Longitudinal Assessment of Antipneumococcal Susceptibility in the United States

Abstract

The prevalence of antimicrobial resistance among 4,940 U.S. pneumococcal isolates collected during 1999 was as follows: penicillin, 16.2%; amoxicillin-clavulanate, 12.2%; cefuroxime, 28.1%; ceftriaxone, 3.6%; trimethoprim-sulfamethoxazole, 30.3%; azithromycin, 21.4%; levofloxacin, 0.6%; and moxifloxacin, 0.1%. Compared to the previous 1997-1998 study (Jones et al., Antimicrob. Agents Chemother. 44:2645-2652, 2000), increases were noted for resistance to penicillin (3.7%; P < 0.001), amoxicillin-clavulanate (3.9%; P < 0.001), cefuroxime (5.7%; P < 0.001), azithromycin (2.4%; P = 0.014), trimethoprim-sulfamethoxazole (15.4%; P < 0.001), and levofloxacin (0.3%; P = 0.017). Resistance to ceftriaxone (0.1%; P = 0.809) and moxifloxacin (0.03%; P = 0.570) decreased. Concurrently, multidrug resistance increased (P < 0.001) from 6.3% to 11.3%.

Community-acquired respiratory tract infections (RTIs), from which Streptococcus pneumoniae is the most frequently isolated bacterial pathogen, are a major cause of disease in the United States (1). Antibiotic resistance in this species is a significant issue (5, 6, 12, 25, 26). Except in some regions (9), resistance to more recently introduced fluoroquinolones used to treat RTIs, such as moxifloxacin, gatifloxacin, and levofloxacin, is rare (6, 12, 19, 25, 26). Some reports suggest resistance to ciprofloxacin (2, 8) is increasing, although this is not always clear (21). We report data on the in vitro activity of moxifloxacin, levofloxacin, and other antimicrobials potentially used against S. pneumoniae from the second phase (1999) of a multiyear surveillance program started in 1997-1998 (12).

During 1999, non-repeat isolates of S. pneumoniae were collected from 290 hospital laboratories throughout the nine U.S. Bureau of the Census regions. There were 156 of 290 (53.8%) sites that had also participated in the 1997-1998 study and had submitted at least 10 S. pneumoniae in each study (12). Protocols for organism collection and reidentification were identical to those described previously (12). Except for the substitution of azithromycin for erythromycin and the addition of vancomycin, the antimicrobials tested remained the same and were susceptibility tested by broth microdilution according to the recommendations of the National Committee for Clinical Laboratory Standards (NCCLS) (15, 17). A subselection of strains included in this study were also retested as part of a faropenem comparative study (3).

The 156 hospital sites participating in both studies were included in the longitudinal data analysis. Statistical analyses were performed with the chi-square test or Wilcoxon's rank sum test (testing the null hypothesis that the median MIC distributions between years for each drug are the same). P values <0.05 were considered statistically significant. Nucleotide sequences of quinolone-resistance-determining regions (QRDRs) of gyrA, gyrB, parC, and parE and pulsed-field gel electrophoresis (PFGE) (11, 13) were undertaken for isolates for which levofloxacin MICs were ≥4 μg/ml or moxifloxacin MICs were ≥2 μg/ml. PFGE results were analyzed with Molecular Analyst software (Bio-Rad, Hercules, Calif.) with gels normalized using S. pneumoniae R6 (ATCC 27336) restricted with SmaI (standard gel image supplied by L. McDougal, Centers for Disease Control and Prevention, Atlanta, Ga.). Percentages of similarity between isolates were calculated with the Dice coefficient.

The overall susceptibility profiles for S. pneumoniae isolates categorized according to penicillin susceptibility from all 290 sites are given (Table 1). Table 2 shows a longitudinal comparison of susceptibility data from the 156 sites that participated in both the 1997-1998 and 1999 studies. Overall increases in nonsusceptibility were recorded for all agents: amoxicillin-clavulanate, 5.3%; cefuroxime, 4.2%; ceftriaxone, 3.4%; azithromycin, 3.4%; trimethoprim-sulfamethoxazole (SXT), 6.7%; penicillin, 1.0%; levofloxacin, 0.4%; and moxifloxacin, 0.3%. Significant increases in resistant isolates were recorded for penicillin (P < 0.001), amoxicillin-clavulanate (P < 0.001), cefuroxime (P < 0.001), SXT (P < 0.001), azithromycin (P = 0.014), and levofloxacin (P = 0.017). For levofloxacin, this represented an increase to 31 resistant strains from 9 strains in 1997-1998. Two isolates showed high-level penicillin resistance (MICs of >32 and 16 μg/ml, respectively) recovered from hospitals in the West North Central and Pacific regions.

