Post-operative Salmonella surgical site infection in a dog

Abstract

Following decompressive surgery for degenerative lumbosacral stenosis, a 6-year-old German shepherd dog developed a subcutaneous infection at the surgical site and discospondylitis at the lumbosacral intervertebral disc. Salmonella enterica subsp. enterica, serotype Dublin was recovered from the surgical site. Salmonella of a different serovar was isolated from a sample of the raw meat-based diet that the owner fed the dog.

A 6-year-old, 35-kg, intact male German shepherd dog was evaluated at the University of Georgia, College of Veterinary Medicine Veterinary Teaching Hospital for a right pelvic limb lameness of approximately 8-months’ duration. Before onset of the chronic lameness, the dog was reported to be very active, working intensively several times a week in Schutzhund training exercises. Initially, the dog was treated conservatively for the lameness with meloxicam (Metacam; Boehringer Ingelheim Vetmedica, St. Joseph, Missouri, USA), 0.1 mg/kg body weight (BW), PO, as needed, and exercise restriction.

At the time of presentation to the veterinary teaching hospital, the dog had a body condition score (BCS) of 5/9 and a normal muscle condition score (MCS). The medical record indicated that the dog was being fed an “agility formula diet, multivitamins, and omega-3 supplements.” The neurological examination revealed a postural reaction deficit, normal patellar tendon reflex, and a reduced withdrawal reflex in the right pelvic limb. The dog was not in pain upon palpation over the lumbosacral vertebral articulation or with manipulation of the pelvic limbs or tail. No other abnormalities were identified on physical examination. Based on these neurological deficits, the anatomic diagnosis was consistent with a lesion involving the right L6 to S1 spinal cord segments, spinal nerve roots, spinal nerves, or sciatic nerve. Differential diagnoses included degenerative lumbosacral stenosis, intervertebral disc (IVD) disease, neoplasia, inflammatory (infectious and noninfectious) meningomyelitis, trauma, or discospondylitis.

Results of a complete blood (cell) count (CBC), serum chemistry profile, urinalysis, and 3-view thoracic radiographs were all normal. Magnetic resonance imaging (MRI) (GE 3.0T Signa HDx; GE Healthcare, Milwaukee, Wisconsin, USA) of the lumbosacral vertebral column was conducted under general anesthesia. The MRI identified protrusion of the lumbosacral (L7-S1) IVD resulting in compression of the cauda equina. Also, there was compression of the cauda equina from the articular processes and yellow ligament. On T2-weighted images, there was a reduction in the signal intensity of the epidural fat in the vertebral canal and of the nucleus pulposus at the lumbosacral IVD (Figure 1). Following intravenous administration of gadolinium-based contrast medium (Magnevist; Bayer Healthcare Pharmaceuticals, Wayne, New Jersey, USA), abnormal contrast enhancement was not observed. The MRI findings were consistent with degenerative lumbosacral stenosis (1).

Figure 1.

The sagittal T2-weighted magnetic resonance image of the caudal lumbar vertebrae and sacrum of a 6-year-old German shepherd dog reveals intervertebral disc herniation (arrowhead), along with a loss of the signal intensity of the nucleus pulposis (arrow) and epidural fat.

After discussion of the MRI findings with the owner, the dog underwent a dorsal laminectomy at the lumbosacral articulation and partial discectomy of the L7-S1 IVD. Aseptic technique was maintained throughout the surgical procedure. Briefly, the hair was clipped over the surgical site and the skin was aseptically prepared using alcohol and chlorhexidine gluconate. Standard draping using 4 towels placed around the surgical area followed by an overlying water-impenetrable disposable paper drape was used. An iodophor-impregnated adhesive drape (Ioban 2 antimicrobial incise drape; 3M, Saint Paul, Minnesota, USA) was placed over the exposed skin in the surgical field. During surgery, cefazolin (Sandoz, Princeton, New Jersey, USA), 22 mg/kg BW, IV, was administered approximately 30 min before making the skin incision; a second dose was administered 90 min after the initial administration.

