Minimal inhibitory and mutant prevention concentrations of azithromycin, clarithromycin and erythromycin for clinical isolates of Streptococcus pneumoniae

Abstract

Background

Previous work showed a higher prevalence of macrolide/azalide resistance in provinces of Canada where azithromycin was the major treatment for Streptococcus pneumoniae as compared with regions where clarithromycin was the dominant treatment. These data provided a way to test the mutant selection window hypothesis, which predicts that the serum drug concentration (AUC₂₄) relative to the mutant prevention concentration (MPC) would be higher for clarithromycin than for azithromycin.

Methods

The MIC and MPC were determined for 191 penicillin/macrolide-susceptible clinical isolates of S. pneumoniae with azithromycin, clarithromycin and erythromycin using agar plate assays.

Results

The MIC_50/90 (mg/L) and MPC_50/90 (mg/L), respectively, were as follows: azithromycin 0.13/0.25 and 1/4; clarithromycin 0.031/0.063 and 0.13/0.5; erythromycin 0.063/0.13 and 0.25/2. We calculated from published pharmacokinetic values that the AUC₂₄/MPC₉₀ for azithromycin was 0.85; for clarithromycin it was 96, and for erythromycin base and estolate it was 4 and 10, respectively. Thus the AUC₂₄/MPC₉₀ was about 50 times higher for clarithromycin than for azithromycin.

Conclusions

The elevated prevalence of azithromycin resistance may derive in part from a low value of AUC₂₄/MPC₉₀ and/or time above MPC, since previous work indicates that the number of prescriptions per person was similar in the geographical regions examined.

Introduction

The macrolide/azalide antimicrobials have been important antimicrobials since the 1950s when the macrolide erythromycin was introduced. During the 1990s, two extended-spectrum compounds, azithromycin (an azalide) and clarithromycin (a macrolide), were added. The new agents offered a broader spectrum of clinical applications and reduced side effects.^1,2 Potent activity against atypical pathogens, such as Mycoplasma spp. and Chlamydia spp., and fastidious Gram-negative bacilli solidified a role for the azalide/macrolides as (i) monotherapy for the treatment of mild to moderate community-acquired respiratory tract infections^3–5 and (ii) components in combination therapy that included agents, such as third-generation cephalosporins, for the treatment of moderate to severe community-acquired pneumonia requiring hospitalization.⁵ Azalide/macrolide compounds are also widely used to treat acute bacterial exacerbations of chronic bronchitis and acute maxillary sinusitis.^4,6 Unfortunately the global escalation of antimicrobial resistance among respiratory tract pathogens extends to the azalide/macrolide agents, with particularly dramatic resistance increases occurring among isolates of Streptococcus pneumoniae.⁷ In Canada the prevalence of S. pneumoniae resistance has been much higher in provinces that preferentially use azithromycin rather than clarithromycin (and erythromycin), even though the total number of prescriptions is similar.^1,8 This observation raises the possibility that members of the macrolide/azalide family differ in their intrinsic propensity to selectively enrich resistant mutant subpopulations.

One way to compare compounds for selective enrichment of mutants is based on measurements of the mutant prevention concentration (MPC). The MPC is essentially the MIC of the least-susceptible, first-step mutant subpopulation present in a large bacterial population. If antimicrobial concentrations are kept above the MPC, mutant subpopulations are unlikely to be enriched.^9,10 Thus the MPC can be used like the MIC in pharmacodynamic considerations: compounds that are less likely to selectively enrich resistant mutants will have higher values of AUC₂₄/MPC. Since MPC has not been studied for the azalide/macrolides with S. pneumoniae, it is not known whether the higher prevalence of resistance seen with the use of azithromycin correlates with a lower value of AUC₂₄/MPC, i.e. a higher intrinsic propensity for enrichment of mutant subpopulations.

