ABSTRACT
The PROTEKT US (Prospective Resistant Organism Tracking and Epidemiology for the Ketolide Telithromycin in the United States) surveillance program was established to determine the prevalence and mechanisms of antibacterial resistance among bacterial pathogens from patients with community-acquired respiratory tract infections. In year 1 of the PROTEKT US study, 10,103 isolates of Streptococcus pneumoniae, including 3,133 erythromycin-resistant strains and 81 levofloxacin-resistant strains, were collected from 206 centers. We report on the molecular analyses of these resistant strains. The resistance genotypes among the 3,044 typed macrolide-resistant isolates overall were mef(A) (n = 2,157; 70.9%), erm(B) (n = 530; 17.4%), mef(A) erm(B) (n = 304; 10.0%), and erm(A) subclass erm(TR) (n = 5; 0.2%). Fifty (1.6%) macrolide-resistant isolates were negative for the mef and the erm resistance genes. Seventy-eight (96.3%) of the 81 levofloxacin-resistant isolates analyzed possessed multiple mutations in the gyrA, gyrB, parC, and/or parE quinolone resistance-determining regions. A total of 43 known multilocus sequence typing (MLST) profiles (or single- or double-locus variants) accounted for 75 of 81 isolates. There was no evidence of dissemination of fluoroquinolone-resistant clones within the United States; however, 12 isolates with the same MLST profile were located in one center in Massachusetts. Almost 90% of the erythromycin-resistant isolates and approximately one-third of the levofloxacin-resistant isolates were multidrug resistant.
Macrolides and fluoroquinolones are antibacterial agents used for the empirical treatment of community-acquired respiratory tract infections (CARTIs). The rate of erythromycin resistance among Streptococcus pneumoniae isolates in the United States was about 0.2% in 1987-1988 (23), but this rate increased to 17% by 1998 and 23% by 1999 (17). The rate of fluoroquinolone resistance among S. pneumoniae isolates remains low worldwide (1%) (14), but reports of increasing rates of resistance are emerging from several countries, such as Hong Kong (16) and the United States (31).
Macrolide resistance in S. pneumoniae is typically achieved through two mechanisms: efflux and target modification. Efflux resistance is mediated by the mef(A) gene, which encodes an efflux pump specific to 14- and 15-membered-ring macrolides. In the case of target modification, mediated by the erm(B) gene, an adenine residue in the 23S rRNA is methylated, leading to reduced binding of 14-, 15-, and 16-membered-ring macrolides, lincosamides, and streptogramin B to their shared target site on the 50S ribosomal subunit (32).
Fluoroquinolone resistance occurs primarily through mutations in the quinolone resistance-determining regions (QRDRs) of the parC and gyrA genes (involved in encoding the A subunits of DNA gyrase and topoisomerase IV) or the gyrB and parE genes (which encode the B subunits of DNA gyrase and topoisomerase IV). Resistance is generally imparted in two steps: a primary mutation reduces fluoroquinolone binding affinity, while a subsequent mutation in a secondary target results in a further reduction in in vitro activity (8, 15, 33).
In 2000, the PROTEKT US (Prospective Resistant Organism Tracking and Epidemiology for the Ketolide Telithromycin in the United States) surveillance program was initiated to longitudinally monitor antibacterial resistance in pathogens across the United States. The geographical distributions of fluoroquinolone- and macrolide-resistant isolates of S. pneumoniae across the United States have been reported in detail by Doern and Brown (7) and Waites and Brown (37). In this paper, we describe the prevalence of resistance genotypes responsible for macrolide and fluoroquinolone resistance among these isolates of S. pneumoniae.
MATERIALS AND METHODS
During 2000 and 2001, 206 centers from 154 cities or metropolitan areas took part in the PROTEKT US study. The participating centers were each requested to collect 50 clinical isolates of S. pneumoniae. Isolates were collected from patients with community-acquired pneumonia, acute bacterial sinusitis, acute otitis media, and acute exacerbation of chronic bronchitis or acute exacerbation of chronic obstructive airway disease. Suitable sources for isolates included blood, sputum, bronchoalveolar lavage fluid, middle ear fluid, nasopharyngeal swab or aspirates, and sinus aspirates. The centers were free to include isolates from patients of any age. Specimens were considered acceptable if they were the first positive specimen from each patient and had been obtained from outpatients with CARTIs or hospitalized patients within 48 h of admission. Isolates from patients hospitalized for 48 h or more or patients with nosocomial lower respiratory tract infections or cystic fibrosis were excluded, as were isolates from existing collections and duplicate strains or isolates from sputum samples with a poor Gram staining result.
