Clinical significance of Proteus mirabilis bacteriuria in dogs, risk factors and antimicrobial susceptibility

Abstract

The objectives of this study were to describe the in vitro antimicrobial susceptibility and clinical significance of Proteus mirabilis in canine bacteriuria and to identify the risk factors associated with P. mirabilis urinary tract infections. This is a retrospective observational study of 48 P. mirabilis-positive canine urinary cultures. Only 22 of the 48 P. mirabilis isolates (45.8%) were non-susceptible to at least one tested antimicrobial. Most P. mirabilis isolates (98%) were susceptible to enrofloxacin, 93.7% to amoxicillin/clavulanic acid, and 85.4% to ampicillin, cephalothin, and trimethoprim-sulfamethoxazole. Five multidrug-resistant isolates were detected (10.4%). A significant increase in antimicrobial resistance was observed over the study period. Positive P. mirabilis cultures were associated with bacterial cystitis in 36 of 39 dogs (92.3%), pyelonephritis in 2 of 39 dogs (5.1%), and one dog had both bacterial cystitis and pyelonephritis (2.5%). There was no subclinical bacteriuria. Most urinary tract infections were complicated as risk factors were identified in 37 of 39 dogs (94.8%). The most commonly identified risk factors were the presence of a contaminated peri-vulvar area with urine/feces or a hypoplastic vulva. To conclude, P. mirabilis bacteriuria was associated with upper and lower urinary tract infections in this study and was found more frequently in complicated bacterial cystitis. Multidrug-resistant isolates and increased P. mirabilis antimicrobial resistance have been identified over the last 10 years, but most isolates remain susceptible to first-line antimicrobials such as amoxicillin and trimethoprim-sulfamethoxazole.

Introduction

Urinary tract infections (UTIs) occur in approximately 14% of dogs in their lifetime (1). The most common uropathogens isolated from dogs are Escherichia coli, Staphylococcus spp., Enterococcus spp., and Proteus spp. (2). Proteus mirabilis is a Gram-negative rod-shaped bacterium that belongs to the Enterobacteriaceae family and is a causative agent of UTIs in humans (3–5) and companion animals (6,7). It has been identified in 5.4% of canine UTI cases (2). Proteus mirabilis is responsible for both uncomplicated and complicated UTIs in humans (5,8) and dogs (2,9).

Proteus mirabilis is intrinsically resistant to tetracyclines, nitro-furantoin, colistin, clindamycin, fusidic acid, glycopeptides, macrolides, rifampin (10), and benzylpenicillin (11). The emergence of multidrug-resistant (MDR) isolates has been described in both human and veterinary medicine (8,12), which highlights the need for continuous surveillance (12).

This study aimed to: i) describe the in vitro antimicrobial susceptibility of P. mirabilis and its evolution over the study period, ii) describe the clinical significance of the identification of P. mirabilis bacteriuria, and iii) identify the risk factors in dogs with P. mirabilis-associated UTIs.

Materials and methods

Case selection criteria

The electronic database of the Faculty of Veterinary Medicine of the University of Montreal was searched for P. mirabilis-positive urine samples that were collected between July 2008 and October 2018. To be included, samples had to be obtained through cystocentesis, aseptic catheterization, or through the operating channel of a cystoscope. Antimicrobial susceptibility results, extracted from the diagnostic service of the faculty, were also required to be included. When several positive cultures were identified in the same dog, only the first positive culture was included. The medical records of the identified cases were reviewed.

Culture, bacterial isolates, and antimicrobial susceptibility monitoring

All samples were cultured aerobically using standard methods and inoculated on plates containing Columbia blood agar base (BD Difco; Fisher Scientific, Ottawa, Ontario) with 5% sheep blood. The plates were incubated at 35°C ± 2°C with 5% carbon dioxide. If the sample was obtained by cystocentesis, urine cultures with ≥ 1000 colony-forming units (CFU)/mL were considered positive and if the sample was obtained by catheterization or transurethral cystoscopy, a threshold of ≥ 10 000 CFU/mL in males and ≥ 100 000 in females was considered positive (13).