In the present work we measured MIC and MPC for more than 190 azalide/macrolide-susceptible clinical isolates of S. pneumoniae with azithromycin, clarithromycin and erythromycin. MPC values were lower for clarithromycin than for either erythromycin or azithromycin. Consideration of MPC data in conjunction with drug pharmacokinetics suggests that clarithromycin has a lower propensity to selectively enrich macrolide-resistant S. pneumoniae, a feature that fits with geographic patterns of clinical resistance.

Materials and methods

Bacterial strains

All isolates were collected through the Clinical Microbiology Laboratory at Royal University Hospital, Saskatoon, Saskatchewan, Canada. Isolates were identified by reference methodology¹¹ and stored frozen in skim milk at −70°C until testing. Duplicate isolates from the same patient were excluded. For inclusion in the study, a strain had to be susceptible to the drugs tested according to the recommended breakpoints of the CLSI.^12,13

Antimicrobial agents

Azithromycin was obtained from Pfizer Canada (Kirkland, QC), and clarithromycin was from Abbott Canada (Saint-Laurent, QC). Erythromycin was purchased commercially (Sigma-Aldrich Canada, Oakville, ON). For testing, the antibiotics were prepared in accordance with the manufacturer's recommendations/specifications. Antibiotic potency was confirmed using two ATCC strains: S. pneumoniae ATCC 49619 and Staphylococcus aureus ATCC 29213. These organisms were included as standards in each set of MIC and MPC measurements.

MIC determination

The guidelines and interpretation of the CLSI were followed for MIC determination.^12,13 Briefly, isolates stored at −70°C were thawed, subcultured using tryptic soy agar plates containing 5% sheep red blood cells (BA) (PML, Richmond, BC, Canada) and incubated for 18–24 h at 35–37°C in 5% CO₂. Isolated colonies were transferred to Todd–Hewitt broth (THB) and cultures were grown to a cell density of approximately 10⁸ cfu/mL. Cultures were diluted into microtitre plates to achieve a final inoculum of 10⁵ cfu/mL in THB containing drugs at various concentrations. Inoculated microtitre plates were incubated for 18–24 h at 35–37°C in 5% CO₂. Concentrations of drug were tested in doubling dilutions. The lowest drug concentration that prevented visible growth was taken as the MIC.

MPC determination

MPC determination with S. pneumoniae for the macrolide/azalide compounds was a modification of a method described previously for fluoroquinolones with large numbers of S. pneumoniae isolates.¹⁴ Briefly, each test isolate was applied to approximately six BA plates and incubated overnight. Then all of the bacterial cells on the plate were transferred to 500 mL THB and incubated overnight. Cells in the resulting culture were concentrated by centrifugation at 5000 g for 30 min at 4°C. The resulting cell pellet was re-suspended in approximately 3 mL of THB and the cell concentration was determined by turbidity measurement. At least 10⁹ organisms were inoculated onto each member of a series of agar plates containing 2-fold concentration increments of drug. After incubation for 24 h, the presence or absence of bacterial growth was determined, and the incubation was continued for another 24 h. The plates were again screened for growth to determine the lowest drug concentration that prevented growth. To confirm the absence of growth, the cells on the surface of the plates were removed, applied to drug-free agar plates, incubated overnight and then transferred to agar plates having the same drug concentration used to initially isolate the strain. For these experiments, drug plates were prepared and used within 7 days; they contained 0.031–32 mg/L for each of the three drugs.

Statistical analysis

The number of strains with MPC values of 1, 2, 4 and ≥8 mg/L for each agent was compared by the χ² test. A P value ≤0.05 was considered significant.

Results

The modal MIC values for azithromycin, clarithromycin and erythromycin with clinical isolates of S. pneumoniae were 0.125, 0.031 and 0.063 mg/L, respectively; MIC₉₀ values were 0.25, 0.063, and 0.125 mg/L, respectively (Table 1). For all agents, the majority of isolates had an MIC ≤0.125 mg/L. Thus all isolates were considered susceptible to the drugs tested by breakpoint criteria (breakpoints were 0.5, 0.25 and 0.25 mg/L for azithromycin, clarithromycin and erythromycin, respectively).^12,13

Table 1.

