Spontaneous Septic Arthritis of Canine Elbows: Twenty-One Cases

Abstract Objective This study provides information on clinical features, diagnosis, treatment and associated risk factors of spontaneous septic elbow arthritis in the dog. Methods Medical records between March 2007 and June 2015 were searched for cases of spontaneous septic elbow arthritis with a diagnosis based on clinical signs, arthrocentesis, cytological and microbiological analysis of elbow joint synovial fluid, radiography and outcome following treatment. Results Twenty-one cases of septic arthritis were identified. Pre-existing osteoarthritis was present in 14/15 elbows for which diagnostic imaging was available. Although all cases had increased neutrophil count on synovial fluid cytology, culture was only positive in 12/21. Despite initial improvement in lameness scores (pre-treatment 9/10 [range: 1–10] versus post-treatment 3/10 [range: 1–5]), 11/12 had residual long-term lameness. Recurrence of infection was noted in 3/12 elbows for which long-term (>8 weeks) follow-up was available. There was an acute mortality rate of 2/21 associated with severe systemic sepsis. Clinical Significance Septic arthritis, even in the absence of pyrexia, should be considered as a major differential diagnosis in middle aged, large breed dogs, with pre-existing elbow arthritis, that suffer an acute onset lameness, with elbow joint effusion and discomfort. Antibiotic therapy alone was effective for treatment with high initial response rates. Chronic lameness post-treatment was common, and a high rate of recurrence was seen with 3/12 dogs suffering more than one episode.


Introduction
Septic arthritis is considered an uncommon condition that can significantly impact the quality of a dog's life. 1 It is an active joint infection, which is usually bacterial in origin and results in an acute inflammation of the joint, with swelling, pain and lameness. 1,2 Bacterial contamination of the elbow may arise from direct inoculation (at surgery or related to trauma), or by the extension of local infections or by haematogenous localization. 3 The term spontaneous is used in this report to describe infections in which there has been no known recent surgical or traumatic episode to the afflicted joint and the infection is presumed to be haematogenous in origin. 4 The majority of bacterial septic arthritides manifest Keywords ► septic arthritis ► elbow ► osteoarthritis ► elbow dysplasia ► infections

Abstract
Objective This study provides information on clinical features, diagnosis, treatment and associated risk factors of spontaneous septic elbow arthritis in the dog.
Methods Medical records between March 2007 and June 2015 were searched for cases of spontaneous septic elbow arthritis with a diagnosis based on clinical signs, arthrocentesis, cytological and microbiological analysis of elbow joint synovial fluid, radiography and outcome following treatment. Results Twenty-one cases of septic arthritis were identified. Pre-existing osteoarthritis was present in 14/15 elbows for which diagnostic imaging was available. Although all cases had increased neutrophil count on synovial fluid cytology, culture was only positive in 12/21. Despite initial improvement in lameness scores (pre-treatment 9/10 [range: 1-10] versus post-treatment 3/10 [range: [1][2][3][4][5]), 11/12 had residual long-term lameness. Recurrence of infection was noted in 3/12 elbows for which long-term (>8 weeks) follow-up was available. There was an acute mortality rate of 2/21 associated with severe systemic sepsis. Clinical Significance Septic arthritis, even in the absence of pyrexia, should be considered as a major differential diagnosis in middle aged, large breed dogs, with pre-existing elbow arthritis, that suffer an acute onset lameness, with elbow joint effusion and discomfort. Antibiotic therapy alone was effective for treatment with high initial response rates. Chronic lameness post-treatment was common, and a high rate of recurrence was seen with 3/12 dogs suffering more than one episode. as a monoarthropathy and may be either acute or chronic in onset. 3 A clear joint predilection of septic arthritis in dogs has not been established for cases of spontaneous infection. In the veterinary literature, when surgical related infection is included, the stifle is the most commonly affected joint 16.1-73.7%, with the elbow representing 12.9-38.7% of cases. 3,[5][6][7] Pre-existing joint diseases, such as osteoarthritis (OA), and concurrent medical conditions (diabetes mellitus, skin disease, urinary tract infection and prosthetic joints) may predispose the joint to opportunistic infection. 2,8 Septic arthritis more often affects larger breeds, with an apparent over-representation of males. 3,[5][6][7] The definitive diagnosis of septic arthritis has traditionally relied on the identification of bacteria from the affected joint by synovial fluid or synovial membrane culture. The difficulty is that the bacterial culture is frequently unsuccessful and the diagnosis must often be based on a degree of suspicion. 1,9 Often a presumptive diagnosis of bacterial infective arthritis is made where the synovial fluid from a monoarthropathy shows very high nucleated cell counts (>50 Â 10 9 cells/mL), predominantly polymorphonuclear cells, the presence of intracellular bacteria on cytology or a combination of these. 1 Despite several retrospective articles on the subject of septic arthritis, there is limited information on the signalment, treatment success, recurrence and long-term outcome of cases of septic arthritis in the elbow joint of dogs that have not had recent surgery. [3][4][5][6][7] This study aimed to review the current literature on septic arthritis and describe cases of septic arthritis including the history, presenting complaint, underlying disease state, response to treatment and outcome.

