Antimicrobial Susceptibility and Selection of Resistance

Transcription

1 ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Sept. 1978, p /78/14-353$2./ Copyright 1978 American Society for Microbiology Vol. 14, No. 3 Printed in U.S.A. Antimicrobial Susceptibility and Selection of Resistance Among Staphylococcus epidermidis Isolates Recovered from Patients with Infections of Indwelling Foreign Devices GORDON L. ARCHER Division of Infectious Diseases, Department ofinternal Medicine, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia Received for publication 31 March 1978 Twenty-seven isolates of Staphylococcus epidermidis from patients with prosthetic valve endocarditis or infected cerebrospinal fluid shunts were examined for susceptibility to antimicrobial agents. Subpopulations resistant to 2 and 1 ttg of methicillin per ml were present in 63% of the isolates (methicillin-resistant isolates). Subpopulations resistant to 2 jig of nafcillin and cephalothin per ml were found in every methicillin-resistant isolate but with frequencies (1-5 ±.5 and 1-64 *.9, respectively) which were not always detectable by susceptibility testing. Resistance to '1.6 jig of penicillin per ml was found in 8% of isolates. Cephalothin, cefazolin, and cefamandole were more active than cefoxitin or cephradine, and gentamicin was more active than tobramycin or amikacin; rifampin was the single most active agent against all isolates. There was no difference in susceptibility between prosthetic valve endocarditis and cerebrospinal fluid shunt infection isolates. Among methicillin-resistant isolates, the phenotypic expression of resistance to methicillin or nafcillin but not to cephalothin could be enhanced by 48 h of incubation with each drug. Isolates containing no methicillin-resistant subpopulations were killed by incubation with methicillin, nafcillin, or cephalothin. High-level resistance to rifampin emerged in both methicillin-resistant and methicillin-sensitive isolates after 8 to 24 h of incubation with this drug. The presence or absence of antibiotic-resistant subpopulations among S. epidermidis isolates and their selection during treatment should be considered when therapy is devised. Staphylococcus epidermidis is one of the most common organisms causing infections of indwelling artificial devices, particularly prosthetic heart valves and cerebrospinal fluid (CSF) shunts (5, 14). Infections of these devices are characteristically difficult to cure with antibiotics alone. Studies have suggested that many S. epidermidis isolates recovered from patients with these infections are resistant to one or more antistaphylococcal antibiotics (9, 16), but no attempt has been made to characterize resistance patterns among these isolates. Investigators who have reported the antimicrobial susceptibilities of large numbers of S. epidermidis isolates have tested organisms recovered from clinical laboratories; many of these isolates may have been contaminants from patient skin (3, 12). We have obtained 27 S. epidermidis isolates from 6 different geographical areas. These isolates were initially recovered on multiple occasions from blood, CSF, or tissue in patients with documented prosthetic valve endocarditis (PVE) or infected CSF shunts. We report the antimicrobial susceptibility and the selection of antibiotic resistance among these isolates. MATERIALS AND METHODS Microorganisms. S. epidermidis isolates from documented infections were obtained from the following institutions and investigators: Medical College of Virginia (G. Archer and R. J. Duma), University of Virginia School of Medicine (M. Sande), Massachusetts General Hospital and Harvard University School of Medicine (A. W. Karchmer), University of Michigan Medical Center (F. R. Fekety, Jr.), University of Alabama Medical School (W. Dismukes), and Stanford University School of Medicine (R. McLeod). Twenty of the isolates were from the blood or perivalvular tissue of patients with PVE, and seven were from the blood or CSF of patients with infections of CSF shunts. Organisms were identified as S. epidernidis by their ability to ferment glucose, their failure to coagulate rabbit plasma, and their inability to use mannitol either aerobically or anaerobically. Of the 27 isolates tested, 13 were typable by specific S. epidermidis typing phages (G. Parisi, University of Missouri School of Medicine). 353 Downloaded from on April 26, 218 by guest

