ANTMCROBAL AGENTS AND CHEMOTHEAPY, Mar. 193, p. 3339 Copyright 193 American Society for Microbiology Vol. 3, No. 3 Printed in U.S.A. Discrepancy Between Carbenicillin and Ampicillin Activities Against Enterococci and Listeria GEORGE H. McCRACKEN, JR., JOHN D. NELSON, AND MARON L. THOMAS Department of Pediatrics, University of Texas Southwestern Medical School, Dallas, Texas 5685 Received for publication 6 November 192 For infections of the newborn, carbenicillin has theoretical advantages over ampicillin, when given with an aminoglycoside, because of its activity against Pseudomonas and many indolepositive Proteus strains. Agar and broth dilution susceptibility studies demonstrated comparable activities of carbenicillin and ampicillin against coliform organisms but greater resistance of enterococci and Li8teria to carbenicillin. However, the significance of the higher inhibitory concentrations is questionable because many serum specimens containing concentrations of carbenicillin lower than the minimal bactericidal values for Listeria and enterococcus strains had demonstrable in vitro bactericidal activity. Carbenicillin and gentamicin combinations had greater antibacterial activities than either drug alone against all enterococci tested and against 5% of Listeria strains. The questions of efficacy raised by contradictory in vitro data may be answered by clinical studies of carbenicillin and gentamicin in infections of the newborn. t is customary to administer a penicillin and an aminoglycoside antibiotic to newborn infants with suspected bacterial infections. Currently, ampicillin and either kanamycin or gentamicin are commonly used in nurseries throughout the United States. Carbenicillin has potential advantages over ampicillin for infections of the neonate because of its activity against Pseudomonas and some indolepositive Proteus strains. ts activity against other bacteria is said to be similar to that of ampicillin. This study of comparative susceptibilities and serum activities of ampicillin and carbenicillin against bacteria encountered in neonatal infections confirmed similar activities against coliform organisms but revealed significant differences between the two drugs with respect to enterococci and Listeria. (This material was presented in part at the 12th nterscience Conference on Antinmicrobial Agents and Chemotherapy, Atlantic City, N.J., 28 September 192.) MATERALS AND METHODS Bacterial isolates. The bacterial strains tested were isolated from blood, cerebrospinal fluid, or infected lesions of newborn and young infants. Enterobacteriaceae were identified by routine bacteriological methods (1). Enterococci were identified on the basis of hemolysis on sheep blood agar, growth in 6.5% sodium chloride broth, mannitolsalt broth and SF medium (Difco), and reduction of litmus milk (Difco). The identity of strains of Listeria monocytogenes was based on morphological and growth characteristics and was confirmed by using specific fluoresceintagged antisera. Antimicrobial susceptibilities. The antimicrobial susceptibilities of 35 strains of enterococcus, 23 of Proteus mirabilis, 25 Shigella species, 5 Escherichia coli strains, 25 Klebsiella species, and 25 Enterobacter species were determined by an agar plate dilution technique with the use of a multiple inoculating apparatus (6). The antibiotics used were ampicillin trihydrate laboratory standard containing 825,pg of activity/mg (Bristol Laboratories) and disodium carbenicillin laboratory standard (J. B. Roerig) with 1, pg of activity/mg. The inoculum was 11 organisms. The minimal inhibitory concentration (MC) was the smallest amount of antibiotic inhibiting visible growth after 18 h of incubation at 3 C. Broth dilution susceptibility studies were performed with 36 strains of enterococcus and 5 strains of L. monocytogenes. Approximately 11 organisms were inoculated into serial dilutions of ampicillin or carbenicillin in Oxoid broth and incubated for 18 h at 3 C. A loopful of broth from each tube showing inhibition of visible bacterial growth was subcultured onto agar media. The lowest concentration of drug inhibiting growth on subculture was taken as the minimal bactericidal concentration (MBC). Serum bactericidal levels. Bactericidal titers were determined in sera from infants being treated 33 with gentamicin and carbenicillin and in pooled normal human serum to which known concentrations of antibiotics were added. The pooled serum without added antibiotics had no bactericidal effect on the test organisms. The specimens were
