|Home | About | Journals | Submit | Contact Us | Français|
β-Lactamase-negative ampicillin-resistant (BLNAR) isolates of Haemophilus influenzae have been emerging in some countries, including Japan. The Clinical and Laboratory Standards Institute has only a susceptible MIC breakpoint (≤1 μg/ml) for piperacillin-tazobactam and a disclaimer comment that BLNAR H. influenzae should be considered resistant, which was adapted without presentation of data. In addition, fluoroquinolone-resistant H. influenzae isolates have recently been occasionally reported worldwide. To address these problems, we examined susceptibilities to β-lactams, including piperacillin-tazobactam, and ciprofloxacin by microdilution and disk diffusion (only for piperacillin-tazobactam) methods, against a total of 400 recent H. influenzae clinical isolates, including 100 β-lactamase-negative ampicillin-susceptible, β-lactamase-positive ampicillin-resistant, BLNAR, and β-lactamase-positive amoxicillin-clavulanate-resistant (BLPACR) isolates each. BLNAR and BLPACR isolates were tested by PCR using primers that amplify specific regions of the ftsI gene. We also detected mutations in quinolone resistance-determining regions (QRDRs) by direct sequencing of the PCR products of DNA fragments. Among β-lactams, piperacillin-tazobactam exhibited potent activity against all isolates of H. influenzae, with all MICs at ≤0.5 μg/ml (susceptible). A disk diffusion breakpoint for piperacillin-tazobactam of ≥21 mm is proposed. We confirmed that all BLNAR and BLPACR isolates had amino acid substitutions in the ftsI gene and that the major pattern was group III-like (87.5%). One ciprofloxacin-resistant isolate (MIC, 16 μg/ml) and 31 ciprofloxacin-susceptible isolates (MICs, 0.06 to 0.5 μg/ml) had amino acid changes in their QRDRs. Piperacillin-tazobactam was the most potent β-lactam tested against all classes of H. influenzae isolates. It is possible that fluoroquinolone-resistant H. influenzae will emerge since several clinical isolates carried mutations in their QRDRs.
Haemophilus influenzae is one of the most common causes of community-acquired respiratory tract infections, such as pneumonia, and serious invasive diseases, including bacteremia and meningitis. The first β-lactamase-positive ampicillin-resistant (BLPAR) H. influenzae isolate was reported in the 1970s in the United States (20, 30), and the first β-lactamase-negative ampicillin-resistant (BLNAR) isolate of H. influenzae was reported in 1980 (23); it caused endocarditis and meningitis in a 79-year-old female patient in 1977, also in the United States. The mechanism of resistance of the BLNAR isolates involves mutations of the ftsI gene encoding penicillin binding protein 3 (PBP3), resulting in decreased affinity of PBP3 (2, 31). Although the frequencies of BLNAR isolates of H. influenzae are still low in the United States (8, 19) and have remained relatively constant in longitudinal European surveillance (18), BLNAR isolates of H. influenzae have been emerging in some countries (14), particularly in Japan (13, 15, 31), Spain (10), and recently Korea (21). In 1997, β-lactamase-positive amoxicillin (amoxicilline)-clavulanate-resistant (BLPACR) isolates of H. influenzae were also reported from the United States (8) and exhibited two characteristics of both BLPAR and BLNAR. Thus, on the basis of production of β-lactamase and susceptibilities to ampicillin and amoxicillin-clavulanate, clinical isolates of H. influenzae can be divided into four types, β-lactamase negative ampicillin susceptible (BLNAS), BLPAR, BLNAR, and BLPACR.
Data on in vitro susceptibilities to β-lactams, including piperacillin-tazobactam, are not sufficient for H. influenzae isolates, particularly in the case of BLNAR and BLPACR types. The Clinical and Laboratory Standards Institute (CLSI; formerly the National Committee for Clinical Laboratory Standards) established H. influenzae susceptibility breakpoints between 1990 and 1995, especially for penicillins, to delineate β-lactamase-positive isolates. Few BLNAR strains were available for analysis at that time, but with their increase in prevalence, modification of breakpoints and surrogate testing guidelines should be considered. Although CLSI has disk and/or MIC susceptibility testing breakpoints for multiple agents, for piperacillin-tazobactam it has only a MIC breakpoint (≤1 μg/ml) and a disclaimer comment that BLNAR H. influenzae should be considered resistant regardless of MIC results, despite the finding that piperacillin alone is very active against BLNAR H. influenzae (24).
