Systematic Review of Plasmid AmpC Type Resistances in Escherichia coli and Klebsiella pneumoniae and Preliminary Proposal of a Simplified Screening Method for ampC

Beta-lactamase (BL) production is a major public health problem. Although not the most frequent AmpC type, AmpC-BL is increasingly isolated, especially plasmid AmpC-BL (pAmpC-BL). The objective of this study was to review information published to date on pAmpC-BL in Escherichia coli and Klebsiella pneumoniae, and on the epidemiology and detection methods used by clinical microbiology laboratories, by performing a systematic review using the MEDLINE PubMed database. The predictive capacity of a screening method to detect AmpC-BL using disks with cloxacillin (CLX) was also evaluated by studying 102 Enterobacteriaceae clinical isolates grown in CHROMID ESBL medium with the addition of cefepime (FEP), cefoxitin (FOX), ertapenem (ETP), CLX, and oxacillin with CLX. The review, which included 149 publications, suggests that certain risk factors (prolonged hospitalization and previous use of cephalosporins) are associated with infections by pAmpC-BL-producing microorganisms. The worldwide prevalence has increased over the past 10 years, with a positivity rate ranging between 0.1 and 40%, although AmpC was only detected when sought in a targeted manner. CMY-2 type has been the most prevalent pAmpC-BL-producing microorganism. The most frequently used phenotypic method has been the double-disk synergy test (using CLX disks or phenyl-boronic acid and cefotaxime [CTX] and ceftazidime) and the disk method combined with these inhibitors. In regard to screening methods, a 1-µg oxacillin disk with CLX showed 88.9% sensitivity, 100% specificity, 100% positive predictive value (PPV), 98.9% negative predictive value (NPV), and 98.9% validity index (VI). This predictive capacity is reduced with the addition of extended-spectrum beta-lactamases, showing 62.5% sensitivity, 100% specificity, 100% PPV, 93.5% NPV, and 94.1% VI. In conclusion, there has been a worldwide increase in the number of isolates with pAmpC-BL, especially in Asia, with CMY-2 being the most frequently detected pAmpC-BL-producing type of microorganism. Reduction in its spread requires routine screening with a combination of phenotypic methods (with AmpC inhibitors) and genotypic methods (multiplex PCR). In conclusion, the proposed screening technique is an easy-to-apply and inexpensive test for the detection of AmpC-producing isolates in the routine screening of multidrug-resistant microorganisms.


Introduction
The loss of susceptibility to beta-lactam antibiotics in Gram-negative bacteria is an emerging problem worldwide and is mainly attributable to the production of beta-lactamases, especially extended-spectrum (ESBL) and AmpC type (AmpC-BL) beta-lactamases and carbapen-and plasmid type AmpC-BL (encoded by genes in transfer plasmids) (12). The former enzymes are expressed constitutively and at low concentrations in Citrobacter spp., Enterobacter spp., Serratia spp., Morganella spp., and Providencia spp. (CESPM group), and in Pseudomonas aeruginosa [5,12,13]. However, exposure of these bacterial species to certain beta-lactams can lead to hyperexpression of the encoding gene and elevated production of cAmpC-BL, with the expression of AmpC-BL being inducible. This is attributed to mutations that affect the enzyme responsible for regulating the AmpC-BL gene [3]. Its constitutive expression at low concentrations has also been documented in other bacteria, such as E. coli, and in Shigella spp., but it is not inducible in these cases because the chromosomal genes of the enzyme lack the natural promoter (ampR). Nonetheless, Pfeifer et al. (2010) described cases of resistance to cephalosporins in E. coli mediated by the inducible expression of cAmpC-BL, caused by mutations that increased expression of the enzyme [14]. The rise in plasmid type AmpC-BL over the past few years has been described as an epidemic by some authors [12]. In 1989, it was discovered that ampC genes may be transmittable by plasmids [4,15], after the finding in South Korea of an isolate highly resistant to FOX, designated plasmid CMY-1 [15]. pAmpC-BLs have traditionally been described in Enterobacteriaceae and other Gram-negative bacilli [3,5] (Table 1). The various plasmid AmpC families have been grouped as follows: CIT group of C. freundii (including LAT-and some CMY-, such as CMY-2 and BIL); EBC group of Enterobacter spp. (MIR-1, ACT-1); DHA group of M. morganii (DHA-1, DHA-2); ACC group of H. alvei (highlighting ACC-1); MOX group of Aeromonas spp. (MOX-and the rest of CMY); and FOX group, observing a very close genetic relationship between these plasmids and their chromosomal origins [4,16]. Table 1. Isolation of the first pAmpC-BLs (modified by Jacoby, 2009) [3].

