In Vitro Synergistic Effects of Double and Triple Combinations of β-Lactams, Vancomycin, and Netilmicin against Methicillin-Resistant Staphylococcus aureus Strains

Abstract

Several studies have previously reported synergistic effects between vancomycin and a given β-lactam or a given aminoglycoside against methicillin-resistant Staphylococcus aureus (MRSA) strains. The aim of our study was to exhaustively compare the effects of different combinations of a β-lactam, vancomycin, and/or an aminoglycoside against 32 clinical MRSA strains with different aminoglycoside susceptibility patterns. The effects of 26 different β-lactam–vancomycin and 8 different aminoglycoside-vancomycin combinations were first studied using a disk diffusion screening method. The best interactions with vancomycin were observed with either imipenem, cefazolin, or netilmicin. By checkerboard studies, imipenem-vancomycin and cefazolin-vancomycin each provided a synergistic bacteriostatic effect against 22 strains; the mean fractional inhibitory concentration (FIC) indexes were 0.35 and 0.46 for imipenem-vancomycin and cefazolin-vancomycin, respectively. The vancomycin-netilmicin combination provided an indifferent effect against all of the 32 strains tested; the mean of FIC index was 1.096. The mean concentrations of imipenem, cefazolin, netilmicin, and vancomycin at which FIC indexes were calculated were clinically achievable. Killing experiments were then performed using imipenem, cefazolin, netilmicin, and vancomycin at one-half of the MIC, alone and in different combinations, against 10 strains. The vancomycin-netilmicin regimen was rarely bactericidal, even against strains susceptible to netilmicin. The imipenem-vancomycin and cefazolin-vancomycin combinations were strongly bactericidal against six and five strains, respectively. The addition of netilmicin markedly enhanced the killing activity of the combination of cefazolin or imipenem plus vancomycin, but only for the MRSA strains against which the β-lactam–vancomycin combinations had no bactericidal effect. It is noteworthy that the latter strains were both susceptible to netilmicin and heterogeneously resistant to methicillin.

Methicillin-resistant Staphylococcus aureus (MRSA) strains represent a worldwide threat because of their virulence and their broad distribution in the hospital setting. Moreover, the MRSA strains are often resistant not only to β-lactam agents but also to fluoroquinolones, chloramphenicol, clindamycin, tetracyclines, and aminoglycosides (20). Vancomycin is almost universally accepted as the drug of choice for the treatment of MRSA infections (13, 20). However, vancomycin used alone kills staphylococci slowly (1), resulting in delayed recovery of patients with life-threatening infections (19, 33). In addition, clinicians now have to face the emergence of strains with reduced susceptibility to vancomycin, i.e., so-called glycopeptide-intermediate S. aureus (GISA) (4, 5, 15, 16). Therefore, there is clearly a need for new antibiotic regimens with strong early bactericidal activity against MRSA. In this field, an alternative to the development of new classes of agents could be the use of combinations of well-known compounds. Some investigators have reported synergistic bacteriostatic and bactericidal (where tested) effects of different β-lactams or different aminoglycosides combined with vancomycin against MRSA strains (3, 9, 11, 24, 28, 29, 30, 36–38), including some GISA isolates (8, 32). However, such findings have usually been observed either for very few of the antibiotic combinations (sometimes only one) tested or for a few strains and thus deserve to be confirmed in larger studies.

The purposes of this investigation were (i) to determine among 26 commonly available β-lactams and among 8 aminoglycosides those that display the strongest beneficial bacteriostatic effect in combination with vancomycin against MRSA strains; (ii) to evaluate the bactericidal effects displayed by the most effective agents used in different β-lactam–vancomycin, vancomycin-aminoglycoside and β-lactam–vancomycin–aminoglycoside combinations; and (iii) to examine whether the effects of the different antibiotic combinations are dependent upon the aminoglycoside susceptibility pattern of the strains.

(Part of this work was presented at the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy [abstr. E58, p. 185], San Diego, Calif., September 1998.)

MATERIALS AND METHODS

Bacterial strains.

