Spontaneous bacterial peritonitis causing Serratia marcescens and Proteus mirabilis ventriculoperitoneal shunt infection

Case report

Full access

✓The authors report their experience treating a polymicrobial ventriculoperitoneal (VP) shunt infection in a developmentally delayed 21-year-old woman. Cerebrospinal fluid (CSF) cultures grew Serratia marcescens and Proteus mirabilis. On admission and throughout her hospitalization, results of physical examination of her abdomen were normal, and radiographic studies showed no evidence of bowel perforation or pseudocyst formation. Contrast-enhanced computed tomography of the abdomen revealed a small fluid collection. After a course of intravenous gentamicin and imipenem with cilastatin in conjunction with intrathecal gentamicin, the infection was resolved and the VP shunt was reimplanted.

Although VP shunt infections are not uncommon, S. marcescens as a causative agent is exceedingly rare and potentially devastating. Only two previous cases of S. marcescens shunt infection have been reported in the literature. Authors reporting on S. marcescens infections in the central nervous system (CNS) have observed significant morbidity and death. Although more common, the presence of P. mirabilis in the CSF is still rare and highly suggestive of bowel perforation, which was absent in this patient. Spontaneous bacterial peritonitis was the likely source from which these bacteria gained entrance into the VP shunt system, eventually causing ventriculitis in this patient.

The authors conclude that in light of the high morbidity associated with S. marcescens infection of the CNS, intrathecal administration of gentamicin should be strongly considered as part of first-line therapy for S. marcescens infections in VP shunts.

Abbreviations used in this paper:CNS = central nervous system; CSF = cerebrospinal fluid; CT = computed tomography; SBP = spontaneous bacterial peritonitis; VP = ventriculoperitoneal.

Abstract

✓The authors report their experience treating a polymicrobial ventriculoperitoneal (VP) shunt infection in a developmentally delayed 21-year-old woman. Cerebrospinal fluid (CSF) cultures grew Serratia marcescens and Proteus mirabilis. On admission and throughout her hospitalization, results of physical examination of her abdomen were normal, and radiographic studies showed no evidence of bowel perforation or pseudocyst formation. Contrast-enhanced computed tomography of the abdomen revealed a small fluid collection. After a course of intravenous gentamicin and imipenem with cilastatin in conjunction with intrathecal gentamicin, the infection was resolved and the VP shunt was reimplanted.

Although VP shunt infections are not uncommon, S. marcescens as a causative agent is exceedingly rare and potentially devastating. Only two previous cases of S. marcescens shunt infection have been reported in the literature. Authors reporting on S. marcescens infections in the central nervous system (CNS) have observed significant morbidity and death. Although more common, the presence of P. mirabilis in the CSF is still rare and highly suggestive of bowel perforation, which was absent in this patient. Spontaneous bacterial peritonitis was the likely source from which these bacteria gained entrance into the VP shunt system, eventually causing ventriculitis in this patient.

The authors conclude that in light of the high morbidity associated with S. marcescens infection of the CNS, intrathecal administration of gentamicin should be strongly considered as part of first-line therapy for S. marcescens infections in VP shunts.

Both Serratia marcescens and Proteus mirabilis are classified as members of the Enterobacteriaceae family of bacteria.11,20 While S. marcescens tends to be an opportunistic pathogen, causing infections of the respiratory and urinary tracts, wounds, and the bloodstream in immunocompromised patients,12 P. mirabilis is commonly present in the gastrointestinal flora, and a P. mirabilis infection may be considered community acquired. Proteus mirabilis typically causes recurrent urinary tract infections in patients with structural abnormalities of the genitourinary system.20 It is uncommon for these species, alone or in combination, to cause CNS infections in patients with or without CSF diversion procedures. The presence of P. mirabilis in the CSF of a patient with a VP shunt is highly suggestive of an interruption in the alimentary system. In contrast, the presence of S. marcescens in the CSF of a patient with a shunt is exceedingly rare and potentially devastating; only two cases have been reported in the literature.5,16

We report our experience in managing a case of infection with a rare combination of pathogens due to SBP. The pharmacological and surgical management is described, and the relevant literature regarding SBP in patients with VP shunts is reviewed along with the two previous reports of S. marcescens infection of VP shunts. Despite the presence of the aforementioned pathogens and their postulated route of entry, we concluded that in the absence of clinical symptoms or radiographic abnormalities, the peritoneal cavity remains an acceptable site for CSF diversion.

