DIAGN MICROBIOL INFECT DIS 1991;14:377-382
377
ANTIMICROBIAL SUSCEPTIBILITY STUDIES
Comparison of the Bactericidal Activity of Clindamycin and Metronidazole Against CefoxitinSusceptible and Cefoxitin-Resistant Isolates of the Bacteroides fragilis Group Charles W. Stratton, Lyndell S. Weeks, and Kenneth E. Aldridge
Time-kill kinetic methodology was used to evaluate the bactericidal activity of cefoxitin, cefotetan, clindamycin, and metronidizole against cefoxitin-susceptible and cefoxitin-resistant isolates of the Bacteroides fragilis group. Overall, metronidazole was the most bactericidal agent, with all isolates being killed with ~ 4 I~g/ml at 24 hr. Clindamycin was the next most bactericidal agent, with 20 of 26 isolates being killed with ~ 16 I~g/ml. Six isolates with clindamycin MICs ~ 64 i~g/ml were not killed at 24 hr, with concentrations as high as 256 ~g/ml. Cefoxitin and cefotetan were the least bactericidal
agents tested. Seven isolates with MICs of ~ 64 ~g/ml to each agent demonstrated a lack of killing at 24 hr, with concentrations of the respective agent as high as 256 i~g/ml. At concentrations with either agent of 32 la,g/ml, the remaining 19 isolates were killed at 24 hr. Of the six B. fragilis isolates resistant to clindamycin, four were also resistant to both cefoxitin and cefotetan. We conclude that in hospitals with cefoxitin-resistant B. fragilis group isolates, metronidazole would provide appropriate therapy.
INTRODUCTION
has been an increase in the incidence of B. fragilis resistant to clindamycin (Soriano et al., 1984; Sosa et al., 1982). Data from these isolates, moreover, indicate that c e p h a m y c i n resistance often is accompanied by increased resistance rates to other antimicrobial agents such as clindamycin. Finally, alt h o u g h rare, resistance to m e t r o n i d a z o l e has b e e n described (Lamothe et al., 1986). In this study, we used time-kill kinetic m e t h o d o l o g y to evaluate the bactericidal activity of two c e p h a m y c i n s (cefoxitin and cefotetan) using isolates of B. fragilis with low m e a n inhibitory concentrations (MICs) to cefoxitin (broth-dilution MIC ~ 32 ~g/ml) a n d isolates with moderate-to-high MICs to cefoxitin (MIC/-- 32 }zg/ml) and c o m p a r e d this activity with that of clindamycin and metronidazole. We also c o m p a r e d this bactericidal activity with those concentrations of antibiotic that w o u l d be expected in vivo.
During the past decade, m e m b e r s of the Bacteroides fragilis g r o u p have s h o w n an increase in resistance to c e p h a m y c i n s such as cefoxitin and cefotetan (Betriu et al., 1990; Tally et al., 1985). In addition, there
From the Department of Pathology (C.W.S., L.S.W.), Vanderbilt University School of Medicine, Nashville, Tennessee; and the Department of Medicine (K.E.A.), Louisiana State University School of Medicine, New Orleans, Louisiana USA. Address reprint requests to Dr. C.W. Stratton, Department of Pathology, C-3217 MCN, Vanderbilt University Medical Center, Nashville, TN 37232-2561, USA. Received 23 July 1990; revised and accepted 1 November 1990. © 1991 Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010 0732-8893/91/$3.50
378
MATERIALS A N D M E T H O D S Microorganisms A total of 12 members of the B. fragilis group with moderate-to-high MICs to cefoxitin were chosen for testing. Each isolate had a broth microdilution MIC value ~ 32 ~g/ml. This test cohort was comprised of the following: B. fragilis, two strains; B. thetaiotaomicron, five strains; B. distasonis, two strains; and B. ovatus, three strains. A second cohort of 12 isolates of B. fragilis with low MICs to cefoxitin (MIC values for cefoxitin ~ 32 p~g/ml) were also tested. In addition, B. fragilis American Type Culture Collection (ATCC) 25285 and B. thetaiotaomicron ATCC 29741 were used for quality control. Each clinical isolate (n = 24) was identified to species level by use of selective media, biochemical profiles, and gas-liquid chromatography (Finegold and Edelstein, 1985). Stock cultures were maintained at - 70°C and cloned twice on prereduced sheep blood agar before use.
C.W. Stratton et al.
has not influenced the bactericidal activity of antimicrobial agents.
Inoculum Portions of five or more colonies on a plate culture were inoculated into prereduced brain-heart infusion broth and incubated. The inoculum was prepared from a logarithmic phase of this culture with a final inoculum size of 106 CFU/ml. The inoculum size was carefully controlled and the final inoculum size was confirmed each time by a modified surface colony count method (Miles and Misra, 1938). For this method, a sample was removed and diluted (undiluted to 10-7) in prereduced saline, and plated (0.05 ml) in duplicate onto prereduced blood-agar plates. These plates were incubated anaerobically for 48 hr, colonies were counted and averaged, and the final inoculum was calculated. If the final inoculum was not between 1 x 10 6 and 1 x 107, then the test was repeated.
Antimicrobial Agents Broth Microdilution Method Standard powders were obtained from their respective manufacturers as follows: cefoxitin (Merck Sharp and Dohme, Rahway, NJ), cefotetan (Stuart, Wilmington, DE), clindamycin (Upjohn, Kalamazoo, MI), and metronidazole (Searle Laboratories, Chicago, IL). Antimicrobial stock solutions were prepared according to the manufacturer's instructions and stored at - 70°C until use. Final concentrations were prepared on the day that they were used.
