From November 1, 1997, to April 30, 1998, 1,601 isolates of S. pneumoniae were recovered in 34 U.S. medical centers. All isolates included in this study were from consecutive patients. With the exception of specimens from the lower respiratory tract, all isolates were from normally sterile body sites (i.e., blood, cerebrospinal fluid, middle ear fluid, sinus aspirates, pericardial fluid, and pleural fluid). Isolates from lower respiratory tract specimens were included only if they were of clinical significance.

In the study centers, isolates were subcultured onto 5% sheep blood agar plates and incubated overnight at 35°C to 37°C in 5% to 7% CO₂. Colony growth was collected on a rayon swab and immediately immersed in a transport tube containing 12 ml of semisolid Ames transport medium with charcoal (Difco Laboratories, Detroit, MI). Transport tubes were then shipped overnight to the University of Iowa College of Medicine for additional analysis (Appendix). The recovery rate from this transport system was 100%. Twelve concentrations each of 23 antimicrobial drugs were tested against 1,601 isolates of S. pneumoniae.

Overall, 17.4% of isolates had intermediate and 12.1% had full resistance to penicillin (Table 1). Overall nonsusceptible rates with ceftriaxone and cefuroxime (intermediate plus fully resistant) were 14.9% and 23.3%, respectively. Because National Committee for Clinical Laboratory Standards (NCCLS)-approved breakpoints are lacking for the six other cephalosporins examined in this study (cefpodoxime, cefixime, ceftibuten, cefprozil, cefaclor, and loracarbef), rates of resistance were not determined for these drugs. However, when MIC values were compared, cefpodoxime was the most active.

Comparison of MIC values of the three macrolides we examined showed that clarithromycin was consistently twice as active as erythromycin, which in turn was consistently twice as active as azithromycin (Table 1). Overall rates of resistance, however, based on NCCLS breakpoints, which differ for these agents, were similar (18%-19%).

Compared with erythromycin as an indicator of macrolide activity, 302 (18.9%) isolates had MICs >1 µg/ml and thus were classified as resistant (Table 1). Among these, 217 (71.9%) had erythromycin MICs <32 µg/ml; the remaining 85 strains (28.1%) had erythromycin MICs >64 µg/ml. Of the 217 strains with erythromycin MICs <32, 214 had clindamycin MICs <0.25 and thus were categorized as clindamycin susceptible. Thirty-five of these strains, randomly selected, were examined by polymerase chain reaction (PCR) for the presence of ermAM and mefE genes. Of the 85 strains with erythromycin MICs >64 µg/ml, 83 had clindamycin MICs >8. Thirty-eight of these isolates, chosen randomly, were ermAM positive; 12 were also positive for mefE (Table 2). Five resistant strains were also characterized for the presence of macrolide resistance determinants (Table 2). Finally, the three isolates (Table 2) negative for both the ermAM and mefE genes were also negative by PCR for other known determinants of macrolide/lincosamide resistance in gram-positive bacteria (ermA, ermC, ereA, ereB, msrA, and linA genes).

Of the 1,601 isolates examined, one had intermediate resistance to trovafloxacin with an MIC 2 µg/ml; three strains (0.2%) were resistant, two with trovafloxacin MICs 4 µg/ml and one with a trovafloxacin MIC 8 µg/ml. The single strain with intermediate resistance had a ciprofloxacin MIC of 16 and an asp83 →tyr substitution in the C subunit of topoisomerase IV, as well as a ser84 →tyr substitution in the A subunit of DNA gyrase. The three trovafloxacin-resistant strains had ciprofloxacin MICs >32 µg/ml and a ser79 →phe substitution in the C subunit of topoisomerase, as well as a ser84 →phe substitution in the A subunit of DNA gyrase. None of these four strains had detectable mutations in the QRDR of par E.

