Journal of Antimicrobial Chemotherapy (1996) 37, 233-242
Quantitative comparison in vitro of mutational antibiotic resistance of
Enterobacter spp. using a spiral plater
Chen M. Yu', Joseph W. Chow' and Victor L. Yu**
°VA Medical Center and University of Pittsburgh, Pittsburgh, PA;bWayne State Medical
Center, Detroit, MI, USA
The presence of spontaneous mutational antibiotic resistance among 18 bacteremic
isolates of Enterobacter spp. to cefotaxime, ceftazidime, gentamicin, amikacin,
ciprofloxacin, and trimethoprim-sulfamethoxazole was determined quantitatively
in vitro using a spiral plater. Each drug was delivered using the device and the agar
plates were inoculated in radial streaks. The degree of resistance was estimated by
dividing the antimicrobial concentration required to inhibit 90% of the colonies
growing in the area beyond the MIC by the MIC itself. The degree of resistance to
third-generation cephalosporins and aztreonam was statistically significantly greater
than that to co-trimoxazole, imipenem, and ciprofloxacin (P < 0.01); the latter three
antibiotics showed virtually no mutational resistance. An intermediate level of
resistance was induced by aminoglycosides, and mutational resistance to piperacillin
varied between this and the higher levels observed for the cephalosporins. By
providing a simple and efficient means of detecting spontaneous mutational
resistance, the spiral plater may prove useful in identifying those antimicrobial agents
which induce few or no mutants and therefore may be more likely to be successful
in treating infections due to Enterobacter spp.
Introduction
The spiral plater is an instrument that can determine the MIC in vitro more precisely
than can conventional dilution methods (Wexler el al., 1991; Spiral System Instruments,
Inc., data on file). The device allows a small amount of sample to be distributed in an
Archimedes spiral on the surface of a rotating agar plate thereby affording continuous,
logarithmic dilution rather than incremental dilution. Although more often used to
enumerate bacteria, the spiral plater can be used to distribute antibiotics such that a
concentration gradient is achieved allowing MICs to be determined. Bacteria are
inoculated as radial streaks onto the agar surface (Figure 1). After incubation, the
endpoint is determined by the transition from growth to no growth and provides the
basis for calculating the MIC. The presence of single colonies within the area of no
growth are likely to represent mutants and can be easily selected for further study.
Moreover, using the spiral plater in this way allows the frequency of spontaneous
mutations to be estimated.
Address for correspondence; Dr Victor L. Yu, University of Pittsburgh, Division of Infectious Disease,
501 Kaufmann Building, Pittsburgh, PA 15213, USA.
0305-7453/96/020233 + 10 $12.00/0
233
-£, 1996 The British Society for Antimicrobial Chemotherapy
234
C. M. Yu et al.
The emergence of resistant Enterobacter spp. and other aerobic Gram-negative
bacilli has been frequently observed to occur during antibiotic therapy, despite initial
susceptibility in vitro. In a national collaborative study conducted in the USA of
consecutive patients with bacteremia due to Enterobacter spp., resistance emerged 4-18
days after starting treatment with third generation cephalosporins. The strains were
shown to be identical by molecular typing, and MICs of the cephalosporins were
4-32-fold greater than those of the original strain (Chow, et al, 1991). The spiral plating
technique seemed ideally suited to investigating strains of Enterobacter spp. for the
presence of spontaneous resistant mutants and this quantitative in-vitro approach may
prove potentially valuable in selecting optimal antibiotic therapy for infections due to
these bacteria.
Materials and methods
Isolates
Eighteen isolates of Enterobacter spp. had been recovered from blood cultures of
individual patients before treatment had begun with a third generation cephalosporin.
Nine Enterobacter cloacae and three Enterobacter aerogenes had been recovered from
the blood of 12 patients who had been treated successfully and a further five isolates
of E. cloacae and one of E. aerogenes had been isolated from blood cultures of six
patients who ultimately experienced a recurrence of infection due to the emergence of
resistance of the original strain. Each isolate had been maintained at — 70°C and was
first recovered on blood agar incubated at 35°C overnight then stored at — 20°C on
Tryptic Soy agar slants.
