Temporal Increase in Antibiotic Resistance in Chronically Treated Non-Healing Ulcers

1. Introduction

Spontaneous chronic corneal epithelial defects (SCCEDs), also known as indolent or non-healing corneal ulcers, represent one of the most common chronic corneal diseases encountered in canine ophthalmology. The condition is characterized by failure of normal epithelial adhesion, abnormalities of the epithelial basement membrane, and formation of an acellular hyaline stromal zone that interferes with normal re-epithelialization, with frequent topical antibiotic treatment for weeks to months until healing is achieved. Histopathologic studies have demonstrated disruption of epithelial-stromal attachment mechanisms, resulting in persistent epithelial defects despite the absence of significant stromal inflammation or neurogenic keratitis component [1,2]. Although SCCEDs are well recognized in veterinary medicine, they share important clinical and biological similarities with human persistent epithelial defects (PEDs). In both species, chronic disruption of the corneal epithelial barrier predisposes the ocular surface to microbial colonization, inflammation, stromal degradation, and delayed wound healing. Human persistent epithelial defects may develop secondary to neurotrophic disease, herpetic keratitis, ocular surface inflammation, limbal stem cell deficiency, graft-versus-host disease, autoimmune disease, or ocular surgery, and remain a significant cause of visual morbidity worldwide [3,4]. Despite differences in underlying etiology, SCCEDs and human PEDs share prolonged epithelial instability, recurrent healing failure, and dependence on supportive therapies designed to restore epithelial integrity [3,5]. A growing body of evidence suggests that chronic epithelial defects should not be viewed solely as disorders of epithelial adhesion but also as disturbances of the ocular surface microbial ecosystem. The normal ocular surface harbors a diverse microbial community that contributes to immune homeostasis and barrier defense [6,7,8]. Loss of epithelial integrity alters this relationship and may permit colonization by opportunistic organisms, particularly coagulase-negative staphylococci and other commensal bacteria. In human ophthalmology, bacterial colonization and secondary infection are recognized complications of persistent epithelial defects and neurotrophic keratopathy, often necessitating prophylactic antimicrobial therapy to prevent stromal ulceration and corneal perforation [4]. Simultaneously, antimicrobial resistance has emerged as one of the most significant challenges facing both human and veterinary medicine. Ocular surveillance programs have documented increasing prevalence of methicillin-resistant staphylococci, multidrug-resistant organisms, and declining susceptibility to commonly used ophthalmic antibiotics [9,10].Companion animals and humans occupy increasingly shared environments, creating opportunities for exchange of antimicrobial-resistant bacteria and resistance determinants [11,12]. Consequently, antimicrobial resistance in ocular pathogens has become an important One Health concern with implications extending beyond individual species [13]. Current treatment recommendations for canine SCCEDs typically include epithelial debridement procedures such as diamond burr keratectomy, grid keratotomy, superficial keratectomy, or thermal cautery, combined with topical antimicrobial therapy [14,15,16,17,18,19,20]. However, despite the routine use of antibiotics, surprisingly little is known regarding the microbiological characteristics of SCCEDs. To date, only a single dedicated study has evaluated bacterial isolates from canine indolent ulcers and reported positive bacterial cultures in approximately 20% of cases [21]. The prevalence of bacterial colonization, distribution of microbial species, antimicrobial susceptibility patterns, and temporal trends in resistance remain largely undefined. The objectives of the present study were therefore to characterize bacterial species isolated from canine SCCED/non-healing corneal ulcers, determine their antimicrobial susceptibility profiles, evaluate temporal changes in resistance patterns, and identify optimal empirical antimicrobial strategies. Additionally, we sought to explore the broader translational relevance of chronic epithelial defects as potential reservoirs of antimicrobial resistance and to compare microbial patterns observed in canine SCCEDs with those reported in human chronic corneal epithelial disease.