Overall rates of resistance, expressed as the percentage of isolates intermediate or resistant, for selected other agents are described in Table 1: tetracycline, 13.2%, TMP/SMX, 31.1%, chloramphenicol, 7.2%, and rifampin, 0.2%. Linezolid was uniformly active over a narrow range of MICs (i.e., 0.12 to 2 µg/ml). Two strains among the 1,601 examined in this study were resistant to quinupristin/dalfopristin; one had an MIC 4 µg/ml and the other 8 µg/ml. Vancomycin was uniformly active against the 1,601 isolates of S. pneumoniae in this survey, with MICs <0.5 µg/ml.

The in vitro activity of all ß-lactams (penicillins, ß-lactamase inhibitor combinations and cephalosporins), macrolides, clindamycin, tetracycline, TMP/SMX, and chloramphenicol was lowest with high-level penicillin-resistant strains of S. pneumoniae and greatest with penicillin-susceptible isolates. This trend was not apparent with linezolid, trovafloxacin, rifampin, quinupristin/dalfopristin, or vancomycin.

The prevalence of resistance to selected agents was assessed according to the specimen source of isolates, the patient's age, and the health-care setting (Table 3). In general, the highest resistance rates for all antimicrobial drugs were observed in middle ear fluid and sinus aspirate isolates and in isolates from patients <5 years old and from patients seen in ambulatory-care settings.

When patterns of multidrug resistance were analyzed, coresistance to penicillin, the macrolides, tetracycline, TMP/SMX, and chloramphenicol was observed in 5.4% of isolates (Table 4). Resistance to the first four of these drugs but not to chloramphenicol was seen with 3.7% of strains. The most common pattern of multidrug resistance (intermediate level or resistant to penicillin and resistant to TMP/SMX, but susceptible to the macrolides, tetracyclines, and chloramphenicol) was observed in 6.9% of isolates.

In the 34 participating medical centers, the lowest and highest overall rates of penicillin resistance (intermediate plus resistant) were 12.8% and 64.6% (Table 5). Eight medical centers had 10% to 18% penicillin resistance; 11 centers had 19% to 26%; 5 had 27% to 36%; 7 had 37% to 45%; 3 had >46%. Although no distinct geographic clustering was identified, the highest overall rates of penicillin resistance were in the South and Southeast. The lowest overall rate of ceftriaxone resistance was 2.9%; the highest was 48.1%. With erythromycin, overall rates of resistance were 3.8% to 43.8%. With tetracycline, chloramphenicol, and TMP/SMX, the rates were 5.7% to 29.6%, 0.0% to 25.9%, and 11.9% to 63.2%, respectively. In general, the highest rates of resistance with most antimicrobial classes were observed in the same centers.

Among the 34 medical centers, 24 had participated in a similar study 3 years before (4). Because the sampling period, the nature of the patients, and the test methods were the same, resistance rates obtained with selected antimicrobial drugs were compared at the 24 centers for the two study periods (Table 6). In 19 of 24 centers, penicillin resistance rates (intermediate and resistant) increased by at least 2% during the 3-year interval. In six cases, the rate of increase was statistically significant (Table 6). The number of centers in which rates of resistance increased by >2% with other selected antimicrobial drugs over the 3-year period were ceftriaxone 13, erythromycin 21, tetracycline 20, chloramphenicol 12, and TMP/SMX 16. In most cases, resistance rates increased; however, with certain antimicrobial agents (i.e., ceftriaxone, TMP/SMX, and chloramphenicol), resistance rates remained the same or decreased. Three centers (Texas Children's Hospital, UNC Hospital, and the University of South Alabama Medical Center) had statistically significant increased rates of resistance to nearly every class of antimicrobial drug.