A total of 32 clinical MRSA strains selected from 32 individual patients attending the Rouen University Hospital between 1995 and 1997 were studied. Twenty-eight isolates were obtained from blood, two were from pleural fluid, one was from pericardial fluid, and one was from joint fluid. They were identified to the species level by conventional methods (colony morphology, Gram stain characteristics, coagulase reactions). All strains were methicillin resistant, as determined by a disk diffusion method with a 5-μg oxacillin disk (10) and by PCR amplification of the mecA gene (2). Strains were considered heterogeneously resistant to methicillin when partial growth within the inhibition zone or the presence of microcolonies around the oxacillin disk was observed. None of the 32 strains had reduced susceptibility to glycopeptides. The study population was chosen according to the strains' patterns of susceptibility to aminoglycosides as determined by a disk diffusion technique described by the Comité de l'Antibiogramme de la Société Française de Microbiologie (10). The aminoglycoside resistance mechanism was determined after applying the aminoglycoside resistance pattern to 12 aminoglycosides by a disk susceptibility test described by Miller et al. (21). Seventeen strains seemed to produce a bifunctional [AAC(6′) + APH(2")] enzyme associated with ANT(4′) (4 strains) or with APH(3′) (13 strains), and 10 seemed to produce the ANT(4′) enzyme alone. The production of the two enzyme combinations cited above notably confers resistance to kanamycin, tobramycin, and gentamicin (i.e., a Km^r Tm^r Gm^r phenotype), whereas production of the ANT(4′) enzyme alone determines resistance to kanamycin and tobramycin (i.e., a Km^r Tm^r phenotype) (31). The five remaining strains were susceptible to aminoglycosides.

Strain ATCC 43300 was used as a reference control strain. Strains were stored frozen in glycerol broth at −70°C and subcultured to ensure purity before testing.

Media and antibiotics.

Mueller-Hinton broth and agar (Becton-Dickinson, Le Pont-de-Chaix, France) and tryptic soy (TS) agar supplemented or not supplemented with blood (Bio-Rad, Marnes-la-Coquette, France) were used. All incubations were at 37°C for 24 h. The following antibiotics were kindly provided by the manufacturers: imipenem (Merck Sharp & Dohme, Paris, France), cefazolin (Panpharma S.A., Fougères, France), vancomycin (Eli Lilly & Co., Saint-Cloud, France), and netilmicin (Schering Plough, Levallois-Perret, France).

Screening for the β-lactam and for the aminoglycoside showing the best beneficial effect in combination with vancomycin.

A one-disk diffusion technique previously described (27) was used to assess the effects of β-lactam–vancomycin and vancomycin-aminoglycoside combinations against all of the 32 MRSA strains tested. Mueller-Hinton agar plates with or without vancomycin at one-fourth of the MIC were flooded with a bacterial suspension of 1.5 × 10⁶ CFU/ml. Twenty-six disks each impregnated with a β-lactam (Bio-Rad) and 8 disks impregnated with an aminoglycoside were placed on top of the agar plates. For each strain, the inhibition zone around the disks in the absence and in the presence of vancomycin in agar plates was compared as described below. An initial score (S1) was established according to the inhibition zone on the plate without vancomycin to take into account the original susceptibility of the strain to the antibiotic tested. This inhibition zone was interpreted as described by the Comité de l'Antibiogramme de la Société Française de Microbiologie (10). S1 was 0 when the strain was resistant to the antibiotic tested, 1 when it was intermediate, and 2 when it was susceptible. A second score (S2) was determined by comparison of inhibition zones in the presence or absence of vancomycin; S2 was 0 in the absence of an increase in the diameter of the inhibition zone, 1 when the zone was increased but the strain remained in the same category, 2 when the strain changed by one category (i.e., moving from resistant to intermediate [R→I] or from intermediate to susceptible [I→S]), and 3 when the zone was increased so that the strain changed by two categories (R→S). Lastly, depending upon the addition of these two scores (S3), the beneficial bacteriostatic effect was defined as absent (S3 = 0), weak (S3 = 1), moderate (S3 = 2), or strong (S3 = 3).

Because of the favorable results described below, cefazolin, imipenem, and netilmicin were chosen for the subsequent studies.