Furthermore, fluoroquinolone-resistant H. influenzae isolates have recently been occasionally reported worldwide (1, 11, 25, 27), though they have rarely been reported in Japan (16, 17). More recently, however, Yokota et al. (32) reported that fluoroquinolone-resistant H. influenzae isolates have been emerging in the elderly in Japan, with some of the strains exhibiting levofloxacin MICs of 16 μg/ml or higher.
We therefore examined the susceptibilities to β-lactams, including piperacillin-tazobactam, and ciprofloxacin of a total of 400 recent H. influenzae clinical isolates gathered from throughout Japan, including 100 each of BLNAS, BLPAR, BLNAR, and BLPACR isolates. We confirmed mutations of ftsI in BLNAR and BLPACR H. influenzae isolates. We also detected mutations in quinolone resistance-determining regions (QRDRs) in isolates with ciprofloxacin MICs of 0.06 μg/ml or higher (susceptibility breakpoint of ≤1 μg/ml) and an additional 50 isolates with ciprofloxacin MICs of 0.03 μg/ml or lower.
(Some of these findings were presented at the Annual Meeting of the American Thoracic Society 2008, Toronto, Canada, May 2008, abstract no. 858.)
In the initial study, a total of 400 recent H. influenzae clinical isolates from 138 hospitals and clinics throughout Japan stored at Mitsubishi Chemical Medience Corporation, Tokyo, Japan, were used. Consecutive isolates were studied without repetition of patients. Table Table11 summarizes numbers of affiliations, duration of isolation, and the age range of patients for each group of H. influenzae isolates. In the additional study, a total of 50 H. influenzae clinical isolates, which were different from the above isolates and were associated with low MICs of ciprofloxacin (≤0.03 μg/ml), were used. H. influenzae strains were identified and confirmed by colony morphology, Gram staining, growth in chocolate agar but not in blood agar, the catalase test, and X and V factor requirements. Production of β-lactamase was examined using a nitrocefinase disk (Nippon Becton Dickinson Co., Ltd., Tokyo, Japan). Prior to the study, the isolates had been kept in litmus milk broth (Difco Laboratories, Detroit, MI) containing 10% glycerol at −80°C.
Antimicrobial susceptibility testing was performed by both the microdilution method and the disk diffusion method. MICs were determined by the reference broth microdilution methods of the CLSI using Haemophilus test medium (3) for ampicillin, ampicillin-sulbactam, amoxicillin-clavulanate, piperacillin, piperacillin-tazobactam, ceftriaxone, imipenem, meropenem, and a non-β-lactam control, ciprofloxacin. Disk diffusion testing was simultaneously performed using the CLSI M2-A9 (4) method for piperacillin-tazobactam. Two strains of H. influenzae, ATCC 49247 (a BLNAR strain) and ATCC 49766 (a BLNAS strain), were used for quality control for both methods. Interpretations of the test results were made using the criteria published in M100-S17 (5).
H. influenzae isolates that had elevated ampicillin MICs (intermediate [2 μg/ml] or resistant [≥4 μg/ml]) but were negative for β-lactamase were judged to be BLNAR. Of β-lactamase-producing isolates, those resistant to amoxicillin-clavulanate were judged to be BLPACR.
H. influenzae isolates with BLNAR and BLPACR phenotypes were tested by PCR using primers that amplify specific regions of the ftsI gene. The resistance-associated mutations in PBP3 were determined according to the method of Hasegawa et al. (12).
