Prevalence and Antibiotic Susceptibility of Bacterial Isolates From Dogs With Ulcerative Keratitis in Midwestern United States

The objective of this study was to describe bacterial culture and antibiotic susceptibility results in 476 dogs presenting with suspected bacterial keratitis in Iowa and surrounding Midwestern states, further detailing trends in patient characteristics, seasonality, and antimicrobial resistance. Corneal swabs yielded 465 bacterial isolates and 220 cultures (46.2%) with no apparent growth (0–5 isolates per culture). The most frequent bacterial genera were Staphylococcus (32.3%), Streptococcus (19.1%), and Pseudomonas (12.5%), while the most common bacterial species were Staphylococcus pseudintermedius (26.7%), Streptococcus canis (12%), and Pseudomonas aeruginosa (7.5%). Compared to mixed-breed dogs, canine breeds most likely to be examined for ulcerative keratitis included Boston terrier, Cavalier King Charles spaniel, miniature pinscher, pug, rat terrier, Saint Bernard, shih tzu, and silky terriers. In summer, the likelihood to yield a negative culture was reduced while the likelihood to culture Pseudomonas species was increased. Bacteria considered multidrug resistant (MDR, resistant to ≥ 3 antibiotic classes) represented 20% of all canine isolates and were most prevalent for Staphylococcus species (33%). An alarming, escalating trend of MDR prevalence was noted between 2016 (5%) and 2020 (34%). Individual ophthalmic preparations (i.e., single antibiotics or commercially available antibiotic combinations) with highest efficacy against all bacterial isolates included chloramphenicol (83%), ceftiofur (79%), amikacin (77%), neomycin-polymyxin B-bacitracin (77%), and gentamicin (74%). Efficacy of systemic antibiotics and combinations of ophthalmic preparations was also evaluated. Based on the present findings, triple antibiotic (Neo-Poly-Bac) is recommended as empirical monotherapy for prophylactic antibiotic therapy in dogs with simple corneal ulcers, while a chloramphenicol-ciprofloxacin combination is empirically recommended for therapeutic management of infected corneal ulcers. Pending culture and susceptibility results, appropriate selection of empiric antibiotic therapy is important to enhance therapeutic outcome and reduce antibacterial resistance in dogs with corneal ulceration.


INTRODUCTION
Bacterial keratitis is a major global cause of ocular discomfort and visual impairment in dogs and other species. Following an injury to the eye from trauma or other causes, corneal wounds in dogs have a high tendency toward infection given the presence of indigenous microflora on the corneal and conjunctival surfaces, including Staphylococcus species, Streptococcus species, Pseudomonas species, and gram-positive bacilli (1)(2)(3). Rapid and appropriate use of antibiotics is critical in mitigating the potential devastating effects of bacterial keratitis in dogs, including keratomalacia (corneal melting), corneal perforation, scarring, and loss of vision or the entire eye (4,5). Successful management requires the clinician to be aware of the most common bacterial isolates and their susceptibility profiles to antibiotics, as there is often a lag period of several days until corneal samples collected from the patient provide results for bacterial culture and sensitivity testing.
The distribution of bacterial populations has been described previously for several geographic regions of the world including Asia (3,6,7), Australia (1,8), Europe (9), North America (10,11), and South America (2). While there are distinct similarities between these regional investigations, there are notable differences in bacterial prevalence and susceptibilities across the world. Further, differences within separate geographic regions of even a single country (10,11) demonstrate that more targeted evaluation of local or regional populations is needed. Most previous reports in veterinary literature are limited to describing the prevalence of bacterial isolates and susceptibility profiles for individual antibiotics (6,7,10). Such information is useful but is generally insufficient to provide optimized recommendations for clinicians managing bacterial keratitis in dogs. For instance, it is common practice to use two different ophthalmic antibiotics in veterinary patients with an infected corneal ulcer (12)(13)(14)(15) or reach for systemic antibiotics in patients with vascularized corneal lesions or corneal perforations. However, to our knowledge, specific comparisons of antibiotic combination efficacies on bacterial isolates has not been described in dogs or other species, and only selected canine reports have described the efficacy of systemic antibiotics for corneal disease (16,17). Using fewer, more effective drugs for an appropriate timeframe allows for better patient outcomes and less antibiotic resistance. Less antibiotic resistance in the environment will benefit both veterinary and human medicine.