MIC/MPC distribution for azalide/macrolide compounds with clinical isolates of Streptococcus pneumoniae (n = 191)

Compound	MIC distribution dataa										MIC50 b	MIC90 b
	≤0.16	0.031	0.063	0.125	0.25	0.5	1	2	4	≥8
Azithromycin	0	15	63	91	20	2					0.125	0.25
Clarithromycin	57	105	28	1							0.031	0.063
Erythromycin	1	23	111	49	7						0.063	0.125
	MPC distribution dataa
	≤0.016	0.031	0.063	0.125	0.25	0.5	1	2	4	≥8	MPC50 c	MPC90 c
Azithromycin				1	10	46	63	37	18	16	1	4
Clarithromycin			49	61	45	17	10	3	5	1	0.125	0.5
Erythromycin			1	20	83	43	20	9	4	11	0.25	2

^aThe heading row shows drug concentrations (mg/L); for each drug, the number of isolates for a given concentration is listed in the body of the table.

^bDrug concentration at which 50% or 90% of strains, respectively, are inhibited.

^cDrug concentration at which growth was inhibited for 50% or 90% of strains, respectively, based on inoculum ≥10⁹ cfu.

The modal MPCs were 1 mg/L for azithromycin, 0.125 mg/L for clarithromycin and 0.25 mg/L for erythromycin (Table 1). For clarithromycin, 155/191 (81.2%) isolates had MPC values less than or equal to the MIC susceptibility breakpoint, as compared with 57/191 (29.8%) for azithromycin and 104/191 (54.5%) for erythromycin. Significantly more isolates had MPC values of 1, 2, 4 or ≥8 mg/L for azithromycin than for clarithromycin (P < 0.0001 for comparisons at all four concentrations). Significantly more strains had MPC values of 1, 2 or 4 mg/L for azithromycin than for erythromycin (P < 0.0001, P < 0.0001, P = 0.004, respectively). Finally, significantly more strains had MPC values of 1, 2, 4 and ≥8 mg/L for erythromycin than for clarithromycin (P < 0.0001, P = 0.007, P = 0.044, P = 0.003, respectively). The MPC₅₀ values for azithromycin, clarithromycin and erythromycin were 1, 0.125 and 0.25 mg/L; MPC₉₀ values were 4, 0.5 and 2 mg/L, respectively. By these criteria, clarithromycin was the most active of the three compounds and azithromycin was the least active.

MPC and MIC values were combined with pharmacokinetic parameters, reported for approved doses of each compound, to determine pharmacodynamic indices (Table 2). If we consider MIC₉₀ and MPC₉₀ to represent the boundaries of the mutant selection window,¹⁵ the window size for azithromycin was 0.25/1 = 4, for clarithromycin it was 0.063/0.25 = 4 and for erythromycin it was 0.125/0.25 = 2. Thus the three compounds are similar by this criterion. However, when time above MPC was determined, azithromycin was 0 h, clarithromycin was 24 h and erythromycin was ∼1 h; erythromycin estolate was ∼5 h (Table 2). Consequently only clarithromycin produces concentrations that are above the mutant selection window for the entire dosing interval. Another way to consider the data is through expression of the AUC₂₄/MPC₉₀. The AUC₂₄/MPC₉₀ was 0.85 for azithromycin, 96 for clarithromycin, and 4 and 10 for erythromycin base and estolate, respectively. A similar consideration can be used with C_max/MPC, which was 0.1 for azithromycin, 7.5 for clarithromycin and 0.42 and 6.2 for erythromycin base and estolate, respectively. Thus clarithromycin has higher values of both AUC₂₄/MPC and C_max/MPC than the other two agents. These serum-based pharmacodynamic considerations indicate that clarithromycin is the least likely to selectively enrich non-susceptible mutant subpopulations.