Susceptibility testing and checkerboard studies.

The MICs of cefazolin, imipenem, netilmicin, and vancomycin were determined by the agar dilution method in accordance with standard guidelines (10, 23). The replicator prong delivered approximately 10⁴ CFU per spot (34). The MICs were interpreted in accordance with the recommendations of the Comité de l'Antibiogramme de la Société Française de Microbiologie (10).

Studies of activities of two-drug combinations comprising vancomycin and either a β-lactam (cefazolin or imipenem) or netilmicin were performed by the checkerboard agar dilution method (12) to obtain a fractional inhibitory concentration (FIC) index. The media, inocula, and conditions were the same as those used for MIC tests. Effects were interpreted as synergistic when FIC indexes were ≤0.5, indifferent when values were >0.5 to 4.0, and antagonistic when values were >4.0 (12).

Killing studies.

Killing experiments were performed to evaluate the bactericidal activities of two-drug and three-drug combinations of antibiotics, including a β-lactam (imipenem or cefazolin) and/or netilmicin and vancomycin, against 10 of the 32 MRSA strains studied. The 10 strains tested were chosen according to their aminoglycoside susceptibility patterns and consisted of 4 strains with the Km^r Tm^r Gm^r phenotype, 4 strains with the Km^r Tm^r phenotype, and 2 strains susceptible to aminoglycosides.

Each antibiotic was tested alone and in combination at one-half of the MIC. The final concentration of the log-phase inocula was approximately 10⁶ CFU/ml. All experiments were performed in duplicate on different days to ensure reproducibility. Viability counts were done at 24 h after incubation at 37°C by plating 25 μl of serial dilutions from each tube onto TS blood agar plates (12). The numbers of viable bacteria were counted after 24 and 48 h of incubation of the plates at 37°C. The detection limit was 40 CFU/ml. In preliminary experiments, antibiotic carryover was ruled out by plating samples of a bacterial suspension containing 10³ to 10⁴ CFU/ml in the presence or absence of antibiotics alone or in combination (12, 25). Bactericidal activity was defined as a ≥3-log₁₀ decrease in the starting inoculum after 24 h of incubation. Synergy was defined as a ≥2-log₁₀ decrease in the number of CFU per milliliter between the combination and its most active single component after a 24-h incubation period, and the number of surviving organisms in the presence of the combination had to be ≥2 log₁₀ CFU/ml below the starting inoculum (12). Antagonism was defined as a ≥2-log₁₀ increase in the number of CFU per milliliter between the combination and the most active single antimicrobial agent. Indifference was defined as an increase or a decrease in killing of less than 100-fold at 24 h with the combination compared with the most active agent used alone.

RESULTS

Screening for the β-lactam and the aminoglycoside showing the strongest beneficial bacteriostatic effect in combination with vancomycin.

As shown in Table 1, the beneficial bacteriostatic effects of β-lactam–vancomycin combinations were variable, depending on the β-lactam tested. The β-lactams which exhibited a strong beneficial effect against a high percentage of strains when combined with vancomycin were meropenem (66% of strains), ampicillin-sulbactam (62%), cefazolin (60%), and to a lesser extent imipenem, cefoxitin, and cefotiam (56% for each). Among these agents, ampicillin-sulbactam, cefazolin, and imipenem were those providing a moderate or strong beneficial effect against the most strains. Among these three drugs, cefazolin and imipenem, whose pharmacodynamic properties and therapeutic indications are markedly different, were both retained for the subsequent bacteriostatic and bactericidal studies.

TABLE 1.