3 McCRACKEN, NELSON, AND THOMAS ANTMCROB. AG. CHEMOTHER. divided and one portion was treated with penicillinase to destroy carbenicillin activity while aminoglycoside activity was retained. (nactivation of carbenicillin was confirmed by assay of penicillinasetreated specimens.) Serial dilutions of these specimens were made in Oxoid broth, and 15 bacteria per ml were inoculated into each dilution and into control tubes. After 18 h of incubation at 3 C, tubes showing inhibition of visible bacterial growth were subcultured onto agar media. The highest dilution with no growth on subculture was taken as the serum bactericidal titer. Studies of antimicrobial synergism. Ten strains of Listeria and enterococcus were studied for susceptibilities to carbenicillin and gentamicin and to ampicillin and gentamicin, separately and in combination. A two dimensional "checkerboard" dilution method with twofold increments of each drug in Oxoid agar was used (2). Approximately 1O organisms per ml were inoculated with a multiple inoculating apparatus. The MC of each drug alone and of all of the combinations of each antibiotic pair was determined. Representative strains of bacteria showing different checkerboard patterns of inhibition were studied by a broth dilution method. n this method, approximately 1O organisms per ml were inoculated into Oxoid broth containing the antibiotics alone and in combinations. Kill curves were constructed (5) from bacterial colony counts performed at,, and 18 h of incubation at 3 C. sobolograms were constructed from the agar dilution susceptibility patterns to compare with kill curves observed with the same strains. By this method, the MC of one drug is plotted on an arithmetic scale on the ordinate and the MC of the other drug is plotted on the abscissa. These two points are connected with a straight line. The MCs of each drug pair are then plotted, and the points are connected with a line called the isobole. f the line is straight and paiallel to the line connecting the MCs of the drugs tested singly, the effect of the combined drugs is additive; if it is concave, the combined effect is synergistic; if the line is convex, the combined drugs are interpreted as antagonistic (2). RESULTS Antimicrobial susceptibilities. Agar plate dilution studies of Proteus, Shigella, and E. coli demonstrated that 88% or greater of strains were inhibited by both ampicillin and carbenicillin at a concentration of 1,ug/ml. For 9% of these strains, the MC of ampicillin and that of carbenicillin were within one dilution of each other (Fig. 1). n contrast, Klebsiella and Enterobacter species were almost uniformly resistant to both carbenicillin and ampicillin; fewer than 1% of the strains were inhibited by ;ig of ampicillin/ml. Although six strains (25%) of Enterobacter were inhibited by,ug of carbenicillin/ ml, the remaining 18 strains required a concentration of 32 Ag/ml or greater for inhibition. Klebsiella species were uniformly resistant to more than 32 jig of carbenicillin/ml. Among 32 strains of enterococcus, 31 (9%) were susceptible to 2.5 Mug or less of ampicillin/ml by agar plate dilution testing, but only 18 of 35 strains (51%) were inhibited by Mg of carbenicillin/ml. n broth dilution studies with 36 strains of enterococcus (Fig. 2), 35 strains (9%) were inhibited and 3 strains (9%) were killed by 2.5 ug or less of ampicillin/ml. By comparison, CP > * E. coil o Shigella. *~~~&Protous mfirobilis/.o2 3 2 _ *o() 5 in,, ' 2.5 e. l C ; A n n ^ ^ ~ ~~Z. D 1u Zu 1u 11U Minimal nhibitory Concentration (pg/mi) (Ampicillin) FG. 1. Comparison of susceptibilities of coliform organisms to carbenicillin and ampicillin. Q5.312 1.25 5 1 16 >32 Concentrotion (pg/mi) FG. 2. Susceptibilities of 86 strains of enterococcus to ampicillin and carbenicillin. MC, minimal inhibitory concentration; MBC, minimal bactericidal concentration.
