In Vitro Activities of Co-Amoxiclav at Concentrations Achieved in Human Serum against the Resistant Subpopulation of Heteroresistant Staphylococcus aureus: a Controlled Study with Vancomycin

Abstract

The effects of concentrations that simulated those in human serum after a single intravenous dose of amoxicillin (2 g), amoxicillin-clavulanic acid (2,000 and 200 mg, respectively), or vancomycin (500 mg), on the viability and β-lactamase activity of two isogenic (β-lactamase and non-β-lactamase producer) heteroresistant Staphylococcus aureus strains were studied in an in vitro pharmacodynamic model. A reduction of ≥97% of the initial inoculum was obtained with vancomycin and amoxicillin-clavulanic acid against both strains, with respect to the total bacterial population and the oxacillin-resistant subpopulation. The same pattern was observed with amoxicillin and the β-lactamase-negative strain. β-Lactamase activity in the β-lactamase-positive strain changed over time parallel to viability, decreasing with amoxicillin-clavulanic acid or vancomycin and increasing in the amoxicillin and control groups. Clavulanic acid concentrations achievable in serum that changed over time allowed amoxicillin to act against the β-lactamase-producing methicillin-resistant S. aureus to a similar extent as vancomycin.

Infections due to methicillin-resistant Staphylococcus aureus (MRSA) became epidemiological and clinical problems in the 1980s (15) and remain so in the current decade (23). MRSA is considered resistant to all penicillins, cephalosporins, carbapenems, and monobactams (24), vancomycin being the standard treatment for infections due to MRSA. Despite the cross-resistance that MRSA exhibits to all β-lactams, the majority of strains exhibit heteroresistance, with only a small subpopulation expressing obvious resistance while the majority remain susceptible (12). Induction by β-lactams of the gene mecA is seen in those strains with inducer-repressors, and the resistance phenotype is induced by environmental conditions (such as temperature and osmolarity) (5); thus, the promotion of growth of the resistant subpopulation is favored by incubation at cooler temperatures (30 to 35°C) and the presence of NaCl (2 to 4%) (15).

The production of PBP 2A (encoded by mecA) with low β-lactam affinity provides, in the absence of β-lactamase production, resistance to most types of β-lactams (3). But interestingly, traditional β-lactams such as penicillin G, amoxicillin, and ampicillin, have relatively good affinity for PBP 2A, amoxicillin showing 10 times greater affinity than methicillin, oxacillin, and clavulanate (which do not overcome the subpopulation resistance) (10). In addition, most MRSA strains also contain the β-lactamase resistance mechanism, contributing to elevated MICs (10).

Due to the higher affinity of amoxicillin for PBP 2A and the β-lactamase-blocking effect exerted by clavulanate, the clinical efficacy of the combination in the treatment of MRSA infections has been hypothesized, this being supported by the reported therapeutic efficacy in rat models of endocarditis (10). The presence of 2 to 4 μg of clavulanic acid per ml allows concentrations of 8 to 25 μg of amoxicillin per ml to inhibit 50 to 100% of MRSA strains, these concentrations being unachievable in humans by the oral route but achievable with the intravenous formulation (25).

In this study, we assessed the in vitro effects of co-amoxiclav, amoxicillin, and vancomycin versus control, using concentrations simulating those in human serum over time, on the viability of the total population and the resistant subpopulation of two isogenic strains (one producing penicillinase and the other not) of MRSA. In addition, the effect on the β-lactamase activity of the corresponding strain was measured.

MATERIALS AND METHODS

The two isogenic strains used in this study were kindly supplied by P. Moreillon (CHUV, Lausanne, Switzerland). The parent strain (1. Paris, 1984) and its penicillinase-negative derivate, together with the method of obtaining it, were fully described by Franciolli et al. (10).

The MICs of amoxicillin, co-amoxiclav (2:1), and vancomycin for these two strains were determined in Mueller-Hinton (MH) broth by a standard method (19) and also in MH broth with 2% NaCl incubated at 35°C.