All isolates were stored in Amies transport medium and were subsequently shipped to a central laboratory (Clinical Microbiology Institute, Wilsonville, Oreg.) for microbiological investigation. Molecular analysis was carried out at GR Micro Ltd. (London, United Kingdom).
MICs were determined by the NCCLS broth microdilution methodology (27). NCCLS breakpoints were used to define susceptibility (28). Tentative NCCLS 2004 breakpoints were applied for telithromycin: susceptible, MIC ≤ 1 mg/liter; intermediate, MIC = 2 mg/liter; and resistant, MIC ≥ 4 mg/liter (29).
Amplification and detection of erm(A), erm(A) subclass erm(TR), erm(B), erm(C), and mef(A) macrolide resistance genes were carried out by a duplex PCR (12).
Extraction and amplification of the gene mutations in gyrA, gyrB, parC, and parE were carried out with a High Pure PCR template kit (Roche, Lewis, United Kingdom). Sequencing was conducted with an ABI Prism 3100 genetic analyzer (Applied Biosystems, Warrington, United Kingdom) (26). Multilocus sequence typing (MLST), as described by Enright and Spratt (9), was conducted with all levofloxacin-resistant isolates to identify the existence of clonality among them. Data available from the Multi Locus Sequence Typing website (http://www.mlst.net) and the Pneumococcal Molecular Epidemiological Network website (http://www.sph.emory.edu/PMEN) were used to analyze and describe these isolates.
RESULTS
A total of 10,103 isolates of S. pneumoniae were collected from 206 centers in 42 states, the District of Columbia, and Puerto Rico during 2000 and 2001. Of these, 3,133 (31.0%) were erythromycin resistant and 81 (0.8%) were levofloxacin resistant (7, 37).
Macrolide-resistant isolates.Of the 3,133 erythromycin-resistant S. pneumoniae isolates collected during year 1 of the PROTEKT US study, genotyping data were available for 3,044. Five resistance genotypes were identified: erm(B), mef(A), erm(B) mef(A), erm(A) subclass erm(TR), and “other” (erythromycin-resistant isolates that were negative for the mechanisms tested for). All but 4 centers collected at least 1 resistant isolate, and of the remaining 202 centers, 51 collected 20 or more isolates (the range data presented below are only for these 51 centers). The mef(A) genotype was the most common resistance mechanism identified and accounted for 70.8% (n = 2,157) of the erythromycin-resistant isolates (median, 71.4% per center; range, 41.7 to 95.0% per center). The erm(B) gene was identified in 17.4% (n = 530) of isolates (median, 16.7% per center) and was the second most prevalent genotype in the majority (36 of 51) of centers (range, 0.0 to 45.5% per center). The third most common genotype overall was erm(B) mef(A), which was detected among a further 10.0% (n = 304) of isolates (median, 7.1% per center) and whose prevalence ranged from 0.0 to 35.0% per center. Five isolates were positive for erm(A) subclass erm(TR) and were located in five separate centers. Fifty (1.6%) erythromycin-resistant isolates were negative for the mechanisms tested for and have been described elsewhere (10).
Fluoroquinolone-resistant isolates.In contrast to erythromycin-resistant isolates, the majority of centers (74.3%; 153 of 206 centers) did not isolate any levofloxacin-resistant S. pneumoniae isolates. Of the remaining 53 centers, 40 collected only one levofloxacin-resistant isolate, 9 centers collected two levofloxacin-resistant isolates, and 2 centers collected three levofloxacin-resistant isolates. Such small sample sizes cannot be relied upon to accurately measure the prevalence of fluoroquinolone resistance within regions or states within the United States, but nevertheless, these results indicate that the rate of fluoroquinolone resistance among pneumococci is low. The two centers with the highest prevalence of levofloxacin-resistant isolates were in New York State and Massachusetts, with 4 and 12 resistant isolates, respectively (resistance rates, 3.1 and 21.8%, respectively).
MLST data for all 81 fluoroquinolone-resistant isolates are presented in Table 1. The most common MLST found was a single-locus variant (SLV) of Tennesee23F-4 clone ST37. Interestingly, this clone represented all 12 levofloxacin-resistant isolates from the center in Massachusetts mentioned above. This indicates that the very high rate of fluoroquinolone resistance in this center was due to the dominance of a single clone. A double-locus variant (DLV) of ST37 was also isolated in Massachusetts, but from a different center, and ST37 itself was found in centers in Washington State and Tennessee (Table 1).