Case Report

History.

A 21-year-old developmentally delayed woman presented with a 24-hour history of nausea, vomiting, lethargy, and difficulty walking. Her medical history was significant for a Grade III subependymal hemorrhage at birth, requiring VP drainage; epilepsy; and idiopathic thrombocytopenia purpura. The patient had undergone a shunt revision procedure to relieve a proximal obstruction 6 months before this presentation; the peritoneal catheter had not been replaced at that time. The patient experienced vague abdominal symptoms, with no fever or chills, 2 weeks prior to the patient’s presentation. These symptoms resolved spontaneously.

Examination.

The patient was afebrile. She had no peritoneal signs or other evidence of abdominal disease. Blood tests revealed a white blood cell count of 5800/mm3. On neurological examination, the patient was noted to be lethargic and to have right sixth cranial nerve palsy; she had no photophobia or other evidence of meningism. Her previous incisions were well healed with no sign of infection.

An unenhanced head CT scan showed marked interval enlargement of the ventricles and a right parietooccipital ventricular catheter in a satisfactory position. Anteroposterior and lateral radiographs of the head, neck, chest, and abdomen showed no discontinuity in the shunt apparatus. The tip of the peritoneal catheter was visible beneath the liver. There was no evidence of bowel obstruction, free air beneath the diaphragm, or cardiopulmonary disease.

Attempts to sample CSF from the shunt valve reservoir were unsuccessful.

First Operation.

Surgery was undertaken for correction of proximal shunt malfunction. The obstructed ventricular catheter was replaced, but CSF obtained from the new catheter was noted to be grossly cloudy and xanthochromatic, prompting immediate laboratory studies. Gram staining revealed a large number of Gram-negative rods. In light of this finding, the planned proximal shunt revision was aborted, the entire VP shunt system was removed, and the new ventricular catheter was externalized.

Postoperative Course.

The results of CSF studies included a glucose level of 27 mg/dl, a protein level of 759 mg/dl, a white blood cell count of 226/mm3, and a red blood cell count of 79/mm3. No gross discoloration or staining of the distal catheter, which was not spring loaded, was seen upon its removal. Gram stains and cultures of specimens from the peritoneal catheter tip were negative for the presence of bacteria. Intravenous treatment with gentamicin and imipenem/cilastatin was initiated. Cultures of CSF revealed the presence of S. marcescens and P. mirabilis. Cultures of blood and urine specimens were nondiagnostic. Sputum was not cultured because the patient had a normal chest radiograph and no respiratory complaint.

Concern over the possibility of an abdominal source of infection prompted a CT scan of the abdomen with oral and intravenous contrast agents. Results of the CT study demonstrated a moderate amount of free fluid in the area of the Douglas cul-de-sac, with surrounding enhancement of the peritoneal surface but no gross abscess cavity (Fig. 1). There was no evidence of fluid collection beneath the liver where the peritoneal catheter had been, or of pelvic or inguinal lymphadenopathy.

f1-1050320

Fig. 1. Oral and intravenous contrast–enhanced CT of the abdomen obtained immediately after removal of the entire VP shunt system. Arrow indicates the fluid collection near the Douglas cul-de-sac. There was no fluid collection beneath the liver where the distal end of the peritoneal catheter had been.

Repeated attempts to obtain a specimen of the abdominal fluid transvaginally were unsuccessful, as were CT- and ultrasonography-guided sampling procedures. Throughout these efforts, the patient remained free of abdominal complaints and the physical examination showed no sign of a pathological condition.

Despite bacterial sensitivities for both gentamicin and imipenem, S. marcescens persisted in the CSF surveillance cultures 1 week after commencement of treatment, although the P. mirabilis infection did resolve. The CSF profiles continued to show an elevated level of protein (> 200 mg/dl) and a low level of glucose. Clinically, the findings on sequential neurological examinations indicated continued deterioration in the patient’s condition. She became persistently febrile and had increasing leukocytosis.

Second Operation and Postoperative Course.