2 354 NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNARCHER Antimicrobial susceptibility testing. For each isolate, minimum inhibitory concentrations (MICs) and miniimum bactericidal concentrations (MBCs) were determined in triplicate by the,*crotiter method with an inoculum size of 15 colony-f6oirminiimigiuts (CFU) per ml, using Mueller-Hinton broth (Baltimore Biological Laboratory [BBL], Cockeysville, Md.) as the diluent. Each MIC was read after both 24 and 48 h of incubation at 37C. Subculturine of wells to Mueller-Hinton agar (BBL) for determination of the MBC was accomplished bvthe use of a replicating ievice which delivered.15 ml on stil prongs. There was a 95% agreement between results obtained by using the replicator and a calibrated.1- ml wire loop. The MBC was read as the lowest dilution yielding no growth on subculture after 24 h of incubation. Susceptibility of the organisms to the following antimicrobial agents was determined: penicillin G potassium (Eli Lilly & Co., Indianapolis, Ind.), sodium nafcillin (Wyeth Laboratories, Philadelphia, Pa.), sodium methicillin (Bristol Laboratories, Syracuse, N.Y.), sodium oxacillin (Bristol), cephalothin sodium (Eli Lilly), cefazolin sodium (Eli Lilly), cefamandole lithium (Eli Lilly), sodium cefoxitin (Merck, Sharp, & Dohme, West Point, Pa.), cephradine (E. R. Squibb and Sons, Princeton, N.J.), vancomycin hydrochloride (Eli Lilly), gentamicin sulfate (Schering Laboratories, Bloomfield, N.J.), tobramycin sulfate (Eli Lilly), amikacin base (Bristol), streptomycin sulfate (Eli Lilly), rifampin (Dow Chemical Co. Indianapolis, Ind.), 7- chloro-7-deoxylincomycin (The Upjohn Co., Kalamazoo, Mich.), erythromycin (Eli Lilly), and sulfamethoxasole-trimethoprim (Burroughs Wellcome Co., Research Triangle Park, N.C.). The susceptibility of isolates to the combination of sulfamethoxasole and trimethoprim was determined in Mueller-Hinton broth supplemented with 9 U of thymidine phosphorylase per ml, kindly provided by S. R. M. Bushby, Burroughs Welcome Co. The MICs and MBCs were expressed as the total activity of the two drugs mixed in a ratio of 19 (sulfamethoxasole) to 1 (trimethoprim). The susceptibility of isolates to cephalothin and nafcillin was further tested in duplicate in plastic tubes (12 by 75 mm; Falcon Plastics, Oxnard, Calif.), using the same media as employed in microtiter testing and an inoculum size of 5 x 15 CFU/ml. Subculture of clear tubes to solid agar was done with a.1-ml wire loop; the MBC was read as that dilution yielding no growth. The MICs and MBCs of nafcillin, methicillin, and cephalothin for all isolates were also determined by microtiter dilution with incubation at,lotk3ands 37Cin brot. containing 5% NaCl. Determination of resistant subpopulations. Four or five colonies of each isolate were picked from agar and incubated overnight at 37 C in Mueller-Hinton broth with shaking. The number of CFU in the overnight cultures was determined by serial dilution resistance was maximized to 1i ANTIMICROB. AGENTS CHEMOTHER. in normal saline, with.1 ml from appropriate dilutions spread on the surface of Mueller-Hinton agar. Resistant isolates were selected by plating.1 ml from the same saline dilutions on agar containing antibiotics. Because overnight cultures yielded only 18 CFU/ml, the population of organisms sampled for CFU/ml by incorporating 1 ml of the overnight culture in pour plates containing antibiotics (2.5 ml in four plates containing 15 ml of agar The antibiotics andconcentrations usedin plates were as follows (pg/ml): nafcillin, methicillin, and cep, each at 2 and1 rifampin, 1;, vancomycin, 2, and gentamicin, 5. These concentrations were chosen to represent levels which would be difficult to attain or maintain in patients. Plate counts were determined after 72 h of incubation, and the number of resistant isolates was determined by the ratio, CFU on plates containing antibiotics/cfu on plates without antibiotics. The fraction was converted to the negative log to indicate the resistance frequency. Antibiotic resistance was further confirmed by picking four colonies from each plate containing antibiotics, incubating them in antibiotic-free broth, and determining the MIC and MBC for each colony of the antibiotic contained in the plate. Time-kill studies. One hundred-fold dilutions of overnight cultures of S. epidermidis were added to 1 ml of Mueller-Hinton broth in 25-ml flasks. The organisms were allowed to grow for 3 