Country of Discovery
Year of Isolation Species that can express pAmpC-BL include K. pneumoniae, Salmonella, E. coli, P. mirabilis, and C. freundii, in which these genes are constitutively expressed at high concentrations [15]. All pAmpC-BLs are expressed constitutively, except for DHA-1, ACT-1, DHA-2, and CMY-13 enzymes. These have been described as inducible because the plasmids that contain them include not only the encoding ampC genes for the enzyme but also ligated ampR genes. These genes are transcription factors responsible for decreasing or increasing the expression of inducible ampC genes depending on the cofactor that interacts with AmpR [3,5,13,15,17,18]. Since the discovery of this type of plasmid, it has been repeatedly observed that the most frequently recorded ampC gene worldwide is CMY-2 [7,14,[17][18][19][20], followed by DHA-1 [18]. Enzymes in the CIT group (CMY-like) are predominant in E. coli, while enzymes of the DHA family predominate in Klebsiella spp. [19][20][21][22][23].
As in the case of Gram-negative bacteria families, AmpC-BL-producing microorganisms produce various types of infection, both nosocomial and community-acquired: urinary tract infections (UTIs, E. coli being the most frequently isolated pathogen in this type of infection), intra-abdominal infections, pneumonias, and soft tissue infections, among others [23]. The literature describes various risk factors associated with infections by pAmpC-BL-carrier microorganisms, observing that they do not differ from those described for ESBLs [24], associating AmpC-BL acquisition with previous hospitalization even more than ESBL acquisition [25].
Independent risk factors have been reported for infection by pAmpC-BL-producing microorganisms, including the previous receipt of fluoroquinolones [2,26,27] and cephalosporins such as cephamycin and FOX [2,3,[26][27][28], demonstrating the possible therapeutic failure of using cephalosporins as empirical treatment (30) even being considered as an independent risk factor [2,25]. In addition, not all AmpC plasmid families are associated with the same mortality rate; thus, according to the report of Pai et al. (2004), infections produced by AmpC-BL DHA-1 have a higher mortality rate than those produced by CMY-2 [29]. Other risk factors are prolonged hospitalization [2][3][4]28,30] or patient institutionalization (in the study by Rodríguez-Baño et al. (2015)) [25], more than 50% of infections were associated with community outbreaks, especially in patients with associated health care), hospitalization in intensive care units (ICUs), and use of central and urinary catheters [3,25,28], mainly in the case of nosocomial infections. Age, presence of diabetes mellitus, hospital admission, institutionalization in care homes, and the use of urinary catheters were associated with community-acquired infections [2].
The worldwide epidemiology of pAmpC-BLs was evaluated by classifying studies according to their target population in the following three groups: hospital (isolates of patients admitted to hospital centers, including ICUs and health institutions), community (isolates from community sampling, studies performed in primary care, infections acquired in the community, and non-hospitalized patients), and hospital and community. The prevalence of AmpC in the reviewed literature was evaluated by considering the total number of isolates in each study. Tables 2-5 display data were obtained from Europe, America, Africa, and countries in Asia, Oceania, and the Middle East, respectively. PRESENCE IN EUROPE (Table 2). Major regional differences can be observed in the percentage of pAmpC-BL positivity. It has remained relatively low in Europe over the past 10 years, ranging from 0.06% Denmark (2010) [31] to 2.6% in Holland (2014) [30,32]. However, higher percentages of positivity have occasionally been observed, such as Holland in 2012 with 5% [33], Spain in 2018 with 14.2% [34], and Germany in 2020 with 11.9% positivity. These higher positivity rates may result from differences in screening method or study population (36). For instance, resistance of clinical isolates to carbapenems was used in the 2018 Spanish study [34] while the 2012 study in Holland evaluated isolates with reduced susceptibility to FOX [33]. pAmpC-BL positivity in E. coli is generally less frequent in Europe than that observed in other parts of the world, with percentages of 0.06% reported in Denmark (2010) [31], 0.46% in France (2010) [35], 0.73% in Holland and Germany (2017) [36], 1.28% in Portugal (2019) [37], and 2.4% in Holland (2018) [38]. The prevalence was slightly higher (7.55%) in the study by Findlay et al. (2020) in England, because all of the clinical isolates evaluated showed CTX resistance, used as a screening method [39]. An elevated prevalence of pAmpC-BL was also obtained in Ireland (19%) because the isolates had a AmpC phenotype (positivity in the phenotypic detection procedure) [40]. In the same way, the elevated percentage (12.5%) described in Switzerland (2013) [41] was obtained in isolates with resistance to third-generation cephalosporins. Reports on the positivity of K. pneumoniae isolates have varied among European countries, with findings of 0.5% in France (2012) [42], 0.47% in Holland (2012) [33], and 1.04% in Portugal (2019) [37]. PRESENCE IN AMERICA (Table 3). Our search of the literature retrieved few publications on the prevalence of pAmpC-BL in America. In general, the global prevalence of pAmpC-BL has been relatively low in the USA, ranging between 1.3% in 2016 [32] and 3.42% in 2019 [52]. Reports on the prevalence of pAmpC-BL in E. coli have ranged widely between 2.23%, as described by Tamma et al. (2019) [52], and 16.33%, reported by Park et al. (2012), who studied the presence of pAmpC-BL in FOX-resistant isolates [53]. In Mexico, Paniagua-Contreras et al. (2018) found a higher prevalence (23.7%) of AmpC-BL among E. coli isolates [54].