After surgery, prednisone (West-Ward Pharmaceutical, Eatontown, New Jersey, USA), 0.4 mg/kg BW, PO, q12h, tramadol (Janssen Pharmaceuticals, Titusville, New Jersey, USA), 2.8 mg/kg BW, PO, q12h, and gabapentin (Amneal Pharmaceuticals of New York, Hauppauge, New York, USA), 0.8 mg/kg BW, PO, q12h, were administered. The dog’s recovery from surgery was uneventful and he was discharged from the hospital 2 d after surgery. The owner was instructed to continue administering tramadol and gabapentin and to gradually reduce the prednisone dose over a 2-week course.

Nine days after surgery, the owner reported that the dog acutely developed lethargy, anorexia, diarrhea, and a right pelvic limb pain and weakness. The dog’s medications at that time included prednisone at 0.4 mg/kg BW, PO, q24h and tramadol and gabapentin at the previously prescribed dosages. At the time of re-evaluation, the dog’s rectal temperature was 39.4°C. He was reluctant to walk and he was paraparetic, but more affected in the right pelvic limb. There was a small non-painful fluctuant swelling at the incision site. Neurological examination revealed normal patellar reflexes, but there were reduced withdrawal reflexes bilaterally. The remainder of the neurological examination was normal. A CBC showed a leukocytosis [15.7 × 10⁹ cells/L, reference interval (RI): 5.5 to 13.9 × 10⁹ cells/L] and mature neutrophilia (13.6 × 10⁹ cells/L, RI: 2.9 to 12 × 10⁹ cells/L). The serum chemistry profile showed hyperal-buminemia (41 g/L, RI: 22 to 39 g/L) and an elevated alkaline phosphatase (420 U/L, RI: 10 to 117 U/L). Radiographs of the lumbosacral vertebral column disclosed a narrowed L7-S1 IVD space and ventral spondylosis deformans (Figure 2). Cytological examination of fluid obtained from the subcutaneous tissues overlying the surgical area showed mixed inflammation (59% non-degenerate neutrophils, 38% large mononuclear cells with foamy cytoplasm, and 3% small lymphocytes). Samples from the aspirate at the surgical site and urine collected via cystocentesis were submitted for aerobic culture at the university’s diagnostic laboratory.

Figure 2.

On the lateral radiograph of the lumbosacral vertebral column obtained 9 days after surgery, the L7-S1 intervertebral disc space is narrowed (arrow) and ventral spondylosis deformans is present. Changes compatible with discospondylitis are not present. The absence of the cranial laminae of the sacrum, and caudal aspect of the laminae and spinous process of the L7 vertebra are a consequence of the surgery.

The dog was treated with amoxicillin/clavulanic acid (Clavamox; Zoetis, Kalamazoo, Michigan, USA), 12.5 mg/kg BW, PO, q12h, and enrofloxacin (Baytril; Bayer Animal Health, Shawnee Mission, Kansas, USA), 8.4 mg/kg BW, PO, q24h. Tramadol and gabapentin were continued, but prednisone was discontinued. Within 24 h, the fever and discomfort resolved and the dog had improved strength in both pelvic limbs; he was discharged from the hospital 24 h after admission with instructions to continue the amoxicillin/clavulanic acid and enrofloxacin at the same doses for a minimum of 8 wk.