h to a concentration of approximately 17 CFU oflog-phase bacteria per ml, and antibiotics were added. Flasks were incubated at 37 C with shaking for 48 h. Samples were removed and serially diluted for viable count before (zero time) and at 2, 4, 8, 24, and 48 h after the addition of antibiotics. Duplicate samples were plated at zero time and at 24 and 48 h on plain agar and on agar containing the antibiotic added to the flask; the number of resistant isolates was determined as described above. The antibiotics added to the flasks were rifampin (1 isg/ml), cephalothin (2 Ag/ml), nafcillin (2 jig/ml), or methicillin (2 jag/ml). Time-kill curves were performed at least twice for each isolate tested. RESULTS Antimicrobial susceptibility. The cumulative percentage of S. epidermidis isolates inhibited (MIC) or killed (MBC) by various antimicrobial agents is shown in Table 1. For 85% of the isolates, the MIC was :1.6 yg of penicillin G per ml; for 78%, the MIC was ug of methicillin per ml; and for 7%, the MIC was -25,ug of streptomycin per ml. There was no difference in methicillin susceptibility between isolates from patients with PVE and CSF shunt infections, nor were there differences in methicillin susceptibility among isolates from different geographic areas. Only 19% of isolates were resistant to nafcillin (MIC isg/ml) when tested in microtiter. Repeat nafcillin susceptibility testing in tubes resulted in a fourfold or greater increase in MIC in four isolates; skipped tubes and highly resistant mutants were observed on MBC testing of each of these isolates. Cefoxitin and cephradine were relatively inactive (7% of isolates for which the MICs were ug/ml). No difference in cephalothin susceptibility was detected between results obtained in microtiter and in tubes. Vancomycin Downloaded from on April 26, 218 by guest

3 VOL. 14, 1978 SUSCEPTIBILITY OF S. EPIDERMIDIS 355 e co e e l Cq el CC I e _ QRO_ CO C M U- CO U3 M M M M ko _P- XOM 2 mtb ste ec clics r4 _ clre cli o ooceodi O t - COoC. -1. C.) C.). 4). 'IO a. C.) m I t C.) ri C. I 54) is eq cs 1 C1 eq " - Olts,-1co C C*r- C* Lat-MP-4 U CD um t-cd O ~ C ' t~-~- tc cu,z]~ :3~ Ct) C t- O ffi O _ 9 9 o m oo b o X b R%5 CD OC 5U cq h W- ko CS q (O LO i C- O P-O- O DOOCq t CD to CD. 4) i eqv- CD eqeo _- _ II _It LO,-4, o o4 -Iq.O t- - - O1V _ 1 '_. COY - o-4 e t-. N (O I"1 ~ eqt~ InN o o CO _ I1 CO t- eqnf 6 4) a *1 4L) 6 1 OD Downloaded from on April 26, 218 by guest eq o vo t- 'q CO 1.4 4).,V C) j. a

4 356 ARCHER showed consistent activity, both inhibiting and killing 8% of the isolates at concentrations of 1.6 to 6.25 pg/ml. Neither nafcillin nor cephalothin gave more than a two-dilution (fourfold) difference between MIC and MBC determinations with either tube or microtiter susceptibility testing. MIC and MBC testing of nafcillin, methicillin, and cephalothin in 5% NaCl at 3 C resulted in a fourfold or greater increase in both MIC and MBC to methicillin and nafcillin for all 17 isolates later found to contain resistant subpopulations (vide infra); there was a change in susceptibility to cephalothin for only a single isolate. Although high-level resistance to streptomycin (MIC - 5 ug/ml) was seen in 65% of isolates, the newer afminoglycosides were highly active. Gentamicin at a concentration of.2 pg/ml killed 89% of isolates, whereas three isolates, two from patients with PVE and one from a patient with a CSF shunt infection, were resistant to 6.25,ug of this antibiotic per ml. Tobramycin was always at least fourfold less active against each isolate than gentamicin, and amikacin was only 1/8 to 1/16 as active as gentamicin. Clindamycin was very active against some isolates, but 25% were highly resistant (MIC 25,ug/ml). Rifampin was the most effective of all antibiotics studied; for no isolate was the MBC >.1,ug/ml. Five antibiotics which might constitute effective therapy of S. epidermidis infections were chosen for further studies on the basis of susceptibility tests reported above. These were nafcillin, cephalothin, vancomycin, gentamicin, and rifampin. Methicillin was also included for comparison. Selection for resistance. The number of solates containing highly antibiotic-resistant colonies and the frequency of occurrence of the resistant colonies are shown in Table 2. Isolates contining subpopulations resistant to 2 ug of methicillin per ml will hereafter be designated methicillin-resistant (MR); those containing no colonies resistant