Microbiology and Infectious Disease / Laboratory Detection of AmpC β-lactamase
Clinical Laboratory Detection of AmpC 𝛃-Lactamase
Does It Affect Patient Outcome?
Kenneth H. Rand, MD,1 Bradley Turner, MD,1 Hilary Seifert,1 Christine Hansen, PharmD,2
Judith A. Johnson, PhD,1,3 and Andrea Zimmer, MD4
Key Words: AmpC β-lactamase; Modified Hodge test; EDTA disk test; Bacteremia outcome
DOI: 10.1309/AJCP7VD0NMAMQCWA
CME/SAM
Upon completion of this activity you will be able to:
• list the major mechanisms of overproduction of AmpC β-lactamases.
• describe laboratory methods for measurement of AmpC β-lactamases.
• discuss the relationship between laboratory testing and clinical
outcome.
Abstract
Plasmid-mediated AmpC-producing Escherichia
coli and Klebsiella pneumoniae have been associated
with poor clinical outcomes, but they are not readily
identified in hospital microbiology laboratories. We
tested 753 gram-negative bloodstream isolates for
AmpC by using the EDTA disk test and the modified
Hodge test (n = 172) and the modified Hodge test
alone (n = 581). The 30-day mortality for the AmpC
group was 9% (2/23) and was 6% (3/51) for the control
group. The clinical response was similar: afebrile on
day 2 (AmpC group, 16/23 [70%]; control group, 32/45
[71%]) and on day 4 (AmpC group, 19/22 [86%];
control group, 37/44 [84%]). Patients with isolates in
the AmpC group were more likely to be in an intensive
care unit at the time of the positive blood culture (P
= .01) and more likely to be intubated (P = .05) than
patients with isolates in the control group. Effective
antibiotic treatment within the first 48 hours was given
to 47 (92%) of 51 patients with isolates in the control
group but to only 14 (61%) of 23 patients with isolates
in the AmpC group (P = .001).The modified Hodge test
and the EDTA disk test did not identify patients at risk
for a poor outcome from AmpC-producing bacterial
infections.
572 Am J Clin Pathol 2011;135:572-576
572 DOI: 10.1309/AJCP7VD0NMAMQCWA
Downloaded from https://academic.oup.com/ajcp/article-abstract/135/4/572/1760867 by guest on 26 October 2018
The ASCP is accredited by the Accreditation Council for Continuing
Medical Education to provide continuing medical education for physicians.
The ASCP designates this educational activity for a maximum of 1 AMA PRA
Category 1 Credit ™ per article. This activity qualifies as an American Board
of Pathology Maintenance of Certification Part II Self-Assessment Module.
The authors of this article and the planning committee members and staff
have no relevant financial relationships with commercial interests to disclose.
Questions appear on p 644. Exam is located at www.ascp.org/ajcpcme.
Escherichia coli and Klebsiella pneumoniae containing
plasmid-mediated AmpC β-lactamases have been associated
with treatment failure in a case-control study of bacteremia
compared with organisms without plasmid-mediated AmpC
β-lactamases.1 Pai et al2 reported a treatment failure rate of
almost 52% for AmpC-containing K pneumoniae bloodstream
isolates at 72 hours.
Unfortunately, plasmid-mediated AmpC β-lactamases are
not reliably detected by standard susceptibility testing methods
in the clinical microbiology laboratory. Black et al3 described
the EDTA disk test, and Yong et al4 described the “modified”
Hodge test for detecting the presence of AmpC β-lactamases
that could be carried out routinely in a busy clinical labora-
tory. A similar test using boronic acid was described by
Coudron.5 However, none of these tests can distinguish plas-
mid-mediated hyperproduction of AmpC from derepressed
chromosomal or any other mechanism of overproduction of an
AmpC β-lactamase. Intuitively, we believe the mechanism of
AmpC overproduction should not matter for clinical outcome
because, either way, such organisms should be resistant to all
extended-spectrum cephalosporins, although perhaps not to the
advanced-spectrum cephalosporins, cefepime and cefpirome.
We compared the clinical outcome for 26 (23 evaluable)
patients with bacteremic gram-negative isolates with over-
production of AmpC as measured by the modified Hodge test
with 52 (51 evaluable) control patients whose isolates did not
produce AmpC.
Materials and Methods
Bacterial Isolates
Between January 2007 and June 2008, we tested 753
consecutive gram-negative bloodstream isolates (excluding
© American Society for Clinical Pathology
 Microbiology and Infectious Disease / Original Article

