Antimicrobial resistance of Escherichia coli isolates from canine urinary tract infections

Abstract

This study determined the antimicrobial resistance profiles of Escherichia coli isolates from dogs with a presumptive diagnosis of urinary tract infection (UTI). Urine samples from 201 dogs with UTI diagnosed through clinical examination and urinalysis were processed for isolation of Escherichia coli. Colonies from pure cultures were identified by biochemical reactions (n=114) and were tested for susceptibility to 18 antimicrobials. The two most frequent antimicrobials showing resistance in Urinary E. coli isolates were oxytetracycline and ampicillin. Among the resistant isolates, 17 resistance patterns were observed, with 12 patterns involving multidrug resistance (MDR). Of the 69 tetracycline-resistant E. coli isolates, tet(B) was the predominant resistance determinant and was detected in 50.9% of the isolates, whereas the remaining 25.5% isolates carried the tet(A) determinant. Most ampicillin and/or amoxicillin-resistant E. coli isolates carried bla_TEM-1 genes. Class 1 integrons were prevalent (28.9%) and contained previously described gene cassettes that are implicated primarily in resistance to aminoglycosides and trimethoprim (dfrA1, dfrA17-aadA5). Of the 44 quinolone-resistant E. coli isolates, 38 were resistant to nalidixic acid, and 6 were resistant to nalidixic acid, ciprofloxacin and enrofloxacin. Chromosomal point mutations were found in the GyrA (Ser83Leu) and ParC (Ser80Ile) genes. Furthermore, the aminoglycoside resistance gene aacC2, the chloramphenicol resistant gene cmlA and the florfenicol resistant gene floR were also identified. This study revealed an alarming rate of antimicrobial resistance among E. coli isolates from dogs with UTIs.

Escherichia coli (E. coli) is the most important causative bacterium of urinary tract infections (UTIs) in both humans and dogs, and strains of these species are often abundant in the gastrointestinal tract at the time of infection [7, 24, 26, 29]. In past studies of canine UTIs, the majority of bacterial isolates were E. coli (33–56%) [2, 12, 20, 29, 34, 39]. Historically, a range of antimicrobial agents has been used to treat UTIs in veterinary medicine, including penicillins, cephalosporins, tetracyclines, chloramphenicols, aminoglycosides, fluoroquinolones and potentiated sulfonamides. The use of antimicrobial drugs has been associated with an increasing trend of antimicrobial resistance among canine E. coli isolates over the last decade [33]. Thus, the management of UTIs in dogs has become more complicated, as the prevalence of antibiotic-resistant strains of E. coli has increased. In addition, the frequent close physical contact between dogs and humans increases the potential for transmission of resistant bacteria between companion animals and humans, as well as the potential for exchange or transfer of antimicrobial-resistant genes to human pathogens [22]. The objectives of this study were to determine (i) the frequency of canine uropathogenic E. coli at the National Taiwan University Veterinary Hospital (NTUVH) and the National Chiayi University Veterinary Teaching Hospital (NCYUVTH), (ii) the percentage of antimicrobial resistance and (iii) the distribution and mechanisms of bacterial resistance to β-lactams, chloramphenicols, tetracyclines, quinolones, aminoglycosides and sulfamethoxazole-trimethoprim. The results should provide useful information for veterinarians in the management of persistent or recurrent UTI in dogs.

MATERIALS AND METHODS

Sampling methods: Urine samples were obtained from 201 non-medicated adult dogs (aged >1 y) of both sexes with a presumptive diagnosis of UTI from July 2010 to June 2011. The inclusion criteria included clinical signs of UTI, such as hematuria and dysuria, urinalysis results that included red blood cell counts >5 under a high-power field (HPF) and proteinuria, and >10³ colony-forming units (CFU) of bacteria per milliliter of urine at the first plating. Urine samples were obtained by catheterization after thorough cleansing of the genital area.