Phenotypic Detection Methods
Since the discovery of AmpC type resistances several decades ago, the approach to their detection has been controversial, attributable to the lack of clear guidelines from CLSI or EUCAST. This has led to an underdiagnosis of AmpC-BLs, contributing to underestimation of the prevalence and global spread of this type of resistance. In 2018, Conejo et al. (2018) called for an improvement in the phenotypic detection of AmpC-BLs in Spanish clinical laboratories [97]. The ability of a laboratory to detect both types of AmpC resistance is essential, and the detection of pAmpC-BLs is of vital epidemiological importance, given their transmission and dissemination capacity and their association with outbreaks of community and nosocomial infections [17,98]. AmpC-BL detection is especially difficult in microorganisms that can produce chromosomal and plasmid AmpC-BL (e.g., E. coli). In these cases, the presence of a plasmid needs to be investigated to monitor their spread more closely [99] and address the clinical, therapeutic, epidemiological, and organizational repercussions, including the isolation of infected patients. For the laboratory detection of these resistances, the presence of pAmpC-BLs should be surveilled in species without chromosomal AmpCs that have proven able to disseminate these enzymes, mainly K. pneumonia [100]. The behavior of antibiotics against pAmpC-BL-producing microorganisms is characterized by a decreased susceptibility to oxyiminocephalosporins (e.g., CTX or CAZ) and methoxy-cephalosporins (e.g., FOX), and a susceptibility to fourth-generation cephalosporins (e.g., FEP) [17,101]. Therefore, isolates showing some of these characteristics in the antibiogram should be suspected of pAmpC-BL production when there is no other apparent cause. However, there have been multiple reports of AmpCproducing bacteria that appeared susceptible to both oxyimino-cephalosporins and FOX in the antibiogram [18,19,24,97,99,[102][103][104][105][106]. This may be attributable to a so-called "inoculum effect" (susceptibility in vitro at low microorganism concentrations but inefficacy in vivo) [24,106] or to the conventional consideration of these isolates as susceptible to cephalosporins in vitro according to now-outdated CLSI or EUCAST cutoff points [103,106]. It should be borne in mind that pAmpC-BL-producing isolates occasionally present with more than one beta-lactamase and are multidrug resistant [4,75,94,101]. The presence of AmpC can mask the coexistence of ESBL, hampering differentiation of the two resistances [22,79,99,105,107,108], because the two enzyme groups are hydrolytically very similar, except that AmpC-BL is not inhibited by CLAV, with reports of an increased cephalosporin MIC in its presence [107].      There appears to be a consensus that AmpC-BL screening should consider reduced susceptibility to cephamycins and/or oxyimino-cephalosporins, specifically FOX resistance, alongside the occasional resistance to conventional inhibitors such as CLAV. However, CLAV resistance should be evaluated with caution, given that some isolates may appear susceptible [4]. Accordingly, Meini et al. (2019) proposed routine laboratory FOX screening in laboratories for the detection of AmpC-BLs, proposing FOX MIC > 8 mg/L combined with resistance to CTX and/or CAZ as phenotypic indicator of the presence of pAmpC-BL [17,30]. EUCAST recommends investigating the presence of AmpC when the aforementioned antibiogram results are obtained [98]. Other authors have also recommended screening for susceptibility to FEP, to which AmpC-BLs are susceptible [109]. However, it should be borne in mind that some E. coli isolates can have reduced susceptibility to FEP (MICs ranging from 0.5 to 12 µg/mL), being known as E. coli producing extended-spectrum AmpC (ESAC) beta-lactamases [110]. Other drawbacks of phenotypic tests are their incapacity to differentiate pAmpC-BL producers if they present inducible cAmpC-BL. Known limitations of phenotypic tests also include confluence with other resistance mechanisms such as ESBLs and porin loss [111]. Specific phenotypic tests are based on the detection of AmpC enzyme hydrolytic activity or utilization of AmpC inhibitors and their capacity to suppress beta-lactamase expression [112]. Various phenotypic methods have been proposed for detecting AmpC-producing microorganisms: -Disk