Pseudomonas and Stenotrophomonas) for AmpC in the
Shands at the University of Florida Hospital (Gainesville)
clinical microbiology laboratory. All isolates were identified
and tested for antibiotic susceptibility by standard microbio-
logical procedures using MicroScan (Siemens HealthCare,
Deerfield Park, IL). All isolates were tested by the modified
Hodge test.4 As described by Yong et al,4 3 mm or more of
“diagonal” growth in the cloverleaf pattern was positive for
AmpC production ❚Image 1❚. Isolates that had no or minimal
distortion of the cefoxitin zone were considered negative.
We also tested 172 isolates by the EDTA disk test3 with
overall agreement of 97.1% (κ = 0.905; 95% confidence
interval, 0.823-0.987).
Patient Selection
The AmpC patient group (n = 26) was selected based on
the degree of AmpC positivity on the modified Hodge test, ie,
growth of the indicator strain along the test isolate streak of 3
mm or more as described by Yong et al.4 These isolates were
considered positive for AmpC production. The control group
was selected from patients with isolates that had no (or only
minimal, ≤0.5 mm) distortion of the circular zone of inhibition
surrounding the cefoxitin disk. These non-AmpC producers

Downloaded from https://academic.oup.com/ajcp/article-abstract/135/4/572/1760867 by guest on 26 October 2018

were matched in a 2:1 ratio by bacterial species (n = 52).
This matching had the desired ratio for E coli, Acinetobacter
species, and Serratia species, but there were insufficient
Enterobacter species lacking AmpC, so Klebsiella species
were substituted. The organisms included in the AmpC and
control groups are shown in ❚Table 1❚. Because the groups
were selected on the basis of AmpC testing in the laboratory, it
was only after review of the patient records that 1 patient was
found to have a mixed blood culture with an AmpC-producing
Klebsiella oxytoca and an AmpC-producing Enterobacter
cloacae. The data for this patient were, therefore, counted
only once in the AmpC group. Another patient in the AmpC
group had 2 bacteremic episodes 6 months apart with an
AmpC-producing E cloacae in both instances. Because the
patient had multiple negative blood cultures in the interval (as
well as a successful liver transplant), the patient was counted
as having 2 separate episodes. This left the AmpC group at 25
patient bacteremic episodes.
❚Image 1❚ Example of a positive and negative EDTA disk test
(BLC disk) and positive and negative modified Hodge test Data Collection
(vertical streaks). A cefoxitin-susceptible Escherichia coli The electronic medical records and paper charts were
indicator strain (American Type Culture Collection 25922) reviewed by 3 of us (K.H.R., B.T., and A.Z.) for the following
is plated and the cefoxitin disk placed. In the case of the data ❚Table 2❚: peak daily temperature, starting with the day
EDTA disk test, the EDTA disks are then placed adjacent to of the positive blood culture (day 1) through day 7; clinical
the cefoxitin disk as shown. The organisms to be tested for
AmpC are inoculated onto the EDTA disks in 20 μL of saline
as described.3 If the organism expresses AmpC, cefoxitin is
❚Table 1❚
hydrolyzed and will give a positive EDTA disk test as shown AmpC-Producing and Control Bacterial Species
by growth of the indicator E coli toward the EDTA disk (right
AmpC Control
side), while a negative test is shown by the lack of growth
toward the disk (left side). For the modified Hodge test, the Enterobacter cloacae 12 9
test organism is streaked toward the cefoxitin disk. If the test Enterobacter agglomerans 0 3
Enterobacter amnigenus 0 1
organism expresses AmpC, it hydrolyzes the cefoxitin and Enterobacter aerogenes 2 0
shows growth along the intersection of the streak and the Serratia marcescens 2 6
Escherichia coli 3 6
zone of inhibition from the cefoxitin disk. The image shows a Klebsiella pneumoniae 0 12
positive modified Hodge test (top), indicated by the 3.6-mm Klebsiella oxytoca 1* 7
Citrobacter freundii 1 0
diagonal area of growth of the indicator E coli toward the
Acinetobacter baumannii 4 8
test organism streak. A negative Hodge test is shown on Providencia stuartii 1 0
the streak at the bottom, with no diagonal growth into the * The patient had AmpC-producing K oxytoca and E cloacae in the same positive
cefoxitin zone. culture.

