprofloxacin. Table 1summarizes the results for the three susceptibility testing methods. The NCCLS breakpoints for levofloxacin classified the 27 isolates with both parC and gyrA mutations as intermediate or resistant. There was only one isolate with aparC mutation but no gyrA mutation that was nonsusceptible (intermediate susceptibility). Interestingly, it did have a very active efflux mechanism (the ciprofloxacin MIC for this isolate decreased by 16-fold in the presence of reserpine). Other than for this isolate, efflux did not contribute to an increase in the levofloxacin MIC that resulted in a strain being classified as nonsusceptible.

Table 2 summarizes the test characteristics of the three different susceptibility testing methods for detecting isolates with mutations in both parC andgyrA. MIC and disk diffusion values that favored a higher sensitivity were chosen. The prevalence used to determine the positive and negative predictive value is 0.4% (27 of 7,551 isolates). For detecting isolates with both gyrA and parCmutations, the levofloxacin E-test and disk diffusion tests were the most sensitive and specific. Using the levofloxacin E-test, a MIC of >2 μg/ml had a sensitivity of 100% and a specificity of 99%. The levofloxacin disk diffusion test had the same sensitivity and specificity using a disk diameter of less than 17 mm. These MIC and disk diffusion values are the same as the NCCLS breakpoints.

Table 3 considers only the levofloxacin-susceptible isolates (levofloxacin MIC of ≤2 μg/ml,n = 73) and summarizes the test characteristics of the three different susceptibility testing methods for detecting isolates with only a mutation in parC from among these isolates. No one MIC or disk diffusion value had a high sensitivity and specificity. For all testing methods, the negative predictive value was 100%. The prevalence used to calculate positive and negative predictive values was 1.0% (77 of 7,551 isolates). For detecting isolates with either aparC mutation or an active efflux mechanism among susceptible isolates, the test characteristics could be slightly improved. For example, in the best case, a disk with a diameter of <22 mm for levofloxacin had a sensitivity of 79% and a specificity of 93% (other data not shown).

S. pneumoniae isolates with mutations in the QRDR regions of both parC and gyrA genes are often not only resistant to levofloxacin but also nonsusceptible to moxifloxacin and gatifloxacin (3). We were able to reliably predict those isolates with both parC and gyrA mutations as well as those with no resistant mechanisms. However, it may be more important clinically and epidemiologically to be able to detect those isolates with first-step mutations. The likelihood of a wild-type isolate acquiring mutations in both of these genes simultaneously is on the order of 1 in 10¹² to 1 in 10¹⁴(6). However, if an infecting organism has a mutation in one of the targets, the likelihood of a mutation occurring in the other target is on the order of 1 in 10⁶ to 1 in 10⁷ (6). In pneumococcal pneumonia, the bacterial burden is >10¹⁰, making the likelihood of a spontaneous mutation in the second target high (5). Unfortunately, no test had both a sensitivity and a specificity adequate for reliably identifying susceptible isolates with single mutations. Future studies should address this deficiency.

The large number of clinical isolates in our study enhances the validity of our findings with recognized fluoroquinolone-resistant mechanisms. A weakness of the study is that all the mechanisms of resistance were selected for during a time when the only fluoroquinolones in clinical use had preferential affinity for ParC over GyrA (i.e., norfloxacin, ofloxacin, ciprofloxacin, and levofloxacin). To further validate our findings, it may be necessary to assess clinical pneumococcal isolates with fluoroquinolone resistance mechanisms that have been selected for both by agents that have preferential ParC affinity and the newer agents with preferential GyrA affinity (i.e., moxifloxacin and gatifloxacin). Finally, these results are based primarily on a selected collection of isolates for most of which the ciprofloxacin MICs were ≥4 μg/ml, and the prevalence values could only be estimated from this small collection.