Table 2.

Pharmacokinetic and pharmacodynamic values for azalide/macrolide agents

Compound	Dosage	Cmax	AUC24	Cmax/MIC90	Cmax/MPC90	AUC24/MIC90	AUC24/MPC90	T > MIC90	T > MPC90	TMSW
Azithromycina	500 mg	0.4	3.4	1.6	0.1	13.6	0.85	>24	0	24
Clarithromycin XLb	2 × 500 mg	3.77	48.09	59.8	7.5	763.3	96.2	24	24	0
Erythromycin basea	500 mg	0.9	8	7.2	0.45	64	4	∼14c	∼1c	13c
Erythromycin estolate32	500 mg	3.1	20.39	12.4	6.2	163.12	10.2	∼18	∼5	13

AUC₂₄ = area under curve over a 24 h time period; C_max = serum maximum concentration; MIC, minimal inhibitory concentration; MPC = mutant prevention concentration; T_MSW, time inside the mutant selection window (h). Concentrations are in mg/L.

^aMean values based on a single 500 mg oral dose; based on values as published in Zuckerman et al.³³

^bSee Biaxin monograph.³⁴

^cDetermined from data in reference 32.

Discussion

Azalide/macrolide antimicrobials remain important for the treatment of moderate to severe forms of community-acquired respiratory tract infections.^3–5 For example, annual Canadian prescriptions number about 3.6 million for the group (clarithromycin, 1.9 million; erythromycin, 0.13 million; azithromycin, 1.6 million) (IMS data, Canada). However, the global escalation of macrolide and penicillin/macrolide cross-resistant strains is compromising the use of these important compounds. For example, azalide resistance has reached 19% in Portugal and almost 47% in France; Canadian data indicate that approximately 10% of more than 6900 S. pneumoniae isolates were macrolide/azalide resistant (1997–2002).¹⁶ More recent data from the Canadian Bacterial Surveillance Network report macrolide resistance for S. pneumoniae to be between 20% and 25% from 2007 to 2009,¹⁷ 2.5 times the rate reported a decade earlier. Thus, maintaining the utility of the azalides/macrolides is a major challenge.

Davidson et al.⁸ made the intriguing observation that in Canadian provinces between 1995 and 2002 the use of particular azalides or macrolides correlated with azalide/macrolide resistance. Provinces that used the highest amount of azithromycin saw the greatest increases in macrolide resistance. In provinces with lower resistance prevalence, azithromycin use was less than 20% of prescribed macrolides, but it was >44% in provinces with the highest resistance rates. No correlation was found between total macrolide consumption and regional resistance rates.⁸ The present work provides microbiological data suggesting that azithromycin is intrinsically most likely to selectively enrich resistant mutant subpopulations, as AUC₂₄/MPC, C_max/MPC, and T _> MPC were lower for this drug than for the two other macrolide compounds examined. Moreover, azithromycin persists inside the mutant selection window for longer times (24 h) than the two other agents (0 h for clarithromycin, 13 h for erythromycin). One way to correct this negative attribute of azithromycin would be to increase the dose, as was done for levofloxacin¹⁸ following comparison with moxifloxacin.¹⁹

Two reasons exist for using MPC rather than MIC for considering resistance. First, the frequency at which mutations occur is on the order of 1 × 10⁻⁷ to 1 × 10⁻⁹, a frequency that would not normally be detected by traditional susceptibility testing in which an inoculum size of 10⁵ cfu is used. Consequently, an isolate deemed susceptible may contain an undetected subpopulation of resistant cells that would require a higher drug concentration to restrict growth. Second, in some respiratory tract infections, such as community-acquired pneumonia, bacteria present at the site of infection may well exceed 10⁵ cfu. Indeed, Frisch et al.²⁰ reported that with community-acquired pneumonia, bacterial load may range between 10¹⁰ and 10¹² organisms, and Fagon et al.²¹ showed that bacterial densities for S. pneumoniae and Haemophilus influenzae could exceed 10⁷ cfu/mL in patients with acute bacterial exacerbations of their chronic bronchitis. Such patients would have a bacterial load of 10⁸ cfu. Thus it is likely that these infections contain subpopulations of resistant mutants that are enriched during treatment.