© American Society for Clinical Pathology Am J Clin Pathol 2011;135:572-576 573

573 DOI: 10.1309/AJCP7VD0NMAMQCWA 573
Rand et al / Laboratory Detection of AmpC β-lactamase

assessment of response during the same period; antibiotics ❚Table 2❚

used to treat the patient in the first 48 hours and the in vitro Patient Characteristics*
susceptibility results; patient survival; associated illnesses; AmpC Group Control Group
patient location at the time of the positive culture, eg, inten- (n = 23) (n = 51) P†
sive care unit (ICU), other inpatient, outpatient; duration of Mean ± SD age (y) 42.4 ± 30.6 53.7 ± 23.1 NS
stay in the hospital before and after the positive blood culture; In intensive care 13 (57) 13 (25) .01
ventilator status; and patient demographics, ie, age and sex. Intubated 7 (30) 6 (12) .05
Effective antibiotic treatment 14 (61) 47 (92) .001
Comorbidities and severity of illness were assessed using in first 48 h
the McCabe score and the simplified acute physiology score Hospital acquired‡ 15 (65) 24 (47) NS
Length of stay before positive 25.4 ± 48 10 ± 16 .047
(SAPS) II. Fever was defined as a peak daily temperature of culture (d)
38°C or more. Effective antibiotic treatment was defined as McCabe score (rapidly 5 (22) 16 (31) NS
fatal/total)
the use of at least 1 antibiotic to which the patient’s isolate Mean ± SD SAPS II score§

Downloaded from https://academic.oup.com/ajcp/article-abstract/135/4/572/1760867 by guest on 26 October 2018

was susceptible by MicroScan using Clinical and Laboratory All patients 37.2 ± 14.9 32.4 ± 14.8 NS
≥1 y 40.8 ± 14.6 32.7 ± 14.8 .042
Standards Institute susceptibility criteria. ≥18 y 39.6 ± 13.2 33.6 ± 15.1 NS

NS, not significant; SAPS, simplified acute physiology score.

* Data are given as number (percentage) unless otherwise indicated.
† Calculated by the χ2 test, except for the length of stay, for which the t test was used.

Results ‡ Blood culture positive >72 h after admission.

§ SAPS II calculator.