Antimicrobial agents
Reagent grade powders of cefotaxime (Hoechst-Roussel, Somerville, NJ, USA),
ceftazidime (Glaxo Pharmaceuticals, Research Triangle Park, NC, USA), amikacin and
aztreonam (Bristol Myers Squibb, Princeton, NJ, USA), imipenem (Merck and
Company, West Point, PA, USA), piperacillin (Lederle Laboratories, Wayne, NJ,
USA), gentamicin (Schering, Kenilworth, NJ, USA), ciprofloxacin (Miles Laboratories,
West Haven, CT, USA), and co-trimoxazole (TMP-SMZ) in a ratio of 1:19 (Burroughs
Wellcome, Research Triangle Park, NC, USA) were prepared according to the
manufacturer's instructions. Stock concentrations were calculated using the Casba II
software supplied with the spiral plater (Spiral Biotech, Bethesda, MD, USA) to cover
an agar concentration range of 1/2 to 2 times the target MIC which was selected on
the basis of in-vitro susceptibilities done previously using broth dilution methods or
from susceptibility data taken from the literature.
Media
Mueller-Hinton agar in 15 cm Petri plates (Remel, Lenexa, KS, USA) as used for the
spiral plater. To test TMP-SMZ, the medium was supplemented with 100 units/L
thymidine phosphorylase in order to eliminate thymidine.
Mutational antibiotic resistance of Enterobacter spp.
235
Procedure
The inoculum was prepared in Mueller Hinton broth from the slants by growing in a
shaker incubator at 37°C for 2-5 h and adjusting the density to c. lO^cfu/mL using
0.5 McFarland's standard. Each antimicrobial agent was delivered onto the surface of
the Mueller Hinton agar using the spiral plater. One hour later, the plates were
inoculated with a swab in a radial streak as described in the manufacturer's guide
(Figure 1). Each strain was tested in quadriplicate for its susceptibility to each agent
by inoculating it twice on each of two plates. The plates were then incubated at 37°C
for 24 h.
Reading and interpretation of results
The point at which confluent growth ceased was taken as the MIC and is conceptually
identical to the minimal activity concentration (MAC) described in the manufacturer's
guide. The number of distinct colonies (outliers) appearing beyond this point was
counted for each strain. The MIC90 was defined as the concentration below which 90%
of outliers occurred and was chosen to minimize any skew that would be created by
an aberrant colony. The degree of resistance (DR) indicates the extent to which outliers
are resistant and was calculated by dividing the MIC90 by the MIC. The smaller the
value of DR, the lower the degree of resistance with the lowest possible value being 1.
The Casba II software package supplied with the spiral plater requires the tail ending
concentration (TEC) to be determined in order to calculate the activity index (AI) by
subtracting the lo& MIC from the Iog2 TEC. Since the tail is the area of growth that
extends beyond confluent growth and includes all the outliers, the AI is essentially
equivalent to the DR. However, the AI is more sensitive to being distorted by the
presence of a single colony growing at an extremely high concentration of drug, whereas
the DR is more representative and stable because it employs the MIG>o
Figure 1. A spiral plate inoculated with radial streaks of Enterobacicr spp after overnight incubation. The
drug has been dispensed in an Archimedes spiral with the highest drug concentration in the center of the
plate.
236
C. M. Yn et al.
Statistical analysis
Differences in the frequency of mutational resistance to each antimicrobial agent as
measured by the DR were compared using the Friedman rank test.
Results
Three patterns of growth were apparent using the spiral plater to test the susceptibility
of Enterobacter spp. Strains exhibiting growth pattern 1 produced numerous large
colonies at concentrations ^ 3 x MIC equivalent to a DR of ^ 3 (Figures 2 and 4(a)).