approximation method. This technique is employed to detect inducible AmpC-BLs. In the case of pAmpC-BLs, it would be valid for the AmpCs of DHA-1, DHA-2, ACT-1, and CMY-13 families. Two disks are used, one with a substrate antibiotic such as an oxyimino-cephalosporin (e.g., CAZ) or piperacillin/tazobactam, and the other with an inducer antibiotic (e.g., FOX, CLAV, or imipenem, etc.). The microorganism produces an inducible BL if the substrate antibiotic inhibition halo is reduced in the area close to FOX [113]. -Methods with AmpC-specific inhibitors. CLX and boronic acid (BA), and their derivatives, have proven to be the most active and effective commercially available inhibitors to detect AmpC-BLs [105,[113][114][115], with CLX being more specific [114,116]. The combination of CTT with other inhibitors, such as Ro48-1220 and LN-2-128 [117,118] or Syn2190 [101,118], especially Syn2190, have demonstrated high sensitivity and specificity to detect AmpC-producing microorganisms; however, they are not commercially available.  [122], uses Tris-EDTA to permeabilize the bacterial membrane and release beta-lactamases. The bacterium suspected of producing pAmpC-BL (study bacterium) is added to the AmpC disks, which contain Tris-EDTA. The medium is inoculated with an isolate known to be susceptible to FOX (control bacterium), and a FOX disk with AmpC disks (containing the studied bacterium) is placed on both sides. A flattening of the FOX inhibition halo indicates antibiotic inactivation (i.e., the presence of AmpC enzyme released into the medium from the studied bacterium) and therefore a positive result for the presence of pAmpC-BL [28,113,122].
Three-dimensional method. A FOX disk is placed in an agar plate inoculated with a strain susceptible to this antibiotic. An incision is made in the agar near the disk for inoculation with the microorganism under study. The result is positive when the inhibition halo is flattened, which is caused by the growth of the AmpC-producing microorganism [73,113,121]. (e) Mast disks (MastDics ® Combi AmpC and ESBL Detection Set, Merseyside, UK) ( Figure 1A-D) [123]. This technique, which can be used to detect both AmpC and ESBL, utilizes cefpodoxime disks, alone and combined with AmpC inhibitor and/or ESBL inhibitor. The result is positive when the difference in halo diameter between disks with versus without inhibitor is >5 mm [112]. (f) E-test ® AmpC (Biomérieux SA, 69280, Marcy-l Etoile, France) ( Figure 1E)  positive when the inhibition halo is flattened, which is caused by the growth of the AmpC-producing microorganism [73,113,121]. (e) Mast disks (MastDics® Combi AmpC and ESBL Detection Set, Merseyside, UK) ( Figure 1A-D) [123]. This technique, which can be used to detect both AmpC and ESBL, utilizes cefpodoxime disks, alone and combined with AmpC inhibitor and/or ESBL inhibitor. The result is positive when the difference in halo diameter between disks with versus without inhibitor is >5 mm [112].

Genotypic Detection Methods
Multiplex PCR has been considered the gold standard technique for the detection of pAmpC-BL since its development by Pérez-Pérez et al in 2002. It permits the detection and differentiation of plasmid AmpC families and of their chromosomal or plasmid origin (valuable for species with both types of AmpC, such as E. coli) [16]. Modifications have been introduced since its invention, including simple PCRs that offer the same diagnostic

Genotypic Detection Methods
Multiplex PCR has been considered the gold standard technique for the detection of pAmpC-BL since its development by Pérez-Pérez et al. in 2002. It permits the detection and differentiation of plasmid AmpC families and of their chromosomal or plasmid origin (valuable for species with both types of AmpC, such as E. coli) [16]. Modifications have been introduced since its invention, including simple PCRs that offer the same diagnostic value and may even be used for screening (Dallenne et al., 2009] [127]; real-time PCR (Brolund et al., 2010) [128] for rapid detection, using affordable reagents; and, more recently, multiplex real-time PCR (Chavda et al., 2016) [129], which offers the identification of AmpC along with other types of resistance, again in a relatively short time.  [132] created a rapid and more costeffective technique that allowed 96 samples to be tested within 2 h using standard real-time PCR equipment.