Two patients in the AmpC group and one in the control

group died within 24 hours of the positive blood culture
because of underlying terminal illness and withdrawal of life
support. Of the evaluable cases, 2 (9%) of 23 patients in the
AmpC group and 3 (6%) of 51 in the control group died within
30 days. There was no difference in the response to treatment
between the 2 groups, as shown by the defervescence of fever
❚Figure 1❚ and the percentage who were afebrile on days 2 and
4 after the day of the positive blood culture ❚Table 3❚. One
patient in the AmpC group had blood cultures positive for the
original organism (Serratia marcescens) on days 1, 3, and 4;
and 1 control patient had a subsequent positive blood culture
for the original organism on day 2. In both patients, infections
cleared without a change in antibiotic therapy.
The AmpC and control groups did not differ significantly
in age (Table 2) or sex distribution (male, AmpC group, 12/23
[52%]; control group, 28/51 [55%]). Table 2 shows that
patients in the AmpC group were significantly more likely to
be in an ICU (P = .01) and to have been intubated (P = .05)
than patients in the control group at the time of the positive
blood culture. Moreover, patients in the AmpC group were
significantly less likely to have received an effective antibiotic
within the first 48 hours after the blood culture was obtained
than were patients in the control group (14/23 [61%] vs 47/51
[92%]; P = .001, respectively). Although there was no dif-
ference in the percentage of each group that acquired the

40.0

39.5 AmpC
39.0 Control

38.5
Temperature (°C)

38.0

37.5

37.0

36.5

36.0

35.5
1 2 3 4 5 6 7
Days After Positive Blood Culture

❚Figure 1❚ Mean ± SD of the peak daily temperature for all patients in each group who remained in the hospital. Day 1 represents
the day of positive blood culture. In the AmpC group, the isolates produced AmpC, whereas in the control group, they did not.