to 2 plg of methicillin per ml will be designated methicillin-sensitive (MS). Susceptibility testing did not uniformly identify those isolates containing highly MR subpopulations. Three isolates (11%) for which the MIC was 6.25,ug of methicillin per ml contained no colonies resistant to 2,ug of that antibiotic per ml, whereas three isolates with that MIC contained colonies resistant to 1 ug of the drug per ml. Likewise, one of four isolates for which the MIC was 12.5 pug/ml contained no resistant colonies. Each MR isolate also contained colonies resistant to 2 pig of nafcillin and cephalothin per ml, but resistance to these antibiotics ANTIMICROB. AGENTS CHEMOTHER. TABLz 2. Antibiotic resistance occurring in populations ofstaphylococcus epidermidisa Patient isolates Antibiotic wita sistant Resistance frequency colonies (%) Methicillin (2) 63 lo-4.3 *.5 Methicillin (1) 63 lo-4.8 *.5 Nafcillin (2) *.5 Nafcillin (1) 52 1lo t.8 Cephalothin (2) 63 lo-4 ±.9 Cephalothin (1) 4 lo-7.7 Rifampmn (1) 1 1o Vancomycin (2) No resistant colonies/19 Gentamicin (5) 11 Total population resistant Values in parentheses indicate concentration in micrograms per milliliter. Other values indicate ratio of resistant colonies/total colonies, and represent the mean ± standard deviation of 27 isolates tested. was expressed at a lower frequency than resistance to methicillin. Colonies resistant to 1 ug of cephalothin per ml were found in only one isolate. There were no colonies resistant to 2,ug of nafcillin or cephalothin per ml selected from any MS isolate. Every isolate contained colonies resistant to 1 ug ofrifampin per ml, and no isolate contained colonies resistant to 2,ug of vancomycin per,ml. Gentamicin resistance was uniformly expressed in those three isolates with high MICs; there were no colonies resistant to 5,ug of gentamicin per ml among gentamicin-susceptible isolates. For 92% of the colonies picked from agar containing methicillin, the MICs and MBCs were -5,g/ml. However, 24% of the colonies picked from nafcillin plates and 33% from cephalothin plates reverted to the antibiotic susceptibility of the parent strain after overnight incubation in antibiotic-free broth. Isolates remaining antibiotic resistant grew slowly, and MICs could only be read after 48 h of incubation at 37C. Emergence of resistance with time. Six MR and five MS isolates were selected for timekill studies with methicillin, nafcillin, cephalothin, and rifampin (Fig. 1). The MBCs of nafcillin, cephalothin, and rifampin were similar for MR and MS isolates:.4 to 1.6 and.2 to 1.6 pg of nafcillin per ml for MR and MS isolates, respectively;.4 to 3.12 and.2 to 1.6,g of cephalothin per ml for MR and MS isolates, respectively; and.125 to.1 pg of rifampin per ml for both MR and MS isolates. Rifampin resistance emerged in all eleven isolates studied. Early killing was rapid, but growth Downloaded from on April 26, 218 by guest

5 VOL. 14, 1978 to the same number of CFU as control occurred between 8 and 24 h. Methicillin, nafcillin, and cephalothin showed a difference in their ability to kill MR versus MS strains. MS isolates were reduced from 17 CFU to <12 CFU by 2,g of any of the three antibiotics per ml after 48 h of incubation, and none of the remaining cells were antibiotic-resistant. MR isolates were not killed by 2 ug of methicillin per ml and were incompletely killed after 24 h of incubation with 2,ug of nafcillin or cephalothin per ml. Between 24 and 48 h of incubation, MR isolates grew two logs in nafcillin, whereas organisms exposed to cephalothin remained at the same concentration (Fig. 1). A comparison of the effects of rifampin, nafcillin, methicillin, and cephalothin on the emergence of resistance after incubation of rapidly growing MR organisms with these antibiotics is shown in Table 3. Rifampin-resistant colonies, although occurring at a low frequency in the parent strain, were rapidly and completely se- Q 7 9 HOURS FIG. 1. Killing of six MR and five MS S. epidermidis isolates with antibiotics over 48 h. Antibiotic concentrations of methicillin, nafcillin, and cephalothin were 2 pg/ml, and rifampin was at 1 pg/ml. Each point represents the mean CFU per milliliter for all designated isolates tested with that antibiotic; no count varied by more than ±.8 log among the isolates at each point. SUSCEPTIBILITY OF S. EPIDERMIDIS 357 lected so that virtually all cells were resistant by 24 h. The number of nafcillin-resistant