The pharmacodynamic simulation method used has been previously described for S. aureus (1, 17). The concentrations used were similar to those obtained in serum after an intravenous dose of 2 g of amoxicillin combined with 200 mg of clavulanic acid (6, 26) or after 500 mg of vancomycin (8, 16). The initial inoculum was obtained by diluting an overnight culture of MH broth (Difco Laboratories, Detroit, Mich.) with 2% NaCl (to allow the growth of the resistant subpopulation) in new fresh broth, which was incubated in a 35°C shaking bath until an A₅₈₀ of 0.3 (Hitachi U 1,100 spectrophotometer) was reached. The culture was further diluted 1:10, resulting in a final inoculum of approximately 10⁷ CFU/ml in a 4-ml volume of MH broth with 2% NaCl containing the initial co-amoxiclav, amoxicillin, or vancomycin concentrations (Table 1). Incubation was performed with Centriprep-10 concentrator tubes (membrane pore size, 10,000 Da; Amicon, Beverly, Mass.) at 35°C in a shaking bath for the first incubation period listed in Table 1. After incubation, aliquots of 200 μl were diluted in 1,800 μl of sterile saline solution, and appropriate decimal solutions were prepared in MH agar plates with 4% NaCl for colony counting of the total population and in MH agar plates with 4% NaCl and 25 μg of oxacillin per ml for colony counting of the resistant subpopulation, with incubation at 35°C for 48 h. The limit of detection was 50 CFU/ml. Centrifugation-filtration for 10 min at 2,000 × g was then performed, and the supernatant above the filter was carefully collected for further determination of β-lactamase activity. A bacterial suspension volume of approximately 500 μl remained below the filter. New MH broth (4 ml) with the subsequent corresponding drug concentration was added to the bacterial suspension. The tubes were incubated under the conditions described above for the ensuing incubation period. This method was used throughout five incubation periods. Experiments were performed three times for each experimental group (co-amoxiclav, amoxicillin, and vancomycin). As a control, the initial inoculum in MH broth without antibiotics was processed according to the same centrifugation-filtration, colony counting, and supernatant collection procedures for the five incubation periods.

TABLE 1.

Concentrations of vancomycin and co-amoxiclav simulating those in human serum for various incubation periods

Incubation period (h)	Concn (μg/ml)
	Vancomycin	Amoxicillin	Amoxicillin-clavulanic acid
0–0.5	15	69	69/10
0.5–2	6.66	28.11	28.11/5.5
2–3	4.07	13.34	13.34/1.99
3–6	2.94	2.75	2.75/0.71
6–8	1.15	0.61	0.61/0.14

The antibiotic carryover effect, which occurred in the 500-μl bacterial suspensions remaining after the centrifugation-filtration steps, was determined. The amounts of antibiotics in the 500-μl bacterial suspensions after each incubation period were calculated by the equation y = 0.5x, where x is the initial amount of antibiotic per milliliter in the 4-ml broth (total antibiotic amount in 4 ml = 4x) and y is the resulting amount in the 500-μl suspension. Concentrations for each incubation period were calculated by adding the carried-over antibiotic amount in the 500-μl suspension to the antibiotic concentration in the new broth that was added. The final concentrations calculated are listed in Table 1. β-Lactamase activity was measured by adding 25 μl of a 500-μl/ml solution of nitrocefin (Glaxo Ltd., Greenford, Middlesex, England) to 225 μl of each broth supernatant previously collected, which was incubated for 30 min at 37°C (21). Afterwards, 1.5 ml of phosphate buffer (0.05 M) was added, and A₄₈₂ was read with the spectrophotometer (17).

Viability rate (percent decrease) with respect to the initial inoculum (I) in terms of CFU per milliliter was calculated by the equation 100 − (I_{t = x h} × 100/I_{t = 0 h}), with the viability of the initial inoculum (t = 0 h) set at 100%.

To detect the different trends over time, the repeated-measures multivariate analysis of variance was used. The time lags used in the total population comparisons lie between 0 and 4. In the subpopulation comparisons, the time lags are between 0 and 2. To check the assumptions of the analysis, the sphericity test of Anderson was used. When the Wilks P value was significant (P < 0.05), contrast between groups was made with the Tukey-Kramer test to adjust the type I error.