574 Am J Clin Pathol 2011;135:572-576 © American Society for Clinical Pathology

574 DOI: 10.1309/AJCP7VD0NMAMQCWA

❚Table 3❚
Outcome of Bacteremia*
AmpC Group
(n = 25)†
Mean ± SD length of stay 34.8 ± 39.3
after positive culture (d)
Survival, died within 24 h 2 1
Subsequent 30-d mortality 2/23§ (9)
Fever response||
Afebrile day 2 16/23 (70) §
Afebrile day 4 19/22 (86) §
*
†
Data are given as number (percentage) unless otherwise indicated.
Denominator of original group.
‡ P = .009.
§ Denominators of evaluable cases.
|| Fever defined as temperature ≥38°C for patients remaining in the hospital.
infection in the hospital, patients in the AmpC group were in
the hospital longer than patients in the control group before
acquiring the infection (mean ± SD, 25.4 ± 48 vs 10 ± 16
days; P = .047; Table 2).
The 2 groups were not statistically different in intensity
of illness as measured by the SAPS II score on admission,
although patients in the AmpC group tended to have higher
scores. Indeed, if infants younger than 1 year were omitted,
the SAPS II score was significantly higher for the AmpC
group, but restricting the scoring system to adults (18 years
or older, as originally intended) showed no significant dif-
ference. Likewise, the McCabe index based on estimated
survival in relation to the underlying comorbid illness was not
significantly different between the groups (Table 2). Patients
in the AmpC group remained in the hospital significantly lon-
ger than did patients in the control group following resolution
of the bacteremic episode (Table 3).
Discussion
We studied the outcome of patients with gram-negative
bacteremia caused by AmpC-producing organisms, as deter-
mined by the modified Hodge test, and the outcome in an
appropriate control group. Compared with the control group,
significantly more patients in the AmpC group were intu-
bated, in an ICU, and received inadequate antibiotic treat-
ment within the first 48 hours after the positive culture was
obtained. Despite this bias of the AmpC group toward greater
morbidity, the clinical response to antibiotics was identical in
the 2 groups as measured by the rate of defervescence of fever
following the bacteremia and by 30-day mortality. Although
patients in the AmpC group remained in the hospital longer
than did patients in the control group after resolution of the
bacteremia, they were also in the hospital more than twice as
long before becoming bacteremic, suggesting that the length
of stay after the bacteremia reflected the generally higher level
© American Society for Clinical Pathology
Microbiology and Infectious Disease / Original Article
of morbidity of patients in the AmpC group rather than an
effect of the bacteremia itself. Indeed, the SAPS II score was
Control Group higher for the AmpC group than the control group, although
(n = 52)† this difference did not reach statistical significance.
15.5 ± 22‡
The AmpC group was obtained from the blood culture
isolates demonstrating the highest degree of “positivity” by
the modified Hodge test using cefoxitin as described by Yong
3/51§ (6)
et al.4 As illustrated in Image 1, it is possible to quantitate
32/45§ (71) the growth of the indicator E coli from the expected circular
37/44§ (84)
zone of inhibition toward the test organism streak, although
this assessment is subjective and only semiquantitative. In the
study by Yong et al,4 the best cutoff distinguishing between
Downloaded from https://academic.oup.com/ajcp/article-abstract/135/4/572/1760867 by guest on 26 October 2018
plasmid-associated AmpC production and non–plasmid-medi-
ated AmpC production in E coli and Klebsiella species was 3
mm. At that level, about 5% of the positive tests were false-
positives, whereas at 4 mm, none were false-positive, but the
sensitivity decreased. In our study, there were 10 patients
with an AmpC diagonal distance of 4 mm or more, and their
rate of defervescence was identical to that of the AmpC and
control groups as a whole (70% [7/10] afebrile by day 2 and
80% [8/10] by day 4).
The modified Hodge test has been compared with other
methods of detecting AmpC expression such as the boronic
acid test and the EDTA disk test.6 The modified Hodge test
was, if anything, less sensitive than the boronic acid test.
Although we found a very good correlation between the
EDTA disk test and the modified Hodge test in a subset of
our isolates, the correlation is not perfect and the EDTA disk
test identified more positives (data not shown). Thus, it is
possible that different methods would have identified a more
accurate group of patients with AmpC; however, in our hands
and in the studies of other investigators, the ability to detect
the strongest positive AmpC-producing strains is virtually
100% among the different methods. Thus, it seems unlikely
we missed any significant number of strong AmpC producers,
and in the subgroup of the strongest AmpC producers we did
identify, there was no difference in outcome compared with
the entire AmpC or control group.
In a study of 27 plasmid-mediated AmpC-producing
bloodstream isolates of Klebsiella, Pai et al2 found a clini-
cal treatment failure rate of 51.9%, which was similar to
that of bacteremic extended-spectrum β-lactamase (ESBL)-
producing Klebsiella isolates. Park et al1 compared the clini-