VOL. 3, 193 CARBENCLLN AND AMPCLLN ACTVTES 35 32 of 35 enterococcus strains (91 %) were inhibited by 8,ug or less of carbenicillin/ml and 2 strains (69%) were killed by this concentration. Among the enterococci tested, 92% were killed by 16,ug or less of carbenicillin/ml. All of 5 strains of Listeria tested by the broth dilution method (Fig. 3) were inhibited and killed by.6,g or less of ampicillin/ml. A concentration of 1 Mug of carbenicillin/ml inhibited all strains of Listeria, but the MBC values were over,ug/ml for 1 strains (3%). Serum bactericidal levels. Sera obtained from three infants receiving carbenicillin and gentamicin and pooled normal human serum to which these drugs were added were studied for their bactericidal effect against strains of enterococcus and Listeria. Each specimen was tested with and without the addition of penicillinase to destroy carbenicillin. Serum bactericidal titers correlated better with the levels of carbenicillin in serum than with those of gentamicin (Fig. ), and serum bactericidal titers dropped significantly after inactivation of carbenicillin by penicillinase (Table 1). The bactericidal titers of specimens treated with penicillinase were 1:2 or less in all cases except infant 3. Direct relationships among the concentrations of carbenicillin in serum, the MC or MBC values for the test organisms, and the bactericidal titers in serum were not consistently observed. When the carbenicillin concentration was equal to or greater than the MBC for the test organism, the bactericidal titer was 1: or greater. On the other hand, 9 of 11 sera with 1 1 " ~~~~~~~~~~~~~~~ 'A so 8 / E*2o C 6~~~~.~~~ 2/ E~~~~ S.3.15.6 2.5 5 1 2' Concentration (pg/mi) FG. 3. Susceptibilities of 5 strains of Listeria monocytogenes to ampicillin and carbenicillin. MC, minimal inhibitory concentration; MBC, minimal bactericidal concentration. 1:128r E 1:8 1: * 1:2 h * Enterococcus o Listeria <1: 2 <1:2 1:2 1: 1:8 1:16 1:32 Treated with Penicil linase FG.. Bactericidal titers of serum specimens containing carbenicillin and gentamicin against enterococcus and Listeria. Each specimen was divided and one portion was treated with penicillinase to destroy carbenicillin activity. carbenicillin levels less than the MBC for the test organism had demonstrable bactericidal effect. Antagonism between gentamicin and carbenicillin in these serum specimens was not observed. Antimicrobial synergism. There is no universally accepted definition of synergism, principally because various test systems have been used by different investigators and there is not always consistency in results. Jawet reviewed this problem at length (). n this study, synergistic effect by kill curves was considered to be present when bacterial colony counts were reduced by 2 logs or more when the organism was incubated for 18 h in the presence of wo drugs compared with the effects on bacterial growth of the same drug concentrations tested separately. The effect of the combination was considered additive if the reduction in bacterial colony counts was equal to or only 1 log greater than the sum of the effects of the same concentrations singly. The susceptibilities of 1 strains each of Listeria and enterococcus to multiple combinations of carbenicillin and gentamicin and ampicillin and gentamicin were determined by agar