We determined the nucleotide sequences of the QRDRs of the quinolone target genes gyrA, gyrB, parC, and parE. Genomic DNA was isolated from bacteria by using InstaGene Matrix (Bio-Rad Laboratories, Hercules, CA). PCR amplification was performed using HotStarTaq DNA polymerase (Qiagen). The primer sets used to obtain DNA fragments of the gyrA, gyrB, parC, and parE QRDRs were previously described (7). Direct sequencing of the PCR products was carried out with an ABI Prism 3130 (Applied Biosystems, Foster City, CA). For the additional 50 isolates with ciprofloxacin MICs of 0.03 μg/ml or lower, mutations in QRDRs were also examined as described above.
In the case of BLNAS H. influenzae isolates, all β-lactams and ciprofloxacin exhibited very strong activities against all 100 isolates. On the other hand, in the case of BLPAR H. influenzae isolates, whereas amoxicillin-clavulanate was active against all isolates, susceptibility to ampicillin-sulbactam was 75%. Piperacillin alone was not active, while piperacillin-tazobactam was highly potent (MIC at which 90% of the isolates tested are inhibited [MIC90], 0.06 μg/ml; MIC range, ≤0.03 to 0.25 μg/ml; 100% susceptible). Although the MIC ranges of ceftriaxone and meropenem were slightly higher than those for BLNAS H. influenzae isolates, their susceptibilities were still 100%. One isolate was resistant to imipenem (MIC, ≥8 μg/ml).
In the case of BLNAR H. influenzae isolates, ampicillin-sulbactam and amoxicillin-clavulanate were not active enough, nor was ampicillin, while piperacillin (MIC90, 0.25 μg/ml; MIC range, ≤0.03 to 0.5 μg/ml) and piperacillin-tazobactam (MIC90, 0.12 μg/ml; MIC range, ≤0.03 to 0.5 μg/ml; 100% susceptible) retained strong activity. The MIC ranges of ceftriaxone, imipenem, and meropenem were higher than those for BLNAS and BLPAR H. influenzae isolates, and four isolates and one isolate were resistant to imipenem and meropenem, respectively. The potency ranking of β-lactams for the BLNAR H. influenzae isolates was as follows (MIC90 in μg/ml): piperacillin-tazobactam (0.12) > piperacillin (0.25) = ceftriaxone (0.25) > meropenem (0.5) > imipenem (4).
In the case of BLPACR H. influenzae isolates, ampicillin, piperacillin, ampicillin-sulbactam, and amoxicillin-clavulanate were not active, while piperacillin-tazobactam retained strong activity (MIC90, 0.25 μg/ml; MIC range, ≤0.03 to 0.5 μg/ml; 100% susceptible). Against imipenem, two isolates were resistant. The potency ranking of β-lactams for the BLPACR H. influenzae isolates was as follows (MIC90 in μg/ml): piperacillin-tazobactam (0.25) = ceftriaxone (0.25) > meropenem (0.5) > imipenem (2).
While ciprofloxacin was consistently very active against H. influenzae isolates of BLNAS, BLPAR, and BLNAR types (MIC, ≤0.25 μg/ml), one BLPACR isolate was fully resistant to ciprofloxacin (MIC, 16 μg/ml) and the range of MICs of ciprofloxacin in BLPACR isolates shifted higher than those of the others tested.
Even though the phenotypes of H. influenzae included BLNAR and BLPACR, piperacillin-tazobactam control ranges were ≤0.03 to 0.5 μg/ml and all zones at ≥21 mm (range, 21 to 50 mm) were considered to indicate high degrees of susceptibility.
All 200 BLNAR and BLPACR isolates had amino acid substitutions in their ftsI gene, corresponding to three different mutation patterns. We followed the BLNAR classifications proposed by Dabernat et al. (6) and Ubukata et al. (31). Of the 200 BLNAR/BLPACR isolates, 100% had amino acid mutations of the Lys-Thr-Gly (KTG) motif, including Asn526 to Lys (97.5%) and Arg517 to His (2.5%), analogous to the previously described original groups IIa and I, respectively (31). However, none of the above BLNAR/BLPACR strains could be classified as group I, since all strains with amino acid mutations of Arg517 to His also had an additional Ser385-to-Thr mutation in the Ser-Ser-Asn (SSN) motif and thus belonged to the group III-like strains (10). The most common (88.5%) pattern also appeared to be group III-like, which had Ser385 to Thr in the SSN motif in addition to Asn526 to Lys in the KTG motif. Consequently, pure group I strains, with Arg517 to His alone, and group IIa, which had Asn526 to Lys alone, accounted for 0% and 11.5% of the total, respectively.