MLST profiles of levofloxacin-resistant isolates of S. pneumoniae collected during year 1 of the PROTEKT US study (2000-2001)a
All 12 isolates of the ST37 SLV clone were from different patients, and 10 were recovered during outpatient visits. The period over which this clone was isolated (3 November 2000 to 10 January 2001) spanned most of the cold season associated with the majority of community-acquired lower respiratory tract infections. The age range of the patients was from 5 to 91 years, with eight elderly patients (>64 years old), three adult patients (15, 51, and 63 years old, respectively), and one child patient (5 years old). The amino acid changes responsible for resistance in all 12 isolates of the ST37 SLV clone were GyrA E85K plus ParC S79F; 7 isolates had a further ParE E474K change, and 1 had a ParE D435N change (data not shown). Therefore, it is possible that multiple separate mutational events may have occurred in the original fluoroquinolone-susceptible ST37 clone rather than the spread of resistance from a single source. The ST37 DLV is even more likely to be from a separate source because it contained resistance mutations (GyrA S81F and ParC S79F) distinct from those seen in the ST37 SLV (data not shown).
Spain23F-1 clone ST81 was the second most prevalent isolate, but it accounted for only 7 of 81 (8.6%) of all fluoroquinolone-resistant pneumococci in the United States. This clone was distributed across the United States and was not concentrated in any one center. An SLV of ST81 was also observed in one center (Table 1). Third in order of prevalence, but found on only four occasions in four separate centers, was Spain9V-3 clone ST156 (Table 1). Other clones observed in year 1 of the PROTEKT US study that have been highlighted by the Pneumococcal Molecular Epidemiology Network were Taiwan19F-14 clone ST236 (two occurrences), Sweden15A-25 clone ST63 (two occurrences), North Carolina6A-23 clone ST376 (one occurrence), Taiwan23F-15 clone ST242 (one occurrence), Spain6B-2 clone ST90 (one occurrence), and an SLV of England14-9 clone ST9 (ST13; one occurrence). The remaining isolates did not possess any distinguishing characteristics. However, three clones plus one SLV were typed and found to be unrelated to known pneumococcal sequence types (Table 1). Interestingly, these isolates had QRDR mutations indicative of the viridans group streptococci (see below), which suggests that either these isolates have undergone considerable genetic transfer with nonpneumococci or they are, in fact, not S. pneumoniae.
The topoisomerase amino acid subunit alterations in the QRDRs for all 81 levofloxacin-resistant isolates are listed in Table 2. The majority of isolates (58 of 81; 71.6%) possessed two amino acid changes. Of these, the GyrA S81F and ParC S79F/T modifications were the most common. One isolate had an unusual ParC S79A change, and two isolates had an S79R change in ParC. A novel Q90H change was found in ParC of one isolate, and a novel S80P change was found in another; both changes appeared to play a role in fluoroquinolone resistance. Some isolates (15 of 81; 18.5%) possessed three QRDR changes, and this additional QRDR change shifted the levofloxacin MICs from 8 to 16 mg/liter for most isolates with two modifications to >16 mg/liter for most isolates with three modifications. A minority of GyrB changes were implicated in resistance, as has been found previously (5, 21, 38). Surprisingly, 13 instances of ParE alterations (D435N and E474K in the main and one novel A325V change) occurred, which is greater than the number found in previous studies (3, 5, 21, 38). One isolate relied only on alterations in GyrB and ParE for resistance (Table 2). Only three levofloxacin-resistant isolates carried just one QRDR alteration, and the alteration was in the DNA gyrase in two isolates (Table 2).
Topoisomerase amino acid alterations and MICs of levofloxacin and gatifloxacin for levofloxacin-resistant S. pneumoniae isolates collected during year 1 of the PROTEKT US study (2000-2001)
Some other less common amino acid substitutions were discovered and are indicated in boldface in Table 2. These changes were detected in those isolates that presented numerous QRDR amino acid substitutions and did not have any major effect on the fluoroquinolone MICs. The GyrA S114G and the ParC N91D changes have previously been found in S. pneumoniae and are a consequence of the uptake of DNA from viridans group streptococci (2). These isolates also produced unusual MLST results, which are mentioned above.
Of the 81 levofloxacin-resistant isolates, 75 (92.6%) were gatifloxacin resistant, indicating almost complete cross-resistance between these two fluoroquinolones.