colonies, however, did not change sigificantly by 24 h, although they comprised a larger percentage of the total population; a further 24 h of incubation was required before they increased in number. No cephalothin-resistant colonies were selected after 48 h of incubation, although almost 14 viable organisms remained. Rifampin-resistant colonies, after passage in antibiotic-free broth, were not killed by 1,ug of rifampin per ml and grew in a manner similar to control when retested by time-kill methods. Nafcillin-resistant colonies, however, were killed in a manner smilar to the parent strain upon reexposure to 2,ig of the antibiotic per ml. DISCUSSION This study has demonstrated the resistance to multiple antibiotics of isolates of S. epiderinidis from documented infections of indwelling foreign devices. None of the 27 isolates was susceptible to all 18 antibiotics studied; 85% were resistant to two, and 67% were resistant to six or more antibiotics. Resistance was defined as an MIC higher than concentrations which could be easily achieved in vivo, or -1.6,ug of penicillin G per ml. Studies of isolates of S. epidermidis from the clinical laboratory show a greater degree of antibiotic susceptibility than was demonstrated with known pathogenic isolates in this study (3, 12). However, susceptibility data reported for isolates of S. epidermidis from cases of PVE confirm their resistance to methicillin and other antistaphylococcal antibiotics (7). Resistance to methicillin as defined by the growth of colonies on agar containing 2 utg of the antibiotic per ml was found in 63% of isolates in this study. This is in contrast to reported incidences of methicillin resistance in 1 to 33% of isolates from clinical laboratories by disc or agar dilution susceptibility testing (11, 15). Methicillin resistance in S. aureus has been extensively studied and defined as intrinic or not due to enzymatic destruction of the antibiotic (17). Characteristics of S. aureus methicillin re- Downloaded from on April 26, 218 by guest TABLE 3. Selection of antibiotic resistance after the addition of antibiotics to growing cultures ofsix MR S. epidermidis isolateea Total CFU at: Resistant CFU at: Antibiotic Oh 24h 48h Oh 24h 48h Methicillin (2) 1i6-8 ± i.7 7. *.6 1o3.5 ± * Nafcilin (2) lo6. * lo5.9 * 1. 1o2.3 ±.2 1o2.5 ±.5 lo6." ± 1.2 Cephalothin (2) 17. ±.3 1o ±.6 2 ± 1 <1 <1 Rifampin (1) 16.8 ± ± <1 164 * 1.2 1o8o *.4 Values in parentheses indicate concentration in micrograms per milliliter. All other values represent mean istandard deviation.

6 358 ARCHER ANTIMICROB. AGENTS CHEMOTHER. sistance were found among the S. epidernidis mase production was found in all penicillin-reisolates in this study. These included the follow- sistant isolates (those for which the MIC was ing: the MR phenotype was heterogeneic, found -1.6,ug of penicillin G per ml [G. Archer and S. in only 1 cell in 14 at 37C; resistant cells were Markowitz, unpublished observations]), the neither inhibited nor killed by 1 jig of methi- characteristics of enzyme-mediated antibiotic cillin per ml; and resistant cells grew slowly at resistance were not seen among MR isolates. 37C. All MR isolates also contained colonies That is, the frequency ofoccurrence ofresistance resistant to 2,ug of nafcillin per ml. The resist- was too low in the population, and the rate of ance characteristics were similar to those of growth of resistant colonies was too slow to be methicillin, although the phenotypic expression typical of f8-lactamase-mediated resistance (13). of nafcillin resistance occurred at a lower fre- Tolerance, a recently described resistance to quency than did methicillin resistance. cell-wall-active antibiotics among S. aureus, also The detection of methicillin and nafcillin re- did not seem to be an important resistance sistance was not always possible on the basis of mechanism. This form of resistance is charactermicrotiter twofold dilution susceptibility tests at ized by the inhibition of bacterial growth withthe standard inoculum of 15 CFU/ml. Of six out bacterial killing (low MIC, high MBC), and isolates for which the MIC was 6.25 /Ag of meth- resistance is lost after prolonged storage or with icillin per ml, only one-half contained a highly prolonged incubation timne (13). A difference beresistant subpopulation; one of four isolates for tween the MICs and MBCs of nafcillin for our which the MIC was 12.5 jig of methicillin per ml isolates