From July 2008 to July 2017, bacteria were identified using conventional biochemical tests. Thereafter, identification was performed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. All isolates were tested for antimicrobial susceptibility using the Kirby-Bauer disk diffusion method according to the guidelines published by the Clinical and Laboratory Standards Institute (CLSI) (10). Isolates were classified as susceptible to an antimicrobial according to the guidelines available at the time of analysis and breakpoints for Enterobacteriaceae (dog and human) issued by CLSI (14–17) were used. Isolates were considered non-susceptible if they were intermediate or resistant. A P. mirabilis isolate was considered MDR if it was non-susceptible to at least one agent in at least 3 antimicrobial classes, including acquired resistance only (18). If polymicrobial cultures were observed, only the antimicrobial susceptibility of P. mirabilis was included in the analysis. To study the evolution in antimicrobial susceptibility over time, we divided the study into 2 equal periods — July 2008 to August 2013 (period 1) and September 2013 to October 2018 (period 2) — and the proportions of non-susceptible isolates in these periods were compared.

Clinical data collection

Historical findings

The following criteria were extracted from medical records: signal-ment, antecedent of UTI with positive culture, antimicrobial treatment 30 d prior to inclusion, concomitant administration of other drugs at inclusion time, and presence of clinical symptoms of UTI at presentation (e.g., inappropriate micturition, polyuria, dysuria, stranguria, pollakiuria, or hematuria) as well as comorbidities.

Retrospective diagnosis

Positive cultures were retrospectively defined as bacterial cystitis, pyelonephritis, or subclinical (19,20), as described in Table I.

Table I.

Description of the diagnostic tools extracted from the medical records to classify the Proteus mirabilis bacteriuria.

Diagnostic tools	Bacterial cystitis	Pyelonephritis	Subclinical bacteriuria
Clinical signs	At least one of the following: pollakiuria, stranguria, hematuria, or dysuria	At least one of the following: elevated body temperature (rectal temperature ≥ 39.4°C), vomiting, lethargy, loss of appetite, pain elicited by renal palpation, or polyuria/polydipsia	No clinical signs suggestive of UTI
Clinicopathologic findings	And pyuria (> 0 to 5 white blood cells per hpf in the sediment when urine was collected by cystocentesis)	And azotemia (not attributed to a prerenal or postrenal cause, with increased blood urea nitrogen and creatinine > 125 μmol/L) ± abnormal leukocyte count or toxogram: leukocytosis (≥ 13.9 × 109 leukocytes/L), neutrophilia (≥ 8.1 × 109 neutrophils/L), or neutropenia (≤ 3.9 × 109 neutrophils/L)	And no clinicopathologic signs suggestive of UTI
Imaging findings	And/or ultrasonographic evidence of cystitis (e.g., diffuse, thickened, or irregular bladder wall)	And ultrasonographic evidence of pyelonephritis (e.g., renal pelvis dilation, retroperitoneal effusion, perineal steatitis, increased echogenicity of the renal parenchyma)	And no imaging findings suggestive of UTI
Evolution	And resolution of the urinary clinical signs after antibiotherapy	And improvement of the azotemia or imaging findings after antibiotherapy

Risk factors of UTI

Risk factors were examined and UTIs were further categorized as complicated if: i) the infection was associated with an anatomical/functional abnormality or comorbidity that could predispose the dog to persistent or recurrent infections, or ii) the treatment was unsuccessful even though it was appropriate (13,21). The risk factors were categorized as follows: altered urothelium, obstructive uropathy, impaired immunity, altered urine composition, or anatomic abnormality of the urinary tract (9,22,23). Presence of urolithiasis and encrusted urothelium or cystitis (i.e., superficial or deep mineralization of the urothelium) (24) was also recorded.

Statistical analysis

Descriptive statistics were used to analyze the epidemiologic data, the clinical findings, and the UTI risk factors. A Fisher’s exact test was conducted to compare the frequency of non-susceptible isolates in the 2 study periods and then to compare the 2 periods in terms of resistance of the isolates to each antimicrobial. Associations between a previous antimicrobial treatment and antimicrobial resistance or MDR were analyzed using contingency tables with Fisher’s exact test. Statistical significance was indicated by P < 0.05 for all comparisons. An R software package (R Core Team, Vienna, Austria) was used for the analysis.