Of 22 low-level BLNAR isolates with an ampicillin MIC of 2 μg/ml, five belonged to group IIa (Asn526 to Lys), while the others were group III-like (16 had Ser385 to Thr and Asn526 to Lys, while one had Ser385 to Thr and Arg517 to His). All 78 fully resistant BLNAR isolates were classified as group III-like; 76 of 78 strains had the amino acid changes Ser385 to Thr and Asn526 to Lys, and the other two had Ser385 to Thr and Arg517 to His. Of 100 BLPACR isolates, 18 had the amino acid change in Asn526 to Lys alone (BLPACR I), 80 had Ser385 to Thr and Asn526 to Lys, and the remaining two had Ser385 to Thr and Arg517 to His (BLPACR II).
We first examined the isolates with a ciprofloxacin MIC of 0.25 μg/ml or higher, but all such isolates had at least one mutation in their QRDRs. We therefore expanded our examination to include isolates with a ciprofloxacin MIC of 0.06 μg/ml or higher, since most clinical isolates of H. influenzae usually exhibit a MIC of ≤0.03 μg/ml for ciprofloxacin. Table Table66 summarizes the results of analysis of mutations in QRDRs for a total of 37 of the initial 400 H. influenzae isolates which had a ciprofloxacin MIC of 0.06 μg/ml or higher, including one BLNAS strain, seven BLPAR strains, four BLNAR strains, and 25 BLPACR strains.
Of these 37 isolates, 32 had amino acid changes in at least one of their QRDRs in gyrA, gyrB, parC, and parE. Two BLPAR isolates with a ciprofloxacin MIC of 0.06 μg/ml, two BLPACR isolates with a ciprofloxacin MIC of 0.12 μg/ml, and a BLPACR with a ciprofloxacin MIC of 0.06 μg/ml did not have mutations in QRDRs with the method that we used.
The frequency of decreases in MICs of ciprofloxacin associated with QRDR mutations was highest for the BLPACR isolates. The BLPACR strain most resistant to ciprofloxacin, with a MIC of 16 μg/ml, had five amino acid changes in its QRDRs, including all those of gyrA, gyrB, parC, and parE. Ser84 to Leu and Asp88 to Asn in the gyrA product were most often found. Mutations in gyrB were rare. BLPACR isolates with ciprofloxacin MICs of 0.06 or 0.12 μg/ml frequently had a Ser84-to-Cys mutation in the parC product and an Ala369-to-Thr mutation in the parE product.
In the additional 50 isolates with ciprofloxacin MICs of 0.03 μg/ml or lower, mutations of QRDRs were found in 13 (26.9%) isolates. These mutations included five patterns: (i) Glu142 to Lys in the gyrA product (four isolates), (ii) Asn138 to Ser in the parC product (three isolates), (iii) Ser133 to Ala in the parC product (two isolates), (iv) Asn138 to Ser and Ser133 to Ala in the parC product (three isolates), and (v) Gly143 to Glu in the parC product (one isolate). Some of these mutations were found in the isolates for the initial analysis summarized in Table Table66.