Bacterial isolation, culture conditions and identification: The fresh urine samples were refrigerated temporarily and were cultured using a calibrated pipette to deliver 10 µl and 100 µl of samples onto Columbia agar supplemented with 5% sheep blood and onto MacConkey agar (Becton Dickinson Microbiology Systems, Cockeysville, MD, U.S.A.) upon receipt. The blood agar plates were incubated with 5% CO₂, and the MacConkey agar plates were incubated aerobically. All samples were incubated at 37°C for 18 to 24 hr until adequate growth was present. Primary plates were carefully inspected for colonies of E. coli, which were plated onto sheep blood agar plates; these plates were incubated at 37°C for 24 hr. Suspected colonies were identified as E. coli using standard techniques, including indole, Methyl red–Voges-Proskauer (MR-VP) and citrate biochemical testing and analysis with an API-20E system (bioMérieux, Marcy l’Etoile, France) [17].

Antimicrobial susceptibility testing: Susceptibility was tested quantitatively by broth microdilution with cation-adjusted Mueller-Hinton broth, according to Clinical and Laboratory Standards Institute (CLSI) guidelines [9]. Eight drug classes were included in this study. These classes represented the most frequently used antibiotics for cases of UTI in our community. Tetracyclines were represented by oxytetracycline and doxycycline. Aminoglycosides were represented by gentamicin and amikacin. Quinolones were represented by nalidixic acid, enrofloxacin and ciprofloxacin. Penicillins were represented by the amoxicillin-clavulanic acid combination, amoxicillin and ampicillin. Lastly, cephalosporins were represented by cefazolin, ceftiofur, ceftazidime and cefotaxime. Sulfamethoxazole-trimethoprim, colistin, florfenicol and chloramphenicol were also tested. E. coli ATCC 25922, Staphylococcus aureus ATCC 29213 and Enterococcus faecalis ATCC 29212 were used as control strains.

Bacterial DNA preparation, PCR assays and DNA sequencing: PCR analyses for preselected resistant genes were performed for all of the isolates. The total DNA of E. coli isolates was extracted by using an InstaGene DNA Purification Matrix kit (Bio-Rad, Hercules, CA, U.S.A.) according to the manufacturer’s instructions. PCR primers specific to these resistant genes were designed based on the gene sequence information from previously published studies [5, 28, 32, 42]. PCR was performed in a 20 µl mixture containing 1 µl of template DNA, 0.5 µl of each primer (10 nmol l⁻¹), 10 µl of 2× PCR Mix (Fermentas, MBI) and 8 µl of ddH₂O. PCR was performed using 30 cycles of denaturation at 94°C and amplification at 72°C for 10 min. Amplicons were visualized by electrophoresis at 100 V in a 2% agarose gel. Amplified PCR products were purified with a QIAquick PCR purification Kit (Qiagen, Valencia, CA, U.S.A.) and sequenced by ABI 3730 × l capillary sequencers (Applied Biosystems, Foster City, CA, U.S.A.). The DNA sequences were compared using the BLAST online search engine from GenBank at the National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov/blast) [1].

The gyrA, gyrB, parC and parE genes were amplified by PCR using the primers and PCR conditions described previously [16]. Both strands of the purified PCR products were sequenced, and the results were compared with the sequences of wild-type E. coli gyrA (NCBI X06373), gyrB (P06982), parC (P20082) and parE (P20083). Comparisons were performed using NCBI BLAST, the ClustalW Multiple Sequence Alignment program and the Lasergene sequence analysis software package (version 4.0, DNASTAR, Madison, WI, U.S.A.) [44].

Statistical analysis: The statistical tests used were the chi-square test and Fisher’s exact test, which were performed using the Mixed Procedure in SAS (version 8.2; SAS Institute, Inc., Cary, NC, U.S.A.). P<0.05 was considered significant.