Beneficial bacteriostatic effects of different β-lactam–vancomycin combinations against 32 MRSA strains

Beneficial bacteriostatic effecta	Activity of vancomycin in combination with indicated β-lactam (% of strains)b
	PEN	AMP	SAM	AMX	AMC	TIC	TIM	PIP	TZP	OXA	IPM	MEM	CEF	CFZ	FAM	CXM	FOX	CTT	CTF	CTX	CRO	CFP	CAZ	FEP	CPO	MOX
Absent	6	10	0	6	0	9	0	3	0	12	0	0	0	0	0	15	0	12	0	9	15	0	19	16	3	3
Weak	22	6	3	10	9	3	0	6	6	25	0	6	0	3	6	16	22	38	6	16	35	16	28	22	9	10
Moderate	59	34	35	47	44	35	47	41	44	19	44	28	47	37	41	34	22	50	38	37	25	31	50	31	44	56
Strong	13	50	62	37	47	53	53	50	50	44	56	66	53	60	53	35	56	0	56	38	25	53	3	31	44	31

^aSee the text for the definition of a beneficial bacteriostatic effect. 

^bActivity was tested by a disk diffusion technique. PEN, penicillin; AMP, ampicillin; SAM, ampicillin-sulbactam; AMX, amoxicillin; AMC, amoxicillin-clavulanic acid; TIC, ticarcillin; TIM, ticarcillin-clavulanic acid; PIP, piperacillin; TZP, piperacillin-tazobactam; OXA, oxacillin; IPM, imipenem; MEM, meropenem; CEF, cephalothin; CFZ, cefazolin; FAM, cefamandole; CXM, cefuroxime; FOX, cefoxitin; CTT, cefotetan; CTF, cefotiam; CTX, cefotaxime; CRO, ceftriaxone; CFP, cefoperazone; CAZ, ceftazidime; FEP, cefepime; CPO, cefpirome; MOX, latamoxef. 

As shown in Table 2, for most of the 32 strains, no or a weak beneficial bacteriostatic effect was demonstrated with the vancomycin-aminoglycoside combinations, except with netilmicin, amikacin, dibekacin, gentamicin, and streptomycin, which frequently displayed a moderate effect. Netilmicin was considered the aminoglycoside showing the best effect, as it was the one providing a strong beneficial effect against the most strains. It is noteworthy that of the 88% of the strains tested against which the vancomycin-netilmicin combination produced a strong or moderate beneficial bacteriostatic effect, 46% harbored the Km^r Tm^r Gm^r phenotype.

TABLE 2.

Beneficial bacteriostatic effects of different aminoglycoside-vancomycin combinations against 32 MRSA strains

Beneficial bacteriostatic effecta	Activity of vancomycin in combination with indicated aminoglycoside (% of strains)b
	Streptomycin	Kanamycin	Tobramycin	Dibekacin	Amikacin	Isepamicin	Gentamicin	Netilmicin
Absent	53	84	81	50	12	12	50	0
Weak	0	0	0	0	22	32	0	12
Moderate	41	13	13	44	44	37	44	56
Strong	6	3	6	6	22	19	6	32

^aSee the text for a definition of a beneficial bacteriostatic effect. 

^bActivity was tested by a disk diffusion technique. 

MIC determinations and checkerboard studies.

The characteristics of the 32 MRSA strains tested are shown in Table 3. The MICs of the antibiotics studied for 90% of the 32 MRSA strains tested were as follows: cefazolin, 256 μg/ml; imipenem, 64 μg/ml; netilmicin, 16 μg/ml; vancomycin, 4 μg/ml. On the basis of the MICs, 15 strains appeared susceptible to cefazolin and 17 appeared susceptible to imipenem. Seventeen of the 32 MRSA strains were susceptible to netilmicin, 9 were intermediate, and 6 were resistant at a low level (MIC, 16 or 32 mg/liter). All of the strains were susceptible to vancomycin.

TABLE 3.