Materials and Methods
The clinical record database of two tertiary-level referral institutions was searched for cases of septic arthritis or bacterial infective arthritis that had been diagnosed between March 2007 and June 2015 to determine relative joint prevalence. Cases identified for septic arthritis were then further stratified to identify cases of spontaneous septic arthritis of the elbow. Inclusion criteria were the diagnosis of a monoarthropathy, where analysis of either the synovial fluid or membranes was consistent with septic arthritis, and there was no recent surgery of the elbow joint within 1 month of presentation or 1 year if implants were placed. 3,10 Analysis of the synovial fluid or synovium was required to fulfil one or more of the following criteria: highly cellular appearance observed subjectively on a direct smear, >40% neutrophil population in the synovial fluid; a total nucleated cell count of more than 50.0 Â 10 9 cells/mL; a positive synovial fluid or membrane bacterial culture. 5,11 The medical records, physical examination, and recent haematology and blood biochemistry results from cases were reviewed. Synovial fluid samples had been obtained from the affected joint by percutaneous arthrocentesis following aseptic preparation in anaesthetized or deeply sedated patients. 12 Synovial fluid samples were submitted for culture and sensitivity after inoculation into blood culture media. Culture was performed as previously described. 3 Lameness of the affected limb was extrapolated from clinical records of patients as assessed and recorded by either Royal College of Veterinary Surgeons or European College of Veterinary Surgeons board-certified veterinarians pre-and post-treatment using a numerical scoring system. 13,14 Because of the variability of recorded information between patient and across the time period of the study, the following grouping was defined: • 0-Sound, no lameness • 1-Occasionally shifts weight off affected limb • 2-Mild lameness at a slow trot, none while walking • 3-Mild lameness visible while walking • 4-Obvious lameness while walking, but places the foot while standing • 4-7-Moderate lameness in degrees of severity • 8-Severe lameness • 9-Places toe when standing, carries limb when trotting • 10-Non-weight bearing When imaging studies of the elbow were available, the plain radiography and computed tomography scans of the elbow joints were reviewed by two authors (BM/RM). Images were assessed for the presence of osteophytes: at the anconeal process, medial and lateral epicondyles, and radial head; ununited anconeal process, fragmented medial coronoid process, incomplete ossification of the humeral condyles and humeral condyle osteochondritis dissecans. A global assessment of OA was given: none (no osteophytes), mild (small numbers of osteophytes less than 1 mm in size, moderate (osteophytes at multiple sites, 1-2 mm) or severe (osteophytes larger than 2 mm) following consensus between the two authors. 5 Short-term (defined as a period less than 8 weeks) outcome was recorded as clinically successful where there was a return to the level of ambulation prior to recent episodes of lameness, clinically unsuccessful if there was continued lameness at the same degree, or greater, than was present prior to intervention but with resolution of infection; and as failed if the synovial fluid cytology was not consistent with resolution of the bacterial infection at the last recorded treatment. 4,15 Longterm outcome (>8 weeks) was reviewed for an ongoing lameness, recurrence of infection or further surgical intervention and was evaluated by both owner telephone calls and assessment of clinical records where available.
The statistical analysis was performed by one of the authors using a statistical software package (SPSS Stat, Version 2.2, IBM Corp). Categorical data are presented as median AE range throughout. The project was ethically reviewed (URN 2015 1359) by the respective institutional Research Ethical Review Boards.