RESULTS

Bacterial isolates and antimicrobial susceptibility testing: A total of 114 E. coli strains were isolated from 201 urine samples from dogs with suspected urinary tract infections. Among them, 69 isolates were collected at the NTUVH, and 45 isolates collected at the NCYUVTH. The minimum inhibitory concentrations at which 50% and 90% of the isolates were inhibited (MIC₅₀ and MIC₉₀) and the antimicrobial susceptibility profiles for the different antimicrobial agents are summarized in Table 1. The percentages of E. coli strains resistant to oxytetracycline, ampicillin, amoxicillin, nalidixic acid, sulfamethoxazole-trimethoprim, chloramphenicol, florfenicol and doxycycline were 60.5% (69/114), 50% (57/114), 44.7% (51/114), 38.6% (44/114), 34.2% (39/114), 31.6% (36/114), 31.6% (36/114) and 28.9% (33/114), respectively. The lowest levels of resistance were found for gentamicin (10.5%, 12/114), enrofloxacin (5.3%, 6/114), ciprofloxacin (5.3%, 6/114) and amoxicillin-clavulanic acid (2.6%, 3/114). Resistance to cephalosporins (cefazolin, ceftazidime, cefotaxime and ceftiofur), amikacin or colistin was not detected. Resistance to ampicillin, amoxicillin, chloramphenicol, oxytetracycline and sulfamethoxazole/trimethoprim was significantly higher in the isolates from NTUVH (43.5–73.9%) than in those from NCYUVTH (13.3–40.0%) (P<0.05). Isolates from NCYUVTH were significantly more resistant to ciprofloxacin and enrofloxacin (13.3%) than those from NTUVH (0%) (P<0.01).

Table 1. Antimicrobial susceptibility profiles of uropathogenic Escherichia coli.

Class	Antimicrobial agents (mg/l)	NTUVH (n=69)			NCYUVTH (n=45)			Overall E. coli (n=114)
		MIC50	MIC90	R%a,b)	MIC50	MIC90	R% a,b)	MIC50	MIC90	R% a,b)
Aminoglycosides	Amikacin	2	8	0	4	8	0	2	8	0
	Gentamicin	1	2	8.7	1	64	13.3	1	64	10.5
Cephalosporins	Cefazolin	2	16	0	2	8	0	2	8	0
	Cefotaxime	1	1	0	1	2	0	1	2	0
	Ceftazidime	2	2	0	1	2	0	2	2	0
	Ceftiofur	1	1	0	0.5	1	0	0.5	1	0
Penicillins	Ampicillin	512	>1024	60.9	16	1024	33.3	16	>1024	50
	Amoxicillin	>1024	>1024	56.5	8	>1024	26.7	16	>1024	44.7
	Amoxicillin-clavulanic acid	8/4	16/8	0	8/4	16/8	6.7	8/4	16/8	2.6
Phenicols	Chloramphenicol	16	512	43.5	8	32	13.3	8	512	31.6
	Florfenicol c)	4	8	34.8	4	16	26.7	4	16	31.6
Polymyxins	Colistin	1	1	0	1	2	0	1	2	0
Quinolones	Ciprofloxacin	<0.125	0.25	0	<0.125	32	13.3	0.125	0.25	5.3
	Enrofloxacin c)	0.25	0.5	0	<0.125	32	13.3	0.125	1	5.3
	Nalidixic acid	8	512	43.5	8	>1024	33.3	8	1024	38.6
Sulfonamides	Sulfamethoxazole-trimethoprim	4/76	>1024	43.5	1/19	>1024	20	2/38	>1024	34.2
Tetracyclines	Doxycycline	8	128	39.1	2	32	26.7	4	128	28.9
	Oxytetracycline	256	1024	73.9	8	1024	40	8	1024	60.5

a) MIC breakpoints according to CLSI guidelines. b) Percentage of resistant isolates. c) Because the MIC breakpoints for UTIs were not provided in the 2007 CLSI guidelines, the values for bovine respiratory pathogens were used [10].

Resistance to three or more antimicrobials (multiresistance) was observed in 52.6% (60/114) of the isolates from the urine samples (Table 2). None of the isolates were resistant to all of the antimicrobials tested. Resistance to oxytetracycline, amoxicillin and ampicillin was the most common combination in the urine samples and was observed in 42 (36.8%) of the 114 isolates.