It was initially considered impossible for a single isolate to express more than one pAmpC-BL for two reasons: AmpC detection is not sufficiently precise, and the amount of AmpC in the bacterium is too limited for it to be a viable pathogen [16,132]. However, it is now considered possible, given that various authors have reported the finding of more than one ampC gene in the same isolate [45,52,61,66,93].

Behavior of Enterobacteriaceae with AmpC-BL in ChromID ® ESBL Medium, Using Disk Diffusion Test with CLX Disks
The behaviors of test 1 ( Figure 2A) and test 2 ( Figure 2B) are exhibited in Tables 6-8.  [132] created a rapid and more cost-effective technique that allowed 96 samples to be tested within 2 h using standard real-time PCR equipment. It was initially considered impossible for a single isolate to express more than one pAmpC-BL for two reasons: AmpC detection is not sufficiently precise, and the amount of AmpC in the bacterium is too limited for it to be a viable pathogen [16,132]. However, it is now considered possible, given that various authors have reported the finding of more than one ampC gene in the same isolate [45,52,61,66,93].

Behavior of Enterobacteriaceae with AmpC-BL in ChromID® ESBL Medium, Using Disk Diffusion Test with CLX Disks
The behaviors of test 1 ( Figure 2A) and test 2 ( Figure 2B) are exhibited in Tables 6-8. Screening test 1 (Figure 2A) to detect AmpC producers, either as sole resistance mechanism or in combination with ESLB production, had a sensitivity of 31.3%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 88.7%. When the test was applied to detect AmpC producers alone (not combined with ESBL), the sensitivity was 55.6%, the specificity 100%, positive predictive value 100%, and negative predictive value 95.6%. Screening test 2 ( Figure 2B) to detect AmpC producers, either as a sole resistance mechanism or in combination with ESLB production, had a sensitivity of 62.5%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 93.5%. When the test was applied to detect AmpC producers alone (not combined with ESBL), the sensitivity was 88.9%, specificity 100%, positive predictive value 100%, and negative predictive value 98.9%.
Comparison of test results using the Youden index showed that test 2 had greater predictive capacity. Notably, a Youden index of 0.889 was obtained for the detection of AmpC-producing isolates without the presence of ESLB producers.    Screening test 1 (Figure 2A) to detect AmpC producers, either as sole resistance mechanism or in combination with ESLB production, had a sensitivity of 31.3%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 88.7%. When the test was applied to detect AmpC producers alone (not combined with ESBL), the sensitivity was 55.6%, the specificity 100%, positive predictive value 100%, and negative predictive value 95.6%.
Screening test 2 ( Figure 2B) to detect AmpC producers, either as a sole resistance mechanism or in combination with ESLB production, had a sensitivity of 62.5%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 93.5%. When the test was applied to detect AmpC producers alone (not combined with ESBL), the sensitivity was 88.9%, specificity 100%, positive predictive value 100%, and negative predictive value 98.9%.
Comparison of test results using the Youden index showed that test 2 had greater predictive capacity. Notably, a Youden index of 0.889 was obtained for the detection of AmpC-producing isolates without the presence of ESLB producers.

Discussion
This review reveals increasing research interest in pAmpC-BLs over the past few years. Numerous scientific publications have described their detection in the environment, including aquatic media, in food production, and in samples from animals and humans [133].
The reviewed data show that the overall prevalence of pAmpC-BL is higher in the region comprising Asia, Oceania, and the Middle East than in the rest of the world, especially India (44.47% in 2012) [72], Nepal (40.26% in 2020) [95], and Iran (20.50% in 2020) [87]. The lowest prevalence has been reported in Europe, followed by America. Nevertheless, it should be taken into account that some studies did not differentiate between cAmpC-BL and pAmpC-BL, and may therefore have overestimated the prevalence of pAmpC-BL. Prevalence data also vary markedly according to the methodology applied, which may at least in part explain the wide differences reported among countries [51,61,72,86,91,93] and among regions, bacterial species, and study dates [121]. For example, the percentage of pAmpC-BL positivity described by Shrestha et al. (2020) in Nepal was very high (40.26%) because they studied a sample of gram-negative bacilli, in general, including non-fermenting gram-negative bacilli [95].