Results

Cultures and antimicrobial susceptibility

Over the 10-year study period, P. mirabilis was cultured in 58 urine samples. Three positive cultures were excluded because of the unavailability of antimicrobial susceptibility results and 7 others were excluded because a positive culture had already been included for the same dog. In those dogs, P. mirabilis isolates had the same antimicrobial susceptibility results. Therefore, only 48 cultures were included in this study. All urinary samples were collected through cystocentesis, except for 2 samples, which were obtained through the operating channel of a cystoscope. Thirty-eight cultures (79.1%) were pure. Nine cultures had a unique concomitant bacterium identified, which was Enterococcus spp. (n = 6), E. coli (n = 2), or a bacterium from the S. intermedius group (n = 1). Only one culture had 2 other concomitant bacteria isolated (Enterococcus spp. and E. coli).

Of the 48 P. mirabilis isolates, 22 (45.8%) were non-susceptible to one antimicrobial and 13 (27%) were non-susceptible to more than one tested antimicrobial. Antimicrobial susceptibility for both periods is presented in Table II. Non-susceptible isolates significantly increased over time (29.6% to 66.7%; P = 0.018). However, these study periods did not significantly differ in terms of the resistance of the isolates when each antimicrobial was considered individually. Multidrug-resistant P. mirabilis isolates were identified in 5/48 dogs (10.4%) (Table III). All MDR isolates were isolated during period 2.

Table II.

Non-susceptible Proteus mirabilis proportions during the study period and comparison of the proportions of non-susceptible isolates between the 2 study periods. P-value is based on Fisher’s exact test to compare the non-susceptible proportion between the 2 study periods (period 1: July 2008 to August 2013; period 2: September 2013 to October 2018).

Antimicrobial class	Tested antimicrobial	2008 to 2018		Period 1		Period 2		Comparison of non-susceptible P. mirabilis proportions between the 2 study periods (P-value)
		Number of tested P. mirabilis	Proportion of non-susceptible P. mirabilis (%)	Number of tested P. mirabilis	Proportion of non-susceptible P. mirabilis (%)	Number of tested P. mirabilis	Proportion of non-susceptible P. mirabilis (%)
β-lactam
β-lactam inhibitor	Amoxicillin/clavulanic acid	48	6.3	27	3.7	21	9.5	0.57
Penam penicillin	Ampicillin	48	14.6	27	3.7	21	28.6	0.07
Cephalosporin	Cephalothin	48	14.6	27	7.4	21	23.8	0.21
Fluoroquinolone	Enrofloxacin	48	2	27	0	21	4.7	0.43
	Orbifloxacin	48	14.6	27	18.5	21	9.5	0.44
Sulfonamide	Trimethoprim-sulfamethoxazole	48	14.6	27	7.4	21	23.8	1
Aminoglycoside	Gentamicin	48	10.4	27	3.7	21	19	0.15
Phenicol	Chloramphenicol	11	36.4	1	0	10	40	1

Table III.

Antimicrobial susceptibility results of multidrug-resistant Proteus mirabilis identified in 5 dogs and the number of resistances to antimicrobial classes.

Dogs	Antimicrobial classes								Number of resistances
	Am	AC	C	E	O	TMS	G	Cl
1	NS	S	S	S	S	NS	NS	N/A	3
2	NS	S	NS	S	S	S	NS	N/A	3
3	NS	S	S	S	NS	S	NS	N/A	3
4	S	S	S	NS	S	NS	NS	S	3
5	NS	NS	NS	S	S	NS	S	S	3

Am — amoxicillin; AC — amoxicillin-clavulanic acid; C — cephalothin; E — enrofloxacin; O — orbifloxacin; TMS — trimethoprim-sulfamethoxazole; G — gentamicin; Cl — chloramphenicol; NS — non-susceptible; S — susceptible; N/A — not assessed.

Clinical results

Study population

Thirty-nine dogs had available medical records (36/39 females and 3/39 males). Thirty-one females (86.1%) were spayed and 2 males were neutered. The median age was 4.5 y (range: 2 mo to 17 y) and the median body weight was 27.1 kg (range: 1.1 to 66.5 kg). No predominant breed was identified.

