A Case of Postoperative Methicillin-Resistant Enterocolitis in an 81-Year-Old Man and Review of the Literature

Background

Before the development of the Clostridium difficile cytotoxin assay in the 1970s, Staphylococcus aureus was frequently implicated as a cause of infectious colitis, particularly in association with recent antibiotic therapy or abdominal surgery [1]. Decreased utilization of stool culture has reduced the recognition of S. aureus as a rare, but historically important, cause of enterocolitis [2]. We present a case of hospital-acquired, C. difficile-negative, diarrhea in a postoperative patient who developed sepsis with methicillin-resistant S. aureus (MRSA) bacteremia and radiographic proctocolitis, which was suspected to be the original source of systemic staphylococcal infection.

Case Report

An 81-year-old man with stage III pancreatic adenocarcinoma without neoadjuvant treatment was admitted for a Whipple procedure, which was expanded intraoperatively with total pancreatectomy, splenectomy, and portal vein resection due to repeatedly positive surgical margins. Portal vein reconstruction was performed with end-to-end anastomosis of remaining portal vein to superior mesenteric vein without vascular graft. He received cefazolin 2 g intravenously prior to surgical incision; additional intraoperative or perioperative antibiotics were not given. He had a medical history of hypertension, hyperlipidemia, gastroesophageal reflux disease, glaucoma, genital herpes, asthma, and a prior renal exophytic mass (fine-needle aspiration showed no evidence of neoplasm) status after cryoablation 2 years prior. It was during annual surveillance imaging for this renal mass that a pancreatic head mass was found. He had no recent diarrhea, travel, antibiotic or proton-pump inhibitor use, or hospitalization, and no history of inflammatory bowel disease. MRSA nares screen (routinely performed on admission at our institution) was positive.

Postoperatively, he was hypotensive and hypovolemic, which was suspected to be secondary to intraoperative fluid losses in addition to further fluid losses from peri-anastomotic Jackson-Pratt drains, and he developed prerenal acute kidney injury that improved with bolus and maintenance intravenous (IV) fluids without requiring vasopressors. No central venous access was required, but the indwelling urinary (Foley) catheter placed intraoperatively was continued postoperatively to closely monitor urine output until it was removed on postoperative day (POD) 5, after which he developed urinary retention that was managed with doxazosin and intermittent straight catheterization. He developed brittle diabetes after total pancreatectomy and required an insulin drip, which was transitioned to scheduled subcutaneous insulin injections on POD4. By POD3, the patient was noted to have waxing and waning encephalopathy consistent with hospitalization-associated delirium, which was managed with quetiapine as needed.

On POD5, the patient developed large-volume, foul-smelling, nonbloody loose stools that were attributed to tube feeding via nasogastric tube and pancreatic insufficiency. He was afebrile with an expected postsplenectomy leukocytosis and without vital sign changes, so he was not started on any empiric antibiotics but was given oral pancreatic enzyme replacement. Blood and urine cultures and C. difficile testing with polymerase chain reaction (PCR) obtained at this time were negative. By POD9, stool output increased to 26 bowel movements per day despite tube feeding cessation and oral pancreatic enzyme replacement. The patient received postsplenectomy vaccinations against Streptococcus pneumoniae, Haemophilus influenzae type B, and Neisseria meningitidis on the morning of POD10. Between POD5 and POD10, his vital signs remained stable, and quick sequential organ failure assessment (qSOFA) scores ranged from 0 to 1 (due to Glasgow coma score of 14 as a result of intermittent confusion). On the evening of POD10, the patient became clinically septic when he developed acute onset of dyspnea, tachycardia (110–130 beats/min), and tachypnea (30 breaths/min), as well as new fever (38.6°C), hypotension (70/40 mmHg), worsening leukocytosis (from 15.9 cells/nL post-splenectomy to 33.4 cells/nL with 98% neutrophils), elevated procalcitonin (0.9 ng/mL), and evidence of end-organ dysfunction (lactate 3.4 mmol/L and troponin 0.08 ng/mL); his qSOFA score rose from 0–1 to 3. Both peripheral IV and Jackson-Pratt sites were clean, Jackson-Pratt drainage was nonpurulent, and no central venous access or indwelling urinary catheter was present. He was empirically heparinized for suspected pulmonary embolism, trialed on bilevel positive airway pressure for increased work of breathing, and started on empiric broad-spectrum antibiotics (IV vancomycin 1 g every 12 hours and IV piperacillin-tazobactam 3.375 g every 8 h) as well as empiric treatment for C. difficile colitis (oral vancomycin 125 mg every 6 h). Vasopressor therapy was not necessary. Computed tomography of the chest, abdomen, and pelvis did not demonstrate pulmonary embolism (empiric heparin drip was subsequently discontinued) or pneumonia; however, it revealed severe sigmoid and rectal colitis evidenced by extensive bowel wall thickening and pericolic fat stranding. Blood cultures at this time grew MRSA in 3 out of 4 bottles within 24 h, after which piperacillin-tazobactam was discontinued. Repeat blood cultures after 24 h of antibiotics also grew MRSA, and a repeat C. difficile PCR on POD12 was negative. Erythrocyte sedimentation rate and C-reactive protein were not checked preoperatively or postoperatively until POD12, at which time both were elevated, to 62 mm/h and 14.7 mg/L, respectively. Procalcitonin was not checked during the patient’s hospital course. MRSA isolates from the patient’s nares and blood were not compared during routine clinical practice to determine whether they were from the same strain. Stool samples adequate for evaluation and culture could not be obtained until POD15, at which time they demonstrated elevated fecal lactoferrin and moderate growth of Candida albicans (antifungal treatment was not started because this organism was favored to be an asymptomatic colonizer rather than a gastrointestinal pathogen in the absence of candidemia) [3,4]. Further infectious source workup ruled out endocarditis, sinusitis, skin and soft tissue infection, spinal and psoas abscesses, and thrombophlebitis. An endoscopy was not performed owing to the patient’s high risk for anastomotic perforation. The patient clinically improved and his diarrhea resolved within 48 h of initiating treatment with IV and oral vancomycin (1 g IV every 12 h for 4–6 weeks [5], and 125 mg orally every 6 hours for 14 days). After roughly 2 weeks of IV vancomycin, the patient developed intrinsic renal injury secondary to vancomycin-associated nephrotoxicity, so he was switched from IV vancomycin to IV daptomycin 8 mg/kg (750 mg) every 24 h to complete 6 weeks of antibiotic therapy. The patient was discharged to a skilled nursing facility and was evaluated in general surgery clinic several weeks later, at which time he was found to be in good clinical condition off antibiotics.