was not seen by either microtiter or contained no resistant cells. Likewise, although tube-dilution testing, nor did fresh clinical isoall MR isolates contained nafcillin-resistant lates exhibit the phenomenon. cells, the MIC for only 4 of 17 (15%) MR isolates Two of the cephalosporin antibiotics tested was >3.12,Lg of nafcillin per ml. The small (cephalothin and cefamandole) had excellent acnumber of bacteria in each microtiter well (15 tivity against most S. epidernidis isolates, one CFU/ml x.2 ml) is obviously too low to detect (cefazolin) had intermediate activity, and one nafcillin resistance which occurs with a low fre- (cephradine) had poor activity. Cefoxitin, a new quency. However, the use of the standard tube cephamycin derivative, was also inactive against dilution susceptibility test and a larger inoculum most isolates. Colonie to 2 ug of ceph- (5 x 1 CFU/ml x 1. ml) detected nafcillin- t d Mthe qey resistant subpopulations in only four additional isolates. was v 14 Unlike resistance to semi- The use of hypertonic media (5% NaCl), low esyntheic penic, resistance to cephalothin incubation temperature (3C), and prolonged could not be detected by changing inoculum size incubation time (48 h) increased the MIC of or physical factors during susceptibility testing. nafcillin and methicillin for all isolates found to Furthermore, cephalothin-resistant subpopulacontain resistant colonies. These factors have tions did not emerge during 48 h of incubation been shown to increase the subpopulations in S. of MR isolates with 2,ug of the antibiotic per aureus resistant to semisynthetic penicillins and ml, although killing was incomplete. The perto enhance their phenotypic expression (2, 17). sistence of cephalothin-susceptible colonies The importance of identifying isolates contain- among MR isolates after 24 to 48 h of incubation ing subpopulations resistant to nafcillin and may be. partially explained by a decrease in methicillin was shown by time-kill studies. Pop- cephalothin concentration in Mueller-Hinton ulations totally resistant to nafcillin and methi- broth of up to 8% after 24 h of incubation at cillin emerged after 48 h of exposure to either 37C (G. Archer, unpublished observations). drug among MR isolates; killing was almost com- However, this does not explain the difference in plete, and no resistant cells emerged among MS killing between MR and MS isolates; further isolates. The use of large inocula to overcome studies are needed to explain this discrepancy. the inoculum effect produced by the low phe- Laverdiere et al. (1) recently confirmed the notypic expression of a resistant trait and the absence of complete cross-resistance between alteration of physical and chemical factors in methicillin and cephalosporins against MR S. order to enhance the detection of resistance epidermidis isolates from a variety of sources. should be routinely employed to identify S. ep- However, whereas only 47% of the isolates from idermidis isolates from serious infections which their study were killed by s6.25 jig of cephaloare resistant to semisynthetic penicillins. thin per ml, 76% (13 of 17) of the MR isolates in Mechanisms of resistance to semisynthetic our study were killed by this concentration of penicillins among S. epidermidis other than in- cephalothin. Thus, on the basis of our data, trinsic resistance were not identified in this cephalothin, cefazolin, and cefamandole appear study. Although methicillin-inducible /3-lacta- to be superior to semisynthetic penicillins for Downloaded from on April 26, 218 by guest

7 VOL. 14, 1978 initial therapy of serious S. epidermidis infections. Rifampin was the single most active antibiotic against S. epidernidis; the MBC for no isolate was >.1,ug/ml. However, resistant cells, although occurring at a low frequency (1-72), were found in every isolate. In contrast to resistance to semisynthetic penicillins, rifampin resistance was rapidly selected, resistant cells grew at a normal rate, and the resistant phenotype was stable and homogenous in the population after selection. Resistance to rifampin has been shown to be due to a one-step mutation whereby the target enzyme, ribonucleic acid polymerase, no longer binds the antibiotic (4). However, because of the uniformly high susceptibility of the majority population of S. epidermidis to rifampin, the demonstration that the addition of a second antibiotic can prevent the emergence of rifampin