Growth pattern 2 was characterized by the presence of outlying colonies which were
smaller and fewer in number than those seen in growth pattern 1 at antimicrobial
concentrations between the MIC and 3 x MIC for which the corresponding DRs were
^1 to ^ 3 (Figure 4b). Lastly, growth pattern 3 was marked by a clean endpoint with
no outlier colonies growing beyond the MIC (Figures 3 and 4(c)). Enterobacter
spp. exposed to cefotaxime and aztreonam produced growth pattern 1; aminoglycosides
produced growth pattern 2, and imipenem, co-trimoxazole, and ciprofloxacin produced
growth pattern 3. When the outlier colonies seen after exposure to cefotaxime and
Figure 2. 15 cm Mueller-Hinton agar spiral plate demonstrating antibiotic mutational resistance of E.
cloacae to cefotaxime. Both organisms and the antibiotic: were applied using the spiral plates and the highest
concentration of drug is in the centre of the spiral. The MIC is the concentration at which confluent growth
of the organism ends (black, flared arrow). Note the number of isolated colonies (outliers) that grow at
concentrations above the MIC and that those nearest the centre of the plate are the most resistant colonies
(black arrowheads).
Mutational antibiotic resistance of Enterobacter spp.
237
Figure 3. 15 cm Mueller-Hinton agar spiral plate demonstrating the lack of antibiotic mutational resistance
of E. cloacae to imipenem (clean endpoint). The highest concentration of imipenem is in the center of the
spiral; the lowest concentration is on the outermost ring of the spiral. The MIC is the concentration at which
confluent growth ends (black, flared arrow). Note the complete lack of any outlier colonies beyond this
concentration.
aztreonam were retested by spiral plating, the resulting MIC became comparable to the
MICso. The growth pattern resulting from exposure to ceftazidime was somewhat
difficult to categorize, although the DRs suggested growth pattern 1. The growth
patterns of strains exposed to piperacillin were nearly equally distributed among the
three patterns of growth.
There was little evidence of spontaneous mutational resistance to imipenem,
trimethoprim-sulfamethoxazole and ciprofloxacin as the median DR was 1.0 in each
case (Table). The almost complete absence of subpopulations resistant to these drugs
was confirmed by the small range of DRs. Some mutational resistance was seen in four,
five and six strains, respectively to imipenem, trimethoprim and ciprofloxacin, but the
highest DR was still only 2.6. In contrast, the DRs to cefotaxime and aztreonam were
significantly higher than the DRs seen with imipenem, co-trimoxazole and ciprofloxacin
with six strains exhibiting an exceptionally high degree of resistance (p < 0.01) (Table)
of which four were clinically resistant. The median DRs to the aminoglycosides were
close to unity and the upper limit was < 2 confirming a relatively low level of resistance
(Table). Varying degrees of resistance to piperacillin were found with DRs ranging from
1 to 58.3. Eleven strains exhibited a very low level of resistance while seven strains
showed a relatively high degree of resistance. There were also varying degrees of
resistance to ceftazidime with DRs of up to 63.5 (Table).
238
C. M. Yu et al.
Figure 4. A strain of Enterobacter cloacae that emerged resistant after treatment with a third-generation
cephalosponn. (a) Note that while the MIC was only 0.2 mg/L of cefotaxime (black arrow), numerous
resistant colonies (outliers) appear beyond the MIC (arrowheads). The 90th percentile outlier (black
arrowhead) and the 100th percentile outlier are noted (open arrow head). The outliers for this strain can
appear at concentrations exceeding 200 mg/L (b) There is a small tail on the strain exposed to amikacin.
Note that the tail consists of smaller outliers than those seen with cefotaxime; these outliers are also relatively
close to the MIC concentration (black arrow). The 90th percentile outlier (black arrowhead) and the 100th
percentile outlier (open arrowhead) are noted In this case, the MIC is 7.6 mg/L while the outliers appear
at a concentration of 16.0 mg/L. (c) In contrast, note the absence of outliers for the strain treated with
imipenem The MIC for this strain to imipenem is 0.1 mg/L (black arrow).
Cefotaxime, ceftazidime and aztreonam induced a significantly greater degree of
spontaneous mutational resistance than did imipenem, ciprofloxacin and co-trimoxazole for each individual comparison (f<0.01; Friedman's test). Cefotaxime and
aztreonam also consistently induced a greater degree of resistance than did the
aminoglycosides, though the difference was not statistically significant. However, the
indices for each individual aminoglycoside were significantly higher than those for
imipenem, ciprofloxacin and co-trimoxazole for each individual comparison (P < 0.01,
Friedman's test). There was no significant difference between the DRs to ceftazidime
and piperacillin and those to the aminoglycosides.