36 McCRACKEN, NELSON, AND THOMAS ANTMCROB. AG. CHEMOTHER. TABLE 1. Carbenicillin and gentamicin concentrations in serum and bactericidal titers against enterococci and Listeria Bactericidal titer Source of serum specimens and interval Test organism and MC/MBC. wtenafter dose Carben Gentam Origial wited icflin icillill icin serum wihpn Drug conen (ug/ml) icil Belwum ~~~~~~~~~~~~icillinas nfant 1.5 h... 2 2.8 Enterococcus 1:2 <1:2 1 h... 5 3.1 Carbenicillin, 16/16 1:2 <1:2 2 h... 3 2.1 Gentamicin, 1/1 1:2 <1:2 3 h... 6 2.9 1:2 <1:2 h... 2 1.8 <1:2 <1:2 nfant 2 O.5 h... 1 2.1 Enterococcus 1: 1:2 i h... 1 1.8 Carbenicillin, /8 1: <1:2 2 h... 99 1. Gentamicin, 1/1 1:8 <1:2 h... 69 1.2 1:2 <1:2 6 h... 6. 1:2 <1:2 nfant 3.5 h... 183. Listeria 1:128 1:8 1 h... 1 3. Carbenicillin, 2.5/2 1:6 1:8 2 h... 16 3. Gentamicin,.3/.6 1:6 1:8 h... 98 2.2 1:6 1:8 6 h... 1.6 1:8 1: Lab sample 1 1... 8. Listeria 1:16 1:2 2.... Carbenicillin, 2.5/> 1:8 1:2 3... 2. Gentamicin,.3/.6 1: 1:2 Lab sample 2 1... 16 5. Enterococcus 1: 1:2 2... 8 5. Carbenicillin, 8/16 1:2 <1:2 3... 5. Gentamicin, 2/2 <1:2 <1:2 plate dilution and broth dilution techniques. The four common checkerboard patterns observed with the agar plate method are shown in Fig. 5 and 6. sobolograms drawn from the data illustrated in these figures were as follows: antagonistic effect for pattern and additive effect for pattern in Fig. 5 and 6. Patterns and V in Fig. 5 and 6 would be interpreted as a synergistic effect by the isobolograms. By our arbitrary definition of synergism for kill curves, there is agreement with the isobologram interpretations for patterns and V with the exception of pattern in Fig. 5, in which the kill curve showed only an additive effect of the combined drugs. Kill curves were intermediate in pattern L. Antagonism could not be demonstrated by the kill curve technique for pattern because colony counts for combined drugs were no greater than those with the drugs tested singly. However, the log concentration of bacteria was near the maximal attainable growth in broth for these bacteria, and it is possible that antagonism might have been demonstrated if the original inoculum of bacteria had been smaller. Against this speculation is the fact that Eickhoff (2) was not able to confirm antagonism by kill curves for drug combinations interpreted as antagonistic by isobologram; thus, his results are consistent with our observations. Five of 1 strains of Listeria demonstrated synergism as interpreted from both isobolograms and kill curves with combinations of carbenicillin and gentamicin and of ampicillin and gentamicin. All 1 strains of enterococcus tested showed synergism with combinations of carbenicillin and gentamicin, and 8 of 1 strains demonstrated synergism with ampicillin and gentamicin. DSCUSSON Carbenicillin and ampicillin are analogues of penicillin which are closely related chemically. Carbenicillin is the first penicillin to have antimicrobial activity against many indolepositive Proteus species and Pseudomona&. With the exception of these two species, carbenicillin and ampicillin have been reported to have similar activities against grampositive and gramnegative bacteria (3, ). This study confirms the
VOL. 3, 193 CARBENCLLN AND AMPCLLN ACTVTES 3 E c C. 1 'U LLo C.) p :L i 2 :L D 5 E CONCENTRATONS CARBENCLLN F~~ (^&g/ml) MC 1.25kg/ml J151.3.6 1.252.5 5..C1 +.3.6 1.25 MC 2.5Fg/ml :5.3J.61.*. 2.515 D8+ ++ +.3 + ++2 1.25 2.5pg/ml MC.15.3.61. 2.5 5. + + 8 6 O 5 Z_ 3 c) i 8 J w6 S 5 3 o 5 8 SNGLE DRUGS BENT.6 CARE.6 BENT.15 COMBNED DRUGS.3 2 3 1 18 s 18 CARS121 BENT.3 BENT.15 k 18 ENT.8 C.6 BENT.39 a.6 18 ~. 1 1.215 a5 M C 5.,ug/ ml E _ Z, m E.15++ ++ _ Z~U 3.6125 2.5 5. Z1 3+ +.6 + + +.2 1L25 6 J 5 3 18.3T. GEX T 32.5 18 L 18 TME N HOURS FG. 5. nteraction of carbenicillin and gentamicin against four strains of Listeria demonstrated by checkerboard (left) and killcurve (right) techniques. sobologram drawn from pattern would be interpreted as antagonistic, that from pattern as additive, and those from patterns and V as synergistic. Kill curves do not confirm antagonism but do demonstrate the addition and synergistic effects. The numbers above the lines on the kill curves to the right refer to the concentrations of drugs used in the numbered boxes in the checkerboard.