Seven days after the initial surgery, we replaced the patient’s external ventricular catheter with a catheter impregnated with rifampin and clindamycin. Intrathecal gentamicin (10 mg twice daily) was added to her antibiotic regimen for a total of 10 days.

After these additional measures were undertaken, the findings on CSF surveillance studies improved. Repeated bacterial cultures were negative, and the patient’s neurological condition improved to her baseline. She completed a 21-day course of intravenous imipenem and cilastatin, a 14-day course of intravenous gentamicin, and a 10-day course of intrathecal gentamicin. In light of the abdominal fluid collection and our uncertainty regarding the source of the patient’s polymicrobial ventriculitis, we considered placement of a ventriculoatrial or ventriculopleural shunt. General surgeons were consulted regarding the possibility of an occult perforation. After a comprehensive metabolic evaluation showed no abnormality, the consensus conclusion was that no further intervention was indicated.

At this point, we concluded that SBP had been the source of the patient’s ascending shunt infection.

Third Operation and Postoperative Course.

After a repeated abdominal CT scan demonstrated interval resolution of the fluid collection and repeated CSF cultures showed no evidence of infection, a diversion procedure was undertaken. Intraoperative examination of the open peritoneal cavity revealed no fluid collection or other abnormality, and a VP shunt was placed 27 days after the patient’s initial surgery.

The patient was discharged home after an uneventful postoperative course. At the most recent follow-up examination, 6 months after surgery, she showed no sign of infection.

Discussion

The concept of Gram-positive bacteria entry into a shunt system is easy to grasp, because the vast majority of these shunt infections are due to otherwise commensal skin flora and are temporally related to recent surgery.3,17,18,23,26 Gram-negative infections that occur in the immediate postoperative period are presumed to be caused by bacteria introduced at the time of surgery (assuming colonization or contamination of the skin), whereas later-onset infections are typically related to clinically or radiographically evident bowel perforation.21,22 The former and the latter scenarios have plausible mechanisms for entry and subsequent infection. These variants of Gram-negative infection typically manifest with outright clinical signs of perforation or the subtle radiographic finding of a pseudocyst.2,6,21,22 The delayed entry of Gram-negative bacteria into the shunt system in the absence of bowel perforation or a pseudocyst is more difficult to understand. It was this ill-defined entry of S. marcescens and P. mirabilis that prompted us to search exhaustively for an abdominal disease process and consider an alternative site for the distal catheter in the patient in the present case.

In their review of Gram-negative bacterial infections of VP shunts, Stamos and colleagues24 acknowledge that the entry of these pathogens into the shunt system remains poorly understood. In the group of patients in their series who presented with a Gram-negative infection at least 3 months after surgery, these authors concluded that a “transient perforation” may have led to those infections for which a mechanism of entry was not identified. Gaskill and Marlin9 further developed the concept of transient perforation in patients with shunts and reported a series of patients who met criteria for SBP-associated shunt infections.

Typically seen in patients with cirrhotic ascites, SBP is an infection of the peritoneal fluid in the absence of an abdominal lesion.8,27 The organisms most frequently cultured in these cases are commonly found in the intestinal flora. Aerobic Gram-negative organisms, primarily Escherichia coli and Klebsiella pneumoniae, account for the majority of infections, while aerobic Gram-positive organisms (Streptococcus, Enterococcus, and Staphylococcus species) account for less than a third. Of note, despite the polymicrobial nature of the gastrointestinal tract, 92% of cases of infection involve only one organism, while the remaining 8% are polymicrobial.8 Gaskill and Marlin9 report their experience with seven cases of SBP in patients with VP shunts. These authors identified criteria for diagnosing SBP in such patients, including: a remote history of shunt revision, cultures consistent with normal intestinal flora, and no recent abdominal surgery, gastrostomy, or passage of wire-impregnated catheters. Our patient met all these criteria, but, interestingly, a species normally found in the intestinal flora (P. mirabilis) was identified in the CSF culture despite the fact that the Gram stain and culture from the peritoneal catheter were negative for the presence of any bacteria. The two cases of S. marcescens infections of a VP shunt that have been previously reported are discussed below. Both of these cases involved abdominal disease processes, which suggests a peritoneal source for this rare pathogen’s entry into the VP shunt system. The fact that 2 weeks before the patient presented to us she experienced vague abdominal symptoms for 2 days suggests that our patient may have suffered from an episode of SBP, which might have allowed P. mirabilis and S. marcescens to ascend the peritoneal catheter and cause obstructive hydrocephalus and eventually ventriculitis.