Media Anaerobic broth MIC (Difco Laboratories, Detroit, MI) was used for the broth microdilution method. This media is supplemented with vitamin K1 (1 ~g/rnl) and heroin (5 ~g/ml) and has a formulation similar to that of the Wilkins-Chalgren medium, but lacks the agar. Previous studies (C. Stratton, unpublished data) have shown good correlation (+ 1 dilution) of MIC determinations done with anaerobic broth MIC and anaerobic broth medium from the BACTEC 7D blood culture bottle (Becton Dickinson, Towson, MD). The time-kill kinetic studies were performed in commercially available anaerobic blood culture bottles (7D bottle, Becton Dickinson). The anaerobic medium from these 7D bottles has been shown to support growth of B. fragilis group isolates to colonyforming units (CFU) of 109-101° over a 24-hr period, and it does not interfere with the bactericidal activity of antimicrobial agents (Stratton et al., 1987). When the anaerobic medium from 7D bottles is compared with anaerobic broth MIC for time-kill kinetic studies, the presence of anticoagulants in the 7D broth
The broth microdilution method was done as described by the National Committee for Clinical Laboratory Standards for anaerobic bacteria (NCCLS, 1985 and 1989). Serial twofold dilutions (0.125-256 ~g/ml) of each agent were prepared in anaerobic broth MIC medium. The final inoculum size was 105 CFU/well (10 6 CFU per ml). All microdilution trays were incubated anaerobically at 35°C for 48 hr. MICs were defined as the lowest concentration of antibiotic that inhibited visible growth and were determined in duplicate.
Time-Kill Kinetic Studies Time-kill kinetics were performed using methodology previously described (Stratton et al., 1987). Briefly, the bactericidal activity of each test antimicrobial agent was determined in commercially available anaerobic blood culture bottles (BACTEC 7D bottle) containing 30 ml of prereduced enriched soybean-casein digest broth. Concentrations studied ranged from 0.25 to 256 ~g/ml in twofold increments. Once killing was achieved (defined as ~3 log10 decrease in CFU/ml), higher concentrations of that agent were not routinely tested out to 256 ~g/ml. After addition of the logarithmically growing inoculum, the bottles were incubated on an orbital shaker (Becton Dickinson) at 35°C for 24 hr; this allowed growth in the control bottles to reach a density of 109-10 l° CFU/ml. Samples (0.5 ml) were taken at regular intervals (0, 4, 12, and 24 hr) and suitably diluted (neat to 10 -7) in prereduced saline, and a micropipetter was used to
Killing of Cefoxitin-Susceptible and -Resistant Bacteroides
TABLE 1
379
MICs (~g/ml) of Bacteroides fragilis Group Isolates Against Cefoxitin, Cefotetan, Clindamycin, and Metronidazole
Antimicrobial Agent
MIC to B. fragilis (ATCC 25285)
MIC to B. thetaiotaomicron ( A T C C29741)
Cefoxitin Cefotetan Clindamycin Metronidazole
4 8 1 1
8 16 2 2
Distribution of MICs Range of MICs (24 Isolates)
remove and drop 0.05-ml samples (five samples per each dilution) onto prereduced blood-agar plates. Drops were allowed to absorb without being streaked, the plates were incubated anaerobically at 35°C for 48 hr, and the colonies were counted and averaged after incubation. This m e t hod used for viable colonycount determinations has been determined to have a minimal accurately detectable number of 50 colonies per ml. Survivors at 24 hr were retested by the broth microdilution m et hod in order to determine whether the MIC has increased. The effect of antibiotic carryover was eliminated by restricting the highest antibiotic concentration tested to 256 ~g/ml and serially diluting test samples up to 10 -7 dilution. In separate experiments, the comparison of washed and unwashed cells showed no significant differences in colony counts. Each isolate was tested in duplicate against each concentration on at least two separate occasions in order to insure reproducibility.
4-128 1-128 0.125-128 0.25-2
64
0 0 7 7
0 1 4 11
0 0 3 6
3 4 4 0
4 3 0 0
5 4 0 0
5 5 0 0
7 7 6 0
Time-Kill Kinetic Results
Time-kill curves revealed metronidazole to be the most bactericidal agent tested, with all 26 isolates being killed by ~ 4 ~g/ml at 24 hr. Metronidazole was also the most rapidly bactericidal agent (Figure 1). There were 20 (77%) of 26 B. fragilis group strains considered susceptible to clindamycin (MICs ~ 4 ~g/ml); all were killed with concentrations of clindamycin ~ 16 ~g/ml at 24 hr. Clindamycin was much less rapidly bactericidal than was metronidazole
1010, 109
Control
10 s
0.25 IcJ/ml
107 106 ~i
0.5 I~g/ml
RESULTS
.~..10 4
Broth Microdilution Results
C
3
The range of MIC values for cefoxitin was 4-128 ~g/ml; nine of 26 isolates tested had MICs < 16 ~g/ml, five had MICs of 16 ~g/ml, five had MICs of 32 ~g/ml, and seven had MICs /> 64 p,g/ml. For cefotetan, the MICs were similar: a range of 1-128 ~g/ml with nine of 26 isolates tested having MICs < 16 ~g/ml, five with MICs of 16 ~g/ml, five with MICs of 32 ~g/ml, and seven with MICs/> 64 ~g/ml. The range of MICs for clindamycin was 0.125-128 ~g/ml with 20 of 26 isolates tested having MICs ~ 4 ~g/ml and six having MICs ~ 64 ~g/ml. Four of the six isolates resistant to clindamycin were also resistant to cefoxitin and cefotetan (MICs /> 64 p,g/ml), for metronidazole, the range of MICs was 0.25-2.0 with 18 isolates of 26 tested having MICs ~ 1 ~g/ml and eight with MICs of 2 ~g/ml. These results are summarized in Table 1.
103
1 pg/ml
102
2 i~g/ml
_o 0 o
4 I~g/ml
0 4 812
24 Time (hours)
FIGURE 1 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 ~g/ml for cefoxitin) of metronidazole (MIC = 0.5) for isolates of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for the average of four determinations is shown.
C.W. Stratton et al.
380
(Figure 2). There w e r e six isolates with clindamycin MICs/> 64 ~g/ml; n o n e were killed with concentrations as high as 256 ~g/ml at 24 hr. The bactericidal activity of cefoxitin and cefotetan was less than that seen with clindamycin and metronidazole. Of the 12 isolates with MICs I> 32 ~g/ml to the two cephamycins, seven (with MICs /> 64 ~g/ml), s h o w e d < 99.9% killing at 24 hr with concentrations as high as 256 p~g/ml for each agent (Figures 3 and 4). Five isolates with MICs equal to 32 ~g/ml showed ~ 99.9% killing at 24 hr with cefoxitin or cefotetan concentrations of 32 ~g/ml. Of the 14 isolates with MICs 16 ~g/ml to the two cephamycins, all s h o w e d 99.9% killing at 24 hr with concentrations of 32 ~g/ml. Cefoxitin and cefotetan for susceptible isolates were m o r e rapidly bactericidal than was clindamycin, but less so than was m e t r o n i d a z o l e (data not shown). Finally, of the six isolates resistant to clindamycin, four were also resistant to the two cephamycins. Table 2 s u m m a r i z e s the cumulative bactericidal activities of the four agents tested. W h e n survivors were retested b y the b r o t h microdilution method, the MICs w e r e essentially identical ( + 1 dilution) to the original MIC.