38 McCRACKEN, NELSON, AND THOMAS ANTMCROB. AG. CHEMOTHER. CONCENTRATON AMPCLLN (1Ag/ml) MC 1.251sg/ml 1.3. 1.25 2 5 1.2 + + +1 1 2.51 + + E 5 _+ 1 f co 2 w CY l o tt 1+ 1 1 1 T 1 2. MC 1.25~g/m 2 XJ51.3.6 1L25 2.5 5 1.25 + ++ (. 5 + E E o t) _ 2 d Z 2. CARBENCLLN (pg/ml) MC 8g/ml 5 112 8 _ 11+ 9 2 _ 1 TS 8 _ MC 8ug/ ml M C,A 8 o 5 o 3 d ~ 8 61 i < 5 co SNGLE DRUGS COMBNED DRUGS GENT 5 MP.3 ASPOiMPO.6 E 5+ +. 1% s + 2 + +w6 CA2 3~~~~ 6 5 3 18 18 2 2 CLONES/ML 18 18 TME N HOURS FG. 6. nteractions of carbenicillin and gentamicin and of ampicillin and gentamicin against four strains of enterococcus demonstrated by checkerboard (left) and killcurve (right) techniques. (See Fig. 5 for description.)
VOL. 3, 193 CARBENCLLN AND AMPCLLN ACTVTES 39 comparable susceptibilities to these two antibiotics of coliform organisms isolated from newborn infants. Both the percentage of strains inhibited and the actual MC values for E. coli, P. mirabilis, Klebsiella and Enterobacter species, and Shigella were almost identical. Listeria and enterococci were more resistant to carbenicillin than to ampicillin by broth dilution testing. Ninetyfive percent of enterococcus strains were killed by ampicillin concentrations of 2.5 Ag or less/ml, and 9% of Listeria strains were killed by ampicillin concentrations of.3,ug or less/ml. n contrast, concentrations of 16 Mug of carbenicillin/ml were necessary for killing 9% of enterococcus strains. Although the carbenicillin MC values for Listeria were 1 Mg or less/ml, only 66% of strains were killed by concentrations of Mg or less/ml. The significance of these differences among in vitro susceptibilities is unknown. Serum concentrations of carbenicillin less than the MBC values for enterococcus and Listeria often produced paradoxical in vitro bactericidal titers of 1:2 or greater. This discrepancy between concentrations of carbenicillin in serum, MBC values of test organisms, and the serum bactericidal effect emphasies the dilemma in correlating in vitro data with clinical application of antimicrobial therapy. This dilemma is compounded by the uncertainties in interpretation of additive or synergistic effects in vitro. A synergistic effect with combinations of carbenicillin and gentamicin was considered to be present with all strains of enterococcus tested and with 5% of Listeria strains, but the relevance of this observation to treatment of infections in patients is uncertain. The best evidence for synergism should come from clinical studies rather than test tube studies, and, as Jawet () pointed out, endocarditis and sepsis in the deficient host should provide the best diseases for studies of drug synergism since the contribution of host mechanisms to eradicating bacteria is minimal. n many respects, the neonate is a deficient host. The efficacy of carbenicillin combined with gentamicin in the treatment of neonatal bacterial infections is presently under investigation. Such studies may help answer some of the questions raised by these sometimes contradictory and confusing in vitro data. LTERATURE CTED 1. Edwards, P. R., and W. H. Ewing. 192. dentification of Enterobacteriaceae, 3rd ed. Burgess Publishing Co., Minneapolis. 2. Eickhoff, T. C. 1969. n vitro effects of carbenicillin combined with gentamicin or polymyxin B against Pseudomona aerugino8a. Appl. Microbiol. 18:693. 3. senberg, H. D., and M. Siegel. 1969. n vitro action of carbenicillin against bacteria isolated from clinical material. Appl. Microbiol. 18:38392.. Jawet, E. 1969. The use of combinations of antimicrobial drugs. Annu. Rev. Pharmacol. 8:151 1. 5. Standiford, H. D., J. B. de Maine, and W. M. M. Kirby. 19. Antibiotic synergism of enterococci. Arch. ntern. Med. 126:255259. 6. Steers, E., E. L. Folt, B. S. Graves, and J. Riden. 1959. An inocula replicating apparatus for routine testing of bacterial susceptibility to antibiotics. Antibiot. Chemother. 9:3311.. Washington, J. A. 192. n vitro susceptibility of gramnegative bacilli to carbenicillin. Mayo Clin. Proc. :33233.