2. Materials and Methods

2.1. Sample Collection and Processing

Samples for microbiology analyses were collected from 221 dogs and a total of 228 eyes with non-healing corneal ulcers in the period January 2019 – February 2023. All patients had a complete eye examination. The antibiotic treatment was initialized with triple antibiotic ointment cream (neomycin-polymyxin-bacitracin) q8h followed by the change of the topical antibiotic based on the antibiogram results of antibiogram. The inclusion criteria for the study were presence of corneal epithelial defect for the duration longer than 10 days despite appropriate medical therapy, absence of corneal stromal involvement, and presence of the loose epithelial margins at the ulcer edges at the time of initial diagnosis. Data from additional 62 patients with non-healing ulcers treated where standard ofloxacin protocol (q6h) without pursuing ocular microbiology were included in the analysis (patients were seen in the period 2012–2013). Additional 3 patients (two patients with a non-healing ulcer, and one patient with corneal stromal ulcer seen in 2025), had corneal confocal microscopy performed (VivaScope 3000, Caliber ID, Rochester, NY, USA) under the topical anesthesia in combination with bacterial culture testing. Clinical canine microbiology samples were collected from corneal non-healing ulcers using a flocked swab and placed in the provided transport media (BD ESwab^TM Collection Kit, COPAN ITALIA SpA, Brescia, Italy). After applying topical anesthetic on the ocular surface (Propracaine 0.05%, Akorn Pharmaceuticals, Lake Forest, IL, USA) samples were harvested and then placed on ice packs until sent to the laboratory or were immediately cultured. Approximately half of each sample in ESwab transport media (roughly 0.5ml, concentrated via centrifugation to increase recovery) was cultured onto MacConkey agar (Hardy Diagnostics, Santa Maria, CA, USA) whereas the other half of the sample was cultured onto Chocolate agar (Hardy Diagnostics Hardy Diagnostics, Santa Maria, CA, USA). The plates were then placed in the incubator at 37⁰C in 5% CO₂, and inspected at 24, 48 and 96 hours. Presence or absence of bacterial colonies was recorded.

2.2. Standard Microbial Culture and Antimicrobial Susceptibility Testing

Standard microbial culture (aerobic) and antimicrobial susceptibility testing were performed as previously described [22]. Briefly, samples were inoculated onto Chocolate and MacConkey agar, incubated aerobically and anaerobically at 37 °C for 24–96 hours. Growth was assessed morphologically, and isolates were identified using Gram staining, and automated MALDI-ToF identification (Bruker BioTyper; MBT Compass database, updated biannually). Antibiotic susceptibility was determined via Kirby-Bauer disk diffusion on Mueller-Hinton agar, interpreting zones as susceptible (S), intermediate (counted as S for analysis), or resistant (R). WHONET database software (World Health Organization) containing 2021 (or later) CLSI breakpoints for dogs was used to manage and analyze microbiology laboratory data and antibiotic susceptibility test results. Other CLSI animal breakpoints were used if breakpoints were not available for dogs. If no other animal breakpoints were available, human CLSI breakpoints were applied. The breakpoints were based upon CLSI VET08 Performance Standards for Antimicrobial Disk and Dilution Susceptibility Tests for Bacteria Isolated From Animals (https://clsi.org/media/2321/vet08ed4_sample.pdf), CLSI M100 Performance Standards for Antimicrobial Susceptibility Testing (updated yearly, https://em100.edaptivedocs.net/Login.aspx) and Hardy Diagnostics Disk Diffusion Zone Diameter Chart (https://www.keyscientific.com/files/Other%20Manufacturers/Hardy%20Diagnostics/AST%20Discs/Hardy%20AST%20Disc%20Insert.pdf). This chart was also used for polymyxin B breakpoints in Pseudomonas aeruginosa (R ≤ 11; S≥12). To detect penicillin-binding protein 2A (PBP2A) in staphylococcal isolates, the Alere^TM PBP2A SA Culture Colony Test was performed following manufacturer’s instructions (Alere Scarborough, Inc., 10 Southgate Road, Scarborough, ME 04074). All bacterial isolates were assigned into resistant and susceptible categories relative to their in vitro response to antibiotics. Susceptible and intermediate interpretations were grouped together to the susceptible class considering the high concentration of antibiotics that can be delivered directly to eyes. Tested antibiotics were grouped into the antibiotic categories of aminoglycosides (amikacin, gentamicin, neomycin, and tobramycin), polypeptides/polymyxins (bacitracin, polymixin B), anti-staphylococcal β-lactams (cephamycins; oxacillin, cefoxitin), tetracyclines (tetracycline), non-extended spectrum cephalosporins (1^st and 2^nd generation cephalosporins; cefazolin), penicillins with and without β-lactamase inhibitors (amoxicillin/clavulanic acid, penicillin G), macrolides (erythromycin), and fluoroquinolones (ofloxacin, moxifloxacin, and ciprofloxacin). Additionally, each clinical isolate was allocated to a not multidrug-resistant (Not MDR), multidrug-resistant (MDR), or possible extensively multidrug-resistant (possible XDR) group based upon its susceptibility data. The MDR group was defined as resistance to at least one antibiotic in three or more antibiotic categories. XDR was defined as resistance to at least one antibiotic in all but two or fewer antibiotic categories (i.e., bacterial isolates remain susceptible to only one or two categories) as previously published [23,24]. Further, if an isolate had intrinsic resistance to a particular antibiotic, then the corresponding resistance was excluded from the analysis [25]. Total of 176 eyes in which ocular microbiology analysis was performed and 62 eyes without microbiology analysis treated with ofloxacin only were treated with diamond burr and placement of the soft contact lens (PureVision, Bauch and Lomb, St Louis, Missouri, USA) as previously described [15]. The healing time was calculated from the time of treatment to the time of the corneal epithelium defect closure, which was determined via slit lamp examination and negative fluorescein staining of the defect.