Six days after re-evaluation at the hospital, the sample from the surgical site grew Gram-negative bacilli identified by the Kauffman-White scheme (2) as Salmonella enterica subsp. enterica, serotype Dublin. The organism was susceptible to enrofloxacin. Urine culture did not yield any organisms. The owner was instructed to continue administering enrofloxacin and to restrict the dog’s activity to short leash walks. Based on the finding of Salmonella spp. in the surgical site, the clinicians attempted to identify the source of the infection. At that time, it came to light in conversation with the primary care veterinarian that the dog was being fed a raw meat-based diet (RMBD). The owner confirmed that the dog (and 2 other dogs in the household) were typically fed a commercial extruded diet (Life’s Abundance Premium Health Food for Adult Dogs; Life’s Abundance, Jupiter, Florida, USA). However, during training exercises, dogs were fed a commercially prepared, frozen RMBD (Natural Mix for Dogs and Beef for Dogs, Blue Ridge Beef, Statesville, North Carolina, USA) as a reward for a good performance. The owner estimated that the last feeding of the RMBD had been several weeks before surgery. No other treats were fed to the dogs. Despite the fact that the dog had been receiving antibiotics for 6 d, a single fecal sample from the dog was submitted for Salmonella polymerase chain reaction (PCR) testing and culture. A single fecal sample from each of the other 2 dogs in the household was submitted for similar testing. In addition, samples of the dog’s current extruded diet (Life’s Abundance Premium Health Food for Adult Dogs) and a sample from each of 2 separate, unopened packages of the RMBD (Natural Mix for Dogs and Beef for Dogs) were cultured for Salmonella spp. using a specific protocol for food material (3). Isolated Salmonella spp. were serotyped using the Kauffmann-White Scheme (2). Both samples of the RMBD were positive for Salmonella enterica serotype Uganda and the sample of the extruded diet was negative. Fecal samples from the 3 dogs were negative for Salmonella organisms with PCR testing and culture.

The dog was re-evaluated approximately 4 wk after surgery. The dog’s body weight, BCS, and MCS were unchanged. The dog’s diet was solely its commercial extruded diet. The owner reported that the dog’s weakness had improved. Radiographs of the lumbosacral vertebral column at this visit showed irregularly marginated lysis surrounded by sclerosis at the caudal endplate and metaphysis of the L7 vertebra and at the cranial endplate and body of the sacrum (Figure 3). The L7-S1 IVD space was narrowed with periosteal new bone formation ventrally. The radiographic findings were consistent with discospondylitis. Based on these findings, the owner was instructed to continue exercise restriction and enrofloxacin therapy for a total of 12 wk.

Figure 3.

Lateral radiographs of the lumbosacral vertebral column 13 days after surgery disclosed irregularly defined lysis of the caudal endplate of the L7 vertebra and the cranial endplate and cranial region of the sacrum (arrowhead) with surrounding sclerosis. The L7-S1 intervertebral disc space remains narrowed. Periosteal new bone formation is present ventrally (arrow). The radiographic findings are consistent with discospondylitis.

At the time of re-evaluation 4 mo after surgery, the dog’s body weight, BCS, MCS, and diet were unchanged. Enrofloxacin had been discontinued after 10 wk of administration. The owner reported persistent subtle weakness and continued circumduction of the right pelvic limb. Radiographs of the lumbosacral vertebral column revealed a subjective increase in sclerosis of the caudal and cranial regions of L7 and the sacrum, respectively. Further lysis was not observed.

Six months after surgery, the dog developed acute right pelvic limb lameness. The dog was evaluated by the primary care veterinarian who prescribed carprofen (Rimadyl; Zoetis), 2.1 mg/kg BW, PO, q12h, and exercise restriction. The dog was re-evaluated at the veterinary teaching hospital 2 wk later, at which time the dog continued to display a weight-bearing lameness and circumduction of the right pelvic limb during the swing phase of the gait. Radiographs of the lumbosacral vertebral column showed changes similar to those observed 4 mo after surgery. Based on the persistent gait deficits, possible future relapses with activity, and the owner’s perception that the dog would never return to its intensive training activity, the owner decided to euthanize the dog at that time. The owner allowed a limited necropsy restricted to evaluation of the L7 vertebra, sacrum, and IVD. Grossly, the IVD space was collapsed with fibrous tissue in the area of the IVD. Microscopically, there were areas with masses of connective tissue and islands of fibrocartilage filling the medullary spaces of the trabecular bone adjacent to the IVD space. Basophilic reveal lines were seen within the trabeculae. Rare macrophages and lymphocytes were observed. Findings were consistent with remodeling of bone.