Descriptions of the 32 strains studied

Strain no.	Aminoglycoside resistance phenotypea	Type of methicillin resistanceb	MIC (μg/ml)c				FIC index
			IPM	CFZ	VAN	NET	IPM-VAN	CFZ-VAN	VAN-NET
1	Kmr Tmr Gmr	HOM	64	256	2	32	0.133	0.254	1.031
2	Kmr Tmr Gmr	HOM	64	256	2	16	0.133	0.5	1.031
3	Kmr Tmr Gmr	HOM	64	256	4	8	0.133	0.266	1.5
4	Kmr Tmr Gmr	HOM	64	256	4	8	0.141	0.254	1.5
5	Kmr Tmr Gmr	HOM	64	256	2	8	0.141	0.5	1.25
6	Kmr Tmr Gmr	HOM	64	256	2	8	0.251	0.266	1.5
7	Kmr Tmr Gmr	HOM	32	256	2	8	0.127	0.5	1.031
8	Kmr Tmr Gmr	HOM	32	256	2	8	0.141	0.266	1.25
9	Kmr Tmr Gmr	HOM	32	256	2	8	0.141	0.266	1.5
10	Kmr Tmr Gmr	HOM	32	256	4	8	0.252	0.28	1.031
11	Kmr Tmr Gmr	HOM	32	256	2	4	0.126	0.375	1
12	Kmr Tmr Gmr	HOM	32	128	4	16	0.252	0.266	1.031
13	Kmr Tmr Gmr	HOM	32	128	4	16	0.252	0.266	1.031
14	Kmr Tmr Gmr	HOM	32	128	4	16	0.141	0.266	1.031
15	Kmr Tmr Gmr	HOM	32	128	2	8	0.141	0.508	1.031
16	Kmr Tmr Gmr	HET	2	1	2	16	0.25	0.375	1.031
17	Kmr Tmr Gmr	HET	0.06	1	1	2	1	1	1
18	Kmr Tmr	HOM	2	64	1	0.5	0.156	0.312	1.062
19	Kmr Tmr	HOM	2	64	1	0.25	0.156	0.375	1.062
20	Kmr Tmr	HET	0.125	8	1	0.5	0.63	0.75	1.062
21	Kmr Tmr	HET	0.125	1	2	0.25	0.373	0.625	1.062
22	Kmr Tmr	HET	0.125	1	1	0.25	0.625	0.75	1.062
23	Kmr Tmr	HET	0.06	8	1	0.25	0.76	0.5	1.062
24	Kmr Tmr	HET	0.06	8	1	0.25	0.61	0.5	1.062
25	Kmr Tmr	HET	0.06	1	1	0.25	0.562	0.75	1
26	Kmr Tmr	HET	0.06	1	1	0.25	0.76	0.75	1
27	Kmr Tmr	HET	0.06	1	1	0.25	0.566	0.75	1
28	S	HOM	0.125	8	1	0.5	0.25	0.312	1
29	S	HOM	0.125	4	2	0.25	0.25	0.312	1.062
30	S	HET	0.125	1	1	0.5	0.625	0.625	0.75
31	S	HET	0.06	4	2	1	0.25	0.5	1.062
32	S	HOM	0.06	1	1	0.25	0.61	0.625	1

^aSee the text for a definition of the aminoglycoside resistance phenotype. 

^bHOM, strain homogeneously resistant to methicillin; HET, strain heterogeneously resistant to methicillin. 

^cIPM, imipenem; CFZ, cefazolin; VAN, vancomycin; NET, netilmicin. S, susceptible to aminoglycosides. 

The results of checkerboard studies of the cefazolin-vancomycin, imipenem-vancomycin, and vancomycin-netilmicin combinations are indicated in Tables 3 and 4. The effects of the cefazolin-vancomycin and imipenem-vancomycin combinations were indifferent (with FIC indexes between 0.5 and 1) against 31% (10 isolates) and synergistic against 69% (22 isolates) of the MRSA strains. For 13 of these 22 isolates, FIC indexes of <0.25 were seen with the imipenem-vancomycin combination while no such low FIC indexes were calculated for the cefazolin-vancomycin combination. Antagonistic FIC values for both combinations were not observed with any of the strains. The FIC indexes of the imipenem-vancomycin combination were inversely correlated with the MICs of imipenem, i.e., most cases of FIC indexes of ≤0.25 occurred with strains for which the imipenem MICs were ≥32 μg/ml. The mean antibiotic concentrations at which FIC indexes were calculated for both combinations were as follows: 7.4 and 0.41 mg/liter for cefazolin and vancomycin, respectively, and 0.19 and 0.24 mg/liter for imipenem and vancomycin, respectively. In contrast, the vancomycin-netilmicin combination exhibited indifferent effects against all of the 32 MRSA strains studied. FIC indexes of 1.0 to 1.5 were observed with all strains but one. The mean antibiotic concentrations at which FIC indexes were calculated were 0.3 mg/liter for vancomycin and 3.6 mg/liter for netilmicin.