Table 2. Distribution of antimicrobial drug resistance patterns among the 114 uropathogenic E. coli strains.

Resistance pattern	No. of isolates
AMP	3
AUG	3
FLO	9
NA	5
CHL, FLO	6
AMP, NA, SUL	3
CHL, NA, OTC	3
AMO, AMP, CHL, OTC	9
AMO, AMP, GEN, SUL	6
DOX, GEN, NA, OTC	3
AMO, AMP, CHL, FLO, OTC	3
AMO, AMP, CIP, EN, NA	3
AMO, AMP, DOX, NA, OTC, SUL	9
AMO, AMP, CHL, DOX, GEN, OTC, SUL	3
AMO, AMP, DOX, FLO, NA, OTC, SUL	6
AMO, AMP, CHL, DOX, FLO, NA, OTC, SUL	9
AMO, AMP, CHL, CIP, DOX, EN, FLO, NA, OTC, SUL	3

AMO, amoxicillin; AMP, ampicillin; AUG, amoxicillin-clavulanic acid; CHL, chloramphenicol; CIP, ciprofloxacin; DOX, doxycycline; EN, enrofloxacin; FLO, florfenicol; GEN, gentamicin; NA, nalidixic acid; OTC, oxytetracycline; SUL, sulfamethoxazole-trimethoprim.

Detections of antibiotic resistance genes: A total of 69 tetracycline-resistant isolates were screened by PCR and sequencing for the 5 tetracycline determinants, tet(A), tet(B), tet(C), tet(D) and tet(E). These results are shown in Table 3. Of the tetracycline-resistant isolates, 79.7% (55/69) were positive by PCR for tet genes, with tet(B) being the most prevalent gene (50.9%, 28/55), followed by tet(A) (25.5%, 14/55), tet(A+B) (14.5%, 8/55), tet(B+C) (5.5%, 3/55) and tet(A+B+C) (3.6%, 2/55); tet(D) and tet(E) were not detected in any isolates.

Table 3. Prevalence of antimicrobial resistance genes among 114 canine E. coli isolates from UTIs.

Antimicrobial agents	Resistance gene (s)	No. of positive isolates
Aminoglycosides	aadA5	18
	aadB	0
	aacC2	10
Amoxicillin/ampicillin	blaTEM-1b	40
	blaSHV-1	14
	blaAmpc	0
Chloramphenicols	Only cmlA	33
	Only floR	15
	cmlA+floR	3
Ciprofloxacin/enrofloxacin	gyrA 83 (Ser-Leu)	6
	parC 80 (Ser-Ile)	6
	aac (6’)-Ib-cr	0
	qnrA, qnrB and qnrS	0
	qepA	0
Nalidixic acid	gyrA 83 (Ser-Leu)	44
Sulfamethoxazole-trimethoprim	dfrA17	30
	dfrA1	3
Tetracyclines	Only tet(A)	14
	Only tet(B)	28
	Only tet(C)	0
	tet(A) + tet(B)	8
	tet(B) + tet(C)	3
	tet(A) + tet(B) + tet(C)	2
	tet(D)	0
	tet(E)	0

The presence of the cmlA and floR genes was investigated in the isolates resistant to chloramphenicol and/or florfenicol; of 36 chloramphenicol- and florfenicol-resistant isolates, all were found to harbor cmlA, and 18 had floR genes. The frequency of detection of the cmlA gene was greater than that of the floR gene. Our results also showed that three E. coli isolates (8.3%, 3/36) were carrying both the floR and cmlA genes.

The fluoroquinolone-resistant isolates (n=6) had 2 point mutations, one in GyrA (Ser-83 to Leu) and one in ParC (Ser-80 to Ile) (Table 3). No mutations were detected in the known resistance regions of GyrB or ParE. Nalidixic acid-resistant isolates (n=44) with a low level of MICs for ciprofloxacin (<0.125–0.25 mg/l) and enrofloxacin (<0.125–1 mg/l) harbored a single mutation in GyrA (Ser-83 to Leu). In this study, the aminoglycoside acetyltransferase variant gene (aac (6’)-Ib-cr) and active drug efflux pump genes (qnrA, qnrB, qnrS and qepA) were not detected in any isolate.