2.3. Statistical Analysis

Commercial software was used for statistical analyses as described in the manuscript and as previously reported (Prism, version 5.0; GraphPad, San Diego, CA) [22].

3. Results

3.1. Growth of Bacteria from Patient Samples

Samples were collected from 221 dogs and a total of 228 non-healing corneal ulcer samples were plated. In 71% (163/228) of plated samples a bacterial growth was observed, while 29% (65/228) of plated samples yielded no growth in the culture media. A total of 170 isolates were obtained and analyzed in this study (some corneal ulcer samples had multiple isolates accounting for the larger number of isolates compared to the number of corneal ulcers with positive growth).

3.2. Distribution of Bacterial Species Isolated from Patients With Corneal Non-Healing Ulcers

The most frequent microbial species identified in the corneal non-healing samples were Staphylococcus epidermidis 21% (36/170), Staphylococcus pseudintermedius 18% (31/170), and Staphylococcus capitis 15% (25/170) followed by coagulase negative staphylococci present in 14% (24/170) of samples (Table 1).

3.3. Antibiotic Susceptibility Profile of Bacteria Isolated From Corneal Non-Healing Ulcers

To determine a suitable antibiotic treatment for canine corneal non-healing ulcers, we examined overall bacterial resistance regarding the use of a single or combination antibiotic between two time points 2019–2020 and 2021–2023. Based upon the general antibiotic susceptibility data, neopolybac alone (94%) or a combination of neopolybac with ofloxacin (98%) or amikacin (96%) appeared to be the optimal antibiotic treatment for canine corneal non-healing ulcers (Figure 1). Importantly, statistically significant increase in bacterial resistance (p = 0.000008, Paired t-test) was observed for the period 2021–2023 in comparison to the period 2019–2020 regarding the use of a single or combination antibiotic (Figure 1).

In samples tested with topical antibiotics, bacterial species were mainly resistant topolymyxin B 69% (117/170) and oxacillin 62% (21/34) followed by cefazolin 55% (94/170), and cefoxitin 51% (85/166). The least resistant species were against bacitracin 16% (27/170), amikacin 18% (31/170), ofloxacin 23% (39/170), and gentamicin 23% (39/169; Table 2). If intrinsic resistance was excluded from the analysis therefore considering only acquired resistance, a similar trend of resistance was observed. Bacterial species were predominantly resistant to polymyxin B 68% (114/167), oxacillin 62% (21/34), cefazolin 52% (81/157), and cefoxitin 48% (76/157). The least resistant species were against amikacin 15% (24/163), bacitracin 15% (25/168), gentamicin 20% (32/162), and neomycin 20% (32/158; Table 2).

Moreover, lower numbers of acquired bacterial resistance are detected comparing to overall resistant numbers. The exception from this pattern was no difference between acquired and total bacterial resistance in the case of oxacillin (Table 2). In samples tested with systemic antibiotics (usually samples from patients with the extensive corneal neovascularization reaching the corneal ulcer region), bacterial species were predominantly resistant to ampicillin 92% (12/13), penicillin G 87% (14/16), cephalexin 62% (10/16), clindamycin 62% (10/16), oxacillin 62% (21/34), cefoxitin 51% (85/166), amoxicillin/clavulanic acid 40% (36/89) followed by sulfamethoxazole/trimethoprim 31% (5/16), ciprofloxacin 31% (25/81), enrofloxacin 25% (4/16), marbofloxacin 25% (4/16), amikacin 18% (31/170), and doxycycline 13% (2/15).