TABLE 1.

Susceptibility of all S. pneumoniae isolates to antimicrobials according to penicillin susceptibility status (n = 4,940) from 1999

Antimicrobial agent and penicillin susceptibility statusa	MIC (μg/ml)g				% of isolatesb:
	Range	50%	90%	Mode	S	I	R
Penicillin
All	≤0.06->32	≤0.06	2	≤0.06	64.6	19.3	16.2
Pen-S	≤0.06	≤0.06	≤0.06	≤0.06	100	0	0
Pen-I	0.12-1	0.5	1	1	0	100	0
Pen-R	2->32	2	4	2	0	0	100
Amoxicillin-clavulanatec
All	≤0.03-16	≤0.03	2	≤0.03	77.9 (94.6)	9.9 (3.4)	12.2 (2.0)
Pen-S	≤0.03-0.25	≤0.03	≤0.03	≤0.03	100 (100)	0 (0)	0 (0)
Pen-I	≤0.03-8	0.5	1	0.5	68.3 (98.5)	23.2 (1.3)	8.5 (0.2)
Pen-R	0.5-16	2	8	2	0.9 (68.4)	33.5 (19.7)	65.6 (11.9)
Cefuroxime
All	≤0.25->16	≤0.25	8	≤0.25	70.1	1.9	28.1
Pen-S	≤0.25-4	≤0.25	≤0.25	≤0.25	99.8	0.1	0.0d
Pen-I	≤0.25->16	2	4	4	29.1	9.3	61.6
Pen-R	2->16	8	16	8	0	0	100
Ceftriaxonee
All	≤0.03-16	≤0.03	1	≤0.03	82.7 (96.4)	13.7 (2.1)	3.6 (1.5)
Pen-S	≤0.03-0.5	≤0.03	0.06	≤0.03	100 (100)	0 (0)	0 (0)
Pen-I	≤0.03-4	0.5	1	0.5	89.7 (99.3)	9.6 (0.6)	0.7 (0.1)
Pen-R	0.5-16	1	2	1	8.1 (78.6)	73.5 (12.4)	21.4 (9.0)
Azithromycin
All	≤0.06->32	≤0.06	6	≤0.06	74.9	3.7	21.4
Pen-S	≤0.06->32	≤0.06	≤0.12	≤0.06	94.5	0.7	4.8
Pen-I	≤0.06->32	1	>32	≤0.06	49.8	7.9	42.3
Pen-R	≤0.06->32	2	>32	>32	26.7	11.0	62.3
SXT
All	≤0.03->4	0.25	>4	0.25	63.3	6.5	30.3
Pen-S	≤0.03->4	0.25	1	0.25	86.7	5.3	8.1
Pen-I	0.06->4	4	>4	>4	33.0	11.0	56.0
Pen-R	0.06->4	4	>4	>4	5.9	5.9	88.2
Levofloxacin
All	≤0.008-32	0.5	1	0.5	99.3	0.1	0.6
Pen-S	≤0.008-16	0.5	1	0.5	99.3	0.1	0.6
Pen-I	0.25-16	0.5	1	0.5	99.4	0	0.6
Pen-R	0.25-32	0.5	1	0.5	99.0	0.1	0.9
Moxifloxacinf
All	≤0.008-4	0.12	0.12	0.12	99.5	0.4	0.1
Pen-S	≤0.008-4	0.12	0.12	0.12	99.5	0.5	0.1
Pen-I	0.03-4	0.12	0.12	0.12	99.7	0.2	0.1
Pen-R	0.06-4	0.12	0.12	0.12	99.2	0.5	0.3
Vancomycin
All	≤0.12-1	0.25	0.5	0.25	100	-h	-
Pen-S	≤0.12-1	0.25	0.5	0.25	100	-	-
Pen-I	≤0.12-1	0.25	0.5	0.25	100	-	-
Pen-R	≤0.12-1	0.25	0.5	0.25	100	-	-

^aA total of 3,189 isolates were penicillin susceptible (Pen-S), 952 were penicillin intermediate (Pen-I), and 799 were penicillin resistant (Pen-R).