The primary goal of the study was to describe results of bacterial cultures and susceptibility testing (ophthalmic and systemic profiles) in canine patients that presented with bacterial keratitis from Iowa and surrounding Midwestern states of the United States. A secondary objective was to report the prevalence and trends over time (2014-2020) of antibacterial resistance in corneal isolates, as well as risk factors associated with positive bacterial growth from corneal samples. We hope this information will provide clinically relevant data for managing bacterial keratitis in dogs and help mitigate the alarming rise of antibacterial resistance in the species.

Data Collection
The database of Iowa State University's Veterinary Diagnostic Laboratory (ISU VDL) was searched for results of bacterial cultures collected from canine corneas, as well as associated antibiotic susceptibility testing when available. The search covered a period from March 2014-when an ophthalmicspecific susceptibility profile was introduced-to the date of manuscript writing (June 2020). Corneal cultures processed by ISU VDL originated from two sources: (i) In-house submissions from the ISU Lloyd Veterinary Medical Center (ISU LVMC), for which information about patients characteristics (i.e., age, breed) was available; and (ii) Mail-in submissions from veterinary practices in Iowa and surrounding states (Minnesota, Missouri, South Dakota).

Sample Identification and Susceptibility Testing
At the ISU LVMC, samples were collected with sterile culturette swabs (BBL TM CultureSwab TM , Becton Dickinson and Company, Sparks, MD) that were pre-moistened prior to contact with the corneal wound and processed for aerobic microbiologic assessment using a non-selective medium (tryptic soy agar with 5% sheep blood [blood agar]) and a Gram-negative selective medium (MacConkey). The blood agar was incubated at 35 ± 2 • C with 5-10% CO 2 for a total length of 4 days while the MacConkey agar was incubated 35 ± 2 • C without CO 2 for a total length of 2 days. Both agar plates were observed for growth every 24 h. Organisms were then identified using Matrix-Assisted Laser Desorption Ionization Time-of-Flight mass-spectrometry (MALDI-TOF MS, Bruker) or conventional microbiology methods when necessary. Minimum inhibitory concentration (MIC) susceptibility testing was performed using an automated broth microdilution system (Sensititre AIM, Trek Diagnostic System Inc.) and susceptibility panels (Thermo Fisher Scientific). Interpretations were determined by the MIC breakpoints, which are based on the VET08 and M100 Clinical and Laboratory Standards Institute (CLSI) documents (18,19). Depending on the clinician's request, susceptibility testing performed by the ISU VDL included an ophthalmic susceptibility profile (JOEYE2 plate, Thermo Scientific Inc.) and/or a systemic susceptibility profile. Antibiotics and drug concentrations evaluated in JOEYE2 plates can be found on the manufacturer's website (assets.thermofisher.com/TFS-Assets/MBD/Specification-Sheets/ Sensititre-Plate-Layout-JOEYE2.pdf).

Data Analysis
Results considered "non-interpretable" by CLSI guidelines were excluded from data analysis, that is, not classified as susceptible nor resistant in calculations of percent sensitivity and multidrug resistance. An isolate was considered susceptible to a combination therapy (e.g., chloramphenicol-ciprofloxacin) if one or both antibiotics yielded a "susceptible" result for the given isolate. Bacterial isolates were considered multidrug resistant (MDR) if resistant to three or more classes of antibiotics (20), removing all known intrinsic resistances from the MDR definition as described by Sweeney et al. (21).