1010
DISCUSSION The susceptible breakpoint for B. fragilis g r o u p isolates to either cefoxitin or cefotetan is defined as an MIC ~ 32 ~g/ml. Yet, differences in resistance rates due to differences in anaerobic susceptibility testing m e t h o d o l o g y are well k n o w n (Aldridge et al., 1987 and 1990). To evaluate cefoxitin-resistant m e m b e r s of the B. fragilis g r o u p m o r e completely, we selected isolates with moderate-to-high MICs to cefoxitin (MICs ~ 32 p~g/ml) and c o m p a r e d the bactericidal activity of cefoxitin and cefotetan against these isolates with that of isolates with low MICs to cefoxitin (MICs ~ 16 ~g/ml). We also c o m p a r e d the activity of cefoxitin and cefotetan with that of clindamycin and metronidazole because of the possibility of multiple resistance to different classes of antibiotics. We used time-kill kinetic m e t h o d o l o g y because we have previously f o u n d this m e t h o d to be a useful m e a n s of assessing and c o m p a r i n g the antimicrobial activity of agents against m e m b e r s of the B. fragilis g r o u p (Stratton et al., 1987). In addition, bactericidal activity, although not essential, has b e e n s h o w n to be useful in the clinical t h e r a p y of serious infections
Control 8 I~g/ml
101°
Control
10 0
10 g 100
10 7
107
10 9
16 IAg/ml
32 r~cj/ml
10 6~
10 s~
0.25 [ug/ml
64 pg/ml
~ '10 5
'10 s
0.5 Ftg/ml
~,,~10 4
1 pg/ml
e.-
2 ug/ml
3 10 3 >., c o o
4 ug/ml
128 ~j/ml
~ 10 4
256 ~g/ml
r-
~ 10 3 co o
o
10 2
10 2
8 IJg/ml
0 4 8 12
24 Time (hours)
FIGURE 2 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 p,g/ml for cefoxitin) of clindamycin (MIC = 0.25) for isolates of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for the average of four determinations is shown.
o-;
12-
2'4 Time (hours)
FIGURE 3 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 ~g/ml for cefoxitin) to cefotetan (MIC -- 128) for isolates of Bacteroidesfragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for four determinations is shown.
Killing of Cefoxitin-Susceptible and -Resistant Bacteroides
1010
Control
lO 9
16 pg/ml
lO e 10 7
32 iJg/ml
lO 6~.
64 pg/ml 128 pg/ml 256 pg/ml
~'10 5 .~10 4 ¢.
8 >, tO o
103
o
lO 2
0 4 of 2
d4 Time (hours)
FIGURE 4 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 i~g/ml for cefoxitin) to cefoxitin for isolates of of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for four determinations is shown.
caused by B. fragilis (Galgiani et al., 1987). Finally, we compared the cumulative percent of isolates that were killed by each agent tested with those antimicrobial concentrations that would be expected in vivo in clinical situations during the pharmacokiTABLE
2
381
netic transition from peak to through concentrations. Overall, the two cephamycins appeared more active than might be predicted by their MIC values. Almost one-half of the isolates tested (12 of 26 strains) had MICs at or above the breakpoint MIC of 32 ~g/ml, yet 19 (73%) of 26 demonstrated t> 99.9% killing at 24 hr with concentrations of 32 t~g/ml. Such concentrations would be achieved clinically (Adam et al., 1981; Gerding et al., 1986). Notably, seven isolates with MICs t> 64 ~g/ml showed 99.9% killing at 24 hr with concentrations of cefoxitin or cefotetan as high as 256 i~g/ml. Clearly, there appears to be a marked difference between the activities of both cephamycins against B. fragilis isolates with strains having MICs I> 64 ~g/ml not being killed at concentrations as high as 256 ~g/ml. We have previously found that there is little correlation between low or moderate MIC values and bactericidal activity at 24 hr for B. fragilis group isolates (Stratton et al., 1987). A possible explanation for this phenomenon is that broth dilution MICs assessed over a 48-hr period are high due to regrowth of microorganisms following inactivation or degradation of antibiotic. In preliminary studies (C. Stratton, unpublished data), we have detected by high-performance liquid chromatography degradation of test antimicrobials in both inoculated and uninoculated test medium that is present at 24 hr and more pronounced at 48 hr. In earlier studies (Stratton et aI., 1987), and in this study, susceptibility testing of survivors has revealed essentially identical MIC values (_+ 1 dilution), suggesting that regrowth is not due to the development of resistance. Clindamycin demonstrated a correlation between
Cumulative Number (%) of 26 Bacteroides fragilis Group Isolates a Having/>99.9% Killingb of the Final Inoculum (5 x 10 6 CFU/ml) After 24 hr of Exposure to Selected Antibiotics at Clinically Expected Concentrations
Concentration c (txg/ml)
Cefoxitin
Cefotetan
Clindamycin
Metronidazole
0.5 1 2 4 8 16 32 64
0 (0%) 0 (0%) 1 (4%) 1 (4%) 5 (19%) 7 (27%) 19 (73%) NS
0 (0%) 0 (0%) 2 (8%) 2 (8%) 5 (19%) 8 (31%) 19 (73%) 19 (73%)
0 (0%) 4 (15%) 6 (23%) 9 (35%) 17 (65%) 20 (77%) 20 (77%) NS
0 (0%) 8 (31%) 23 (88%) 26 (100%) ND ND ND ND
ND, not determined. qncludes 24 clinicalisolates and two ATCCstrains. bKillingdetermined by time-killkinetic methodology. cConcentrationsof >32 or 64 I~g/mlnot shown (NS) as these would not be expected clinically.