Antibacterial susceptibility among erythromycin- and levofloxacin-resistant isolates.The antibacterial susceptibilities of all S. pneumoniae isolates from year 1 of the PROTEKT US study are compared to the susceptibilities of the erythromycin-resistant and levofloxacin-resistant isolates in Table 3. The majority of the erythromycin-resistant pneumococci (irrespective of their mechanism of resistance) were also resistant to cefuroxime, co-trimoxazole, and penicillin. As would be expected, coresistance did not occur with clindamycin among isolates carrying mef(A) and almost complete cross-resistance occurred with those isolates carrying erm(B). Fewer mef(A) isolates than erm(B) isolates were coresistant to tetracycline (tetracycline resistance rates, 28.1 and 93.6%, respectively). In contrast, among the erythromycin-resistant isolates, no coresistance between erythromycin and levofloxacin or telithromycin was detected. Among the levofloxacin-resistant isolates, approximately 40 to 50% were also resistant to erythromycin, cefuroxime, penicillin, and co-trimoxazole. Resistance to tetracycline or clindamycin was not as prevalent among the levofloxacin-resistant isolates than among the erythromycin-resistant population.
Percent susceptibility among all S. pneumoniae isolates, erythromycin-resistant isolates by genotype, and levofloxacin-resistant isolates of S. pneumoniae collected during year 1 of the PROTEKT US study (2000-2001)
An analysis of multidrug resistance is given in Table 4. Overall, 34.4% of the S. pneumoniae isolates were multidrug resistant, which is defined by the U.S. Food and Drug Administration to be resistance to two or more of the five classes of antibacterials represented by erythromycin, cefuroxime, co-trimoxazole, penicillin, and tetracycline (http://www.fda.gov/cder/foi/label/2004/21144_ketek_lbl.pdf). Among the levofloxacin-resistant population, a total of 53.1% of the isolates were multidrug resistant. In contrast, almost 90% of the erythromycin-resistant isolates were multidrug resistant, with virtually all erm(B) and erm(B) mef(A) isolates resistant to two or more agents. An alternative analysis of multidrug resistance was also made by replacing cefuroxime susceptibility with levofloxacin susceptibility, on the basis of the fact that β-lactam agents are represented twice by the inclusion of both penicillin and cefuroxime in the U.S. Food and Drug Administration definition. The results of this analysis showed little difference in the prevalence of isolates resistant to two or more or three or more agents, but fewer isolates with resistance to four or more or five or more agents were observed (results not shown). Isolates with double mechanisms of macrolide resistance are of particular concern, as more than 80% were resistant to four or more agents according to either multidrug resistance definition. No isolate was found to be resistant to all eight antibacterials tested overall.
Multidrug resistance among all S. pneumoniae isolates, erythromycin-resistant isolates by genotype, and levofloxacin-resistant isolates of S. pneumoniae collected during year 1 of the PROTEKT US study (2000-2001)
DISCUSSION
Traditional treatment options for CARTIs have been threatened in recent years by increasing levels of antibacterial resistance among S. pneumoniae isolates (6, 14, 22), an observation supported by the results of the PROTEKT US study for 2000-2001 (7). An understanding of local resistance patterns is essential to ensure that appropriate antibiotics are used for the treatment of CARTIs. The PROTEKT US study was undertaken not only to monitor national and local resistance patterns as a guide to the prescription of antibacterials but also to determine the prevalence of resistance mechanisms. Although there is often no proven direct link between in vitro resistance and clinical treatment failure, we have followed the principles laid out by the NCCLS and others who provide MIC breakpoints to help clinicians make informed decisions about appropriate patient therapy.
The prevalence of macrolide resistance mechanisms varies widely geographically. As this report shows, the mef(A) gene clearly predominates in the United States; other research has suggested that the same is true in Canada (20). In contrast, for Europe overall, erm(B) prevails among macrolide-resistant S. pneumoniae isolates (although this varies widely from country to country) (13). In the Far East, the prevalence of macrolide resistance mechanisms among isolates of S. pneumoniae is more diverse. In Hong Kong, mef(A) positive isolates are prevalent (72.8%) (19). In South Korea, mef(A) is the least common genotype (18.4%), with isolates positive for erm(B) and for both erm(B) and mef(A) occurring at similar levels (43.3 and 38.3%, respectively) (13). In Japan, isolates with the mef(A) (40.3%) and the erm(B) (43.5%) genes are equally prevalent, but isolates with both erm(B) and mef(A) are relatively common (16.1%) (30). Interestingly, most erm(B) mef(A) isolates appear to belong to a multidrug-resistant clonal complex (11). Telithromycin is one of the few antibacterials to have good activity against this clone (11). In this study, resistance to multiple antibacterial agents was shown to be highly prevalent among all erythromycin resistance genotypes, with the number of antibacterial agents to which isolates were resistant being higher for those that possessed the erm(B) gene or the erm(B) and the mef(A) genes.