^bS, susceptible; I, intermediate; R, resistant.

^cNCCLS 1999 breakpoints (16) were used to interpret data for amoxicillin-clavulanate. Interpreted data utilizing NCCLS 2002 breakpoints (18) are shown in parentheses.

^dActual percent resistance was 0.02%.

^eNCCLS 1999 breakpoints (16) were used to interpret data for ceftriaxone. Interpreted data utilizing NCCLS 2002 interpretations (18) are shown in parentheses.

^fNCCLS 2002 breakpoints (18) were used to interpret data for moxifloxacin.

^g50% and 90%, MIC₅₀ and MIC₉₀, respectively.

^h-, no NCCLS criteria (18) available for I and R category intrepretations.

TABLE 2.

Longitudinal comparison of susceptibility of pneumococcal isolates from 1997-1998 (n = 3.352) and 1999 (n = 3.315)^a

Antimicrobial agent	Yr	% of isolates (% change)b			P valuec	MIC (μg/ml)
		S (change)	I (change)	R (change)		Mode	90%f
Penicillin	1997-1998	65.4	21.9	12.7		≤0.03	2
	1999	64.4 (−1.0)	19.2 (−2.7)	16.4 (+3.7)	<0.001	≤0.06	2
Amoxicillin-clavulanate	1997-1998	83.0	8.6	8.4		≤0.01	1
	1999	77.7 (−5.3)	10 (+1.4)	12.3 (+3.9)	<0.001	≤0.03	2
Cefuroxime	1997-1998	73.6	3.4	23.0		≤0.12	4
	1999	69.4 (−4.2)	1.9 (−1.5)	28.7 (+5.7)	<0.001	≤0.25	8
Ceftriaxone	1997-1998	85.8	10.6	3.6		≤0.01	1
	1999	82.4 (−3.4)	14.1 (+3.5)	3.5 (−0.1)	0.809	≤0.03	1
Azithromycind	1997-1998	78.0	2.9	19.1		≤0.03	4
	1999	74.6 (−3.4)	3.9 (+1.0)	21.5 (+2.4)	0.014	≤0.06	8
SXT	1997-1998	69.1	15.5	15.4		0.12	4
	1999	62.4 (−6.7)	6.8 (−8.7)	30.8 (+15.4)	<0.001	0.25	>4
Levofloxacin	1997-1998	99.9	0	0.1		0.5	1
	1999	99.5 (−0.4)	0.1 (+0.1)	0.4 (+0.3)	0.017	0.5	1
Moxifloxacin	1997-1998	99.9	<0.1	<0.1		0.12	0.25
	1999	99.6 (−0.3)	0.4 (>+0.3)	<0.1 (−0.0)e	0.670	0.12	0.12

^aOnly data from the 156 laboratories that participated in both studies and submitted a minimum of 10 isolates of pneumococci were included in the longitudinal analysis.

^bValues illustrate the percent change in the susceptible (S), intermediate (I), or resistant (R), population between 1997-1998 and 1999.

^cP values were calculated based on the number of isolates resistant to each antimicrobial in each study period.

^dErythromycin was the macrolide tested in 1997-1998; azithromycin was tested in 1999.

^eActual percent change, −0.03%.

^f90%, MIC₉₀.

Similar to the 1997-1998 study (12), fluoroquinolone resistance remained independent of resistance to other agents (including penicillin and macrolides) (Table 1). Of the 31 isolates resistant to levofloxacin, 6 remained susceptible to moxifloxacin (20 intermediate, 5 resistant). Comparative fluoroquinolone MIC distributions for both studies are shown (Fig. 1). Statistical tests show that the differences in MIC distributions per drug per year appear significant for both agents (P < 0.001), although for levofloxacin, this represents a shift towards a raised median MIC, and for moxifloxacin, it represents a shift towards a lower median MIC. While the number of levofloxacin-resistant isolates appeared to increase significantly, no interpretive differences were noted for moxifloxacin, and modal MICs or MICs at which 90% of the isolates tested are inhibited (MIC₉₀s) for either fluoroquinolone remained constant (or decreased in the case of moxifloxacin) during both years (Table 2). Shifts are not likely to be clinically relevant for either fluoroquinolone. In addition, a recent study showed for S. pneumoniae that no strains for which levofloxacin MICs were 0.5 μg/ml and approximately only 5% of strains for which levofloxacin MICs were 1 μg/ml carried key mutations in QRDRs (4). As such, the MIC shifts measured (Fig. 1) are not likely to reflect mutational events in QRDRs and likely demonstrate the inherent weakness of statistical analysis in studies that incorporate a large number of isolates. In addition, the accepted artifact of ±1 dilution in MIC test results can also complicate comparative analysis of susceptibility test results. Comparison of MIC distributions from future years will confirm whether trends are real or artifactual.