Discussion

Discospondylitis is an infection involving the IVD that extends to the adjacent vertebral endplates and vertebral bodies (4–6). Discospondylitis is usually the consequence of hematogenous spread of bacteria (7), and can occur secondary to cardiovascular infections (8), foreign bodies (9), epidural injection (10), as well as surgery (4,11,12). However, direct infection of the vertebra or disc space also can occur (7).

In most reports of discospondylitis, the etiologic agent is identified or suspected based on a positive blood or urine culture and response to therapy. Culture of the infected IVD material provides definitive identification but is rarely performed (13). In the present case, Salmonella was the organism suspected to be responsible for discospondylitis as it was cultured from the subcutaneous tissue overlying the surgical site at the time of clinical signs of infection; however, Salmonella was not cultured from the affected IVD space so confirmation of Salmonella spp. as the cause of the discospondylitis was not possible. Salmonella spp. are not a common cause of discospondylitis; however, discospondylitis and epidural empyema in a dog have been reported secondary to Salmonella spp. infection (14). In that case, the diagnosis was established through positive blood cultures alone (cultures of the subcutaneous tissues were negative and IVD aspirates were not obtained) (14). A complete diet history was not provided, nor was the diet cultured, so diet could not be linked as a possible source of Salmonella infection. In that case report and in the current case, Salmonella as the cause of discospondylitis remains presumptive since cultures of the affected IVDs were not obtained.

Salmonella enterica subsp. enterica, serotype Dublin was cultured from the subcutaneous tissues overlying the surgical site. This bacterium is a host-adapted bovine strain that can infect young calves and adults; some infected calves fail to clear the infection and serve as carriers (15). Given its host adaptation, clinical infection by S. enterica subsp. enterica, serotype Dublin in dogs is uncommonly reported, however, one study identified S. enterica subsp. enterica, serotype Dublin as the second most common Salmonella serotype isolated from dogs in the United Kingdom (16). Although limited clinical data were reported, clinical disease included diarrhea, meningoencephalitis, and a vaginal infection, along with its isolation from clinically healthy dogs (16). Since the RMBD fed to the dog in the current report was beef-based, this is additional evidence that the RMBD may have been the source of the surgical site infection, although it is important to note that S. enterica subsp. enterica, serotype Dublin was not isolated from the dog’s food. The fact that 2 different Salmonella serotypes were isolated from the surgical site and diet — S. Dublin from the surgical site and S. enterica subsp. enterica, serotype Uganda from the RMBD — does not preclude a link between the diet and the surgical site infection. When fed Salmonella-contaminated diets, some dogs shed 1 or more than 1 serotype different from those isolated from the diet, suggesting infection with multiple serotypes (17). In the present case, the RMBD may have been contaminated with multiple serotypes despite isolation of a single type. The inability to culture or identify that serotype in the feces also could be related to the long lag period after the last ingestion of a RMBD, to antibiotic therapy before analysis, or to intermittent shedding of Salmonella spp.

The presence of Salmonella spp. in the surgical site of the dog, as well as in the dog’s diet are concerning for the dog and for public health. Although the RMBD and the infection of the surgical site in the present case cannot be definitively linked, the contaminated RMBD is important for the dog, the owner, and the veterinary healthcare team. The identification of Salmonella in the RMBD is not surprising since studies show a prevalence rate of 21% to 48% of Salmonella contamination of commercially available RMBDs (17–20). While less information is available for home-prepared RMBD, Salmonella-positive culture results were obtained in 8 of 10 home-prepared RMBDs (21). A variety of other pathogens also can be found in commercial or home-prepared RMBD, including Escherichia coli O157:H7 (22), Clostridium spp. (20), Listeria monocytogenes (18), Campylobacter jejuni (23), and Toxoplasma gondii (24–27). Both home-prepared and commercially available RMBDs are at risk for contamination since many of these pathogens are not completely destroyed by freezing and freeze-drying which are the forms in which most commercially available diets are sold. Even commercial RMBDs which have been treated by high-pressure pasteurization may contain pathogens as susceptibility to high-pressure pasteurization varies among pathogens (28), and the shipping or storage conditions of these diets may put them at risk for cross-contamination.