3.4. Temporal Increase of Acquired Resistance In Isolates From Corneal Non-Healing Ulcers

To analyze dynamics of acquired resistance over time, we examine antibiotic susceptibility profile of isolates between two time points 2019–2020 and 2021-2023. In the period 2019-2020, isolates were predominantly resistant to polymyxin B (65%), oxacillin (45%), and cefazolin (41%). A similar trend was observed for the period 2021-2023 (polymyxin B (70%), oxacillin (70%), and cefazolin (59%; Figure 2). The smallest percentage of isolates was resistant to bacitracin (7%), amikacin (12%), gentamicin (17%), and neomycin (17%) in the period 2019-2020. In the period 2021–2023, the lowest percentage of isolates was resistant to amikacin (16%), bacitracin (20%), gentamicin (22%), and neomycin (23%). Overall, apparently higher percentages of resistance were noticed in the period 2021-2023 than in the period 2019–2020. Accordingly, statistically significant changes (p = 0.000504, Paired t-test) in acquired resistance were observed between these two time points (Figure 2).

3.5. Distribution of Bacterial Species In Patients Diagnosed with Corneal Non-Healing Ulcers Relative to Previous Patient’s Antibiotic Treatment

The most frequent bacteria identified in the corneal non-healing samples relative to the patient’s previous antibiotic treatments were Staphylococcus pseudintermedius, coagulase negative staphylococci, Staphylococcus capitis, Staphylococcus epidermidis, and Pseudomonas aeruginosa (Table 3). In comparison to the overall data in Table 1, most of isolates were present in relatively similar percentages. Consequently, no statistically significant difference (p = 0.313141, paired t-test) was observed in the distribution of bacterial species relative to the patient’s previous antibiotic treatments.

3.6. Susceptibility Profile of Isolates From Corneal Non-Healing Ulcers Relative To Patient’s Previous Antibiotic Treatments

To address a possible temporal change of acquired resistance relative to previous the patient’s antibiotic treatments, we analyzed isolates within an approximately four-year period from 2019 to 2023. In the group of patients without previous antibiotic treatments, the highest percentage of isolates was resistant to oxacillin, polymyxin B, cefazolin, and cefoxitin with a similar trend in the group previously exposed to antibiotics (Figure 3). The lowest percentage of isolates was resistant to amikacin/ofloxacin and bacitracin/amikacin in the group of patients without previous antibiotic treatments and the group previously exposed to antibiotic, respectively. Moreover, no statistically significant changes (p = 0.439938, paired t-test) in acquired resistance were observed relative to the patient’s previous antibiotic treatments within the examined four-year period (Figure 3).

3.7. Resistance Profile Based on The Number of Resistant Antibiotics in Canine Corneal Non-Healing Ulcers

The highest percentage of bacteria was resistant to 1 antibiotic 17% (29/170) and 2 antibiotics 15% (25/170). The lowest percentage of bacteria was resistant to 12 antibiotics 1% (1/170), 10 antibiotics 2% (4/170), 9 antibiotics 3% (5/170), and 11 antibiotics 4% (6/170; Figure 4).

3.8. Distribution of Multidrug-Resistant Bacteria in Clinical Corneal Non-Healing Ulcers

Overall, in the corneal non-healing ulcer samples, 47% (80/170) were not MDR, 20% (34/170) belong the MDR group while 33% (56/170) were possible XDR. Thus, over a half of isolates belong to a general multi-drug resistant category. In the period 2019–2020, 57% (40/70) were not MDR isolates, 20% (14/70) were MDR isolates while 23% (16/70) were possible XDR isolates. In the period 2021–2023, 40% (40/100) were not MDR isolates, 20% (20/100) were MDR isolates while 40% (40/100) were possible XDR isolates. Moreover, no statistically significant difference (p = 0.411720, Paired t-test) was observed in multidrug resistance although a decrease of not MDR and increase of XDR percentages were noticed between 2019–2020 and 2021–2023.