RESULTS

MICs of amoxicillin, co-amoxiclav (2:1), and vancomycin for the β-lactamase-positive strain were 32, 1, and 1 μg/ml, respectively, in the absence of NaCl and with 37°C incubation and ≥128, 16, and 1 μg/ml, respectively, in the presence of 2% NaCl and with 35°C incubation. For the β-lactamase-negative strain, values were 1, 1, and 1 μg/ml, respectively, in the absence of NaCl and with 37°C incubation and 16, 16, and 1 μg/ml, respectively, in the presence of 2% NaCl and with 35°C incubation.

After an overnight incubation in MH broth with 2% NaCl at 35°C to allow expression of the PBP 2A and at the appropriate dilutions, initial inoculum at time zero of the pharmacodynamic simulation was 7.2 × 10⁶ and 9.8 × 10⁶ CFU/ml for the β-lactamase-positive and -negative strains, respectively. Initial counts for the resistant subpopulation were 6.6 × 10³ and 2.8 × 10⁴ CFU/ml for the β-lactamase-positive and -negative strains, respectively. At time zero, the rates of resistant CFU per milliliter with respect to the total population were 1/1,000 for the β-lactamase-positive strain and 3/1,000 for the β-lactamase-negative strain.

The viability rates at the different incubation times for the β-lactamase-positive strain are shown in Table 2 for the total population and in Table 3 for the resistant subpopulation. Significant differences were found among amoxicillin (P = 0.03), vancomycin (P = 0.01), and co-amoxiclav (P = 0.01) versus control as well as between vancomycin and amoxicillin (P = 0.03) and co-amoxiclav versus amoxicillin (P = 0.01), due to a regrowth in the latter group from 2 h on. The same growth patterns were observed in the subpopulation, with regrowth in the amoxicillin group from 2 h on. Significant differences were found in the initial inoculum reduction rate over time with respect to the resistant subpopulation between amoxicillin and vancomycin (P = 0.03). As in the total population, significant differences were also found between antibiotic groups and control (P ≤ 0.003), where bacterial counts rose from 6.6 × 10³ CFU/ml at time zero to 4 × 10⁵ CFU/ml at 8 h. The resistant subpopulation counts were below the limit of detection in the vancomycin and co-amoxiclav groups from 3 h on.

TABLE 2.

Viability rates (percent decreases) of β-lactamase-positive S. aureus at the different incubation times

Incubation period (h)	Mean viability rate (%) ± SD
	Control	Vancomycin	Amoxicillin	Amoxicillinclavulanic acid
0–0.5	−287 ± 47.32	77.53 ± 15.82	57.64 ± 7.76	50.03 ± 18.48
0.5–2	−572 ± 146.4	97.28 ± 0.19	87.18 ± 1.20	93.11 ± 0.48
2–3	−1,877 ± 255.2	97.92 ± 0.23	79.79 ± 2.47	97.15 ± 0.75
3–6	−74,000 ± 8,943	99.72 ± 0.09	−462 ± 12.95	99.49 ± 0.09
6–8	−120,000 ± 6,851	99.81 ± 0.03	−761 ± 100.7	99.80 ± 0.04

TABLE 3.

Viability rates (percent decreases) of β-lactamase-positive MRSA subpopulation at the different incubation times

Incubation period (h)	Mean viability rate (%) ± SD
	Control	Vancomycin	Amoxicillin	Co-amoxiclav
0–0.5	−505 ± 184.7	73.40 ± 4.88	52.68 ± 7.01	45.73 ± 4.99
0.5–2	−847 ± 260.7	89.58 ± 0.93	77.08 ± 6.13	92.64 ± 0.26
2–3	−1,346 ± 309.1	96.39 ± 1.57	−79.6 ± 14.59	98.12 ± 0.51
3–6	−3,100 ± 805.1		−1,358 ± 233.6
6–8	−6,068 ± 1,815		−4,774 ± 1,298