382
MIC values a n d bactericidal activity with 20 of 20 susceptible isolates (MIC ~ 4 ~g/ml) s h o w i n g / > 99.9 killing at 24 hr with a concentration of 16 ~,g/ml. Six isolates with M I C s / > 64 ~g/ml, on the other h a n d , lacked killing at clindamycin concentrations as high as 256 ~g/ml. Concentrations of 16 ~,g/ml are readily achievable in clinical situations (Adam et al., 1981; Gerding et al., 1986). Interestingly, clindamycin d e m o n s t r a t e d slightly better bactericidal activity (albeit less rapid) against B. fragilis isolates than did the two cephamycins. We (Stratton et al., 1987) and others (Sutter and Finegold, 1975) have previously f o u n d that clindamycin does exhibit better bactericidal activity against B. fragilis than does cefoxitin. Finally, of the six isolates resistant to clindamycin, four were also resistant to the two c e p h a m y c i n s as j u d g e d by b o t h MICs i> 64 ~g/ml and the lack of killing at c e p h a m y c i n concentration as high as 256 ~g/ml. This confirms the previously described (Aidridge et al., 1988) multiple resistance to different classes of antibiotics. Metronidazole was the most active agent as judged by either MICs or by time-kill curves. In addition,
C.W. Stratton et al.
it was the most rapidly bactericidal agent tested. All isolates s h o w e d /> 99.9% killing at 24 hr with metronidazole concentrations of ~ 4 ~g/ml. Such concentrations can be achieved e v e n with oral metronidazole t h e r a p y (Adam et al., 1981; Gerding et al., 1986). Although m e t r o n i d a z o l e resistance has b e e n described (Lamothe et al., 1986), n o n e of these isolates were resistant b y either MIC values or by timekill curves. Of our isolates resistant to either cephamycins or clindamycin (or both), n o n e s h o w e d as m u c h as an increase in their MIC values for metronidazole. In s u m m a r y , high-level MICs (MICs i> 64 ~g/ml) for both cefoxitin and cefotetan as well as clindamycin are associated with a lack of killing by concentrations of d r u g well above those achieved in a clinical setting. Isolates that are resistant to the cephamycins are also likely to be resistant to clindamycin. Metronidazole remains the most active anaerobic agent of these four as j u d g e d by either MIC values or bactericidal activity and is an appropriate agent for use in B. fragilis infections w h e r e resistance to cephamycins or to clindamycin is suspected.
REFERENCES Adam D, Wilhelm K, Chysky V (1981) Antibiotic concentrations in blood and tissue. Arzneimittelforschung 31:19721976. Aldridge KE, Henderberg A, Schiro DD, Sanders CV (1988) Susceptibility of Bacteroidesfragilis group isolates to broadspectrum f~-lactams, dindamycin, and metronidazole: rates of resistance, cross-resistance, and importance of ~-lactamase production. Adv Ther 5:273-282. Aldridge KE, Sanders CV (1987) Antibiotic- and methoddependent variation in susceptibility testing results of Bacteroides fragilis group isolates. J Clin Microbiol 25:23172321. Aldridge KE, Wexler HM, Sanders CV, Finegold SM (1990) Comparison in vitro antibiograms of Bacteroides fragilis group isolates: differences in resistance rates in two institutions because of differences in susceptibility testing methodology. Antimicrob Agents Chemother 34:179181. Betriu C, Campos E, Cabronero C, Rodrignez-Avial C, Picazo JJ (1990) Susceptibility of species of Bacteroides fragilis group to 10 antimicrobial agents. Antimicrob Agents Chemother 34:671-673. Finegold SM, Edelstein MAC (1985): Gram-negative, nonspore-forming anaerobic bacilli. In Manual of Clinical Microbiology, 4th ed. Eds, EH Lennettee, A Balows, WJ Hausler, and HJ Shadomy. Washington DC: American Society for Microbiology, pp 450-460. Galgiani JN, Busch DF, Brass C, Rumans LW, Mangels JI, Stevens DA (1987) Bacteroides fragilis endocarditis, bacteremia and other infections treated with oral or intravenous metronidazole. Am J Med 65:284289. Gerding DN, Peterson LR, Hughes CE, Bamberger DM (1986) Extravascular antimicrobial distribution in man. In Antibiotics in Laboratory Medicine, 2nd ed. Ed,
V Lorian. Baltimore, MD: Williams and Wilkins, pp 938-994. Lamothe F, Fijalkowski C, Malovin F, Bourgault AM, Delorme L (1986) Bacteroides fragilis resistant to both metronidazole and imipenem. J Antimicrob Chemother 18:642643. Miles AA, Misra SS (1938) The estimation of the bactericidal activity of the blood. J Hygiene 38:732-739. National Committee for Clinical Laboratory Standards (NCCLS) (1985) Alternative Methods for Antimicrobial Susceptibility Testing of Anaerobic Bacteria. Proposed Guideline. NCCLS document M17-P. Villanova, PA: NCCLS. National Committee for Clinical Laboratory Standards (NCCLS) (1989) Methods for Antimicrobial Susceptibility Testing of Anaerobic Bacteria. Tentative Standard. NCCLS document M11-T2. Villanova, PA: NCCLS. Soriano F, Ponte C, Wilhelmi I (1984) Increasing incidence of clinical isolates of Bacteroides fragilis resistant to clindamycin. J Antimicrob Chemother 13:395. Sosa A, Tally FP, Jacobs NV, Gorbach SL (1982) Bacteroides fragilis resistant to clindamycin in vitro. Antimicrob Agents Chemother 22:771-774. Stratton CW, Weeks LS, Aldridge KE (1987) Comparison of kill kinetic studies with agar and broth microdilution methods for determination of antimicrobial activity of selected agents against members of the Bacteroides fragilis group. J Clin Microbiol 25:645-649. Sutter VL, Finegold SM (1975) Susceptibility of anaerobic bacteria to carbenicillin, cefoxitin, and related drugs. J Infect Dis 131:417-422. Tally FP, Cuchural GJ Jr, Jacobus NV et al. (1985) Nationwide study of the susceptibility of the Bacteroides fragilis group in the United States. Antimicrob Agents Chemother 28:675-677.