3.9. Methicillin-Resistant Staphylococcus Spp. in Clinical Corneal Non-Healing Ulcers

To address methicillin resistance in Staphylococcus species, the isolates were probed with cefoxitin disks ¹⁵. Per CLSI VET08 guidelines (https://clsi.org/media/2321/vet08ed4_sample.pdf) for Staphylococcus pseudintermedius and Staphylococcus schleiferi, the results were confirmed with oxacillin disks. Additionally, the penicillin-binding protein 2A (PBP2A) antibody test was used to determine a status of methicillin resistance in a few species. Overall, in the 2019-2023 period, 61% (68/112) isolates were methicillin-resistant. Regarding patient’s previous antibiotic treatments, a conspicuous increase in the percentage of methicillin resistant isolates was detected in the group with no previous antibiotic treatment 74% (14/19). In the group with previous antibiotic treatment 58% (54/93) of isolates were methicillin resistant. Over time, a noticeable increase in the percentage of methicillin-resistant Staphylococcus species was observed from 2019-2020 (47%; 20/43) to 2021-2023 (70%; 48/69).

4.0. Difference in Healing Time Between Non-Healing Ulcers With And Without Bacterial Growth

There was no statistically significant difference in healing times between ulcers with bacterial growth vs. no growth observed in this study (Welch’s t-test, p = 0.948): no growth group (51 eyes) = 17.25+/-1.16 days (mean+/-SEM) and bacterial growth group (125 eyes) 17.34+/=0.74 days. Some patients did not elect to pursue treatment or present for the recheck evaluations resulting in the discrepancy in numbers between total dogs with corneal microbiology isolates, and number of dogs for which healing time data was available. We have further evaluated whether there is a statistical difference in healing time between non-healing ulcers in dogs with bacterial growth, and non-healing ulcers (57 eyes) treated with generic ofloxacin protocol (q6h) without pursuing corneal microbiology testing. Ofloxacin treated dogs had significantly longer healing time at 20.39 +/- 1.58 (Welch’s t-test, p = 0.02) when compared to non-healing ulcers with bacterial growth group. Conversion to corneal stromal ulcer occurred in 0/125 (0%) eyes in non-healing ulcer group with bacterial growth and initially treated with neopolybac, and 5/62 (8.1%) eyes in ofloxacin group, representing a significantly increased risk in ofloxacin group (Fisher’s exact test, p = 0.003; OR 24.0, 95% CI 1.3–438.6). Confocal corneal imaging in limited number of patients (Figure 5) revealed numerous presence of inflammatory cells in ulcerated corneas with positive bacterial growth.