Table. Mutation resistance of Enterobacter spp. to nine antimicrobial agents.
Antimicrobial
agent
Azetreonam
Cefotaxime
Ceftazidime
Piperacillin
Imipenem
Amikacin
Gentamicin
Ciprofloxacin
Co-trimoxazole
Number of outliers
mean median low high
9.6
8.7
7.5
4.2
0.5
5.3
4.6
0.9
1.0
5.1
3.1
4.4
0.9
0
5.0
3.8
0
0
0
0
0
0
0
0
0
0
0
35.5
50.0
26.5
14.5
7.0
13.5
11.5
11.0
9.0
Degree of resistance
Activity index
mean1 median low high mean median low high
8.0
9.1
3.8
2.9
1.0
1.5
1.6
1.1
1.1
8.6
7.8
1.8
1.4
1.0
1.6
1.6
1.0
1.0
.0 228.5 3.0
.0 295.1 3.2
.0 63.5 2.3
.0 58.3 1.3
1.2 0.1
.0
.0
1.9 0.7
.0
2.4 0.8
.0
2.6 0.2
.0
1.6 0.2
2.4
2.3
1.3
0.5
0
0.7
0.8
0
0
0
0
0
0
0
0
0
0
0
9.0
8.6
6.5
6.1
0.3
1.1
1.3
1.2
0.9
"Geometric mean
The entries represent a mean value of all duplicate or triplicate streaks. The cephalosporins and azetreonam have
significantly higher values for Degree of Resistance and the Activity Index than imipenem, ciprofloxacin and co-trimoxazole.
The Activity Index was calculated using the Casba II Software (Spiral Biotech, Bethesda, MD, USA).
240
C. M. Yu et al.
When the six strains that emerged resistant after treatment with a third-generation
cephalosporins were compared to the 12 strains that were successfully eliminated, no
significant difference could be found for degree of mutational resistance (data not
shown). The degree of resistance to the aminoglycosides, imipenem, ciprofloxacin,
aztreonam, and co-trimoxazole was similar for both groups of strains. The median DRs
for the strains which had emerged resistant to cefotaxime, ceftazidime, and piperacillin
were 10.2, 5.8, and 3.9, respectively, compared with 7.2, 1.6, and 1.1, respectively, for
the strains eradicated successfully in the study of Chow et al. (1991).
Discussion
The inadequacies associated with traditional methods of antimicrobial susceptibility
testing are well-known and alternative approaches that involve testing for resistance
have been advocated (Sanders, 1991). This requires the detection of resistant
subpopulations which can be achieved using the spiral plater since an extended 'tail'
of resistant colonies occurs when Enterobacter spp. was exposed to cephalosporins.
Furthermore, the single colonies that appeared in the 'tail' proved to be resistant in vitro
to high concentrations of the antibiotic when these individual colonies were retested.
That is, when these colonies were subcultured and retested using the spiral plater, the
resulting MIC became comparable to the MIC*), the concentration of antibiotic
required to inhibit 90% of the colonies growing in the area beyond the MIC. We
performed susceptibility tests using the spiral plater on Enterobacter isolates collected
from bacteraemic patients before treatment (Chow et al., 1991) in an attempt to show
that mutational resistance in vitro correlated with the emergence of clinical resistance.
The pattern of growth of Enterobacter spp. exposed to seven of the nine antimicrobial
agents could be allocated to one of three patterns. The pattern of growth in the presence
of imipenem, co-trimoxazole and ciprofloxacin (growth pattern 3) indicated the absence
of spontaneous mutants, whereas the growth pattern that resulted from exposure to
cefotaxime and aztreonam (growth pattern 1) was associated with a high degree of
resistance due to the presence of numerous mutants (Figures 2 and 4(a)). The growth
pattern produced in the presence of ceftazidime was closest to that produced by
cefotaxime and aztreonam but not identical, and those resulting from exposure to
piperacillin could be of any type.