Results
Twenty-seven cases of septic arthritis of the elbow joint were initially identified during the data collection period. Five elbows were excluded due to a history of recent surgery involving the septic elbow. One case was excluded based on repeat synovial fluid analysis consistent with an immune-Veterinary and Comparative Orthopaedics and Traumatology Vol. 31 No. 6/2018 mediated process (polyarthropathy with non-degenerate neutrophils on synovial fluid analysis), resulting in a total of 21 elbows meeting the inclusion criteria for spontaneous septic arthritis (summary of case details is provided in ►Supplementary Appendix A, available in online version only). Breeds included Labrador Retrievers (n ¼ 11), and one each of English Springer Spaniel, Cross Breed, Munsterlander, Golden Retriever, Bull Mastiff, Rottweiler, German Shepherd, Saint Bernard, Cavalier King Charles Spaniel, Patterdale Terrier and Staffordshire Bull Terrier. The mean age of dogs was 6.8 years AE 2.3. The median body weight was 35.5 kg (9-83 kg). The right elbow joint was involved in 10/21 cases, left in 9/21 and 2/21 cases were bilateral.
A history of prior orthopaedic surgery was identified in 11 cases that met our prior inclusion criteria. Three of these had a history of surgery at a site distant to the infected elbow, (tibial plateau levelling osteotomy with implants in place). The remaining eight had a history of surgery on the septic elbow joint; however, it was outside of the time frame for exclusion as a surgical site infection. Seven did not have implants (elbow arthroscopy [n ¼ 5, 3-8 years prior], bilateral forelimb angular limb deformity and ulnar osteotomy [n ¼ 1, 9 years prior], bilateral elbow hygroma [n ¼ 1, 2 years prior]). The eighth case had a stainless steel transcondylar lag screw for incomplete ossification of the humeral condyles [1 year prior]. The median time since prior surgery was 3 years (range: 2 months to 8 years). Of the three cases that had a tibial plateau levelling osteotomy procedure performed, two had surgery within 2 months of presentation for forelimb lameness. Both of these cases had evidence of surgical site infection of the distant original surgical site suggesting the possibility of a haematogenous spread to the elbow.
At presentation, physical examination findings included joint effusion (n ¼ 21), pain upon manipulation of the affected joint (n ¼ 21), lethargy (n ¼ 8), muscle atrophy (n ¼ 6), regional lymphadenopathy (n ¼ 5), pyrexia (>39.2°C n ¼ 5) and systemic leucocytosis (n ¼ 6). Sixteen cases were referred as an emergency consultation due to an acute deterioration in lameness. Of these 16 cases, 12 had a chronic (>2 months) history of forelimb lameness prior to deterioration. The remaining five dogs were presented for an investigation of chronic lameness through routine referral consultation. The duration of deterioration in clinical signs in all dogs was median 4.5 days (1-120 days) and a lameness score on presentation was median 7.5/10 (range: 1-10). The group of dogs (n ¼ 5) presenting for investigation of chronic forelimb lameness had clinical signs of greater than 2 months and lameness score of median 5/10 (range: 1-10). Routine haematology and serum biochemical results were available for 13/21 cases. A leucocytosis was present in 4/13 cases with neutrophilia in 5/13. A thrombocytopenia (<150 Â 10 9 /L) was present in three cases, two of which had concurrent neutropenia (<3 Â 10 9 /L). Of these two cases, one (case 14) was receiving chemotherapy for ALL, and the other (case 15) was euthanatized due to clinical deterioration and signs of suspected sepsis (pyrexia, tachycardia and neutropenia). 16 Alkaline phosphatase was elevated in three dogs.
Imaging available for evaluation included orthogonal radiographs in six and the computed tomography of the elbow joint in nine elbows. Osteophytosis was present in 14/ 15 elbows, fragmentation of the medial coronoid was diagnosed in nine elbows, ununited anconeal process in one, incomplete ossification of the humeral condyle in one and humeral condyle osteochondritis dissecans in two. Global OA assessment was severe 11/15, moderate 1/15, mild 2/15 and absent 1/15.
