Human Salmonella and Concurrent Decreased Susceptibility to Quinolones and Extended-Spectrum Cephalosporins

For complicated infections, decreased susceptibility could compromise treatment with drugs from either antimicrobial class.

A lthough antimicrobial agents are not indicated for uncomplicated Salmonella infections, fl uoroquinolones and extended-spectrum cephalosporins are potentially lifesaving treatments for extraintestinal infections (1). The National Antimicrobial Resistance Monitoring System (NARMS) has monitored antimicrobial drug resistance among enteric pathogens since 1996. NARMS has docu-mented decreased susceptibility to each of these drug classes, in most instances among separate serotypes (2). Historically, decreased susceptibility to fl uoroquinolones, which can be monitored by tracking resistance to nalidixic acid, has been noted among Salmonella serotypes (ser.) Typhi, Senftenberg, and Virchow (2,3). More recently, decreased susceptibility to fl uoroquinolones has been noted among Salmonella ser. Enteritidis (4). Decreased fl uoroquinolone susceptibility has also been seen among nalidixic acid-susceptible isolates (5). Extended-spectrum cephalosporin resistance was noted among 15 non-Typhi Salmonella NARMS isolates (including 12 ser. Typhimurium) during 1996-1998 (6). In all instances, extended-spectrum cephalosporin resistance was the result of bla CMY-2 , a class C plasmid-encoded ampC gene (7). In addition to conferring resistance or decreased susceptibility to extended-spectrum cephalosporins such as ceftiofur and ceftriaxone, this gene also confers resistance to ampicillin (AMP), amoxillin-clavulanate, cephalothin, and cefoxitin. This AmpC resistance phenotype has been seen in strains of Salmonella ser. Newport along with resistance to other drugs including chloramphenicol, streptomycin, sulfamethoxazole, and tetracycline. This MDRAmpC strain rose from 1% (1/77) of Salmonella ser. Newport submissions in 1998 to 25% (31/124) in 2001 (4). CMY β-lactamases are largely re-sponsible for extended-spectrum cephalosporin resistance among Salmonella ser. Newport, Typhimurium, and others isolated in North America (6,8).
To understand coresistance to both antimicrobial classes among Salmonella isolates obtained from humans in the United States, we studied the NARMS human collection from 1996 through 2004, looking for decreased susceptibility to quinolones and extended-spectrum cephalosporins. Information for some of the isolates has been presented elsewhere (3,13,(16)(17)(18). We present the molecular epidemiology of this phenotype and mechanisms responsible for its decreased susceptibility.

Isolates and Antimicrobial Drug Susceptibility Testing
NARMS-participating state and local public health laboratories submitted non-Typhi Salmonella isolates to the Centers for Disease Control and Prevention (CDC) for antimicrobial susceptibility testing: every 10th isolate from 1996 through 2002 and every 20th isolate from 2003 to present. Serotypes were determined by the submitting laboratory and, for this study, were confi rmed by the CDC National Salmonella Reference Laboratory according to the Kaufmann-White scheme as described (19). MICs were determined by using broth microdilution (Sensititre, Westlake, OH, USA). Isolates exhibiting an amikacin MIC >4 μg/mL were confi rmed by Etest (ABBiodisk, Piscataway, NJ, USA). Criteria for decreased susceptibility to quinolones and extendedspectrum cephalosporins were as follows: MIC >32 μg/mL for nalidixic acid or >0.12 μg/mL for ciprofl oxacin and >2 μg/mL for ceftiofur or ceftriaxone. Susceptibility testing was performed according to manufacturer's instructions by using control strains Escherichia coli ATCC25922 and ATCC35218, and Klebsiella pneumoniae ATCC700603 (for extended-spectrum β-lactamase [ESBL] confi rmation only). When available, Clinical Laboratory Standards Institute (CLSI) guidelines were used for interpretation (20).

Pulsed-Field Gel Electrophoresis (PFGE)
PFGE was performed as previously described (32). Isolates that produced indistinguishable patterns with XbaI (Roche Molecular Biochemicals, Indianapolis, IN, USA) were restricted with BlnI. Patterns were analyzed by using the BioNumerics version 4.0 software (Applied Maths, Sint-Martens-Latem, Belgium) and compared by unweighted pair group method with averages by using the Dice coeffi cient with a 1.5% band position tolerance window. The DNA sequence and deduced amino acid sequence for the Salmonella ser. Senftenberg bla CMY-23 gene were assigned GenBank accession no. DQ463751.
The mechanisms responsible for resistance and decreased susceptibility are shown in Table 4. Some mechanisms for some of the isolates are presented elsewhere (3,17,18). At least 1 gyrA mutation was found in 26 of 27 isolates. A gyrA mutation at codon 83 only was found for 11 isolates; a mutation at codon 87 only was found for 3; mutations at both codons were found for 12. No functional mutations were detected in gyrB or parE genes. All Senftenberg isolates had parC mutations (S80I and T57S), and 6 other isolates had the T57S mutation. In addition to the T57S mutation in parC, the Mbandaka isolate contained a plasmid-mediated qnrB2 gene and has been described (13). Four isolates contained aac(6′)-Ib, but none contained the ciprofl oxacin-modifying aac(6′)-Ib-cr variant. Nine AmpC phenotype isolates produced β-lactamase with a pI >8.4 (Table 4); 8 contained bla CMY-2 , but the Senftenberg strain contained a bla CMY-23 gene (GenBank accession no. DQ463751) identical to that found in an E. coli isolate (GenBank accession no. DQ438952). This gene differs from bla CMY-2 by 1 amino acid. Three of the bla CMYpositive isolates, including the strain positive for bla CMY-23 , also contained bla TEM-1b . The Mbandaka isolate was positive for bla SHV-30 with pI 7.0 (33) and also produced an enzyme with a pI 7.6, the nature of which is still under study. Two isolates (1 Senftenberg and 1 Typhimurium) contained bla SHV-12 , and both also contained bla OXA and bla TEM-1 genes. Of the 11 Senftenberg isolates, 10 contained bla OXA-1 (n = 9) or bla OXA-9 (n = 1). No isolates contained bla CTX-M genes.