Tables 4 and 5 show the viability rates at the different incubation times for the total population and the resistant subpopulation of the β-lactamase-negative strain, respectively. Both populations showed similar patterns of reduction of the initial inoculum with significant differences with respect to control (P = 0.01 for the total population and any of the three antibiotics and P = 0.0001 for the resistant subpopulation and any of the three antibiotic groups). In the control group, the counts of the resistant subpopulation increased from 2.8 × 10⁴ to 2 × 10⁶ CFU/ml. No differences were found between the different antibiotic groups with respect to the total population reduction rates over time, but differences became statistically significant in comparing the change over time of the resistant subpopulation inocula between vancomycin and amoxicillin (P = 0.004) or co-amoxiclav (P = 0.007) and between co-amoxiclav and amoxicillin (P = 0.0003). In the amoxicillin and vancomycin groups, the colony counts of this subpopulation were under the limit of detection from 3 h on, whereas in the co-amoxiclav group this occurred from 2 h on.

TABLE 4.

Viability rates (percent decreases) of β-lactamase-negative S. aureus at the different incubation times

Incubation period (h)	Mean viability rate (%) ± SD
	Control	Vancomycin	Amoxicillin	Co-amoxiclav
0–0.5	−273 ± 22.59	82.99 ± 2.71	64.18 ± 2.71	52.24 ± 26.84
0.5–2	−1,726 ± 107.9	97.18 ± 0.45	96.57 ± 0.27	93.85 ± 3.13
2–3	−2,286 ± 117.3	97.80 ± 0.29	98.47 ± 1.51	98.72 ± 0.34
3–6	−13,000 ± 1,098	99.66 ± 0.08	99.78 ± 0.07	99.78 ± 0.06
6–8	−13,000 ± 1,439	99.98 ± 0.00	99.98 ± 0.02	99.94 ± 0.02

TABLE 5.

Viability rates (percent decreases) of β-lactamase-negative MRSA subpopulation at the different incubation times

Incubation period (h)	Mean viability rate (%) ± SD
	Control	Vancomycin	Amoxicillin	Co-amoxiclav
0–0.5	−1.68 ± 41.45	95.20 ± 0.13	59.37 ± 7.96	96.68 ± 0.23
0.5–2	−149 ± 124.1	98.01 ± 0.11	89.40 ± 13.61	99.35 ± 0.19
2–3	−408 ± 57.03	99.44 ± 0.18	99.48 ± 0.10
3–6	−702 ± 111.3
6–8	−5,522 ± 854.1

Comparative β-lactamase activities (means ± standard deviations) for the four groups at the different incubation times are shown in Table 6. From 0.5 h on, all antibiotic groups presented statistically significant differences versus control (P < 0.05). From 2 h on, a significant difference was found between co-amoxiclav or vancomycin and amoxicillin; there were no statistically significant differences between co-amoxiclav and vancomycin.

TABLE 6.

β-Lactamase activities of the penicillinase-producing MRSA at different incubation times

Incubation period (h)	β-Lactamase activity (mean A482 ± SD)
	Control	Vancomycin	Amoxicillin	Co-amoxiclav
0–0.5	0.154 ± 0.005	0.143 ± 0.004	0.146 ± 0.002	0.133 ± 0.009a,b
0.5–2	0.178 ± 0.005	0.139 ± 0.007a	0.134 ± 0.008a	0.119 ± 0.004a,c
2–3	0.203 ± 0.005	0.132 ± 0.007a	0.160 ± 0.009a,d	0.118 ± 0.008a,b
3–6	0.307 ± 0.006	0.116 ± 0.008a	0.178 ± 0.001a,d	0.114 ± 0.002a,b
6–8	0.434 ± 0.006	0.106 ± 0.004a	0.301 ± 0.011a,d	0.109 ± 0.001a,b

^aStatistically significant versus control. 

^bStatistically significant (co-amoxiclav versus amoxicillin). 

^cStatistically significant (co-amoxiclav versus vancomycin). 

^dStatistically significant (amoxicillin versus vancomycin). 

DISCUSSION

Vancomycin is the treatment of choice for infections due to MRSA, but resistance could emerge (13), and so alternatives to vancomycin should be explored (22).

