INTRODUCTION
In recent years, the prevalence of infections with multidrug-resistant
Enterobacteriaceae has steadily increased (
18).
Enterobacteriaceae producing AmpC beta-lactamases (AmpCs) have become a major therapeutic challenge. The detection of AmpC-producing
Klebsiella spp.,
Escherichia coli,
P. mirabilis, and
Salmonella spp. is of significant clinical relevance since AmpC producers may appear susceptible to expanded-spectrum cephalosporins when initially tested (
13,
27,
28). This may lead to inappropriate antimicrobial regimens and therapeutic failure (
24). Thus, a simple and reliable detection procedure for AmpC producers is needed.
Many Gram-negative bacteria harbor chromosomal
ampC beta-lactamase genes, which are constitutively expressed at low level. In general, the expression of chromosomally located
ampC genes is inducible by beta-lactam antibiotics, such as cefoxitin, cefotetan, and imipenem, and mediated by the regulator AmpR. Mutations in the repressor gene
ampD are the most common cause of constitutive (hyper-)production of AmpC beta-lactamases (
23). AmpC beta-lactamases degrade penicillins, expanded-spectrum cephalosporins (with the exception of cefepime and cefpirome), cephamycins, monobactams, and beta-lactam inhibitors. In contrast to expanded-spectrum beta-lactamases (ESBLs), AmpC beta-lactamases are inhibited by boronic acid and cloxacillin (
2,
9,
25). In
E. coli, regulation of chromosomal
ampC expression differs significantly from that of other
Enterobacteriaceae. In
E. coli ampC is regulated by a weak promoter and a strong attenuator resulting in a constitutive low-level
ampC expression (
11). Diverse mutations in the
ampC promoter region leading to overexpression have been described (
3,
4,
7,
11,
12,
24,
29). In addition to chromosomal
ampC,
Enterobacteriaceae can acquire plasmid-encoded
ampC genes (
9). In general, plasmid-encoded AmpC beta-lactamases are expressed constitutively and are readily detected by a multiplex PCR (
17).
Different phenotypic AmpC detection tests have been described in the literature (
9). A standardized diagnostic approach integrating screening and confirmation tests for the detection of AmpC beta-lactamase-producing
Enterobacteriaceae has not been established to date. We sought here to develop a comprehensive diagnostic flow chart integrating a simple phenotypic screening and confirmation for implementation in the routine diagnostic laboratory.
DISCUSSION
Detection of AmpC production in pathogens might be important for ensuring effective antibiotic therapy (
20) since the presence of an AmpC beta-lactamase frequently seems to result in therapeutic failure when broad-spectrum cephalosporins are used (
14,
24). However, further studies are required to assess whether AmpC production is an independent risk factor for clinical outcome. Several methods have been evaluated for phenotypic screening and confirmation of AmpC beta-lactamase production (
9,
25). However, a comprehensive diagnostic algorithm integrating both screening and confirmation has not been established. In the present study we evaluated individual screening and confirmation methods for AmpC production. Subsequently, we developed a diagnostic algorithm that (i) combines the most efficient and accurate methods, (ii) is simple, and (iii) can be implemented in the diagnostic laboratory (
Fig. 2).
When cefoxitin and cefotetan (both cephamycins) were compared as the primary screening marker, cefoxitin was clearly superior to cefotetan regarding sensitivity (see
Table 2). Our results for cefoxitin are in agreement with those of other authors (
20,
25). However, the specificity in the present study was significantly lower, e.g., 78.7% versus the 95% reported by Tan et al. (
25). In contrast to MIC determination by automated systems, the determination of drug susceptibility by disc diffusion may further enhance sensitivity since synergy and antagonism phenomena are readily observed, e.g., when placing a cefoxitin disc near a expanded-spectrum cephalosporin disc. For example, the presence of DHA type enzymes will lead to flattening of inhibition zones (antagonism phenomena) of expanded-spectrum cephalosporins toward inducers such as cefoxitin, carbapenems, or clavulanic acid. Otherwise, ACC-type enzymes are characteristically inhibited by cefoxitin visible as enhancement of the inhibition zones (synergy phenomena) of expanded-spectrum cephalosporins and cefoxitin. With this strategy, the detection of ACC-type AmpC enzymes is possible, although ACC enzymes appear to be cefoxitin susceptible (
1,
22). In contrast, cefoxitin screening by MIC alone would miss ACC types. Other authors recommend additional screening criteria for ACC enzymes such as critical inhibition zone diameters for amoxicillin-clavulanic acid or expanded-spectrum cephalosporins (
26). To date, the ACC types seem to be the only known enzymes that can be missed by cefoxitin screening. The isolation numbers of ACC enzymes are still significantly lower than those of CIT (CMY), FOX, and DHA types (
10,
14,
19,
25,
26). No ACC-type AmpC was detected in the present study.