TABLE 4.

FIC indexes of β-lactam–vancomycin and vancomycin-netilmicin combinations for 32 MRSA strains

Antibiotic combinationa	Mean FIC index	No. of strains
		Synergy (FIC ≤ 0.5)	Indifference (0.5 < FIC ≤ 4)	Antagonism (FIC > 4)
CFZ-VAN	0.46	22	10	0
IPM-VAN	0.35	22	10	0
VAN-NET	1.10	0	32	0

^aCFZ, cefazolin; VAN, vancomycin; IPM, imipenem; NET, netilmicin. 

Killing studies.

The rates of killing of the 10 strains by antibiotics used alone or in double or triple combinations, determined at one-half of the MIC, are presented in Table 5. At this concentration, none of the monotherapies tested produced bactericidal effects after 24 h, except vancomycin against strain 4. It is noteworthy that vancomycin alone produced a >2-log₁₀ reduction in the counts of three additional strains. Regimens of netilmicin combined with either vancomycin or a β-lactam (cefazolin or imipenem) were infrequently or never bactericidal. In contrast, regimens consisting of imipenem or cefazolin plus vancomycin were bactericidal after 24 h against 6 and 5 of the 10 strains studied, respectively, and were often synergistic. Both β-lactam–vancomycin combinations were bactericidal against all of the strains harboring the Km^r Tm^r Gm^r phenotype, while only the imipenem-vancomycin combination produced a bactericidal effect against one strain with the Km^r Tm^r phenotype. It is noteworthy that all cases but one of the absence of a bactericidal effect of the β-lactam–vancomycin regimen occured with strains for which FIC indexes of ≥0.5 was calculated. The addition of netilmicin markedly enhanced the killing activity of the combinations of cefazolin or imipenem plus vancomycin, but only for three of the four MRSA strains against which the β-lactam–vancomycin combinations failed to produce a bactericidal effect. It is noteworthy that these strains (no. 21, 22, 27, and 30) were both susceptible to netilmicin and heterogeneously resistant to methicillin. As a result, the three-drug combinations each demonstrated a bactericidal effect against 8 of the 10 strains studied. The imipenem-vancomycin-netilmicin combination produced a greater bacterial count reduction and was more frequently synergistic than the cefazolin-vancomycin-netilmicin regimen.

TABLE 5.

Killing activities of imipenem, cefazolin, vancomycin, and netilmicin alone and in double or triple combinations against 10 MRSA strains after 24 h of incubation

Strain no.	Change in viable count (log10 CFU/ml) after 24 h (effect of antibiotic combination)a
	Control	IPM	CFZ	VAN	NET	IPM + VAN	CFZ + VAN	VAN + NET	IPM + NET	CFZ + NET	IPM + VAN + NET	CFZ + VAN + NET
3	+2.34	+1.87	+2.10	+2.22	+2.11	−5.92 (SB)	−3.84 (SB)	−0.75 (−b)	−0.78 (−b)	−1.81 (−b)	−4.32 (SB)	−4.02 (SB)
4	+2.09	+2.36	+2.14	−3.87	+1.96	−4.25 (−B)	−3.95 (−B)	−2.70 (−b)	+1.40 (—)	−0.14 (−b)	−4.25 (−B)	−3.55 (−B)
7	+2.48	+2.13	+1.97	+2.40	+2.09	−5.93 (SB)	−5.93 (SB)	−2.46 (Sb)	+1.85 (—)	+1.61 (—)	−5.93 (SB)	−4.03 (SB)
12	+2.41	+1.60	+2.23	−2.91	+2.19	−5.95 (SB)	−5.95 (SB)	−2.11 (−b)	−0.07 (−b)	−0.72 (Sb)	−5.95 (SB)	−3.57 (−B)
21	+1.87	+1.57	+1.50	−2.27	+1.62	−4.03 (−B)	−2.87 (−b)	−4.81 (SB)	−0.85 (−b)	+1.50 (—)	−5.42 (SB)	−3.82 (−B)
22	+1.88	+1.46	+1.45	+1.56	+1.27	−2.28 (Sb)	+1.13 (—)	+0.85 (—)	−0.74 (−b)	+1.02 (—)	−6.61 (SB)	−6.61 (SB)
24	+1.97	+1.34	+1.66	+1.47	+1.81	−2.13 (Sb)	−0.85 (−b)	+1.25 (—)	−0.13 (−b)	−0.89 (−b)	−2.70 (Sb)	−1.94 (−b)
27	+2.92	+2.32	+1.75	+2.57	+2.54	−2.39 (Sb)	−1.68 (−b)	−0.56 (−b)	+1.54 (—)	+0.35 (—)	−5.66 (SB)	−5.66 (SB)
30	+2.13	+1.80	+1.92	+2.16	+1.81	+1.67 (—)	+1.79 (—)	+1.70 (—)	−1.70 (−b)	+0.57 (—)	−2.85 (Sb)	−1.66 (−b)
31	+2.88	+0.86	+1.78	−2.23	+2.92	−5.47 (SB)	−5.47 (SB)	−3.87 (−B)	−3.57 (SB)	−2.33 (Sb)	−5.47 (SB)	−3.87 (−B)