However, there can be no doubt that the worldwide prevalence of pAmpC-BL has increased, rising globally in the USA from 1.3% (2016) [32] to 3.42% (2019) [52] and in Egypt from 2.91% in E. coli isolates in 2019 [69] to 4.09% in 2020 [70]. In India, the prevalence in E. coli isolates increased from 4.44% in 2012 [81] to 11.10% in 2020 [96]. Major efforts are needed to control the spread of this type of beta-lactamase and avoid hospital/community outbreaks and endemic dissemination, such as that observed between 2010 and 2012 in Hungary, the first European country with endemic dissemination of a pAmpC-BL, specifically DHA-1 plasmid [134].
Although the most frequent type of plasmid worldwide is CMY-2 of the CIT family, a high percentage of DHA-1 isolates has been reported in Asia, especially in China, where DHA-1 is the most commonly isolated plasmid [28,75,84,132]. DHA-1 is also the most frequent plasmid in K. pneumoniae isolates, whereas CMY-2 is the most common in E. coli isolates, as noted above. A striking finding is the predominance of FOX/MOX plasmids recently observed in Mozambique (2021), surpassing both CMY-2 and DHA-1 plasmids [60].
The presence of pAmpC-BL should be suspected when isolates have a BL resistance pattern that differs from their wild phenotype [121], given the well-documented validation of AmpC-BL screening detecting a reduced susceptibility to FOX [17]. Importantly, however, this screening procedure cannot detect AmpC-BLs of the ACC family, which are incapable of hydrolyzing FOX, and microorganisms that express these appear as susceptible to it in the antibiogram [5,17,103,114,116]. Reduced susceptibility to FOX may be due to not only AmpC-BL production but also to a reduced permeability of the external bacterial membrane [20,106]. Among the phenotypic methods reviewed, the "AmpC disk method" is a highly sensitive and specific method that can differentiate resistance to FOX due to the presence of AmpC-BL from that caused by a reduction in external membrane permeability [122], which cannot be achieved by the BA disk synergy method [20].
Phenotypic techniques are simple, generally inexpensive, rapid, and readily interpreted [112,114], favoring their incorporation in routine laboratory analyses to detect possible resistance-producing microorganisms. However, they are not capable of differentiating between pAmpC-BLs and cAmpC-BLs [3,17,19,98,115,116,119] or between different pAmpC-BL families [3]. Hence, a study of E. coli based on phenotypic methods alone would have high sensitivity but low specificity, because it may detect many E. coli isolates that are hyperproducers of chromosomal AmpC [98]. In Klebsiella spp., which lacks cAmpC-BL, a positive phenotypic result for AmpC would be confirmatory because it would imply the presence of pAmpC-BL [120,121]. One limitation of phenotypic methods using BA and its derivatives is that dimethyl sulfoxide is usually employed for their dilution, and this toxic substance needs to be handled with caution. However, their dilution in distilled water has also been described, resolving this problem [113,135]. A further limitation is that BA is not specific for AmpC, and positive results can result from the presence of ESBL and carbapenemases [121].
Few recommendations have been published on the application of these techniques, which require clear guidelines for their interpretation [104]. A good approach to the detection of AmpC may be to apply lower cutoff points than those habitually used for CAZ, ceftriaxone, and CTX. For instance, Agyekum et al. (2016) [136] described all isolates with CTX MIC > 1 mg/L as non-susceptible. Another shortcoming of phenotypic methods is that they are not all equally effective at detecting all AmpC families [20,137], and Ingram et al. (2011) found that inhibitors such as CLX or BA are more sensitive to DHA than CMY [114].
Proposed phenotypic methods that have demonstrated greatest diagnostic usefulness are those based on inhibitors. Thus, acceptable sensitivity and specificity values (>90%) are obtained using the double-disk synergy method with CLX + FOX [20,30,114] and the AmpC disk method [114,122]. The AmpC E-test has obtained the worst results, possibly because it contains CTT, showing lower sensitivity but higher specificity to detect AmpC [114].
The rapid fluorogenic method (1.5 h) recently developed by Park et al. (2020) has demonstrated high sensitivity and specificity to detect pAmpC-BL. It utilizes an antimicrobial bound to a fluorogenic substance that emits fluorescence in the presence of a beta-lactamase capable of its hydrolyzation. This method can also be combined with direct diagnostic techniques (Vitek2 ® or MALDI-TOF ® ) for the rapid detection of the bacterial species [138]. This represents an advancement in the development of novel methods for pAmpC-BL detection.