Odds ratios were calculated with SigmaPlot 14.0 (Systat software, Point Richmond, CA), and P < 0.05 were considered significant unless another α-value is described: (i) Likelihood of a pure breed vs. mixed breed dog to present to the ISU LVMC Ophthalmology service with a corneal infection-as compared to other reasons for the visit such as keratoconjunctivitis sicca, glaucoma, or cataracts-evaluating the canine population presented to Ophthalmology over the 6-years study period; (ii) Likelihood of a pure breed vs. mixed breed dog with a corneal infection (i.e., positive corneal culture) to yield a bacterial isolate classified as MDR; (iii) Likelihood of selected antibiotics vs. all others to provide higher efficacy against all bacterial isolates, adjusting the α-value to 0.0026 (0.05/19) to account for multiple pairwise comparisons with the Bonferroni correction; and (iii) Likelihood of not detecting bacteria (i.e., negative corneal culture) or detecting selected bacterial genera in different seasons. For the latter, a corneal culture performed between March and May was considered as the spring season, June to August for summer, September to November for fall, and December to February for winter.
Compared to all bacterial isolates, Pseudomonas species were significantly less likely to be isolated in the spring (OR = 0.34, 95% CI = 0.15-0.77, P = 0.011), but were significantly more likely to be isolated in the summer (OR = 2.04, 95% CI = 1.18-3.53, P = 0.015). Compared to all bacterial cultures, a negative culture result (no growth) was significantly less likely to occur in the summer (OR = 0.57, 95% CI = 0.39-0.83, P = 0.004).

DISCUSSION
In the present study, the three most common bacterial genera isolated from dog corneas were Staphylococcus (32.3%), Streptococcus (19.1%), and Pseudomonas (12.5%). These findings are generally in agreement with other canine studies across the globe (2, 3, 9-11), with subtle geographic differences such as a lower prevalence of Staphylococcus species in Australia (8) or a relatively higher prevalence of Streptococcus species in the Midwestern United States (present study) compared to other locations (6,7). Staphylococcus pseudintermedius was the most common Staphylococcus species (26.7% of all isolates), as recognized in most canine reports (2, 3, 7, 10, 11), followed by non-specified Coagulase-negative Staphylococcus species (3%) and Staphylococcus epidermidis (0.6%). Surprisingly, Staphylococcus aureus only accounted for 0.4% of isolates in the present study, while it had previously been reported as more prevalent in canine bacterial keratitis (6,9). Streptococcus canis (12%) and Pseudomonas aeruginosa (7.5%) were the most common species cultured for each respective genus, consistent with previous reports (3,8,10,11). As compared to mixed-breed dogs, one canine breed (Labrador retriever) was found to be less likely to present to the Ophthalmology service with bacterial keratitis-a finding presumably related to the higher proportion of eye certification exams in Labradors-while eight canine breeds were found to be at higher risk. Over-represented dogs included brachycephalic breeds (Boston terrier, Cavalier King Charles spaniel, pug, and shih tzu), as previously reported  (2, 8,9), but also non-brachycephalic breeds such as Saint Bernard, miniature pinscher, rat terrier, and silky terrier. Further, the study assessed the seasonality of bacterial isolates and found that summer increased the risk of yielding a positive bacterial culture from canine corneas, notably Pseudomonas species. Summer was also shown to affect the prevalence of selected bacterial isolates in humans (22). UV light is a known risk factor for potentiating infections in bovine eyes (23), resulting in degenerative changes to corneal epithelial cells that allow for easier bacterial colonization (24,25), and the same may be true in dogs. Awareness of seasonal variation might inform clinical recommendations and prevention strategies, and it is therefore advised to report seasonality in bacteriological studies. Appropriate management of corneal ulcers requires an understanding of common bacterial isolates and associated susceptibilities to antibiotics. Monotherapy (i.e., individual drug or combination drug such as triple antibiotic) is generally advised for ulcers characterized as "simple" (superficial, no gross signs of infection), while combination therapy with 2 antibiotic formulations is generally suggested for "complicated" corneal ulcers with active signs of infection (e.g., stromal loss, keratomalacia, cellular infiltrates) to provide broad spectrum coverage for both gram-positive and gram-negative bacteria (12)(13)(14)(15). Further, the complementary use of a systemic antibiotic can be considered in well-vascularized corneal lesions and corneal perforations (at risk for endophthalmitis), or when systemic administration achieves tear film concentrations above minimal inhibitory concentrations for common bacterial isolates. The latter was demonstrated for various antibiotics administered parenterally to cows (i.e., oxytetracycline, chloramphenicol, gentamicin, and erythromycin) (26) but has not been documented in companion animals to date. The summary information detailed in the manuscript's tables provide a quick reference for clinicians that manage bacterial keratitis in canine patients.