The aim of the experiment was to study the effect of concentrations simulating those after an intravenous dose of 500 mg of vancomycin or 2,000/200 mg (amoxicillin and clavulanic acid, respectively) of co-amoxiclav on the viability of the methicillin-resistant subpopulation. Dosing intervals for these doses (8 h for amoxicillin or co-amoxiclav and 6 h for vancomycin) fit the experimental time (8 h to include both dosing intervals). These doses fulfil the following two conditions: (i) peak concentration/MIC ratio of ≥10 and (ii) time over MIC of ≥75% (6 h) of dosing interval (when the MIC is determined with standard conditions: MH broth and 37°C incubation). These are important pharmacodynamic facts because the two antibiotics tested exhibit a time-dependent action. In the case of amoxicillin, levels are over the MIC during 6 h (8 h for vancomycin); theoretically, the rest of the experiment time (dosing interval) would be covered due to the postantibiotic effect on susceptible S. aureus (7). To detect the effect of the antibiotics, with these achieved pharmacokinetic-in vitro susceptibility values, on the resistant subpopulation viability, the medium osmolarity and the temperature of incubation were changed in order to allow the expression of PBP 2A. A high inoculum (approximately 10⁷ CFU/ml) was used in order to obtain measurable numbers of CFU of the resistant subpopulation, despite the inoculum effect having been described for aminopenicillins against S. aureus but not for vancomycin (2); the effect might influence the results with the total population due to the favorable conditions for the glycopeptide. On the other hand, allowing PBP 2A expression through osmolarity and temperature of incubation increases the MIC of co-amoxiclav for both strains from 1 to 16 μg/ml but not the MIC of vancomycin, which remains 1 μg/ml.

With respect to the total population, with the β-lactamase-negative strain, no differences were found between antibiotic groups due to the pharmacodynamic conditions being fulfilled and the standard MICs being the same. Differences in MICs became relevant when, with the expression of gene mecA, only vancomycin maintained its value. Due to this, as expected, significant differences in the reduction of the resistant subpopulation initial inoculum over time were found between vancomycin and amoxicillin, because of the higher and earlier killing with vancomycin (59 versus 95%) in the first incubation period. Interestingly, when clavulanic acid is added, amoxicillin acquires an initial killing capability similar to that of vancomycin (97 versus 95%). This increase in earlier killing by clavulanic acid is also extended to the time needed to reduce the initial inoculum 100-fold (9) in the resistant subpopulation: 2 h for co-amoxiclav and 3 h for vancomycin.

In evaluating the effect on the total population, with respect to the β-lactamase-positive strain, significant differences were found between amoxicillin and vancomycin or co-amoxiclav, because of the regrowth obtained in the amoxicillin group due to the regrowth of the resistant subpopulation. No differences were found between vancomycin and co-amoxiclav in considering either the total population or the resistant subpopulation, with 100-fold reduction in the initial inoculum being obtained after 3 h in all cases. β-Lactamase activity over time changed in parallel with viability, increasing with the increase of viability (amoxicillin or control) and decreasing with the decrease of viability in the co-amoxiclav and vancomycin group, without statistically significant differences between them.

With both strains, amoxicillin or co-amoxiclav concentrations from 6 h on are subinhibitory despite a maintenance of the more than 99% reduction of the initial inoculum in both groups except in the amoxicillin group and with the β-lactamase-producing strain. This postantibiotic effect, detected and maintained due to a post-β-lactamase inhibitory effect in the co-amoxiclav group with the β-lactamase-producing strain, has been previously described for methicillin-susceptible S. aureus (1, 17).

These findings on the effects of concentrations achievable in serum may contribute to explaining the efficacy in vivo of the combination aminopenicillin–β-lactamase inhibitor in infections caused by MRSA, both in animal models (4, 10, 18, 27) and in the clinical setting (11, 20). Despite vancomycin remaining the drug of choice for MRSA infections, the increase of vancomycin resistance and the Centers for Disease Control and Prevention recommendation (14) for prudent use of vancomycin have encouraged the search for alternatives. Aminopenicillin–β-lactamase inhibitor may be one of them, adding to its wide spectrum this activity, which may be considered for empirical treatment in the clinical setting, if further in vitro studies and animal models confirm these results with other MRSA strains.