Synovial total nucleated cell count was available for 13/21 elbows, with a mean of 102.2 AE 55.8 Â 10 9 cells/L (range: 13.7-183). The total nucleated cell count was below the inclusion level defined in this study for septic arthritis of 50 Â 10 9 /L in 2/13 cases (cases 14 and 21). In both these cases, the polymorphonuclear differential was greater than > 90%. Case 14 was included due to resolution in clinical signs following antibiotic therapy and case 21 subsequently had a positive bacterial culture. Cytological assessment was available for 20/21 elbows. Based on the differential cell count, polymorphonuclear cells predominated in all cases (mean: 91.4 AE 5.1% of the total nucleated cell count population). Degenerate neutrophils were present in only one case (1/20) and the intracellular bacteria were seen in five cases (3/5 subsequently having a positive culture result). Synovial fluid was submitted for culture in 21 cases with a positive culture obtained in 11/21 cases. Bacteria cultured included Staphylococcus aureus (n ¼ 4), Staphylococcus pseudintermedius (n ¼ 3), Streptococcus canis (n ¼ 2), Streptococcus agalactiae (n ¼ 1) and a multi-organism culture (Escherichia coli, Enterococcus faecalis and Staphylococcus pseudintermedius) (n ¼ 1). Antibiotic therapy had been given in 3/21 cases prior to referral and subsequent culture and sensitivity results; two of these (both post-tibial plateau levelling osteotomy infection), subsequently had a positive synovial fluid culture. Urinalysis was performed in 5/21 cases with a positive (S. aureus and E. coli) urine culture in two of these (cases 1 and 7). In case 7, bacteria isolated from the bladder (S. aureus) matched the synovial fluid suggesting a haematogenous origin. Dogs were treated either medically with antibiotic medications only (n ¼ 16), or surgically by joint lavage and antibiotic medications (n ¼ 2, cases 5 and 7), or arthroscopy, joint lavage and antibiotic medications (n ¼ 3, cases 6, 9 and 10). Joint lavage involved placement of an Veterinary and Comparative Orthopaedics and Traumatology Vol. 31 No. 6/2018 ingress and egress needle and flushing of the joint with 1-2 L of isotonic solution. The decision in treatment strategy was determined by the clinician at the time of diagnosis. In cases for which arthroscopy was performed (cases 6, 9 and 10), this was justified to manage concurrent medial compartment disease of the elbow. Antibiotic therapy included amoxicillin/clavulanic acid (Noroclav, Norbrook Laboratories Ltd., North Ireland) (n ¼ 12), amoxicillin/clavulanic acid (Noroclav, Norbrook Laboratories Ltd., North Ireland) and enrofloxacin (Baytril, Bayer plc, United Kingdom) (n ¼ 6), cephalexin (Cephacare; Animalcare Ltd, United Kingdom) and enrofloxacin (Baytril; Bayer plc, United Kingdom) (n ¼ 1), cephalexin (Cephacare) (n ¼ 1). For all 11 elbows with a recorded antibiotic sensitivity, the instigated empirical antibiotic therapy was appropriate. Antibiotic therapy was continued for a mean of 6 weeks AE 1.7 weeks.
Medium-to long-term follow-up (>8 weeks) information was available for 12/21 cases (median: 57 weeks; range: 14 weeks-7 years). Recurrence of infection was recorded in 3/12 occurring at 14 weeks (case 4), 1.2 years (case 1) and 3.8 years (case 6) after original diagnosis. Initial treatment in these three cases had included antibiotic therapy only in cases 1 and 4, and arthrotomy, joint lavage and antibiotic therapy in case 6. Residual lameness attributable to the elbow joint based on owner follow-up was reported in 11/12 cases. The median lameness score was 3/10 (range: 2-5). Case 8 had progressive ongoing lameness that was treated with total elbow replacement at another referral institution.