The coexistence of ESBL and AmpC poses a diagnostic challenge, given that ESBL resistance may go unnoticed in AmpC-producing organisms by presenting this resistance to CLAV. It is also possible that FOX resistance may be produced by a combination of ESBL production and reduced external membrane permeability [120]. A possible solution to the diagnostic challenge posed by the coexistence of ESBL and AmpC is to include FEP, which is not affected by the presence of AmpC, in ESBL screening along with CLAV [17,99]. In cases of ESBL and AmpC coexistence, BA is a more diagnostically valuable inhibitor [128].   [139] modified the CLSI ESBL detection technique, which uses CLAV, by adding BA to CTX or CAZ disks and to CTX/CAZ disks with CLAV, producing CTX/CAZ+BA disks and CTX/CAZ+CLAV+BA disks. An increase of ≥3 mm in the halo of the CTX/CAZ+CLAV+BA versus CTX/CAZ+BA disk is considered positive for ESBL [113,135,140].
Genotypic methods are considered the gold standard techniques for the detection of AmpC resistances [3,30,114] and are able to differentiate between chromosomal and plasmid AmpC-BLs [98]. However, most of them are expensive, technically complex, and time-consuming methods that are mainly used in research and are reserved for doubtful cases in a clinical setting (e.g., E. coli isolates) [140]. Another major drawback is that they can detect ampC genes that are already known but not new mutations or AmpC families [16,98,112,115,132].
Based on the data gathered in this review, we developed an algorithm similar to that depicted in Figure 3. Accordingly, the presence of AmpC-BL should be suspected when there is resistance to cephamycins (MIC > 8 mg/L for FOX) and oxyimino-cephalosporins (pattern of resistance to CTX or CAZ). This helps avoid a search for AmpC-BL in ESBLproducing bacteria, which would meet the second but not the first criterion because they are susceptible to FOX. A phenotypic method should then be applied to confirm the presence of AmpC, with the double-disk synergy test being the most highly recommended approach. Finally, a PCR or a genotypic analysis should be carried out in doubtful cases to verify the presence/absence of AmpC-BL encoding genes, bearing in mind that some isolates present positivity in the double-disk method that even PCR cannot detect [102,135]. Regarding our screening proposal (see Section 2.2) for the phenotypic detection of AmpC-producing Enterobacteriaceae, test 2 obtained the best results. In comparison, Reuland et al. [30] obtained 91% sensitivity and 96% specificity using the double-disk synergy method with CLX in 66 isolates with reduced susceptibility to FOX and third-generation cephalosporins, and 85% sensitivity and 95% specificity when they applied the same method but with BA [30]. The use of CTT disks alone, and with the addition of BA as phenotypic confirmation, was recommended in a study with 635 isolates of Enterobacteriaceae not susceptible to FOX (MIC ≥ 32 mg/dL) [141]. In another study using the doubledisk synergy method to test 255 isolates, the addition of CLX to a FOX disk, obtaining the best predictive values when the halo increase ≥4 mm, was considered positive, achieving 95% sensitivity and 95% specificity [20]. Finally, Polsfuss et al. used the double-disk synergy method in 305 isolates and described 97.2% sensitivity and 100% specificity [114]. With regard to other techniques, Black et al. used the AmpC disk to screen 140 isolates not susceptible to FOX and obtained 100% sensitivity and 98% specificity [122]. Ingram et al. compared different screening and confirmatory methods in a study of 246 isolates, concluding that the screening method with the AmpC disk obtained the best result, offering 95% sensitivity and 98% specificity [109]. In a sample of 125 pAmpC-BL-positive isolates, the CTT/CXT E-Test showed 98.6% sensitivity and 35.4% specificity, while the AmpC disk method obtained slightly lower sensitivity (96%) but higher specificity (58%) values [44]. Hence, the results achieved with test 2 are comparable to the best results described for screening phenotypic tests, with the added advantage of integrating this fourth disk in the screening test with FEP, FOX, and ETP disks, previously proposed by our group for the detection of microorganisms with ESBL and/or carbapenemase [9].