FIG. 1.

Comparison of distributions of levofloxacin and moxifloxacin MICs from the 1997-1998 and 1999 studies. The achievable maximum concentrations of both drugs in serum following a single 400-mg (moxifloxacin) or 500-mg (levofloxacin) oral dose are 4.5 ± 0.53 and 5.1 ± 0.8 μg/ml, respectively (according to the package inserts).

Defining multiple drug resistance (MDR) as resistance to ≥3 of the agents penicillin, ceftriaxone, erythromycin, SXT, vancomycin, and either levofloxacin or moxifloxacin and considering only isolates included in the longitudinal analysis, 212 of 3,352 isolates (6.3%) from the 1997-1998 study, and 374 of 3,315 isolates (11.3%) from 1999 exhibited MDR phenotypes, an increase of 5.0% (P < 0.001). The number of organisms testing as susceptible to all compounds decreased from 57.2% (n = 1,919) to 53.8% (n = 1,784) during the same period (P < 0.005). Only four MDR isolates demonstrated resistance to levofloxacin, of which none were moxifloxacin resistant. Of the 35 isolates (0.7%) nonsusceptible to levofloxacin (4 intermediate, 31 resistant), 26 (0.5%) were nonsusceptible to moxifloxacin (21 intermediate, 5 resistant). Similar to the 11 isolates (0.2%) from the 1997-1998 study for which levofloxacin MICs were ≥4 μg/ml, single or combinations of classic QRDR alterations (7, 10, 11, 14, 20, 22, 23), namely GyrA (Ser81-Phe/Tyr or Glu85-Lys) with or without ParC alterations (Ser79-Phe/Tyr, Asp83-Asn, Lys137-Asn) and with or without the ParE Ile460-Val alteration, were detected in each fluoroquinolone-resistant isolate. No novel mutations were detected in the QRDRs in gyrA or parC in any isolates. The contribution of efflux was not measured, since all fluoroquinolone-resistant isolates possessed key QRDR mutations.

The 35 fluoroquinolone-nonsusceptible isolates varied considerably with respect to PFGE band profiles (difference of more than five bands) and were identical in only three instances. Two of these were isolates submitted by the same site. The third set consisted of two isolates submitted by hospitals in Illinois and New York. Aside from these isolates, few of the levofloxacin-resistant isolates demonstrated similarities of >80% with the Dice coefficient to interpret isolate banding patterns (data not shown), and none could be considered closely related (24).

The 22.1% rate of nonsusceptibility recorded for penicillin reflects recent U.S. studies reported, including the TRUST (24, 25) and SENTRY (5, 19) studies. Similarly, the high levels of nonsusceptibility to macrolides (25.1%) and SXT (36.8%) reported confirm recent observations (5, 6, 19, 26). The two high-level β-lactam-resistant isolates represent unusual strains rarely isolated in the United States and suggest the need to test for higher-endpoint MICs. Longitudinal trending shows the continued emergence of penicillin-, amoxicillin-clavulanate-, cefuroxime-, macrolide-, and SXT-resistant S. pneumoniae. The increased incidence of MDR phenotypes demonstrates that several antimicrobial classes, including penicillin derivatives, cefuroxime, azithromycin, and SXT, are potentially able to select MDR isolates, an increasing problem that needs careful study. In contrast, fluoroquinolone resistance was rarely associated with MDR phenotypes and occurred in only four MDR isolates, suggesting that, at least for now, fluoroquinolones may play a role in strategies to avoid the selection of MDR types. Apart from vancomycin, moxifloxacin and levofloxacin were the most active compounds in vitro. Analysis by PFGE of the 35 levofloxacin- or moxifloxacin-nonsusceptible isolates showed little evidence of clonal expansion of a fluoroquinolone-resistant strain of pneumococci, such as the so-called “type A” strain we identified as predominating during the 1997-1998 study (12).