Individual antibiotics with the highest efficacy rates included chloramphenicol (83%), ceftiofur (79%), amikacin (77%), neomycin (76%), and gentamicin (74%). When considering commercially available formulations, the best option remains triple antibiotic (neomycin-polymxin B-bacitracin) with 77% efficacy, consistent with general guidelines to use this medication for prophylactic use in non-infected ulcers (4,5,13,27), followed by gentamicin (4,13). Other options for empirical monotherapy include chloramphenicol, ceftiofur, and amikacin (≥77% efficacy), however chloramphenicol and amikacin require compounding in the US, while ceftiofur is only marketed for horses and production animals as an injectable formulation for parenteral use, with proven clinical efficacy for cattle with infectious bovine keratoconjunctivitis (28). Chloramphenicol and ceftiofur were highly effective against Streptococcus and Staphylococcus species (≥88%), but their efficacy against Pseudomonas was limited (20%). Preferred antibiotics for Pseudomonas species included ciprofloxacin (95%) and aminoglycosides (amikacin, gentamicin, neomycin) with ≥93% efficacy. It is important to note that another 2nd-generation fluoroquinolone (ofloxacin) had relatively limited efficacy against Pseudomonas species in the present study (45%), consistent with the trend described in some veterinary and human studies (29,30). As expected, the MIC 50 and MIC 90 values for Pseudomonas sp. and cephalosporins (cefazolin and ceftiofur) were higher than MIC values detected for Staphylococcus and Streptococcus species; further, MIC 50 and MIC 90 values for Streptococcus sp. and aminoglycosides (amikacin, gentamicin, neomycin, and tobramycin) were higher than MIC values identified for Staphylococcus or Pseudomonas isolates, as recently described in canine and equine patients (31). Combination antibiotic therapies are routinely used in practice to provide broad-spectrum coverage for suspected infected corneal ulcers (12)(13)(14)(15), yet the efficacy of combined therapy is not addressed in previous bacteriological studies. A strong combination therapy highlighted in the present study is chloramphenicol (Staphylococcus, Streptococcus, Corynebacterium, and "others" coverage) and ciprofloxacin (Pseudomonas coverage), two drugs with good penetration into corneal tissues (32) and excellent efficacy against common bacterial isolates in dogs when used together (97%). Other strong combinations include chloramphenicol-amikacin (98%), chloramphenicol-tobramycin (97%), amikacin-ceftiofur (97%), chloramphenicol-gentamicin (96%), and ciprofloxacingentamicin (96%). Importantly, adjustments to combination therapies should be considered when culture and sensitivity results are obtained, avoiding the unnecessary use of antibiotics, and reducing ocular surface toxicity from excessive use of preservative-containing ophthalmic solutions.
Systemic antibiotics can be used to complement topical antibiotic therapy in dogs with bacterial keratitis if the lesion is vascularized or the cornea is perforated. Amoxicillinclavulanic acid was very effective for gram-positive organisms (≥83% efficacy) (16), while aminoglycosides and other selected antibiotics (ceftazidime, marbofloxacin, and piperacillin) were most effective against gram-negative organisms (≥76% efficacy). Antibiotics generally considered "last-line" therapies such as vancomycin (90%) and imipenem (86%) were highly effective against most bacterial isolates.