377
ANTIMICROBIAL SUSCEPTIBILITY STUDIES
Comparison of the Bactericidal Activity of Clindamycin and Metronidazole Against CefoxitinSusceptible and Cefoxitin-Resistant Isolates of the Bacteroides fragilis Group Charles W. Stratton, Lyndell S. Weeks, and Kenneth E. Aldridge
Time-kill kinetic methodology was used to evaluate the bactericidal activity of cefoxitin, cefotetan, clindamycin, and metronidizole against cefoxitin-susceptible and cefoxitin-resistant isolates of the Bacteroides fragilis group. Overall, metronidazole was the most bactericidal agent, with all isolates being killed with ~ 4 I~g/ml at 24 hr. Clindamycin was the next most bactericidal agent, with 20 of 26 isolates being killed with ~ 16 I~g/ml. Six isolates with clindamycin MICs ~ 64 i~g/ml were not killed at 24 hr, with concentrations as high as 256 ~g/ml. Cefoxitin and cefotetan were the least bactericidal
agents tested. Seven isolates with MICs of ~ 64 ~g/ml to each agent demonstrated a lack of killing at 24 hr, with concentrations of the respective agent as high as 256 i~g/ml. At concentrations with either agent of 32 la,g/ml, the remaining 19 isolates were killed at 24 hr. Of the six B. fragilis isolates resistant to clindamycin, four were also resistant to both cefoxitin and cefotetan. We conclude that in hospitals with cefoxitin-resistant B. fragilis group isolates, metronidazole would provide appropriate therapy.
INTRODUCTION
has been an increase in the incidence of B. fragilis resistant to clindamycin (Soriano et al., 1984; Sosa et al., 1982). Data from these isolates, moreover, indicate that c e p h a m y c i n resistance often is accompanied by increased resistance rates to other antimicrobial agents such as clindamycin. Finally, alt h o u g h rare, resistance to m e t r o n i d a z o l e has b e e n described (Lamothe et al., 1986). In this study, we used time-kill kinetic m e t h o d o l o g y to evaluate the bactericidal activity of two c e p h a m y c i n s (cefoxitin and cefotetan) using isolates of B. fragilis with low m e a n inhibitory concentrations (MICs) to cefoxitin (broth-dilution MIC ~ 32 ~g/ml) a n d isolates with moderate-to-high MICs to cefoxitin (MIC/-- 32 }zg/ml) and c o m p a r e d this activity with that of clindamycin and metronidazole. We also c o m p a r e d this bactericidal activity with those concentrations of antibiotic that w o u l d be expected in vivo.
During the past decade, m e m b e r s of the Bacteroides fragilis g r o u p have s h o w n an increase in resistance to c e p h a m y c i n s such as cefoxitin and cefotetan (Betriu et al., 1990; Tally et al., 1985). In addition, there
From the Department of Pathology (C.W.S., L.S.W.), Vanderbilt University School of Medicine, Nashville, Tennessee; and the Department of Medicine (K.E.A.), Louisiana State University School of Medicine, New Orleans, Louisiana USA. Address reprint requests to Dr. C.W. Stratton, Department of Pathology, C-3217 MCN, Vanderbilt University Medical Center, Nashville, TN 37232-2561, USA. Received 23 July 1990; revised and accepted 1 November 1990. © 1991 Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010 0732-8893/91/$3.50
378
MATERIALS A N D M E T H O D S Microorganisms A total of 12 members of the B. fragilis group with moderate-to-high MICs to cefoxitin were chosen for testing. Each isolate had a broth microdilution MIC value ~ 32 ~g/ml. This test cohort was comprised of the following: B. fragilis, two strains; B. thetaiotaomicron, five strains; B. distasonis, two strains; and B. ovatus, three strains. A second cohort of 12 isolates of B. fragilis with low MICs to cefoxitin (MIC values for cefoxitin ~ 32 p~g/ml) were also tested. In addition, B. fragilis American Type Culture Collection (ATCC) 25285 and B. thetaiotaomicron ATCC 29741 were used for quality control. Each clinical isolate (n = 24) was identified to species level by use of selective media, biochemical profiles, and gas-liquid chromatography (Finegold and Edelstein, 1985). Stock cultures were maintained at - 70°C and cloned twice on prereduced sheep blood agar before use.
C.W. Stratton et al.
has not influenced the bactericidal activity of antimicrobial agents.
Inoculum Portions of five or more colonies on a plate culture were inoculated into prereduced brain-heart infusion broth and incubated. The inoculum was prepared from a logarithmic phase of this culture with a final inoculum size of 106 CFU/ml. The inoculum size was carefully controlled and the final inoculum size was confirmed each time by a modified surface colony count method (Miles and Misra, 1938). For this method, a sample was removed and diluted (undiluted to 10-7) in prereduced saline, and plated (0.05 ml) in duplicate onto prereduced blood-agar plates. These plates were incubated anaerobically for 48 hr, colonies were counted and averaged, and the final inoculum was calculated. If the final inoculum was not between 1 x 10 6 and 1 x 107, then the test was repeated.
Antimicrobial Agents Broth Microdilution Method Standard powders were obtained from their respective manufacturers as follows: cefoxitin (Merck Sharp and Dohme, Rahway, NJ), cefotetan (Stuart, Wilmington, DE), clindamycin (Upjohn, Kalamazoo, MI), and metronidazole (Searle Laboratories, Chicago, IL). Antimicrobial stock solutions were prepared according to the manufacturer's instructions and stored at - 70°C until use. Final concentrations were prepared on the day that they were used.
Media Anaerobic broth MIC (Difco Laboratories, Detroit, MI) was used for the broth microdilution method. This media is supplemented with vitamin K1 (1 ~g/rnl) and heroin (5 ~g/ml) and has a formulation similar to that of the Wilkins-Chalgren medium, but lacks the agar. Previous studies (C. Stratton, unpublished data) have shown good correlation (+ 1 dilution) of MIC determinations done with anaerobic broth MIC and anaerobic broth medium from the BACTEC 7D blood culture bottle (Becton Dickinson, Towson, MD). The time-kill kinetic studies were performed in commercially available anaerobic blood culture bottles (7D bottle, Becton Dickinson). The anaerobic medium from these 7D bottles has been shown to support growth of B. fragilis group isolates to colonyforming units (CFU) of 109-101° over a 24-hr period, and it does not interfere with the bactericidal activity of antimicrobial agents (Stratton et al., 1987). When the anaerobic medium from 7D bottles is compared with anaerobic broth MIC for time-kill kinetic studies, the presence of anticoagulants in the 7D broth
The broth microdilution method was done as described by the National Committee for Clinical Laboratory Standards for anaerobic bacteria (NCCLS, 1985 and 1989). Serial twofold dilutions (0.125-256 ~g/ml) of each agent were prepared in anaerobic broth MIC medium. The final inoculum size was 105 CFU/well (10 6 CFU per ml). All microdilution trays were incubated anaerobically at 35°C for 48 hr. MICs were defined as the lowest concentration of antibiotic that inhibited visible growth and were determined in duplicate.