Test 2 yields lower values in the presence of ESBL. As noted above, the coexistence of AmpC and ESBL production hampers the phenotypic detection of both resistance mechanisms, and phenotypic methods are recommended with the addition of CLAV (AmpC inducer and ESBL inhibitor) and even genotypic methods for a definitive diagnosis [17]. Song et al. used BA as AmpC inhibitor in their study of 182 isolates, comparing CTX/CA/BA disks with CTX and/or CAZ/CA/BA with CAZ disks as well as CTX/CA/BA disks with CTX/BA disks and/or CAZ/CA/BA disks with CAZ/BA disks, reporting that both approaches markedly improved sensitivity and specificity values in comparison to the utilization of CA alone [139]. However, a review proposed the double-disk synergy Regarding our screening proposal (see Section 2.2) for the phenotypic detection of AmpC-producing Enterobacteriaceae, test 2 obtained the best results. In comparison, Reuland et al. [30] obtained 91% sensitivity and 96% specificity using the double-disk synergy method with CLX in 66 isolates with reduced susceptibility to FOX and third-generation cephalosporins, and 85% sensitivity and 95% specificity when they applied the same method but with BA [30]. The use of CTT disks alone, and with the addition of BA as phenotypic confirmation, was recommended in a study with 635 isolates of Enterobacteriaceae not susceptible to FOX (MIC ≥ 32 mg/dL) [141]. In another study using the double-disk synergy method to test 255 isolates, the addition of CLX to a FOX disk, obtaining the best predictive values when the halo increase ≥4 mm, was considered positive, achieving 95% sensitivity and 95% specificity [20]. Finally, Polsfuss et al. used the double-disk synergy method in 305 isolates and described 97.2% sensitivity and 100% specificity [114]. With regard to other techniques, Black et al. used the AmpC disk to screen 140 isolates not susceptible to FOX and obtained 100% sensitivity and 98% specificity [122]. Ingram et al. compared different screening and confirmatory methods in a study of 246 isolates, concluding that the screening method with the AmpC disk obtained the best result, offering 95% sensitivity and 98% specificity [109]. In a sample of 125 pAmpC-BL-positive isolates, the CTT/CXT E-Test showed 98.6% sensitivity and 35.4% specificity, while the AmpC disk method obtained slightly lower sensitivity (96%) but higher specificity (58%) values [44]. Hence, the results achieved with test 2 are comparable to the best results described for screening phenotypic tests, with the added advantage of integrating this fourth disk in the screening test with FEP, FOX, and ETP disks, previously proposed by our group for the detection of microorganisms with ESBL and/or carbapenemase [9].
Test 2 yields lower values in the presence of ESBL. As noted above, the coexistence of AmpC and ESBL production hampers the phenotypic detection of both resistance mechanisms, and phenotypic methods are recommended with the addition of CLAV (AmpC inducer and ESBL inhibitor) and even genotypic methods for a definitive diagnosis [17]. Song et al. used BA as AmpC inhibitor in their study of 182 isolates, comparing CTX/CA/BA disks with CTX and/or CAZ/CA/BA with CAZ disks as well as CTX/CA/BA disks with CTX/BA disks and/or CAZ/CA/BA disks with CAZ/BA disks, reporting that both approaches markedly improved sensitivity and specificity values in comparison to the utilization of CA alone [139]. However, a review proposed the double-disk synergy method using CTX and CAZ with and without the addition of CLX as the optimal phenotypic confirmation test for AmpC in the presence of ESBL [119]. The diagnostic usefulness of test 2 in these isolates is enhanced by increasing the amount of CLX on the oxacillin disk.

Limitations
Studies with larger samples of AmpC-producing isolates are required to obtain more accurate predictive values. In common with other phenotypic techniques, our method cannot differentiate between the presence of plasmid or chromosomal AmpC except when the isolate is known to produce plasmid AmpC producer alone, as in the case of K. pneumoniae. In addition, the detection of pAmpC-BL is hampered by confluence with other resistance mechanisms such as ESBL production, porin loss, or E. coli producing ESAC beta-lactamases. Reference methods used for the detection of AmpC producers were the cefotetan/cefotetan-CLX E-Test and an increased CTX and CAZ MIC in the presence of CLAV, because the means required for genotypic identification of the AmpC resistance mechanism were not available.

Conclusions
There has been an increase in pAmpC-BL-producing isolates over the past 10 years, especially in the Asian continent, and CMY-2 producers are the most frequently responsible. Prevention of their spread requires the implementation of routine surveillance procedures that combine phenotypic and genotypic approaches (multiplex PCR). Among phenotypic screening methods, double-disk synergy and AmpC disk methods can be especially recommended for their predictive capacity. Our proposed screening method, which involves the addition of CLX on an oxacillin disk, is an easy-to-use and inexpensive test for the detection of AmpC-producing isolates, especially when there is no other resistance mechanism. Moreover, it can be combined on a single plate with a screening method for the detection of Enterobacteriaceae with ESBL and/or carbapenemases through the addition of FOX, FEP, and ETP disks.