PCRs with primers specific for the bla_TEM, bla_SHV and bla_AmpC genes were performed for the 57 ampicillin- and/or amoxicillin-resistant isolates in this study, and the results are shown in Table 3. A bla_TEM gene, identified by sequencing as bla_TEM-1b, was detected in 40 isolates (70.2%, 40/57). In 14 isolates (24.6%), bla_SHV was detected, and all of these genes were identified by sequencing as bla_SHV-1. The bla_AmpC genes were not detected in any of the isolates.

Overall, 83.3% (10/12) of the gentamicin-resistant isolates were PCR positive for the aacC2 gene. The aacC2 gene was detected in 10 isolates with PCR and confirmed based on DNA sequence. The results showed widespread presence of the aacC2 gene among gentamicin-resistant uropathogenic E. coli isolates.

Of the 114 isolates tested, 28.9% (n=33) carried class 1 integrons. Although the gene cassettes present within each integron ranged in size from 1.0 to 2.0 kb, the most prevalent gene cassette size was around 1.7 kb. The 1.7 kb gene cassette contained the spectinomycin-resistant gene aadA5 and the trimethoprim resistant gene dfrA17. No genes encoding resistance to tetracycline, chloramphenicol or ampicillin were found in the cassette arrays.

DISCUSSION

E. coli has been generally sensitive to antimicrobial agents, and this has made treatment simple [31]. First-line antibiotics are antibiotics that may be chosen empirically or based on culture and susceptibility results targeting a specific bacterium (E. coli in this case) with minimal impact on other bacteria. The first-line antibiotics studied here included amoxicillin and ampicillin, which showed cross-resistance. Other generally effective antibiotics empirically used to treat first-time urinary tract infections are potentiated sulfonamides, cephalosporins, fluoroquinolones and chloramphenicols. They are bacteriostatic agents, which generally demonstrate good efficacy against Gram-positive and/or negative bacteria; however, more recently, several resistant strains have emerged, including strains resistant to aminopenicillins (ampicillin and amoxicillin), chloramphenicols (chloramphenicol and florfenicol), tetracyclines (oxytetracycline and doxycycline) and sulfamethoxazole-trimethoprim [15, 18]. Overusing the same type of antimicrobial agents, including prophylactic use to prevent surgical site infections or infections associated with other urogenital diseases, may have caused the emergence of resistant strains.

A high prevalence of oxytetracycline-resistant strains was observed in this study. The major mechanisms of tetracycline resistance are known to have efflux pump activity, ribosomal protection and enzymatic inactivation. Various tet genes confirm resistance via these mechanisms. Among the 40 tetracycline-resistant genes discovered to date, only the tet(A), tet(B), tet(C), tet(D) and tet(E) genes were associated with the efflux mechanism [8, 43]. Most tetracycline-resistant genes have been found on mobile elements, either as plasmids or transposons [35]. The detection of tet(A), tet(B) and tet(C) genes in tetracycline-resistant uropathogenic E. coli isolates in pets indicated the primary mechanism to be active efflux. The presence of multiple tetracycline-resistance determinants was detected in 11 tetracycline-resistant strains (20%), which carried both tet(A) and tet(B) or both tet(B) and tet(C). Two strains (3.6%) harbored the genes for the tet(A), tet(B) and tet(C) genes. Although 13 of our isolates carried more than one tet gene, they did not exhibit higher MIC values. Bryan et al. explained this phenomenon with the theory that harboring more than one tet gene is due to strong selective pressures rather than a selective advantage [6]. In this study, there were 14 strains with a high level of resistance to oxytetracycline, but none of the 5 tetracycline resistance genes or combinations of them were found. One possibility might be the existence of an additional resistance gene, a potential efflux mechanism or other unidentified factors contributing to the resistance phenotype. The mechanism of resistance in these isolates is the subject of a further study.