Time-Kill Kinetic Studies Time-kill kinetics were performed using methodology previously described (Stratton et al., 1987). Briefly, the bactericidal activity of each test antimicrobial agent was determined in commercially available anaerobic blood culture bottles (BACTEC 7D bottle) containing 30 ml of prereduced enriched soybean-casein digest broth. Concentrations studied ranged from 0.25 to 256 ~g/ml in twofold increments. Once killing was achieved (defined as ~3 log10 decrease in CFU/ml), higher concentrations of that agent were not routinely tested out to 256 ~g/ml. After addition of the logarithmically growing inoculum, the bottles were incubated on an orbital shaker (Becton Dickinson) at 35°C for 24 hr; this allowed growth in the control bottles to reach a density of 109-10 l° CFU/ml. Samples (0.5 ml) were taken at regular intervals (0, 4, 12, and 24 hr) and suitably diluted (neat to 10 -7) in prereduced saline, and a micropipetter was used to
Killing of Cefoxitin-Susceptible and -Resistant Bacteroides
TABLE 1
379
MICs (~g/ml) of Bacteroides fragilis Group Isolates Against Cefoxitin, Cefotetan, Clindamycin, and Metronidazole
Antimicrobial Agent
MIC to B. fragilis (ATCC 25285)
MIC to B. thetaiotaomicron ( A T C C29741)
Cefoxitin Cefotetan Clindamycin Metronidazole
4 8 1 1
8 16 2 2
Distribution of MICs Range of MICs (24 Isolates)
remove and drop 0.05-ml samples (five samples per each dilution) onto prereduced blood-agar plates. Drops were allowed to absorb without being streaked, the plates were incubated anaerobically at 35°C for 48 hr, and the colonies were counted and averaged after incubation. This m e t hod used for viable colonycount determinations has been determined to have a minimal accurately detectable number of 50 colonies per ml. Survivors at 24 hr were retested by the broth microdilution m et hod in order to determine whether the MIC has increased. The effect of antibiotic carryover was eliminated by restricting the highest antibiotic concentration tested to 256 ~g/ml and serially diluting test samples up to 10 -7 dilution. In separate experiments, the comparison of washed and unwashed cells showed no significant differences in colony counts. Each isolate was tested in duplicate against each concentration on at least two separate occasions in order to insure reproducibility.
4-128 1-128 0.125-128 0.25-2
64
0 0 7 7
0 1 4 11
0 0 3 6
3 4 4 0
4 3 0 0
5 4 0 0
5 5 0 0
7 7 6 0
Time-Kill Kinetic Results
Time-kill curves revealed metronidazole to be the most bactericidal agent tested, with all 26 isolates being killed by ~ 4 ~g/ml at 24 hr. Metronidazole was also the most rapidly bactericidal agent (Figure 1). There were 20 (77%) of 26 B. fragilis group strains considered susceptible to clindamycin (MICs ~ 4 ~g/ml); all were killed with concentrations of clindamycin ~ 16 ~g/ml at 24 hr. Clindamycin was much less rapidly bactericidal than was metronidazole
1010, 109
Control
10 s
0.25 IcJ/ml
107 106 ~i
0.5 I~g/ml
RESULTS
.~..10 4
Broth Microdilution Results
C
3
The range of MIC values for cefoxitin was 4-128 ~g/ml; nine of 26 isolates tested had MICs < 16 ~g/ml, five had MICs of 16 ~g/ml, five had MICs of 32 ~g/ml, and seven had MICs /> 64 p,g/ml. For cefotetan, the MICs were similar: a range of 1-128 ~g/ml with nine of 26 isolates tested having MICs < 16 ~g/ml, five with MICs of 16 ~g/ml, five with MICs of 32 ~g/ml, and seven with MICs/> 64 ~g/ml. The range of MICs for clindamycin was 0.125-128 ~g/ml with 20 of 26 isolates tested having MICs ~ 4 ~g/ml and six having MICs ~ 64 ~g/ml. Four of the six isolates resistant to clindamycin were also resistant to cefoxitin and cefotetan (MICs /> 64 p,g/ml), for metronidazole, the range of MICs was 0.25-2.0 with 18 isolates of 26 tested having MICs ~ 1 ~g/ml and eight with MICs of 2 ~g/ml. These results are summarized in Table 1.
103
1 pg/ml
102
2 i~g/ml
_o 0 o
4 I~g/ml
0 4 812
24 Time (hours)
FIGURE 1 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 ~g/ml for cefoxitin) of metronidazole (MIC = 0.5) for isolates of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for the average of four determinations is shown.
C.W. Stratton et al.
380
(Figure 2). There w e r e six isolates with clindamycin MICs/> 64 ~g/ml; n o n e were killed with concentrations as high as 256 ~g/ml at 24 hr. The bactericidal activity of cefoxitin and cefotetan was less than that seen with clindamycin and metronidazole. Of the 12 isolates with MICs I> 32 ~g/ml to the two cephamycins, seven (with MICs /> 64 ~g/ml), s h o w e d < 99.9% killing at 24 hr with concentrations as high as 256 p~g/ml for each agent (Figures 3 and 4). Five isolates with MICs equal to 32 ~g/ml showed ~ 99.9% killing at 24 hr with cefoxitin or cefotetan concentrations of 32 ~g/ml. Of the 14 isolates with MICs 16 ~g/ml to the two cephamycins, all s h o w e d 99.9% killing at 24 hr with concentrations of 32 ~g/ml. Cefoxitin and cefotetan for susceptible isolates were m o r e rapidly bactericidal than was clindamycin, but less so than was m e t r o n i d a z o l e (data not shown). Finally, of the six isolates resistant to clindamycin, four were also resistant to the two cephamycins. Table 2 s u m m a r i z e s the cumulative bactericidal activities of the four agents tested. W h e n survivors were retested b y the b r o t h microdilution method, the MICs w e r e essentially identical ( + 1 dilution) to the original MIC.