cal response of 30 patients infected with E coli and Klebsiella
isolates carrying AmpC-producing plasmids with control
isolates and found a higher treatment failure rate at 72 hours
and 7 days in the AmpC group, although the response rate at
30 days was similar (87% in the AmpC group and 97% in the
control group). They noted that 70% of the Klebsiella isolates
(and 1 E coli isolate) carried AmpC and ESBL plasmids but
did not state what percentage of the control isolates produced
ESBL. Thus, if their control patients did not have very many
Am J Clin Pathol 2011;135:572-576 575
575 DOI: 10.1309/AJCP7VD0NMAMQCWA 575
Rand et al / Laboratory Detection of AmpC β-lactamase
isolates with ESBLs, their interpretation of the relationship
between clinical failure and AmpC could have been related to
the presence of ESBLs, as they state in their discussion. In our
institution, we see few ESBL-containing E coli and Klebsiella
isolates (<2%), and there were no ESBL-producing E coli or
Klebsiella isolates in the AmpC group or control group.
The modified Hodge test, EDTA disk test, and boronic
acid test all fail to distinguish between plasmid-mediated
AmpC production and derepressed hyperproduction of chro-
mosomal AmpC. Characterization of the AmpC-producing
plasmid status is complex and was beyond the scope of the
study. If organisms with AmpC production due to a dere-
pressed chromosomal mechanism produced less AmpC than
organisms with AmpC-producing plasmids, it is possible
that our results reflect the clinical outcome of gram-negative
bacteremia by organisms with derepressed chromosomal
AmpC production, while the studies of Pai et al2 and Park et
al1 reflect the outcome of organisms with plasmid-mediated
AmpC production.
Several groups have noted that Enterobacteriaceae may
test susceptible to extended-spectrum cephalosporins (eg,
ceftazidime and cefotaxime), despite the presence of AmpC-
producing plasmids and, thus, potentially may be reported
as falsely susceptible. In fact, 9 (36%) and 14 (56%) of our
25 AmpC-producing isolates were reported as susceptible or
intermediate to cefotaxime and ceftazidime, respectively, by
standard broth dilution semiautomated methods (MicroScan).
Although nearly 40% of our AmpC patient group did not
receive an effective antibiotic within the first 48 hours (com-
pared with 8% of the control group), 18 (78%) of 23 received
a full treatment course with cefepime or meropenem. The
remaining patients received various combinations that most
often included a fluoroquinolone and an aminoglycoside.
Thus, in our institution, false susceptibility to extended-spec-
trum cephalosporins might not have the same adverse effect
on patient outcome as it otherwise could at institutions where
extended-spectrum cephalosporin treatment was continued,
particularly if ESBL-producing plasmids were common.
Despite the fact that patients in the AmpC group tended
to be more ill (more likely to be intubated, in the ICU, and
not to have received an effective antibiotic within the first
576 Am J Clin Pathol 2011;135:572-576
576 DOI: 10.1309/AJCP7VD0NMAMQCWA
48 hours), the clinical response and overall outcome of the
AmpC and control bacteremic episodes were essentially iden-
tical. The modified Hodge test and the EDTA disk test did not
seem to identify a subset of patients at risk for a poor outcome
from AmpC with the antibiotic use patterns in our hospital.
From the Departments of 1Pathology, Immunology, and
Laboratory Medicine and 4Medicine and the 3Emerging Pathogens
Institute, University of Florida; and 2Shands at the University of
Florida Hospital, Gainesville.
Supported in part by the Department of Pathology,
Immunology, and Laboratory Medicine, University of Florida,
College of Medicine.
Downloaded from https://academic.oup.com/ajcp/article-abstract/135/4/572/1760867 by guest on 26 October 2018
Address reprint requests to Dr Rand: Dept of Pathology,
Immunology, and Laboratory Medicine, PO Box 100275,
University of Florida, College of Medicine, Gainesville, FL
32601-0275.
References
1. Park YS, Yoo S, Seo MR, et al. Risk factors and clinical
features of infections caused by plasmid-mediated AmpC
beta-lactamase-producing Enterobacteriaceae. Int J
Antimicrob Agents. 2009;34:38-43.
2. Pai H, Kang CI, Byeon JH, et al. Epidemiology and clinical
features of bloodstream infections caused by AmpC-type-
beta-lactamase-producing Klebsiella pneumoniae. Antimicrob
Agents Chemother. 2004;48:3720-3728.
3. Black JA, Moland ES, Thomson KS. AmpC disk test for
detection of plasmid-mediated AmpC beta-lactamases in
Enterobacteriaceae lacking chromosomal AmpC beta-
lactamases. J Clin Microbiol. 2005;43:3110-3113.
4. Yong D, Park R, Yum JH, et al. Further modification of
the Hodge test to screen AmpC beta-lactamase (CMY-1)-
producing strains of Escherichia coli and Klebsiella pneumoniae.
J Microbiol Methods. 2009;34:38-43.
5. Coudron PE. Inhibitor-based methods for detection of
plasmid-mediated AmpC beta-lactamases in Klebsiella
spp, Escherichia coli and Proteus mirabilis. J Clin Microbiol.
2005;43:4163-4167.
6. Lee W, Jung B, Hong SG, et al. Comparison of 3
phenotypic-detection methods for identifying plasmid-
mediated AmpC beta-lactamase-producing Escherichia coli,
Klebsiella pneumoniae, and Proteus mirabilis strains. Korean J
Lab Med. 2004;29:448-454.
© American Society for Clinical Pathology