Historical findings

Of the 39 dogs, 18 (46.1%) had a history of UTIs (bacterial cystitis or pyelonephritis) and P. mirabilis had been previously isolated in 3/39 dogs (7.6%). Sixteen of the 39 (41%) dogs had received at least one course of oral or topical antimicrobial treatment 30 d prior to their enrolment. The administered antimicrobials included cephalexin (n = 4), amoxicillin (n = 3), amoxicillin/clavulanic acid (n = 3), polymyxin B (n = 2), doxycycline (n = 1), cefazolin (n = 1), or an unknown antibiotic (n = 2). Two patients were still receiving antimicrobials at the time of inclusion. Proteus mirabilis isolates that were non-susceptible to at least one tested antimicrobial were identified in 6/39 dogs that had received an antimicrobial 30 d prior to inclusion and 2 of these isolates were characterized as MDR. The administration of an antimicrobial 30 d prior to inclusion was not associated with the presence of antimicrobial resistance (P = 1) or with the characterization of P. mirabilis as MDR (P = 1).

Thirty-seven dogs (94.8%) had obvious clinical signs related to upper or lower urinary tract infection (inappropriate micturition, polyuria, dysuria, stranguria, pollakiuria, or hematuria) with a median duration of 7 d (interquartile range: 27.5 d). One dog had signs of urinary incontinence only and the last dog had polyuria and polydipsia only.

Retrospective diagnosis

Bacterial cystitis was diagnosed in 36/39 dogs (92.3%), pyelonephritis in 2/39 dogs (5.1%), and both bacterial cystitis and pyelonephritis in 1/39 dog (2.5%). None had a subclinical bacteriuria. Urine analysis results were available in all dogs, bloodwork (biochemical profile and hematology) in 25/39 dogs, abdominal ultrasound in 22/39, and cystoscopy in 6/39 dogs. Only relevant results associated with a diagnosis of bacterial cystitis or pyelonephritis are reported. The most relevant abnormalities detected by urinalysis were pyuria, alkalinuria (pH > 7) (median: 9, range: 6 to 9), and struvite crystalluria, which were recorded in 38/39 (97.4%), 36/39 (92.3%), and 12/39 (33.3%) dogs, respectively. Azotemia secondary to acute kidney injury was reported in the 3 dogs diagnosed with pyelonephritis (mean creatinine: 263 μmol/L, range: 197 to 301 μmol/L; mean blood urea nitrogen: 27.61 mmol/L, range: 18.39 to 33.06 mmol/L). Abdominal ultrasound was available for 22/39 dogs. Ultrasonographic evidence for cystitis and pyelonephritis was detected in 16/22 (72.7%) and 3/22 dogs (13.6%), respectively. Moreover, evidence for polypoid and pseudomembranous cystitis was recorded in 3/22 dogs (13.6%), and evidence for encrusted cystitis (in the superficial or deep layer of the bladder wall) was recorded in 2/22 dogs (9%). Urolithiases were detected on abdominal ultrasound in 11/22 dogs (50%).

UTI risk factors

Of the 39 dogs with a UTI (bacterial cystitis or pyelonephritis), 37 (94.8%) displayed potential risk factors. Among them, 22/37 (59.5%) had multiple potential risk factors; therefore, most UTIs were classified as complicated. Details concerning the identified potential risk factors are presented in Table IV.

Table IV.

Distribution of UTI risk factors in dogs with a Proteus mirabilis-associated complicated UTI. Total number of risk factors exceeds number of dogs as some dogs had multiple predisposing factors listed.