Chloramphenicol and florfenicol are broad-spectrum antibiotics that have rarely been used in companion animals. Of the phenicols, only chloramphenicol was used infrequently at NTUVH and NCYUVTH. In this study, resistance to chloramphenicols was moderate, with 31.6% (36/114) of isolates resistant to chloramphenicol and florfenicol. Active efflux pumps (cmlA and floR) played important roles in intrinsic and acquired chloramphenicol and/or florfenicol resistances. Overexpression of efflux pumps affecting chloramphenicol or florfenicol has become increasingly common in E. coli [3]. However, the high incidence of florfenicol resistance in the E. coli isolates analyzed herein was somewhat unexpected. This finding suggested that the selective pressure of chloramphenicol, as well as that the other antimicrobials, may have a relevant role in the emergence and dissemination of florfenicol resistance in E. coli. In addition, the floR gene was not detected in 18 isolates in which the MICs of florfenicol ranged from 8 to 256 mg/l. It may be that other mechanisms, such as overproduction of the AcrAB-TolC multidrug efflux system, may be responsible for florfenicol resistance in these isolates [38].

Multiple mechanisms were involved in the resistance to fluoroquinolones in E. coli. In addition to mutations in chromosomal genes encoding DNA gyrase and topoisomerase IV, a plasmid-mediated quinolone resistance (PMQR) mechanism has also been reported, including Qnr-mediated protection of DNA from quinolone binding, expression of a QepA-encoded efflux pump and aac(6’)-Ib-cr mediated FQ acetylation [19, 36, 45, 48]. In the present study, E. coli urinary isolates from 6 dogs were resistant to enrofloxacin (MIC 32–128 mg/l). All enrofloxacin-resistant isolates had two point mutations, one in GyrA (Ser-83 to Leu) and one in ParC (Ser-80 to Ile) (Table 3). Each enrofloxacin/ciprofloxacin-resistant isolate in this study was also resistant to at least three other antibiotics that were commonly used to treat UTIs. Further characterization of the mechanisms of resistance to enrofloxacin/ciprofloxacin and to the other antibiotics assessed in this study is needed.

Our results demonstrate that canine uropathogenic E. coli isolates have a higher level of resistance to aminopenicillins (50.0% to ampicillin and 44.7% to amoxicillin) and a lower level of resistance to amoxicillin-clavulanic acid (2.6%). This result was consistent with beta-lactam antibiotics, mostly aminopenicillins, being the most commonly used class of antimicrobials in dogs. The results of the study indicated that bla_TEM-1 was the most common β-lactamase among the ampicillin-resistant E. coli isolates (70.2%), in agreement with that results of previous studies that reported a high prevalence of the bla_TEM-1b gene among E. coli isolates [4, 46]. In addition, bla_SHV-1 was found in 14 ampicillin-resistant isolates. The TEM-1 and SHV-1 enzymes were the predominant plasmid-mediated β-lactamases found in gram-negative Enterobacteria [5]. Thus, production of the TEM-1 and SHV-1 enzymes in uropathogenic canine E. coli is of concern. This study confirms the importance in β-lactamase detection in animals. The data indicate that the clinical usage of β-lactams in canine therapy imposes a strong selective pressure in the emergence of resistant bacterial isolates.

In addition to resistance to penicillins, fluoroquinolones, tetracyclines and chloramphenicols, the uropathogenic E. coli isolates were also resistant to other antimicrobials, such as spectinomycin and sulfonamide. Resistance to multiple antimicrobials, especially to spectinomycin and sulfonamides, was often associated with integrons, which are genetic elements that can acquire and spread antibiotic resistance genes [41]. Sulfonamides have been used alone or in combination with trimethoprim for the treatment of UTIs in humans. However, sulfonamides have been infrequently used in dogs, because of susceptibility to adverse effects [47]. Thus, according to our study, 28.9% (33/114) of the E. coli isolates causing UTIs possessed class l integrons, a result that correlated with other reports [40]. However, many studies have documented a higher proportion of class 1 integrons among E. coli clinical isolates from farm animals: 52% of isolates from farm animals [11], 64% of swine diarrhea isolates [27] and 82% isolates from chickens [28]. Since E. coli strains from food production animals were more frequently exposed to antimicrobial pressure than those from companion animals [11], the differences between companion animals, such as cats and dogs, and farm animals may reflect differences in selective pressure due to antibiotic use [37].