Given the high prevalence of contamination and the fact that between 3% and 50% of dogs fed RMBD intermittently or as a primary diet shed Salmonella organisms in their feces (17,21,29,30), a gastrointestinal (GI) source of contamination of the surgical site appears most likely. There are at least 2 possible mechanisms of spread via the GI tract. One could have been contamination of the surgical site from the dog’s feces. Another possible route could have been via bacterial translocation from the GI tract with hematogenous spread to the surgical site. In humans, bacterial translocation from the GI tract occurs during surgical procedures, particularly those involving the GI tract (31). The dog’s perioperative corticosteroids also could have increased the risk for bacterial translocation (32). Contamination also may have occurred via other routes including direct contact with contaminated RMBD, cross-contamination of owner’s hands, food bowls, counters, freezers, or the feces of other dogs. Although a GI route of transmission is most likely, a recent report of Salmonella bacteriuria in a cat fed a Salmonella-contaminated diet raises the possibility of a urinary route of contamination (33), although the urine culture of the dog herein was negative. Other environmental sources of Salmonella infection also are possible.

The pathophysiology of discospondylitis may involve slow blood flow through the vertebral epiphyses, allowing for hematogenous spread of bacteria into the adjacent IVD (7). While such pathophysiology may underpin discospondylitis, it highlights the need for a primary site of infection as a source of bacteria. Although urinary tract infections are commonly implicated as the primary source of bacteremia in dogs with discospondylitis, other sources include skin, dental, and cardiovascular infections (7,34). For dogs undergoing surgery involving the vertebral column, the prevalence of postoperative discospondylitis ranges from < 1% to 2.2% of cases (12). In the case herein, the combination of a laminectomy and partial discectomy of the L7-S1 IVD and Salmonella infection of the surgical site could have predisposed the dog to Salmonella infection of the IVD via a hematogenous route or via direct extension along the surgical field. Alternatively, it is possible that the infection was present before surgery as 23% of dogs undergoing lumbosacral decompressive surgery for degenerative lumbosacral stenosis had positive cultures when the IVD material was tested (although only 52/156 of the dogs in that retrospective study had disc material evaluated) (35).

In this dog, the authors hypothesize that ingestion of a Salmonella-contaminated commercial RMBD led to colonization of the GI tract, Salmonella shedding in the feces, and subsequent infection of the surgical site. The dog’s perioperative corticosteroids also could have increased the risk for infection. While a causal relationship between the RMBD and discospondylitis in the current case is not proven but surmised, this case highlights another potential hazard of feeding RMBDs (36). The veterinary healthcare team should obtain a diet history for every companion animal during every evaluation to help with diagnosis and optimal treatment (37,38). The diet history can not only help to identify if the animal’s diet is optimized, but may also identify potential health risks. If a diet history reveals that owners are feeding a RMBD (or raw dried or freeze-dried pet treats, which would carry similar risks to RMBD), those owners should be counseled on the health risks to themselves and their pets as a result of this feeding strategy. The diet history also should make clinicians consider salmonellosis in dogs fed a RMBD that develop post-operative complications such as surgical site infection, diarrhea, or other GI illness. The diet history also can help to determine precautions during handling and hospitalization for the animals themselves, other animals in the hospital, and for staff. Given the potential association between a RMBD and a serious post-operative infection, future research is warranted to evaluate the risk of post-operative infections in animals consuming RMBDs. CVJ