Discussion
This is the first retrospective case series to focus solely on spontaneous septic arthritis of the canine elbow. It was the authors' experience that the elbow is one of the most common joints to spontaneously develop septic arthritis, when excluding surgical site associated infections (<1 month prior if no implants, <1 year if implants present). 10 A preliminary review of all cases of septic arthritis was performed during data collection for this manuscript. Fifty cases of spontaneous septic arthritis were identified during the study period and the elbow had the highest prevalence within this group (21/50, 42%). In a similar smaller retrospective series, when recent surgical cases were removed, the elbow was again the predominant joint, 8/14 cases (57%). 5 In people, certain conditions are considered risk factors including rheumatoid arthritis or OA, old-age, skin infection, cutaneous ulcers, diabetes, joint prosthesis, intra-articular corticosteroid injection and intravenous drug abuse. 4,[17][18][19] These risk factors appear to be in accordance with our findings in dogs in that 85% of dogs were middle aged or older (mean age: 6.8 years), and pre-existing OA was present in 93% of cases in which imaging of the elbow was available, and concurrent medical conditions in 43% of our case population. Both ALL and anal furunculosis were treated with immunosuppressive therapy and it is likely the conditions and/or the treatment had contributed to the risk of septic arthritis developing in cases 6 and 14. 17 The presence of a transcondylar screw in case 17 potentially contributed to the development of infection. Surgical implants can act as a nidus for infection and subsequent removal of the implant and prolonged antibiotic therapy resulted in clinical improvement in case 17.
The main clinical signs seen in dogs with spontaneous septic arthritis of the elbow joint were joint effusion (21/21), pain on joint manipulation (21/21) and acute deterioration in lameness (16/21). Pyrexia was an inconsistent clinical finding (6/21-29%), similar to a previous case series (19.4%), 5 although notably lower than post-surgical stifle sepsis (75%). 6 In this study, large breed dogs and breeds with a susceptibility to elbow dysplasia were most common (17/21), likely reflecting a higher degree of underlying joint disease and OA in these groups. In non-immunocompromised people, pre-existing joint disease is often identified, with OA accounting for 33% of joint disorders. 8,19 Radiographic evaluation was available for 15 cases in the present series, and of these 14/15 had evidence of OA. The high prevalence (11/15 cases) of severe radiographic OA, as found in this study, is in accordance with previous reports in which severe OA was present in 5/8 elbows. 5 A positive bacterial culture was obtained for 11/21 of cases, consistent with previous reports of variable positive culture rates (20-80%). 3,5,9,20 Interestingly, the use of antibiotic therapy prior to culture in three cases did not appear to affect outcome. In 2/3 cases given antibiotic therapy prior to sampling, two still had a positive culture. There is conflicting information in the literature regarding the influence of antibiotic medications on culture success. Pre-culture antibiotic therapy has been linked with false-negative results in several studies, while others have reported no difference in culture success. 3,5,9 Despite this, current recommendations are to perform arthrocentesis prior to initiation of antibiotic therapy. In this study, single isolates of Staphylococcus spp. were the most common bacteria isolated (7/11) followed by Streptococcus spp. (3/11), which is similar to previous reports ranging from 42-59% 5,6 and 16-24% respectively. 3  bacteria isolated from both urine and synovial fluid. This finding highlights the importance of evaluating all potential sources of bacteria when a haematogenous origin is suspected. Importantly, 10/21 of elbows in this study had a negative culture and relied on a presumptive diagnosis based on high total nucleated cell count, predominance of polymorphonuclear cells and response to therapy. In two elbows, the total nucleated cell count was below the cut-off value for septic arthritis (50 Â 10 9 cells/L); however, they were included in the study based on other criteria: a high percentage of neutrophils, response to antibiotic therapy and subsequent culture results in one elbow. 11 The presence of a monoarthropathy with a predominantly neutrophilic cytology from synovial fluid sampling may be the only indication of septic arthritis/infection. This can make diagnosis and ruling out conditions such as immune-mediated polyarthropathies challenging. To that end, other diagnostic tests have been sought, such as molecular methods (bacterial ribosomal RNA [rRNA] gene sequencing), analysis of synovial lactate concentration and use of leukocyte esterase and glucose reagent strips. [20][21][22][23] However, even these new avenues for diagnosis are not without constraints, with comparisons between synovial fluid culture and rRNA polymerase chain reaction (PCR) analysis not being able to demonstrate improved accuracy in diagnosis, and a wide reported 95% confidence interval (CI) in the sensitivity of lactate to predict septic arthritis (sensitivity 1.00, 95% CI: 0.63-1.00). 