A total of 1,601 clinical isolates of S. pneumoniae obtained from November 1997 to April 1998 from patients in 34 U.S. medical centers were characterized with respect to the in vitro activity of 23 antimicrobial drugs. The overall rate of penicillin resistance was 29.5% (17.4% with intermediate resistance; 12.1% fully resistant). Penicillin, amoxicillin, and amoxicillin/clavulanate had essentially equivalent MICs for the pneumococcal isolates examined. Among the cephalosporins tested, the rank order of activity based on a comparison of MICs was ceftriaxone > cefpodoxime = cefuroxime > cefprozil > cefixime > cefaclor = loracarbef > ceftibuten. NCCLS has developed MIC interpretive breakpoints for two of these compounds, ceftriaxone and cefuroxime. On the basis of these breakpoints, overall resistance rates were 14.9% with ceftriaxone (10.9% intermediate, 4.0% resistant) and 23.3% with cefuroxime (1.3% intermediate and 22.0% resistant). Although no NCCLS-approved breakpoints have been developed for cefaclor, loracarbef, and ceftibuten versus S. pneumoniae, the high MICs we obtained with these drugs indicate that they would be poor choices for treating pneumococcal infections.

We observed a direct relationship between penicillin activity and the activity of all the other ß-lactam drugs we examined. This relationship, which has been observed by others, results from the principal mechanism of penicillin resistance in S. pneumoniae, alterations in penicillin binding proteins (4-6). The same proteins to which penicillin binds are necessary for the expression of the activity of all other ß-lactam antimicrobial drugs (8-11).

Among the macrolides examined, clarithromycin was generally one dilution more active than erythromycin, which in turn was one dilution more active than azithromycin. However, since NCCLS MIC interpretive breakpoints differ for these agents and given the actual distribution of MICs, overall rates of resistance for these three agents were similar (19.0%).

Two previous studies have indicated that macrolide resistance with S. pneumoniae exists primarily in one of two forms: strains with altered ribosomal targets due to expression of the ermAM gene and strains with an active efflux pump due to expression of the mefE gene (13,14). ErmAM-positive isolates have constitutive expression of resistance and typically have high levels of resistance to both clindamycin (MICs >8 µg/ml) and erythromycin (MICs >64 µg/ml). Efflux mutants are susceptible to clindamycin (MICs <0.25 µg/ml) and typically have erythromycin MICs 1 to 32 µg/ml.

In a recent survey in Canada, approximately 60% of macrolide-resistant strains of S. pneumoniae appeared to be efflux mutants (19). In our study, 214 (70.8%) of 302 macrolide-resistant strains were characterized by the efflux phenotype. When a random sample of 35 of these isolates was examined by PCR, 31 (88.6%) lacked the ermAM gene but had the mefE gene responsible for efflux. Among the remaining four strains in this group that were characterized at a molecular level, one was ermAM positive and mefE negative, two were positive for both ermAM and mefE, and one was negative for all macrolide-resistance markers. Based on their phenotype (i.e., erythromycin MICs >64 µg/ml, with clindamycin MICs >8 µg/ml), 83 (27.5%) of the 302 macrolide-resistant strains appeared to have constitutive expression of ermAM-mediated methylation of ribosomal targets. All 38 randomly selected isolates from this group were ermAM positive; 12 were also mefE positive by PCR.

Five macrolide-resistant strains were not readily placed into either the efflux or the target modification groups based on phenotype. Two strains with high-level clindamycin resistance but erythromycin MICs 2 and 32 µg/ml were ermAM positive but mefE negative. Three strains with erythromycin MICs >64 had clindamycin MICs 0.5, 1, and 4 µg/ml. One of these strains, which had a clindamycin MIC of 4 and a high erythromycin MIC, appeared to be an efflux mutant since it was mefE positive but ermAM negative. The mechanism of resistance in the last two isolates in this group is unclear. Both isolates were negative by PCR for ermAM and mefE, as well as other genetic markers for macrolide and lincosamide resistance in gram-positive bacteria.

Several conclusions can be drawn from these observations. Macrolide resistance in S. pneumoniae is nearly always characterized either by efflux or ribosomal target modifications. The phenotype of isolates in both categories is highly predictable. Efflux mutants usually have erythromycin MICs 1 - 32 µg/ml and clindamycin MICs <0.25 µg/ml. Target modified strains are characterized by constitutive expression of resistance and typically have erythromycin MICs >64 µg/ml and clindamycin MICs >8 µg/ml. Although rare exceptions to these patterns exist, clindamycin activity, as assessed by an MIC determination, is a reliable indicator of the nature of macrolide resistance.