Two veterinary hospitals in northern (NTUVH) and southern (NCYUVTH) Taiwan were selected for this study. The 2 hospitals were chosen for their similarities and dissimilarities important for a meaningful comparison of antimicrobial resistance of uropathogenic E. coli isolates. Both are university-affiliated teaching hospitals with innovative equipment and capable faculty/staff providing quality full inpatient and outpatient services. However, the geographical and demographic differences are distinct enough to show inequality in average incomes, animal breeds and density of private practices such that clinical case distributions and drug use could result in diversified antimicrobial resistances and yet be representative of Taiwan’s current status.

Strong evidence supports an association between antibiotic use and resistance in hospitals [25, 30]. In this study, penicillin was the most often used antibiotic class in both animal hospitals, but the penicillins used were quite different. At NTUVH, older and cheaper penicillins with extended spectrum (amoxicillin and ampicillin) predominated, whereas the combination of amoxicillin and clavulanic acid, a commonly used antibiotic to treat sinusitis, bronchitis, otitis media, skin and tissue infections and UTIs, was most commonly used at NCYUVTH. This result suggested a direct link between the higher prevalence of resistance and continuous selective pressure by a single antibiotic. The resistance to sulfamethoxazole-trimethoprim, nalidixic acid and oxytetracycline among the most common pathogens causing urinary tract infections was higher at NTUVH than at NCYUVTH (Table 1). Gentamicin was also very commonly used at NCYUVTH due to its broad antimicrobial spectrum, low cost and convenient once-daily administration. The persistent use of gentamicin and other aminoglycosides may be related to the perception that they are essential components of therapy for life-threatening infections. Unfortunately, the possible toxicity of gentamicin is sometimes overlooked. The rate of resistance to enrofloxacin (13.3%) was also higher in NCYUVTH, as fluoroquinolones are now the most commonly prescribed therapy for UTIs [14, 23] and previous studies have shown cross-resistance to fluoroquinolones in E. coli isolated from UTIs [12, 13, 21]. The results of the present study indicated that all fluoroquinolone-resistant isolates (n=6) from NCYUVTH had higher MICs against enrofloxacin (≥32 mg/l) and that ciprofloxacin resistance (≥16 mg/l) was closely related to the number of topoisomerase mutations in GyrA (Ser-83 to Leu) and ParC (Ser-80 to Ile) (Tables 1 and 3).

The potential role of different infection control practices was not investigated in this study. However, this and a longer hospital stay, which increases the risk for colonization of resistant hospital clones from other small animal patients in the same environment, probably contributed to the higher resistance figures for NTUVH. Other factors that must be considered and are subject to further studies include the total antibiotic pressure in the community, adherence to guidelines for doses and duration of antibiotic treatment and the spread of resistant clones related to good infection control practices.

In conclusion, monitoring antibiotic usage and resistance patterns in a veterinary teaching hospital may serve as an early indicator of changes in the antibiotic susceptibility of clinical isolates. The combination of bacterial culturing and analysis of antibiotic susceptibility and resistance genes provided useful information for veterinarians in the management of persistent or recurrent UTI in dogs. Such extensive evaluation would be unnecessary in most less complicated UTIs in dogs, but the bacteria responsible for persistent infection often eventually incorporate nosocomial resistance plasmids and become increasingly difficult to manage. The techniques used in this study were straightforward and could be applied retrospectively, because the isolates had been maintained. Thus, retaining isolates early is advisable when persistent E. coli infection may become problematic, particularly when a dog has suspected impairments in resistance to colonization and when information about the pattern of E. coli strains is desired.