That enrichment of resistant subpopulations during treatment occurs is supported by several observations. First, Cornick and Bentley²² noted a strong association between macrolide resistance and previous macrolide or azithromycin use.²³ Second, Vanderkooi et al.²⁴ reported that patients were four times more likely to have macrolide-, fluoroquinolone- or trimethoprim/sulfamethoxazole-resistant pathogens associated with invasive pneumococcal infection if they had received the same drug dose within the previous 3 months (P < 0.001, P < 0.001, P < 0.04 for the three drugs, respectively). Regarding patients with macrolide-resistant strains, resistance was selected in the presence of azithromycin and clarithromycin; however, >50% of these patients had received azithromycin rather than other macrolides (i.e. clarithromycin), and a significant difference was seen between the likelihood for macrolide-resistant pneumococcus in patients exposed to azithromycin than to clarithromycin (P = 0.03).

We note that the choice of dose is not the only factor involved in the emergence of resistance: the level of macrolide use is clearly important.^24,25 For example, a reduction in the use of erythromycin reduced the prevalence of macrolide resistance.²⁶ Moreover, horizontal transfer of resistance genes may lower the relevance of MPC measurements to azalide/macrolide resistance,²⁷ just as clonal spread of a strain reduces the importance of the emergence of new resistant strains. However, the initial acquisition of resistance alleles from mobile elements presumably arose from spontaneous events. Thus, identifying ways to restrict the emergence of resistance is likely to be important.

The present measurements of azalide/macrolide MPC with S. pneumoniae add to previous work that compared fluoroquinolones for their propensity to select resistant subpopulations within high-density bacterial populations (i.e. inocula ≥10⁹ cfu).^14,19,28 MPC measurements have also been reported for β-lactams (including cephalosporins and carbapenems), aminoglycosides, glycylcyclines, oxazolidinones, glycopeptides and metronidazole with a wide variety of bacteria that include S. aureus, S. pneumoniae, H. influenzae, Escherichia coli and other members of the Enterobacteriaceae, Pseudomonas aeruginosa, and Clostridium difficile.^29,30 Thus we may be approaching a point at which MPC can be combined with pharmacokinetics to identify agents and doses that will severely restrict the emergence of resistance.³¹

In summary, the mutant selection window hypothesis¹⁵ predicts that the utility of the macrolides/azalides will be eroded by continued use of agents that place concentrations inside the selection window for long periods, as is the case for azithromycin with S. pneumoniae. That may be part of the reason for geographical differences in resistance reported for Canada.⁸ If so, prescribing practices and doses may require adjustment.

Funding

This work was funded, in part, by an unrestricted research grant from Abbott Laboratories Canada. Work by K. D. is supported by NIH grant AI73491.

Transparency declarations

The authors have received funding for antimicrobial research, meeting attendance and/or honoraria from In Site, Abbott Laboratories, Alcon, Allergan, Aventis, Bayer Healthcare, Bausch & Lomb, Bristol-Myers Squibb, Cooper Vision, Cubist, Eli-Lilly, GlaxoSmithKline, Hoechst Marion Roussel, Hoffman LaRoche, Janssen-Ortho, Oscient Pharmaceuticals, Merck Frosst, Pfizer, Rhone-Poulenc Rorer, Royal University Hospital Foundation, Smith-Kline Beecham, Wyeth-Ayerst, YM Biosciences, Novartis and Zeneca.