20,23 Currently, synovial fluid inoculation into blood culture media, synovial biopsy and cytology examination are recommended. 14,16,24,25 The vast majority of septic elbows were treated by antibiotic therapy alone. The initial response to treatment was very good (94% resolution) and there was no difference in the response between cases treated with antibiotic medications alone compared with cases that had joint lavage and/or arthrotomy. This finding concurs with previous studies suggesting non-surgical management with antibiotic therapy alone 3,6,26 is sufficient due to the excellent blood supply in joints. Failure of treatment in case 9 was likely to have been due to insufficient antibiotic treatment duration (4 weeks) or inappropriate initial antibiotic implementation. Selection of surgical management may also have been reserved for more severely affected cases increasingly the risk of recurrence. However, the retrospective nature of this report does not allow further investigation of this potential bias. Subsequent extended treatment in case 9 with cephalexin resulted in clinical resolution and a significant improvement in lameness (20 mg/kg orally twice daily for 8 weeks). Longterm follow-up in 12 cases revealed a 25% recurrence of infection which is higher than that found in a previous smaller case series. 5 The high rate of recurrence in the elbow contrasts to that reported for septic arthritis of other joints such as the hip joint (0%), stifle (7%) or hock (0%). 4,26 In this series, recurrence occurred 14 weeks (case 4), 1.2 years (case 1) and 3.8 years (case 6) after initial diagnosis. The long periods between remission and recurrence are less suggestive of recrudescence of incompletely resolved infection and more likely a result of renewed inoculation of a vulnerable and compromised joint. However, it does remain possible that the recurrence of infection may be a result of quiescent bacteria remaining in the joint post-antibiotic treatment, or could represent haematogenous reseeding from the same or a new focus elsewhere in the body and an underlying predisposition to infection. 27,28 Case 4 may represent a late relapse due to insufficient antibiotic therapy duration (6 weeks), or represent a recurrence of infection since deterioration in lameness occurred following a period of 8 weeks of minimal reported lameness. Both cases 1 and 6 had predisposing factors for joint infection (diabetes mellitus and skin/urinary infection) and likely represent true recurrence in a predisposed joint. It is postulated that synovial vascular changes in OA joints predispose them to initial colonization, and re-colonization post-treatment. 11,27,29 In rheumatoid patients and OA human patients, altered joint structure, including thinner vascular canals associated with increased subchondral plate thickness, increased osteochondral vascular density may contribute to bacterial seeding and an increased risk of infection. 29,30 Analysis of both the migratory and phagocytic function of polymorphonuclear cells in the synovial fluid of humans with OA has shown a decreased function compared with rheumatoid patients. The altered function and potential anomalous joint structure may help to explain a component of the increased susceptibility of osteoarthritic patients to joint infections, 8,29,31 although we do not know if this is the case in the clinical canine patient.
A major limitation in this study is the retrospective design and reliance on assessing outcome from clinical records and low case numbers due to the relatively uncommon nature of this condition. The assessment of outcome is further compounded by the presence of pre-existing joint disease in the majority of dogs. Inclusion criteria were chosen to avoid the possible inclusion of non-infective cases based on previously described criteria. 32 However, due to the low positivity from synovial culture, diagnosis of infection is often presumptive and may have resulted in inclusion of aseptic joints.
In conclusion, middle aged, large breed dogs, with preexisting arthritis, that suffer an acute onset lameness, with elbow joint effusion and discomfort, even in the absence of pyrexia, should be considered for septic elbow arthritis. Antibiotic treatment was effective when prolonged treatment was instigated (6-8 weeks) appropriately; however, owners and veterinarians need to be aware of the potential for recurrence. 3,5,6,27 Although there is evidence supporting a good early or short-term response to medical therapy for septic arthritis, further evaluation of the long-term outcome and recurrence rates for dogs treated medically or surgically is warranted. In addition, improving the ability to rapidly and accurately diagnose cases is critical to allow appropriate and early implementation of therapy to our patients.

Funding
No funding was provided for this manuscript.

Conflict of Interest
None.

Author Contributions
Ben Mielke and Richard Meeson contributed to conception of study, study design, and acquisition of data and data analysis and interpretation. All authors drafted, revised and approved the submitted manuscript.