^aIPM, imipenem; CFZ, cefazolin; VAN, vancomycin; NET, netilmicin. S, synergy; −B or −b, no synergy; b, bacteriostatic effect; B, bactericidal effect; —, absence of bacteriostatic or synergistic effect. 

DISCUSSION

In this study, we analyzed the activities of 8 aminoglycoside-vancomycin and 26 β-lactam–vancomycin combinations against a set of 32 clinical MRSA strains. This represents 1,088 individual analyses, a work which would have been tremendously laborious with conventional methods for the study of antibiotic combinations such as checkerboard synergy testing. To circumvent such technical problems, we used a previously described disk diffusion method (27) that is easy to perform, reproducible, and easier to interpret than the double-disk potentiation methods (8). Compared with our original description (27), the calculation of the final score that analyzes the beneficial bacteriostatic effect observed was slightly modified to take into account the intrinsic β-lactam activity against MRSA isolates.

Among the eight aminoglycosides tested, netilmicin was considered the best agent to combine with vancomycin, frequently showing a moderate or strong beneficial bacteriostatic effect, particularly against strains with a Km^r Tm^r phenotype and, interestingly, even against those with a Km^r Tm^r Gm^r phenotype (data not shown), despite the cross-resistance between gentamicin and netilmicin.

The screening test also revealed that most β-lactam–vancomycin combinations exhibited a beneficial effect against MRSA strains and that the β-lactams most often demonstrating a moderate or strong beneficial effect were ampicillin-sulbactam, imipenem, and cefazolin. This may be related to the binding affinities of these three β-lactams for penicillin-binding protein 2a of MRSA, which are stronger than that of methicillin and allow the classification of these three agents among the β-lactam antibiotics most active against MRSA in vitro (6, 7). Using a similar screening test, Climo et al. (8) recently reported a synergistic bacteriostatic effect of a combination of vancomycin and either oxacillin, ceftriaxone, ceftazidime, cefpodoxime, or amoxicillin-clavulanate against three clinical GISA isolates.

The favorable bacteriostatic effect of the imipenem- or cefazolin-vancomycin combinations was also demonstrated by the checkerboard studies. The concentrations at which synergism was observed were easily clinically achievable with both combinations. It is noteworthy that the imipenem-vancomycin regimen produced the best results, in terms of both FIC values and the antibiotic concentrations generating bacteriostatic synergy. By checkerboard studies, synergistic bacteriostatic effects of β-lactam–vancomycin combinations against MRSA have been previously reported for imipenem and cephalosporins such as cefotiam, cefoperazone, and cefpirome (3, 9, 30, 35, 36), but not for oxacillin (8). However, unlike our results, the synergistic FIC values usually observed in such studies were frequently high, close to the 0.5 limit value for synergy (9, 30).