resistance (1), and the successful use of rifampin-containing regimens to treat serious infections caused by MR S. aureus (8), rifampin should be considered a potentially useful antibiotic for the treatment of serious S. epidermidis infections. Vancomycin and gentamicin may also be valuable antibiotics for the treatment of infections caused by S. epidermidis. Although isolates had only internediate susceptibility to vancomycin, no resistant subpopulations were found. This antibiotic is of proven value in the treatment of severe infections caused by S. aureus (6). Gentamicin-susceptible isolates had the lowest MICs and MBCs for this antibiotic of any of the drugs studied, and they were at least fourfold lower than for the other available newer aminoglycosides (tobramycin and amikacin). No resistant subpopulations were found, and the three resistant isolates were easily identified by routine susceptibility testing. This resistance may be plasmid-mediated, as has been described for S. aureus (16). The therapy of S. epidernidis infections of indwelling foreign devices should be carefully guided by antimicrobial susceptibility tests. However, routine tests for resistance to semisynthetic penicillins may be misleading, and these antibiotics should not be used as initial therapy. Antibiotic combinations chosen from the cephalosporins, vancomycin, gentamicin, and rifampin may improve the cure of these refractory infections. More studies are necessary to define resistance mechanisms among S. epidernidis. ACKNOWLEDGMENTS I am grateful to J. Linda Johnston for her excellent technical assistance and to Sheldon Markowitz for his review of SUSCEPTIBILITY OF S. EPIDERMIDIS 359 the manuscript. This work was supported by Public Health Service grant number 5 RO 1 HL from the National Heart, Lung, and Blood Institute. LITERATURE CITED 1. Arioli, V., M. Berti, G. Carniti, E. Rossi, and L. G. Silvestri Interaction between rifampin and trimethoprim in vitro and in experimental infections. J. Antimicrob. Chemother. 3: Barry, A. L., and R. E. Badal Reliability of the microdilution technic for the detection of methicillinresistant strains of Staphylococcus aureus. Am. J. Clin. Pathol. 67: Corse, J., and R. E.. Williams Antibiotic resistance of coagulase-negative staphylococci and micrococci. J. Clin. Pathol. 21: di Mauro, D., L. Snyder, P. Marino, A. Lamberti, A. Coppo, and G. P. Tocchini-Valentini Rifampicin sensitivity of the components of DNA-dependent RNA polymerase. Nature (London) 222: Dismukes, W. E., A. W. Karchmer, M. J. Buckley, W. G. Austen, and M. N. Swartz Prosthetic valve endocarditis: analysis of 38 cases. Circulation 48: Esposito, A. L, and R. A. Gluckman Vancomycin: a second look. J. Am. Med. Assoc. 238: Johnson, W. D Prosthetic valve endocarditis p In D. Kaye (ed.), Infective endocarditis. University Park Press, Baltimore. 8. Jensen, K Methicillin-resistant staphylococci. Lancet ii: Keys, T. F., and W. L. Hewitt Endocarditis due to micrococci and Staphylococcus epidermidis. Arch. Intern. Med. 132: Laverdiere, M., P. K. Peterson, J. Verhoef, D. N. Williams, and L D. Sabath In vitro activity of cephalosporins against methiciflin-resistant, coagulasenegative staphylococci. J. Infect. Dis. 137: Sabath, L D., F. F. Barrett, C. Wilcox, D. A. Gerstein, and M. Finland Methicillin resistance of Staphylococcus aureus and Staphylococcus epidermidis, p Antinicrob. Agents Chemother Sabath, L. D., C. Garner, C. Wilcox, and M. Finland Susceptibility of Staphylococcus aureus and Staphylococcus epidermidis to 65 antibiotics. Antimicrob. Agents Chemother. 9: Sabath, L D., N. Wheeler, M. Laverdiere, D. Blazevic, and B. J. Wilkinson A new type of penicillin resistance of Staphylococcus aureus. Lancet i: Schoenbaum, S. C., P. Gardner, and J. Shillito Infections of cerebrospinal fluid shunts: epidemiology, clinical manifestations, and therapy. J. Infect. Dis. 131: Slaughter, L., J. E. Morris, and A. Starr Prosthetic valvular endocarditis: a 12-year review. Circulation 47: Smith, H. B. H., and H. Farkas-Himsley The relationship of pathogenic coagulase-negative staphylococci to Staphylococcus aureus. Can. J. Microbiol. 15: Sutherland, R., and G. N. Rolinson Characteristics of methicillin-resistant staphylococci. J. Bacteriol. 87: Wood, D. O., M. J. Carter, and G. K. Best Plasmid-mediated resistance to gentamicin in Staphylococcus aureus. Antimicrob. Agents Chemother. 12: Downloaded from on April 26, 218 by guest