Although based on a small series, these data suggest that, whereas strains of
Enterobacter spp. produce numerous spontaneous mutants resistant to cefotaxime, and
aztreonam, only certain strains will do so in the presence of piperacillin, ceftazidime and
amikacin and virtually none will do so when exposed to imipenem, ciprofloxacin, and
co-trimoxazole. Use of the DR index allowed the degree of resistance to be quantified
and to be compared statistically. Cefotaxime, ceftazidime, and aztreonam showed a
significantly greater degree of mutational resistance than did imipenem, co-trimoxazole
and ciprofloxacin. Furthermore, the range of mutational resistance expressed for these
three antimicrobial agents was relatively large.
While outlying colonies of some strains occurred at concentrations that were a
100-fold higher than the original MIC, other strains showed no evidence of resistance.
Ceftazidime demonstrated a moderate to high degree of mutational resistance.
Piperacillin showed a widely varying degree of mutational resistance depending on the
strain, with 11 strains showing a low degree of mutational resistance comparable to that
induced by the aminoglycosides and the remainder demonstrating a high degree of
Mutational antibiotic resistance of Enterobacter spp.
241
mutational resistance similar to that induced by the cephalosporins. Imipenem,
co-trimoxazole and ciprofloxacin marginally induced mutational resistance in only four,
five and six strains respectively. Importantly, these strains included those organisms
which contained the subpopulations which were highly resistant to the third generation
cephalosporins.
We found that the strains exposed to the aminoglycosides consistently showed an
intermediate degree of mutational resistance and that there were fewer outliers and
lower DRs than was found after exposure to the cephalosporins and aztreonam.
However, the aminoglycosides showed a higher degree of mutational resistance than did
imipenem, co-trimoxazole and ciprofloxacin.
Complete correlation does not exist between any in-vitro test of antimicrobial
susceptibility and clinical outcome since other factors play a role, including the immune
status of the patients, the site of infection, and the pharmacokinetics of the drug. The
degree of mutational resistance found in the six strains which ultimately emerged
resistant to the third-generation cephalosponn used to treat bacteraemia proved not to
be statistically significantly different from that of the 12 strains that were eradicated
successfully. However, there was a discernible trend for cefotaxime and ceftazidime
since median DRs to cefotaxime and cetazidime were significantly higher for the six
strains that ultimately emerged resistant than were those for the twelve strains
eradicated successfully. This suggests that mutational resistance might indeed be
responsible for the recurrence of bacteraemia observed in the six patients.
Since resistance among Enterobacter spp. appears to be a growing problem, it may
be important to choose antimicrobial agents that achieve blood and tissue levels which
are not only well in excess of the MIC, but are high enough to prevent the emergence
of resistance. Interestingly, none of the 17 patients who received imipenem during the
US national collaborative study of bacteraemia due to Enterobacter spp. experienced
bacteriological failure with relapse (Chow el al, 1991) whereas relapse occurred in six
(19%) of the 31 patients receiving third generation cephalosporins. Other reports have
also noted both the efficacy of carbapenems against Enterobacter spp. (Pechere, 1991)
and the unusually high frequency of emergent resistance in patients treated with
cephalosporin antibiotics (Murray et al., 1983; Olson et al, 1983; Follath et al, 1987;
Milatovic & Braveny, 1987, Quinn, DiVencenzo, & Foster, 1987).
Although based on a limited number of strains, the results of this study show that
the use of the spiral plater for antimicrobial susceptibility testing may have distinct
advantages over conventional tests which give only a undimensional view of
susceptibility by declaring a strain resistant or susceptible according to whether or not
the MIC falls above or below a rigid breakpoint. Besides, the precision of dilution
susceptibility methodology is compromised by the traditional use of two-fold dilutions
(Hsieh et al, 1993; Woolfrey et al, 1982), whereas the continuous antimicrobial
concentrations provided by the spiral plater offer a marked improvement in
reproducibility and precision (Wexler et al, 1991). Furthermore, conventional broth and
agar dilution tests do not provide the means of quantifying emergent resistance. In
contrast, the spiral plater technique provides more information on the interaction
between the antimicrobial agent and the bacterial population by allowing individual,
resistant colonies to be both detected and enumerated. Hitherto, the detection of
mutants of Enterobacter spp. resistant to aminoglycosides, trimethoprim and
cephalosporins was essentially the preserve of research laboratories (Guttman et al.,
1985; Pechere, 1991) but the simplicity afforded by the spiral plater makes it feasible