These results show that therapeutic efficacy can be predicted for patients with pneumococcal infections who are treated with macrolides. Using current NCCLS interpretive criteria, 18% to 19% of clinical isolates of S. pneumoniae in the United States would be categorized as resistant. Approximately three-fourths of the resistant strains, however, are efflux mutants and have mid-range resistant macrolide MICs. Patients infected with mefE versus ermAM-positive strains of S. pneumoniae may differ in their response to macrolide therapy. The answer to this important question would define true clinical rates of macrolide resistance. If mefE strains respond to therapy with macrolide drugs, then true macrolide resistance rates may not be 18% to 19%, but closer to 4% to 5%, and high-level clindamycin MICs could be used to accurately identify strains highly resistant to macrolide drugs.

Multidrug resistance was noted with 16.0% of 1,601 isolates, a substantial increase over the 9.1% prevalence of multidrug resistant S. pneumoniae reported in our 1994-95 study (4). Among multidrug resistant strains, four resistance patterns were the most common: penicillin and TMP/SMX (n = 111); penicillin, macrolide, and chloramphenicol (n = 75); penicillin, macrolide, tetracycline, and TMP/SMX (n = 57); and penicillin, macrolide, tetracycline, TMP/SMX, and chloramphenicol (n = 87). No obvious clustering of multidrug resistant strains was observed in either individual medical centers or geographic regions.

That we found resistance to ß-lactam antimicrobial agents and numerous non-ß-lactam agents in the same organism is of interest. The mechanisms of resistance to the different antimicrobial classes are distinct and do not appear to be linked genotypically. One explanation, therefore, for the increasing prevalence of multidrug resistant strains is clonal spread of organisms resistant to more than one class of antimicrobial drugs. Under such circumstances, regardless of the antimicrobial drug used to select for resistance, multidrug resistant organisms may become more prevalent. Clonal spread warrants further study using appropriate techniques for establishing clonal relationships among multiple geographically distinct isolates of S. pneumoniae.

Isolates were frozen at -70°C at the University of Iowa College of Medicine until further characterization. After two subcultures, the identity of isolates was confirmed as S. pneumoniae by conventional tests and criteria. MICs were determined in Mueller-Hinton broth supplemented with 3% lysed horse blood, according to the broth microdilution method recommended by the National Committee for Clinical Laboratory Standards (NCCLS) (15). Microdilution trays (final volume 100 µl per well) were inoculated with approximately 5 x 10⁵ colony-forming units (CFU)/ml (final concentration) of test organism and incubated 22 to 24 hours at 35°C in ambient air before MICs were determined.

S. pneumoniae ATCC 49619 and Haemophilus influenzae ATCC 49247, ATCC 49766, and ATCC 10211 were used as controls. Calculations of the percentages of isolates resistant to individual agents were based on the most recent MIC interpretive criteria published by NCCLS (16).

Selected isolates were examined for mutations in the quinolone resistance-determining region of the parC and par E genes encoding the C and E subunits of topoisomerase IV, respectively, and the quinolone resistance-determining region of the gyrA gene encoding for the A subunit of DNA gyrase. Polymerase chain reaction (PCR) amplification with subsequent sequencing of PCR products was done by using the method and primers described by Pan et al. (17) and Perichon et al. (18). Other isolates were screened for various macrolide resistance determinants by PCR amplification as described by Shortridge et al. (13). Primers were chosen by using Oligo 5.0 software (NBI, Plymouth, MI) from sequences deposited in Genbank. Strains negative for both mefE (efflux mutants) and ermAM (ribosome methylase) were characterized further for the presence of other methylase genes (ermA and ermC), erythromycin esterase (ereA and ereB), MS efflux (msrA) and the gene for lincosamide resistance (linA), again as described by Shortridge et al. (13).