Risk factors class	Specific risk factors	Number of dogs	Number of dogs/ risk factors class
Anatomical abnormality of the urogenital tract	Hypoplastic vulva	15	36
	Vaginal stenosis	5
	Prominent dorsal vulvar skin fold	5
	Vestibulovaginal remnant	4
	Ectopic ureter	3
	Polypoid cystitis or pseudomembranous cystitis	3
	Pelvic bladder and short urethra	2
	Urethral diverticulum or septum	1
	Urachal remnant	1
	Vaginal mass	1
Alteration of urine composition	Unspecified low USG*	14	19
	Chronic kidney disease	3
	Diabetes mellitus	2
Alteration of the urothelium	Urolithiasis in the urinary bladder	8	18
	Dystrophic mineralization (kidney)	3
	Urolithiasis in the urethra/ureter*	2
	Encrusted urothelium	2
	Urolithiasis in the kidney	1
	Ureteritis	1
	Transitional cell carcinoma	1
Abnormal micturition
Obstructive uropathy	Urolithiasis in the urethra/ureter	1	17
	Ureteral stricture/stenosis	1
	Transitional cell carcinoma	1
	Urethral stenosis with hydronephrosis	1
Urinary incontinence	Urethral sphincter mechanism incompetence	5
	Neurogenic micturition disorders	4
	Unspecified urinary incontinence	4
Impaired immunity	Corticosteroids/immunosuppressive therapy	9	14
	Chemotherapeutics	3
	Hyperadrenocorticism	1
	Hypothyroidism	1
Others	Contaminated peri-vulvar area with urine or feces	19	24
	Peri-vulvar dermatitis	7
	Prostatitis/prostatic hyperplasia	1
	Perineal hernia previously treated with herniorrhaphy	1
	Balanoposthitis	1
	Vaginitis	1
	Indwelling urethral catheter	1
	Urinary implant (stent/SUB)	2

^*Non-obstructive.

UTI — urinary tract infection; USG — urine specific gravity; SUB — subcutaneous ureteral bypass.

Discussion

Proteus mirabilis-associated UTI has been well-described in human medicine (4,8,25), but it remains rarely reported in the veterinary literature (2,9,21,26). This study describes the in vitro antimicrobial susceptibility of P. mirabilis and the changes in antimicrobial susceptibility over the study period in 39 dogs. Moreover, this study describes the clinical significance of P. mirabilis bacteriuria in dogs that presented to a veterinary teaching hospital.

In this study, 45.8% of the P. mirabilis isolates were non-susceptible to at least one antimicrobial and MDR was detected in 10.4% of the isolates. However, most P. mirabilis isolates were susceptible to commonly used antimicrobials. The in-vitro susceptibility of the P. mirabilis isolates in this study was higher than the in-vitro susceptibility among all isolates identified in the canine urinary tract described in a previous study based on 1028 UTI cases (2). That is, 98% of P. mirabilis isolates were susceptible to enrofloxacin (versus 74.6% of all isolates in the previous study), 93.7% to amoxicillin/clavulanic acid (versus 76.7%), and 85.4% to ampicillin (versus 59.4%) (2). The susceptibility to trimethoprim-sulfamethoxazole was similar (85.4% versus 85.7%) (2) and 85.4% of P. mirabilis isolates were susceptible to cephalothin. Therefore, in accordance with the current recommendations for empirical treatment of uncomplicated low UTIs, the use of amoxicillin or trimethoprim-sulfamethoxazole as first-line treatment (20) is appropriate for P. mirabilis-associated bacterial cystitis. In cases of pyelonephritis, an empirical treatment is not recommended and an antimicrobial should be chosen based on antimicrobial susceptibility results. However, the use of amoxicillin is not recommended and fluoroquinolones or cefpodoxime may be preferred (20).

A significant increase in antimicrobial resistance was observed over the study period, which is consistent with the temporal increase in resistance in Enterobacteriaceae described in a veterinary medicine study (27). However, despite the observed increasing trend in antimicrobial resistance for the tested antimicrobial (especially ampicillin), no significant increase in resistance was observed for each individual antimicrobial. This discrepancy might be due to the small study population. Companion animals are considered potential reservoirs for the dissemination of P. mirabilis to humans (6). Thus, the increase in resistance and the detection of MDR raise public health concerns and may cause future therapeutic limitations in veterinary medicine (27). The precautious use of antimicrobials is still recommended and should be based on the results of antimicrobial susceptibility tests.

Surprisingly, no association was observed between the administration of an antimicrobial treatment 30 d prior to inclusion to the study and the presence of antimicrobial resistance. A study has shown that administration of amoxicillin, doxycycline, and enrofloxacin 30 d before the presentation of an uncomplicated UTI was associated with resistance to these antimicrobials in dogs (2). These conflicting results could be explained by the small number of cultures included in the present study as well as by the retrospective study design (i.e., inappropriate keeping of medical records).