In contrast, the combination of vancomycin with netilmicin resulted in an indifferent effect, even against MRSA strains susceptible to netilmicin. Such results are in accordance with recent studies performed with gentamicin (17, 18) or netilmicin; D. Ince and H. Eraksoy, 8th Int. Congr. Infect. Dis., abstr. 12.038, p. 20, 1998).

The potential bactericidal effects of double or triple combinations including a β-lactam (imipenem or cefazolin) and/or vancomycin and/or netilmicin were further evaluated by killing experiments with 10 strains. The antibiotics were tested at one-half of the MICs to improve the detection of synergy (12) and also to reflect clinical conditions. As expected from the antibiotic concentrations used in these experiments, the monotherapies and the β-lactam–netilmicin combinations generally failed to produce bactericidal effects after 24 h. The bacterial count reductions sometimes observed in broth with vancomycin alone were maybe related to a lack of accuracy of the vancomycin MICs, which were determined by agar and not by broth dilution.

As expected from the high FIC indexes, the vancomycin-netilmicin combination was found to be poorly active against the MRSA strains studied, whatever their aminoglycoside susceptibility pattern. Previous studies examining the bactericidal activity of vancomycin-aminoglycoside combinations against MRSA most of the time tested gentamicin instead of netilmicin, but their authors globally reported similar indifferent results (22; Ince and Eraksoy, 8th Int. Congr. Infect. Dis.). Synergism was shown not to be predictable from the aminoglycoside MIC (22). On the other hand, some studies have reported that gentamicin (14, 17) or netilmicin (26) can enhance the bactericidal activity of vancomycin but these results were observed at inhibitory antibiotic concentrations and against only one to three strains.

The most interesting results of our study are the frequent and strong bactericidal effects of the β-lactam (cefazolin or imipenem)–vancomycin combination and, moreover, of the β-lactam–vancomycin–netilmicin combination. The results of our bitherapies compared favorably with those of previous works evaluating the killing activity of combinations of vancomycin and either cefotiam (35) or cefpirome (29) against MRSA: for the cefotiam-vancomycin combination, only a weak synergistic bacteriostatic effect was reported; for the cefpirome-vancomycin regimen, the bactericidal effect at 24 h was observed only against the two strains homogeneously resistant to methicillin. In a previous work performed with a single MRSA strain (36), the bactericidal effect of the imipenem-vancomycin combination has been reported to be dependent on the imipenem concentration.

To our knowledge, this is the first work studying the efficacy of triple combinations of a β-lactam, a glycopeptide, and an aminoglycoside against MRSA strains. For the strains both homogeneously resistant to β-lactams and resistant to netilmicin, the triple combinations had no advantage over the β-lactam–vancomycin combinations in terms of bactericidal effect. For four of the six strains both heterogeneously resistant to methicillin and susceptible to netilmicin, the β-lactam–vancomycin combinations failed to provide a bactericidal effect, probably because of the heterogeneous methicillin resistance. As expected for these strains, the addition of netilmicin markedly increased the killing effects of the β-lactam–vancomycin combinations, except for one strain. However, as all of the four Km^r Tm^r Gm^r strains tested were homogeneously resistant to methicillin and all of the four Km^r Tm^r strains were heterogeneously resistant, these results do not allow determination of what is significant in the phenotype of the strains to predict the bactericidal effects of the combinations.

In conclusion, this study shows that vancomycin combined with imipenem or cefazolin, and even with netilmicin in a triple combination (depending on the aminoglycoside susceptibility pattern of the strains), at subinhibitory concentrations can be bactericidal after 24 h at higher rates than vancomycin-netilmicin. Further studies are needed before generalizing the concept of the usefulness of adding a β-lactam such as imipenem or cefazolin, and potentially an aminoglycoside such as netilmicin, to vancomycin for the treatment of MRSA infections. Interestingly, the choice between cefazolin and imipenem could be based on in vitro tests but also on the clinical context: cefazolin in cases of proven monomicrobial infection due to MRSA and imipenem in cases of suspected or proven mixed infections. Lastly, the recent description of synergistic activities of different β-lactams and vancomycin against some GISA isolates (8, 32) indicates that further studies in this field are warranted.