Of the 3,044 erythromycin-resistant isolates discussed here for which genotyping data were available, 50 isolates were negative for the mechanisms of resistance for which tests were performed. Subsequent testing of these isolates has shown that resistance is imparted through ribosomal mutations (10).
The majority of S. pneumoniae isolates collected during year 1 of the PROTEKT US study were susceptible to fluoroquinolones: 10,014 (99.1%) were susceptible to levofloxacin at ≤2 mg/liter. Of the resistant isolates identified, 64.2% were resistant to one or more other antibacterials. Typically, resistance was mediated through mutations at three of the QRDRs: gyrA, parC, and parE. Only 2 of the 81 fluoroquinolone-resistant isolates identified in this study possessed mutations in the gyrB QRDR. Low incidences of mutation in the gyrB QRDR among fluoroquinolone-resistant isolates of S. pneumoniae have been reported previously (5). However, this study has highlighted the possibility that ParE may play a greater role in fluoroquinolone resistance than was previously thought. Four isolates possessed a large number of amino acid substitutions (5 to 10) within the QRDRs. The nature of these changes is evidence of transfer of genetic material from viridans group streptococci to pneumococci, as described previously (2). It would appear that this does not have any effect on fluoroquinolone susceptibility.
Of the levofloxacin-resistant isolates collected in this study, most possessed MLST profiles that have been identified previously, and some were recognized international clones. The profiles of 7 of the 51 isolates with previously identified MLST profiles matched the profile of clone Spain23F-1, which has been highlighted as an important clone causing high-level fluoroquinolone resistance in Hong Kong (16, 18, 26). However, unlike in Hong Kong, the fluoroquinolone-resistant Spain23F-1 clone did not dominate at any U.S. center. Overall, the MLST data provided in this study indicate that the rare occurrences of fluoroquinolone resistance in the United States have occurred as independent mutational events in a diverse number of clones. The exception to this occurred in one center in Massachusetts, where relatively high rates of fluoroquinolone resistance can be directly attributed to the dominance of the SLV of the Tennessee23F-4 clone. Although all 12 isolates were the same clone, resistance may still have evolved independently. Interestingly, the four isolates with viridans group streptococci-like QRDR sequences also possessed unusual MLST profiles. This suggests that these isolates had undergone considerable DNA transformation or were not true S. pneumoniae strains (even though they were identified as such by three separate laboratories). The study of the MLST profiles of levofloxacin-resistant isolates will continue in the coming years as part of the PROTEKT US study to monitor the epidemiology of fluoroquinolone resistance in the United States.
The results of this study have highlighted the extent of coresistance among macrolide-resistant pneumococci. This resistance was greatest with the erm(B)-carrying subpopulation, which showed coresistance to clindamycin, β-lactams, tetracycline, and co-trimoxazole. The mef(A) isolates retained susceptibility to clindamycin (as would be expected), and considerably fewer isolates were resistant to tetracycline. This phenomenon may occur because erm(B) and mef(A) are associated with different conjugative transposons (Tn1545 and Tn916, respectively) (25, 34, 36). It is possible that the insertion of erm(B) into transposon Tn1545 already associated with tet(M) and other resistance genes has evolved over a long time period, whereas the association of mef(A) and tet(M) on Tn916 has occurred more recently (1, 25, 35). It will be interesting to see if any further evolutionary changes are observed as the PROTEKT US surveillance study continues.
Isolates carrying both mef(A) and erm(B) have been reported previously (4, 11, 24) and are highly multidrug resistant (11), as the results of the present study have confirmed. This high-level resistance is most likely due to the presence of more than one mobile genetic element (11), but this hypothesis was not tested in the present study. The prevalence of these isolates in the United States may have increased slightly from 7% in 1996-1997 (4) to 10% in 2000-2001 (this study). However, as different centers were used for these two studies, it is difficult to form any firm conclusions. As stated above, we await the results of future PROTEKT US surveillance studies to see if this upward trend continues. It is noteworthy that only telithromycin and levofloxacin showed activity against these antibiotic-resistant S. pneumoniae isolates.
ACKNOWLEDGMENTS
We thank all investigators involved in the PROTEKT US study. We also thank the researchers at the Clinical Microbiology Institute for analysis of the isolates and researchers at GR Micro Ltd. for genotyping analysis. Preliminary data analysis was carried out by Micron Research Ltd., Upwell, United Kingdom.
The PROTEKT US study is funded by Aventis.
FOOTNOTES
- Received 10 May 2004.
- Returned for modification 8 July 2004.
- Accepted 30 July 2004.
- American Society for Microbiology