1010
DISCUSSION The susceptible breakpoint for B. fragilis g r o u p isolates to either cefoxitin or cefotetan is defined as an MIC ~ 32 ~g/ml. Yet, differences in resistance rates due to differences in anaerobic susceptibility testing m e t h o d o l o g y are well k n o w n (Aldridge et al., 1987 and 1990). To evaluate cefoxitin-resistant m e m b e r s of the B. fragilis g r o u p m o r e completely, we selected isolates with moderate-to-high MICs to cefoxitin (MICs ~ 32 p~g/ml) and c o m p a r e d the bactericidal activity of cefoxitin and cefotetan against these isolates with that of isolates with low MICs to cefoxitin (MICs ~ 16 ~g/ml). We also c o m p a r e d the activity of cefoxitin and cefotetan with that of clindamycin and metronidazole because of the possibility of multiple resistance to different classes of antibiotics. We used time-kill kinetic m e t h o d o l o g y because we have previously f o u n d this m e t h o d to be a useful m e a n s of assessing and c o m p a r i n g the antimicrobial activity of agents against m e m b e r s of the B. fragilis g r o u p (Stratton et al., 1987). In addition, bactericidal activity, although not essential, has b e e n s h o w n to be useful in the clinical t h e r a p y of serious infections
Control 8 I~g/ml
101°
Control
10 0
10 g 100
10 7
107
10 9
16 IAg/ml
32 r~cj/ml
10 6~
10 s~
0.25 [ug/ml
64 pg/ml
~ '10 5
'10 s
0.5 Ftg/ml
~,,~10 4
1 pg/ml
e.-
2 ug/ml
3 10 3 >., c o o
4 ug/ml
128 ~j/ml
~ 10 4
256 ~g/ml
r-
~ 10 3 co o
o
10 2
10 2
8 IJg/ml
0 4 8 12
24 Time (hours)
FIGURE 2 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 p,g/ml for cefoxitin) of clindamycin (MIC = 0.25) for isolates of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for the average of four determinations is shown.
o-;
12-
2'4 Time (hours)
FIGURE 3 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 ~g/ml for cefoxitin) to cefotetan (MIC -- 128) for isolates of Bacteroidesfragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for four determinations is shown.
Killing of Cefoxitin-Susceptible and -Resistant Bacteroides
1010
Control
lO 9
16 pg/ml
lO e 10 7
32 iJg/ml
lO 6~.
64 pg/ml 128 pg/ml 256 pg/ml
~'10 5 .~10 4 ¢.
8 >, tO o
103
o
lO 2
0 4 of 2
d4 Time (hours)
FIGURE 4 Representative killing curve (Bacteroides ovatus, strain PN233, MIC = 128 i~g/ml for cefoxitin) to cefoxitin for isolates of of Bacteroides fragilis group that are cefoxitin resistant. Survivors at 24 hr had an MIC within one dilution of the original MIC. Reproducibility for four determinations is shown.
caused by B. fragilis (Galgiani et al., 1987). Finally, we compared the cumulative percent of isolates that were killed by each agent tested with those antimicrobial concentrations that would be expected in vivo in clinical situations during the pharmacokiTABLE
2
381
netic transition from peak to through concentrations. Overall, the two cephamycins appeared more active than might be predicted by their MIC values. Almost one-half of the isolates tested (12 of 26 strains) had MICs at or above the breakpoint MIC of 32 ~g/ml, yet 19 (73%) of 26 demonstrated t> 99.9% killing at 24 hr with concentrations of 32 t~g/ml. Such concentrations would be achieved clinically (Adam et al., 1981; Gerding et al., 1986). Notably, seven isolates with MICs t> 64 ~g/ml showed 99.9% killing at 24 hr with concentrations of cefoxitin or cefotetan as high as 256 i~g/ml. Clearly, there appears to be a marked difference between the activities of both cephamycins against B. fragilis isolates with strains having MICs I> 64 ~g/ml not being killed at concentrations as high as 256 ~g/ml. We have previously found that there is little correlation between low or moderate MIC values and bactericidal activity at 24 hr for B. fragilis group isolates (Stratton et al., 1987). A possible explanation for this phenomenon is that broth dilution MICs assessed over a 48-hr period are high due to regrowth of microorganisms following inactivation or degradation of antibiotic. In preliminary studies (C. Stratton, unpublished data), we have detected by high-performance liquid chromatography degradation of test antimicrobials in both inoculated and uninoculated test medium that is present at 24 hr and more pronounced at 48 hr. In earlier studies (Stratton et aI., 1987), and in this study, susceptibility testing of survivors has revealed essentially identical MIC values (_+ 1 dilution), suggesting that regrowth is not due to the development of resistance. Clindamycin demonstrated a correlation between
Cumulative Number (%) of 26 Bacteroides fragilis Group Isolates a Having/>99.9% Killingb of the Final Inoculum (5 x 10 6 CFU/ml) After 24 hr of Exposure to Selected Antibiotics at Clinically Expected Concentrations
Concentration c (txg/ml)
Cefoxitin
Cefotetan
Clindamycin
Metronidazole
0.5 1 2 4 8 16 32 64
0 (0%) 0 (0%) 1 (4%) 1 (4%) 5 (19%) 7 (27%) 19 (73%) NS
0 (0%) 0 (0%) 2 (8%) 2 (8%) 5 (19%) 8 (31%) 19 (73%) 19 (73%)
0 (0%) 4 (15%) 6 (23%) 9 (35%) 17 (65%) 20 (77%) 20 (77%) NS
0 (0%) 8 (31%) 23 (88%) 26 (100%) ND ND ND ND
ND, not determined. qncludes 24 clinicalisolates and two ATCCstrains. bKillingdetermined by time-killkinetic methodology. cConcentrationsof >32 or 64 I~g/mlnot shown (NS) as these would not be expected clinically.