4. Discussion

The present study provides the largest microbiological characterization of canine spontaneous chronic corneal epithelial defects (SCCEDs) reported to date and offers several observations that may have significance beyond veterinary ophthalmology. Although SCCEDs have traditionally been regarded as sterile or minimally contaminated lesions, bacterial growth was identified in 71% of cases, a prevalence substantially higher than previously reported in dogs and remarkably similar to frequencies reported in human persistent epithelial defects (PEDs) [4,21]. These findings suggest that chronic epithelial defects, regardless of species, may create a shared biological niche that promotes bacterial colonization of the ocular surface. The predominance of coagulase-negative staphylococci, including Staphylococcus epidermidis and Staphylococcus capitis, is particularly noteworthy. Historically regarded as relatively benign commensals, these organisms are now increasingly recognized in human medicine as important opportunistic pathogens and major reservoirs of antimicrobial resistance genes [26,27]. Coagulase-negative staphylococci frequently harbor mecA-containing SCCmec elements and other mobile resistance determinants capable of horizontal transfer to more pathogenic species [28]. The observation that approximately half of all isolates in this study belonged to this bacterial group suggests that chronic corneal epithelial defects may serve as localized reservoirs of resistance-associated microorganisms within the ocular microbiome. Despite the high prevalence of bacterial isolation, culture-positive ulcers did not heal more slowly than culture-negative ulcers following diamond burr debridement and bandage contact lens placement. This finding supports the hypothesis that many recovered organisms may function primarily as colonizers rather than invasive pathogens. Similar observations have been reported in human persistent epithelial defects, neurotrophic keratopathy, and ocular graft-versus-host disease, where microbial colonization frequently accompanies epithelial breakdown without necessarily serving as the primary driver of delayed epithelial healing. Instead, the underlying epithelial dysfunction appears to be the dominant pathological mechanism, while bacterial colonization develops as a secondary consequence of prolonged barrier disruption [3,4,5]. From a translational perspective, the most concerning finding was the progressive increase in antimicrobial resistance observed during the study period. Overall resistance and acquired resistance both increased significantly between 2019–2020 and 2021–2023. Furthermore, methicillin-resistant staphylococci increased from 47% to 70% during the same interval, while more than half of all isolates met criteria for multidrug-resistant (MDR) or possible extensively drug-resistant (XDR) phenotypes. These trends parallel resistance patterns increasingly reported in human ophthalmology, where methicillin-resistant Staphylococcus aureus (MRSA), methicillin-resistant coagulase-negative staphylococci, and fluoroquinolone-resistant ocular pathogens have emerged as major therapeutic challenges [10]. The similarities between resistance trends observed in veterinary and human ocular isolates suggest that common ecological and antimicrobial selection pressures may be operating across species. The current findings also support the growing importance of One Health approaches to antimicrobial resistance surveillance. Companion animals share close environmental and physical contact with humans, creating opportunities for bidirectional exchange of bacterial organisms and resistance determinants. Previous studies have demonstrated genetic similarities between antimicrobial-resistant staphylococcal isolates recovered from dogs and their owners, suggesting that companion animals may participate in shared resistance ecosystems [11,29]. Although transmission dynamics were not evaluated in the present study, the rapid increase in methicillin resistance and multidrug resistance observed within a relatively short time frame raises important questions regarding environmental, healthcare-associated, and household factors contributing to resistance emergence. An additional translational implication involves the utility of canine SCCEDs as a naturally occurring model of chronic epithelial disease. Human persistent epithelial defects remain a significant therapeutic challenge and are associated with neurotrophic keratopathy, herpetic disease, autoimmune ocular surface disorders, limbal stem cell deficiency, ocular graft-versus-host disease, and post-surgical epithelial dysfunction. While SCCEDs differ in etiology, both conditions share prolonged epithelial barrier disruption, susceptibility to bacterial colonization, recurrent healing failure, and dependence on supportive therapies to restore epithelial integrity. The comparable prevalence of bacterial recovery observed between canine SCCEDs and human PEDs suggests that naturally occurring canine disease may provide a valuable translational platform for studying host-microbe interactions, ocular surface microbiome dynamics, antimicrobial stewardship strategies, and novel approaches to epithelial regeneration. The antibiotic susceptibility findings further reinforce the importance of evidence-based antimicrobial stewardship. Although neomycin-polymyxin B-bacitracin provided the broadest empirical coverage in this population, the rapid increase in resistance observed during the study period emphasizes that empirical antibiotic selection should not replace microbiological evaluation in refractory or atypical cases. The goal of antimicrobial therapy in chronic epithelial defects may be less about eradication of invasive infection and more about preventing secondary bacterial overgrowth while preserving the effectiveness of available antibiotics. This concept mirrors contemporary recommendations in human ophthalmology, where increasing resistance rates have prompted greater emphasis on culture-guided therapy and antimicrobial stewardship. Several limitations should be acknowledged. The study was retrospective and performed at a single referral center, potentially limiting geographic generalizability. Anaerobic organisms and molecular microbial characterization were not evaluated. In addition, whole-genome sequencing and resistance gene analysis were not performed, preventing assessment of specific resistance mechanisms or comparison with human ocular isolates. Future studies integrating metagenomics, microbiome profiling, and genomic epidemiology may help clarify whether resistant organisms identified in canine ocular disease share common evolutionary pathways with those observed in human ophthalmology.

5. Conclusions

Canine SCCEDs appear to represent a chronic ocular surface disorder characterized by frequent bacterial colonization, increasing antimicrobial resistance, and substantial methicillin-resistant staphylococcal involvement. The similarities between canine SCCEDs and human persistent epithelial defects suggest that naturally occurring canine disease may serve as a useful translational model for studying chronic epithelial failure, ocular surface microbiology, and antimicrobial resistance. Continued microbiological surveillance and antimicrobial stewardship efforts are likely essential for preserving therapeutic efficacy in both veterinary and human ophthalmology.