Discussion

The history of staphylococcal enterocolitis – from its recognition in the 1940s to its eclipse by C. difficile colitis in the 1970s – is well summarized in prior reports [1,2]. Staphylococcus aureus was previously a commonly recognized cause of infectious colitis; however, the increased recognition of C. difficile and expansion of nucleic acid amplification testing has dramatically altered the workup and treatment of nosocomial diarrhea to favor testing and treatment for C. difficile without utilizing stool culture. As such, the diagnosis of S. aureus enterocolitis has become infrequent or delayed and can result in progression to severe systemic infection (e.g., sepsis with bacteremia according to Sepsis-3 [6] and Surviving Sepsis Campaign [7] guidelines), which in our case ultimately suggested the diagnosis.

Prior cases of S. aureus enterocolitis in adults (Table 1), at least in the English literature, have been primarily reported from the United States (n=383) and the European Union (n=300, including the United Kingdom); however, the US and EU numbers are dwarfed by the 432 cases reported from Japan (which includes the 91 Japanese cases we identified in addition to the 341 summarized by Iwata and colleagues [8]). The vast majority of prior S. aureus colitis cases (74%, 581/782) were reported to be antibiotic associated, and the next most common risk factors were recent gastrointestinal surgery (18%, 140/782) and inflammatory bowel disease (2%, 12/782). Of the 281 patients with reported outcome (excluding 2 patients reported by Sommers and colleagues in 2 separate reports [9,10]), death occurred in 66 (23%), which may be presumed to be at least partially attributable to S. aureus infection due to the variable follow-up intervals reported, which were often brief and limited to hospital stay. Based on aggregating these previously published reports, the case fatality rate for S. aureus enterocolitis has declined over time, from 74% (1950–1960) to 30% (1960– 1970) to 11% (2010–2020), which may correlate with expanded access to antibiotics such as vancomycin (1954), which is a first-line treatment for C. difficile colitis and also has a specific indication approved by the US Food and Drug Administration for the treatment of staphylococcal enterocolitis.

Our patient’s putative diagnosis of MRSA proctocolitis is a controversial one, especially in the absence of definitive endoscopic evidence (due to high risk of anastomotic compromise) or confirmatory stool culture, which is infrequently performed in current clinical practice and is often limited by delayed results and low negative predictive value [11–14]. However, our diagnosis is supported by the following lines of evidence. The likelihood of clinically significant C. difficile colitis was low given the absence of C. difficile antigen and toxin on 2 consecutive tests (each with sensitivity and specificity >90% and negative predictive value >95%), and toxin-negative C. difficile infection may not require antibiotic treatment as our patient did [15–18]. Nasal colonization with S. aureus (which can occur in about 30% of people, with MRSA colonization occurring in about 1–2% of people) and placement of a nasogastric feeding tube increased his likelihood for gastrointestinal staphylococcal colonization (which can also be present in about 15–25% of people) [2,19–23]. Several prior studies and reviews have also described clinical features and risk factors (large-volume and high-frequency stool output, age over 70 years, tube feeding), all of which were present in our case, that may favor S. aureus enterocolitis over C. difficile colitis [2,24,25]. Bacteremia with C. difficile has been infrequently reported compared with bacteremia associated with MRSA (although nosocomial diarrhea has been associated with increased risk of nosocomial infection, especially urinary tract infection), so it may be possible that gastrointestinal infection with subsequent MRSA bacteremia may be more characteristic of MRSA colitis than C. difficile colitis [26–28]. Therefore, in the absence of other more common sources, the likely source of his bacteremia was his colon, due to either translocation of nondiarrheagenic MRSA due to inflammation caused by non-infectious diarrhea, or MRSA as the principal cause of intestinal inflammation.

Conclusions

The management of diarrhea in hospitalized patients is complex and requires both the investigation of multiple noninfectious etiologies, especially in postsurgical and oncologic patients, and the consideration of pathogens not included in routine laboratory testing. Our case highlights the potential for staphylococcal enterocolitis or translocation of colonizing staphylococcal species into the bloodstream to cause severe S. aureus infection. At present, empiric treatment for C. difficile is common in clinical practice, and while vancomycin currently treats both S. aureus and C. difficile, the emergence of antibiotic resistance may require distinct treatments for these 2 pathogens. As such, recognition of S. aureus enterocolitis as a distinct clinical entity may become more important over time because it would necessitate dedicated antibiotic coverage. Although the clinical significance of staphylococcal colonization in the gastrointestinal tract and its potential to cause enterocolitis remains controversial, our case and others summarized in Table 1 should prompt physicians to consider this rare diagnosis, which has a high case fatality rate, in situations in which infectious nosocomial diarrhea is suspected but Clostridium assays are negative.