382
MIC values a n d bactericidal activity with 20 of 20 susceptible isolates (MIC ~ 4 ~g/ml) s h o w i n g / > 99.9 killing at 24 hr with a concentration of 16 ~,g/ml. Six isolates with M I C s / > 64 ~g/ml, on the other h a n d , lacked killing at clindamycin concentrations as high as 256 ~g/ml. Concentrations of 16 ~,g/ml are readily achievable in clinical situations (Adam et al., 1981; Gerding et al., 1986). Interestingly, clindamycin d e m o n s t r a t e d slightly better bactericidal activity (albeit less rapid) against B. fragilis isolates than did the two cephamycins. We (Stratton et al., 1987) and others (Sutter and Finegold, 1975) have previously f o u n d that clindamycin does exhibit better bactericidal activity against B. fragilis than does cefoxitin. Finally, of the six isolates resistant to clindamycin, four were also resistant to the two c e p h a m y c i n s as j u d g e d by b o t h MICs i> 64 ~g/ml and the lack of killing at c e p h a m y c i n concentration as high as 256 ~g/ml. This confirms the previously described (Aidridge et al., 1988) multiple resistance to different classes of antibiotics. Metronidazole was the most active agent as judged by either MICs or by time-kill curves. In addition,
C.W. Stratton et al.
it was the most rapidly bactericidal agent tested. All isolates s h o w e d /> 99.9% killing at 24 hr with metronidazole concentrations of ~ 4 ~g/ml. Such concentrations can be achieved e v e n with oral metronidazole t h e r a p y (Adam et al., 1981; Gerding et al., 1986). Although m e t r o n i d a z o l e resistance has b e e n described (Lamothe et al., 1986), n o n e of these isolates were resistant b y either MIC values or by timekill curves. Of our isolates resistant to either cephamycins or clindamycin (or both), n o n e s h o w e d as m u c h as an increase in their MIC values for metronidazole. In s u m m a r y , high-level MICs (MICs i> 64 ~g/ml) for both cefoxitin and cefotetan as well as clindamycin are associated with a lack of killing by concentrations of d r u g well above those achieved in a clinical setting. Isolates that are resistant to the cephamycins are also likely to be resistant to clindamycin. Metronidazole remains the most active anaerobic agent of these four as j u d g e d by either MIC values or bactericidal activity and is an appropriate agent for use in B. fragilis infections w h e r e resistance to cephamycins or to clindamycin is suspected.
REFERENCES Adam D, Wilhelm K, Chysky V (1981) Antibiotic concentrations in blood and tissue. Arzneimittelforschung 31:19721976. Aldridge KE, Henderberg A, Schiro DD, Sanders CV (1988) Susceptibility of Bacteroidesfragilis group isolates to broadspectrum f~-lactams, dindamycin, and metronidazole: rates of resistance, cross-resistance, and importance of ~-lactamase production. Adv Ther 5:273-282. Aldridge KE, Sanders CV (1987) Antibiotic- and methoddependent variation in susceptibility testing results of Bacteroides fragilis group isolates. J Clin Microbiol 25:23172321. Aldridge KE, Wexler HM, Sanders CV, Finegold SM (1990) Comparison in vitro antibiograms of Bacteroides fragilis group isolates: differences in resistance rates in two institutions because of differences in susceptibility testing methodology. Antimicrob Agents Chemother 34:179181. Betriu C, Campos E, Cabronero C, Rodrignez-Avial C, Picazo JJ (1990) Susceptibility of species of Bacteroides fragilis group to 10 antimicrobial agents. Antimicrob Agents Chemother 34:671-673. Finegold SM, Edelstein MAC (1985): Gram-negative, nonspore-forming anaerobic bacilli. In Manual of Clinical Microbiology, 4th ed. Eds, EH Lennettee, A Balows, WJ Hausler, and HJ Shadomy. Washington DC: American Society for Microbiology, pp 450-460. Galgiani JN, Busch DF, Brass C, Rumans LW, Mangels JI, Stevens DA (1987) Bacteroides fragilis endocarditis, bacteremia and other infections treated with oral or intravenous metronidazole. Am J Med 65:284289. Gerding DN, Peterson LR, Hughes CE, Bamberger DM (1986) Extravascular antimicrobial distribution in man. In Antibiotics in Laboratory Medicine, 2nd ed. Ed,
V Lorian. Baltimore, MD: Williams and Wilkins, pp 938-994. Lamothe F, Fijalkowski C, Malovin F, Bourgault AM, Delorme L (1986) Bacteroides fragilis resistant to both metronidazole and imipenem. J Antimicrob Chemother 18:642643. Miles AA, Misra SS (1938) The estimation of the bactericidal activity of the blood. J Hygiene 38:732-739. National Committee for Clinical Laboratory Standards (NCCLS) (1985) Alternative Methods for Antimicrobial Susceptibility Testing of Anaerobic Bacteria. Proposed Guideline. NCCLS document M17-P. Villanova, PA: NCCLS. National Committee for Clinical Laboratory Standards (NCCLS) (1989) Methods for Antimicrobial Susceptibility Testing of Anaerobic Bacteria. Tentative Standard. NCCLS document M11-T2. Villanova, PA: NCCLS. Soriano F, Ponte C, Wilhelmi I (1984) Increasing incidence of clinical isolates of Bacteroides fragilis resistant to clindamycin. J Antimicrob Chemother 13:395. Sosa A, Tally FP, Jacobs NV, Gorbach SL (1982) Bacteroides fragilis resistant to clindamycin in vitro. Antimicrob Agents Chemother 22:771-774. Stratton CW, Weeks LS, Aldridge KE (1987) Comparison of kill kinetic studies with agar and broth microdilution methods for determination of antimicrobial activity of selected agents against members of the Bacteroides fragilis group. J Clin Microbiol 25:645-649. Sutter VL, Finegold SM (1975) Susceptibility of anaerobic bacteria to carbenicillin, cefoxitin, and related drugs. J Infect Dis 131:417-422. Tally FP, Cuchural GJ Jr, Jacobus NV et al. (1985) Nationwide study of the susceptibility of the Bacteroides fragilis group in the United States. Antimicrob Agents Chemother 28:675-677.