The AmpC flow chart (
Fig. 2) can be combined with a flow chart for ESBL detection (unpublished data). If cefoxitin is not routinely tested, an alternative branch may be chosen that substitutes the cefoxitin screening criteria by CLSI screening criteria for ESBL (
Fig. 3). With a combined ESBL/AmpC screening strategy, ACC enzymes will readily be detected. ACC confers high resistance to expanded-spectrum cephalosporins, which serve as primary screening markers for ESBL detection (
19,
21). Thus, corresponding isolates will be assigned to a combined ESBL/AmpC confirmation test via the CLSI screening criteria for ESBL (
5,
6).
The single false-negative result for the cefoxitin screening test in the present study (
Fig. 2) resulted from the presence of a CIT-type
ampC detected by multiplex PCR. MICs of this isolate for cefoxitin and cefotetan were well within the susceptible range, and both phenotypic confirmation tests were clearly negative (Etest AmpC ratio of 1.0; CC-DDS, no difference). Sequence analysis of the CIT
ampC gene did not reveal any mutation affecting the structure and/or function of the enzyme. However, mutations in the regulatory regions may result in very low expression or no expression of the structural gene (
8). If the CIT type enzyme in this isolate were nonfunctional, the sensitivity of the cefoxitin screening procedure would be close to 100%.
We compared the performance of the Etest AmpC and the cefoxitin-cloxacillin CC-DDS as a phenotypic confirmation test (
25). Boronic acid can be used as alternative to cloxacillin as AmpC inhibitor, since it was found to be almost as sensitive and specific as cloxacillin by Tan et al. (
25). However, boronic acid may produce false-positive results in isolates carrying class A carbapenemases, whereas cloxacillin does not (
15,
16). Therefore, we chose cloxacillin as AmpC inhibitor in our algorithm. Regarding sensitivity, the CC-DDS was clearly superior to the Etest AmpC (97.2% versus 77.4%, respectively, see
Table 2). This result may be explained by the use of cefotetan in the Etest AmpC. Cefotetan has a lower sensitivity than cefoxitin concerning the detection of AmpC production. This is also apparent when cefotetan disc diffusion was used as a screening test (see
Table 2). Ten inconclusive results were obtained with the Etest AmpC, due to MICs exceeding the Etest scale of cefotetan with or without cloxacillin (
Table 2). In routine use, this may hamper the sensitivity and practicability of this method. In contrast, with CC-DDS only two inconclusive results were obtained. For both isolates with an inconclusive result, no inhibition zone for cefoxitin was observed both with or without cloxacillin. Eventually, AmpC enzymes of the CIT type were found in both strains. The results for the CC-DDS are in agreement with other studies that reported a high sensitivity and specificity for this test (
25).
Combining the high sensitivity of cefoxitin screening with the high specificity of the cefoxitin-cloxacillin CC-DDS confirmation test, we propose a flow chart for the phenotypic detection and characterization of AmpC beta-lactamases (
Fig. 3). In the case of (rarely occurring) inconclusive results, molecular methods are used for resolution. The proposed flow chart would have a calculated sensitivity and specificity of 97.4 and 100%, respectively, with respect to the isolates in the present study. Phenotypic AmpC screening and confirmation tests are inexpensive but nevertheless highly sensitive and specific. Therefore, it can be performed in all types of clinical laboratories, whereas the implementation of molecular methods is often complex, requires specially trained personnel, and is associated with higher costs.
In conclusion, the proposed flow chart for detection of AmpC is simple to use and easy to implement in a diagnostic laboratory. If molecular methods are not available, the few inconclusive isolates can be submitted to a reference laboratory for further investigations. In parallel, we have developed a flow chart for ESBL detection (unpublished), which in combination with the AmpC detection flow chart, covers a broad spectrum of beta-lactamases, facilitating therapeutic decisions and epidemiological surveillance.