The presence of P. mirabilis bacteriuria in dogs must be considered clinically relevant given that subclinical bacteriuria was not diagnosed in our study. Most dogs presented clinical signs of urinary tract infection (either acute or chronic), alkaline urine, or pyuria, and the most frequent retrospective diagnosis associated with P. mirabilis bacteriuria was bacterial cystitis. The proportion of subclinical bacteriuria might have been underestimated due to our study design, which, in most cases, involved submitting a urine culture when clinicians suspected a UTI. The identification of potential UTI risk factors was essential to avoid persistent or recurrent infections and treatment failure since most P. mirabilis-associated UTI cases were considered complicated. The proportion of P. mirabilis-associated complicated UTI (94.8%) in our study was higher than previously reported in dogs. In a study based on 1028 dogs with a UTI, 58% (52/89) of P. mirabilis-associated UTIs were classified as complicated (2). Because both studies were conducted in a tertiary care hospital, which received referrals for recurrent UTIs, it is unlikely that the discrepancy was due to selection bias; however, the small study population in the present study could possibly explain it. In both studies, the proportion of complicated P. mirabilis-associated UTIs was higher than the proportion of uncomplicated P. mirabilis-associated UTIs (2). Moreover, as urine cultures were not systematically submitted to the laboratory by the clinicians, especially when an uncomplicated bacterial cystitis was suspected and when the administration of an empirical antimicrobial treatment was chosen, the number of P. mirabilis isolates might have been underestimated.

The most frequent risk factors identified in dogs with a P. mirabilis-associated UTI were conditions that could predispose these dogs to local urine retention, maceration of the peri-vulvar area due to the presence of urine or feces, or a hypoplastic vulva. It is likely that the major contamination route in dogs was the cross-contamination of the peri-urethral area with the host intestinal tract flora, similar to that described in humans (28). Understanding the most common contamination route may be useful in preventing the occurrence of P. mirabilis-associated UTIs.

Although they might be direct consequences or complications of P. mirabilis-associated UTIs, urolithiasis and encrusted urothelium were identified as UTI risk factors in our study because they can predispose dogs to recurrent infection or treatment failure. Proteus mirabilis is a potent urease enzyme-producing bacterium (29) and it has rarely been identified as the cause of encrusted cystitis in human medicine (25). In this study, urolithiasis was observed in 11 dogs and encrusted cystitis was strongly suspected in 2 cases based on abdominal ultrasound (24,30) and one case based on the mineralization observed during cystoscopy. Detection of both conditions (urolithiasis and encrusted urothelium) was necessary because each required a specific therapeutic approach, such as the use of an adequate antimicrobial and adequate treatment duration, the use of an acidifying diet, and sometimes, surgical resection of encrustation (30–32). In human medicine, in-vitro studies have shown that the presence of other uropathogens promotes the survival of P. mirabilis and it is believed that this synergistic effect could be responsible for the persistence and recurrence of urolithiasis in the urinary tract (33). In the present study, most of the urine samples (79.1%) revealed monobacterial positive cultures consistent with the findings reported in the literature (2,7). The detection of polymicrobial cultures should not be ignored and should be treated adequately. However, in polymicrobial cultures with Enterococcus spp., the infection may resolve when the other organisms are successfully treated (13).

The limitations of the present study are related to its retrospective nature. Antimicrobial susceptibility monitoring tests were performed and antimicrobial susceptibility was interpreted using the most recent update from the Clinical and Laboratory Standards Institute at the time. Over the 10-year study period, there were 3 updates — July 2013, June 2015, and June 2018 — and all were taken into account in our laboratory. A relevant update that would modify the results or the interpretation presented in this study was not identified over the study period. The second relevant limitation is the lack of standardization of the available medical records used to establish a retrospective diagnosis of bacterial cystitis, pyelonephritis, or subclinical bacteriuria. Finally, the small population included in the present study might have led to type II errors.

In conclusion, Proteus mirabilis can be involved in complicated bacterial cystitis, especially in the case of urine or fecal incontinence that causes contamination in the peri-vulvar area. Multidrug-resistant isolates and an increase in antimicrobial resistance of P. mirabilis have been identified over the last 10 y, but most of these isolates are susceptible to commonly used antimicrobials. Cautious use of antimicrobials is recommended, but trimethoprim-sulfamethoxazole or amoxicillin may be appropriate empirical antimicrobials. Identification of UTI risk factors is strongly recommended to avoid recurrent UTIs.