Antimicrobial resistance is a serious concern in veterinary ophthalmology (33). The overall prevalence of MDR isolates was relatively high (20%) in the present study, with Pomeranians and Saint Bernards being significantly more likely to yield MDR isolates when presented to our Ophthalmology service with corneal ulcers. The high prevalence of MDR isolates in these two canine breeds is puzzling and may simply be related to low sample size; alternatively, it is possible the microbiome differs among canine breeds due to peculiarities in ocular surface anatomy and physiology. Interestingly, none (0%) of Pseudomonas isolates were considered MDR in this study, likely due to extensive intrinsic resistance profile for this bacterial genera in veterinary medicine (18,21); of note, most (>90%) Pseudomonas isolates were susceptible to aminoglycosides and ciprofloxacin as previously described (7,34). Further, we noted an alarming, escalating trend of multi-drug resistance over the last 5 years of the study (5% in 2016 to 34% in 2020). MDR is also prevalent in human patients with bacterial keratitis (35,36). The "Antibiotic Resistance Monitoring in Ocular Microorganisms" (ARMOR) studies in humans have provided invaluable information for clinicians managing bacterial keratitis in practice, and the same collaborative effort is critically needed in veterinary medicine to mitigate the rise of MDR in animals. Such work could promote antimicrobial stewardship at a regional-or clinic-level, decreasing the rate of MDR through judicious use of ophthalmic antibiotics in practice. The present work focused on aerobic bacterial cultures in dogs with ulcerative keratitis. Depending on the clinical appearance of the lesion, clinicians should also consider anaerobic and/or fungal cultures to increase the likelihood of identifying the causative agent (37,38). In fact, diagnostic tests other than culture-based methods will soon become the gold standard for microbial species identification as the field of clinical microbiology is rapidly evolving. Unlike culture-based methods, technologies such as mass spectrometry and nucleic acid sequencing provide rapid and sensitive tools to probe the microbiome in clinical patients (39)-as recently described for the ocular surface of veterinary species (40,41)-enabling clinicians to optimize the antibiotic treatment sooner and thereby improve clinical outcomes (42).
Similar to previous reports of antibiotic susceptibility profiles in veterinary medicine, the main limitation of the present study is the reliance on incomplete veterinary specific CLSI guidelines to determine whether a bacterial isolate is susceptible or resistant to a given antibiotic. In veterinary CLSI guidelines, the lack of interpretative breakpoints for selected bacteria/antibiotic combinations yields a "non-interpretable" clinical interpretation in the antibiogram. Such missing information may be due to intrinsic resistances, or the absence of pharmacokineticspharmacodynamics studies that assess the concentration of a given antibiotic at the target tissue and determine whether such levels are above the minimal inhibitory concentration (MIC) of a given bacterial species. Findings presented herein likely underestimate the true susceptibility rates as some of the "noninterpretable" results may in fact be sensitive in clinical practice. For instance, moxifloxacin susceptibility rates were generally low in the present study because 64% of moxifloxacin results were reported as "non-interpretable" by the CLSI guidelines, while in practice this fourth-generation fluoroquinolone is considered superior to ofloxacin and ciprofloxacin for most bacterial isolates (43). Similarly, polymyxin B, cefazolin, and bacitracin also had high numbers of "non-interpretable" reports (72, 71, and 55%, respectively). On the other hand, caution must be exercised when clinical guidelines provide an actual interpretation of "susceptible" or "resistant." A closer look at the MIC data provided by the ophthalmic panel (JOEYE2 plate) shows that the concentration tested for selected antibiotics (e.g., 4 µg/mL for chloramphenicol/neomycin/tobramycin, 16 µg/mL for amikacin/ticarcillin) is relatively high when compared to drug levels achieved in canine tear film. Indeed, topical drug delivery achieves high concentrations in the short-term but concentrations on the canine ocular surface rapidly decrease due to efficient drainage through the nasolacrimal duct (44)(45)(46).
In conclusion, the bacterial profile from corneal cultures in Iowa and surrounding Midwestern United States followed world-wide trends with high proportion of Staphylococcus, Streptococcus, and Pseudomonas species. Characteristics of bacterial keratitis in dogs were influenced by season and canine breeds. The rate of multi-drug resistance was relatively high, notably for Staphylococcus isolates, with an alarming escalating trend over time. Appropriate selection of empiric antibiotic therapy is important to enhance therapeutic outcome and reduce antibacterial resistance in dogs with corneal ulceration, whether using individual or combination drug therapy. Subsequently, clinicians' selection of antibiotics should be guided by the antibiogram received for each given patient, requiring adjustments to the empirical therapy initiated earlier in selected cases.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
Ethical review and approval was not required for the animal study because the work was a retrospective analysis of bacterial cultures collected in dogs as part of the patients' standard of care. Written informed consent for participation was not obtained from the owners because the work was a retrospective analysis of bacterial cultures collected in dogs as part of the patients' standard of care.