242
C. M. Yu et at.
for any clinical microbiology laboratory to look for mutants. Clearly, our findings
require further large scale in-vitro studies of many more strains to validate them, but
these preliminary results should provide an incentive for developing in-vitro tests better
suited to address the clinical problem of antimicrobial resistance.
Acknowledgements
We thank Tzielan Lee for preliminary studies, Marilyn Wagener for statistical analysis,
Richard M. Vickers for photographic consultation, Saul Weiss for medical media
assistance, Shirley Brinker for secretarial assistance and Professor David Livermore for
his critical review.
References
Chow, J. W., Fine, M. J., Shales, D. M., Quinn, J. P., Hooper, D. C , Johnson, M. P. et al. (1991).
Enlerobacter bacteremia: clinical features and emergence of antibiotic resistance during
therapy. Annals of Internal Medicine 115, 585-90.
Follath, F., Costa, E., Thommen, A., Frei, R., Burdeska, A. & Meyer, J. (1987). European Journal
of Clinical Microbiology and Infectious Diseases 6, 446-50.
Gutmann, N. L., Williamson, R., Moreau, N., Kitzis, M. D., Collatz, E., Acar, J. F. et al. (1985).
Cross resistance to halidixic acid, trimethoprim, and chloramphenicol associated with
alteration in outer membrane proteins of Klebsiella, Enterobacter, and Serratia. Journal of
Infectious Diseases 151, 501-7.
Hsieh, M. H., Yu, C. M., Yu, V. L. & Chow, J. W. (1993). Synergy assessed by checkerboard.
A critical analysis. Diagnostic Microbiology and Infectious Diseases 16, 343-9.
Milatovic, D. & Braveny, I. (1987). Development of resistance during antibiotic therapy.
European Journal of Clinical Microbiology and Infectious Diseases 6, 234-44.
Murray, P. R., Granich, G. G., Krogstad, D. J. & Niles, A. C. (1983). In-vivo selection of
resistance to multiple cephalosportins by Enterobacter cloacae. Journal of Infectious Diseases
147, 590.
Olsen, B., Weinstein, R. A., Nathan, C. & Kabins, S. A. (1983). Broad-spectrum fi-lactam
resistance in Enterobacter: emergence during treatment and mechanisms of resistance.
Journal of Antimicrobial Chemotherapy 11, 299-310.
Pechere, J.-C. (1991). Why are carbapenems active against Enterobacter cloacae resistant to third
generation cephalosporins? Scandinavian Journal of Infectious Diseases, Suppl. 78, 17-21.
Quinn, J. P., DiVencenzo, C. A. & Foster, J. (1987). Emergence of resistance to ceftaidime during
therapy for Enterobacter cloacae infections. Journal of Infectious Diseases 155, 942-7.
Sanders, C. C. (1991). A problem with antimicrobial susceptibility tests. ASM News 57, 187-90.
Spiral Systems Instruments Inc. (1993). User Guide. Determination of antimicrobial susceptibility
by the spiral gradient endpomt test. Bethesda, MD.
Wexler, H. M., Molitoris, E , Jashonian, F. & Finegold, S. M. (1991). Comparison of spiral
gradient and conventional agar dilution for susceptibility testing of anaerobic bacteria.
Antimicrobial Agents and Chemotherapy 35, 1196-202.
Woolfrey, B. F., Fox, J M. K , Lally, R. T. & Quail, C. O. (1982). Broth microdilution testing
of Pseudomonas aerugmosa and aminoglycosides: need for employing dilutions differing by
small arithmetic increments. Journal of Clinical Microbiology 16, 663-7.
(Received 15 July 1994; returned 8 September 1994; revised 7 November 1994;
accepted 15 August 1995)