Discussion
Fluoroquinolone and extended-spectrum cephalosporin coresistance is rare; however, the appearance of this phenotype in 2 commonly isolated serotypes from humans (Typhimurium and Newport) is concerning. Sporadic in-fections are alarming, but if clonal expansion of an isolate with this phenotype were to take place, as occurred with Salmonella ser. Typhimurium DT104 and Newport-MDRAmpC, the clinical consequences could be dramatic. Statistically signifi cant increases in resistance to nalidixic acid (odds ratio [OR] 6.7, 95% confi dence interval [CI] CI 2.6-17.7) and ceftiofur (OR 43.2, 95% CI 10.5-177.4) have been documented among non-Typhi Salmonella of human origin submitted to NARMS during 1996-2003 (4). Of 202 nalidixic acid-resistant non-Typhi Salmonella collected by NARMS during 1996-2003, most were ser. Enteritidis (31%) or Typhimurium (10%). Most of the 324 ceftiofur-resistant non-Typhi Salmonella collected by NARMS during the same time period were ser. Newport (56%) or Typhimurium (23%). A slightly broader geographic representation can be found in the SENTRY surveillance project, which analyzed 786 Salmonella isolates (blood and stool) from medical facilities in Latin America and North America (including Canada) during 2001-2003 (8). Of these, 11% were resistant to nalidixic acid, and 2% exhibited decreased susceptibility to ceftazidime, ceftriaxone, or aztreonam.
Extended-spectrum cephalosporin-resistant Newport and Typhimurium isolates are typically obtained from community-acquired infections. Newport-MDRAmpC infections have been associated with consumption of contaminated beef and unpasteurized dairy products (34). Salmonella containing bla CMY genes have been isolated from ground chicken (Typhimurium DT208), turkey (Agona), and beef (Agona) purchased from retail outlets in the Washington DC area (35). In addition, cattle, chickens, turkeys, pigs, horses, and dogs have all been sources of bla CMYcontaining Salmonella, including common serotypes such as Typhimurium, Newport, and Heidelberg (26,36,37). Decreased susceptibility to fl uoroquinolones among Salmonella serotypes that typically carry bla CMY genes warrants exploration of factors that could select for decreased susceptibility to fl uoroquinolones in animal reservoirs and in the human host.
PFGE showed diversity within some serotypes and indistinguishable strains within others. PFGE diversity among 2 serotypes commonly associated with extendedspectrum cephalosporin resistance (Newport and Typhimurium) is not surprising, given that CMY-producing strains have been seen at least since the late 1990s. Isolates of ser. Enteritidis are highly clonal; therefore, PFGE-indistinguishable patterns among isolates with no apparent epidemiologic link are not unusual. All PFGE-indistinguishable Senftenberg isolates from group 1 were isolated in the same state. Results for the Florida Senftenberg isolates are described elsewhere (16,18).
Salmonella ser. Senftenberg exhibiting decreased susceptibility to fl uoroquinolones has been associated with nosocomial infections in healthcare facilities in the United States (18). All 11 isolates contained identical gyrA mutations (S83Y and D87G) and parC mutations (T57S and S80I). These parC mutations have been identifi ed in several Salmonella serotypes including Senftenberg (38). Ten Senftenberg isolates included in this study contained bla OXA genes; the bla OXA -negative Senftenberg strain contained a bla CMY-23 mechanism of extended-spectrum cephalosporin resistance. Acquisition of a bla CMY gene by a traditionally nalidixic acid-resistant serotype warrants further epidemiologic and laboratory investigation. The bla OXA-1 gene has been identifi ed in Salmonella ser. Typhimurium and is reported to be carried by an integron (39); bla OXA-9 has been associated with Tn1331 (40). The epidemiology of Salmonella with decreased susceptibility to fl uoroquinolones is relatively well characterized, as is that of Salmonella with bla CMY -mediated extended-spectrum cephalosporin resistance. Conversely, little is known about the events leading to quinolone and extendedspectrum cephalosporin coresistance and the epidemiology of these infections in humans. Patients with Salmonella infections who exhibit decreased susceptibility to both antimicrobial drug classes should be interviewed to determine risk factors and the effects of antimicrobial drugs and other potential selective factors on this phenomenon.