Among β-lactams, piperacillin-tazobactam clearly exhibited very strong activities against all classes of H. influenzae isolates (MIC ranges, MIC50s, and MIC90s [μg/ml] were as follows: BLNAS isolates, ≤0.03 to 0.12, ≤0.03, and 0.06; BLPAR isolates, ≤0.03 to 0.25, ≤0.03, and 0.06; BLNAR isolates, ≤0.03 to 0.5, 0.06, and 0.12; and BLPACR isolates, ≤0.03 to 0.5, 0.12, and 0.25, respectively). A total of 400 isolates, including four different phenotypes, were all susceptible according to the current CLSI breakpoint (5). The warnings in CLSI documents that BLNAR H. influenzae strains should be considered resistant to some β-lactam compounds are found in M2-A2/M100-S17 tables (amoxicillin-clavulanate, ampicillin-sulbactam, cefaclor, cefetamet, cefonicid, cefprozil, cefuroxime, and loracarbef); in M7-A7/M100-S17 tables, piperacillin-tazobactam and cefamandole were additionally listed; many of those statements have been present since the early 1990s. The current findings are presented to stimulate reevaluation of the H. influenzae breakpoints in the era of escalation of BLNAR and BLPACR problems worldwide. This problem appears to be greatest in Asia and Western Europe but may be underestimated in other geographical areas. The high potency of piperacillin-tazobactam has been demonstrated against these strains, but resistance disclaimers are found in the CLSI tables and in U.S. FDA package inserts. Piperacillin-tazobactam, previously considered inactive against BLNAR H. influenzae due to limited data, has now been found to have great potency against these strains and should be assigned accurate susceptibility breakpoints.
In the present study, we confirmed that all BLNAR and BLPACR isolates had amino acid substitutions in their ftsI gene products and that the major pattern was group III-like (87.5%), followed by the isolates belonging to group IIa.
Moreover, we found that many isolates susceptible to ciprofloxacin had mutations in their QRDRs, particularly in the case of BLPACR isolates, and that one ciprofloxacin-resistant isolate (MIC, 16 μg/ml) had five mutations in its QRDRs. Mutations Ser84Ile and Ser84Arg in ParC and Ser84Leu and Asp88Asn in GyrA are previously known mutations associated with quinolone resistance in H. influenzae (11, 22, 26, 33). Asp420Asn in ParE corresponds to resistance-associated mutation sites in Staphylococcus aureus ParE (Asp432) and in Streptococcus pneumoniae ParE (Asp435) (28, 29). The Asn138Ser mutation in ParC was also found in susceptible H. influenzae strains and was suggested not to contribute to fluoroquinolone resistance (32), though it is possible that this is the first mutation in a pattern of stepwise mutation. It is thus still unclear whether all mutations in QRDRs found in the present study contribute to the decreased susceptibility to ciprofloxacin. However, since the MICs of ciprofloxacin for most clinical isolates are usually ≤0.03 μg/ml, those for the isolates associated with QRDR mutations appear to be elevated. Since the isolates with higher ciprofloxacin MICs tended to have several mutations in their QRDRs, it is possible that numerous H. influenzae isolates have silent mutations as first mutations and could acquire full resistance to quinolones by stepwise QRDR mutation, as found in S. pneumoniae (9). In an additional 50 clinical isolates of H. influenzae with ciprofloxacin MICs of 0.03 μg/ml or lower, 26.0% had mutations in their QRDRs, which had not been reported as the causes of ciprofloxacin resistance in H. influenzae, and some of them were found in the initial 400 clinical isolates as summarized in Table Table6.6. These mutations would not contribute to ciprofloxacin resistance in H. influenzae, if additional significant mutations do not occur in their QRDRs.
In conclusion, piperacillin-tazobactam, a penicillin-β-lactamase inhibitor combination, was the most potent β-lactam tested, with all MICs being ≤0.5 μg/ml against all groups of H. influenzae isolates, including BLNAS, BLPAR, BLNAR, and BLPACR isolates. The susceptible disk diffusion breakpoints for piperacillin-tazobactam are proposed to be ≥21 mm. It is possible that quinolone-resistant H. influenzae strains may emerge, since several clinical isolates with elevated MICs of ciprofloxacin were found to have specific mutations in their QRDRs, while some isolates with low ciprofloxacin MICs carried mutations in their QRDRs, which do not contribute to ciprofloxacin resistance in H. influenzae. Checking for β-lactamase alone is performed for clinical isolates of H. influenzae, and antimicrobial susceptibilities are not accurately determined in several laboratories. This may result in lack of detection of BLNAR, BLPACR, and silent quinolone-resistant isolates. Susceptibility testing should be performed in order not to underestimate the frequency of such emerging resistant strains.
This project was jointly funded by Taiho Pharmaceuticals research grants.
Published ahead of print on 3 August 2009.