It has been postulated that SBP results from the failure of the gut to contain the bacterial flora and the failure of the immune system to kill the virulent bacteria once they are in the peritoneum. Whereas patients with advanced cirrhosis are vulnerable to peritoneal fluid infection by translocation of bacteria from their own gut, patients with VP shunts should generally have intact protective mechanisms.27 This may explain the absence of peritoneal signs, the negative culture results, and nondiagnostic abdominal evaluation in our patient, despite the presence of intestinal bacteria and S. marcescens in her CSF. Gaskill and Marlin9 reported similar findings in three patients in their series of patients with SBP and VP shunts. Two patients presented without peritoneal signs, and in one patient there was no evidence of infection in the ascitic fluid, similar to the findings in our patient.

As there is no other plausible alternative source for the pathogens in this case, we conclude that SBP was the cause of our patient’s shunt infection. To our knowledge, polymicrobial SBP leading to ventriculitis has not been previously reported in a patient with a VP shunt.

After much deliberation, we decided to place the distal end of the catheter into the peritoneal cavity again, instead of placing it into the pleural or vascular space. Various authors3,21 have recommended using an alternative site after an abdominal infection. Several factors went into our decision to continue the use of the peritoneal cavity as the CSF diversion site in this patient: the interval resolution of the abdominal fluid collection, the results of the general surgical consultation, the normal findings on physical examination of the patient’s abdomen, and the multiple failed attempts to access the fluid collection. We concluded that the absence of clinical and radiographic findings in the abdomen left the peritoneal cavity an acceptable location for CSF diversion.

Our patient’s polymicrobial infection involved a rare combination of pathogens. Serratia marcescens is an opportunistic Gram-negative bacillus primarily associated with infections of the respiratory tract, urinary tract, and bloodstream in immunocompromised patients.4,11,28 It is an infrequent cause of CNS infection and an even less common cause of VP shunt infection, with only two previous reports in the literature.5,16,25 Given that one of the largest series of S. marcescens infection in neurosurgical patients demonstrated a substantial mortality rate despite antibiotic therapy, its presence in a VP shunt is cause for alarm.12,13,25

The two previous reports of S. marcescens shunt infections are vague in regard to treatment and outcome. Nourisamie and colleagues16 reported the case of a 10-month-old girl who had presented with symptoms of sepsis 2 months after surgery for shunt revision. The CSF specimen obtained from her shunt reservoir grew S. marcescens, and a CT scan of her abdomen demonstrated intraluminal migration of the shunt tubing. These authors reported that the shunt was removed and replaced at a later date, but they did not comment on treatment or outcome.

Bremer and Darouiche5 reported the second case of an S. marcescens infection of a VP shunt. The patient in their case was a 58-year-old woman who had been treated with methotrexate for severe rheumatoid arthritis and had required shunt placement after an aneurysmal subarachnoid hemorrhage. She was found to have a fluid collection in the right upper quadrant of her abdomen, which, in conjunction with evidence of hydrocephalus on an unenhanced head CT scan, prompted CSF evaluation. The shunt reservoir was accessed and S. marcescens was isolated from a specimen. The patient underwent VP shunt removal. Culture of a specimen from the peritoneal catheter also revealed the presence of S. marcescens. The patient was treated with intravenous ciprofloxacin and also underwent percutaneous drainage of the abdominal abscess. The authors gave no further information concerning the duration of treatment or outcome.

On the basis of previous reports, S. marcescens is known to be an alarming pathogen in a neurosurgical patient, and when it is documented to have infected a VP shunt, surgical intervention and aggressive antibiotic therapy are required.1,4,7,12–15,25 Our experience with the case presented in this report further substantiates this view. Despite documented sensitivity to both imipenem and gentamicin, S. marcescens persisted in our patient’s CSF. Even though the results of subsequent sensitivity studies did not show the development of resistance to these antibiotics, the patient’s declining neurological condition, leukocytosis, and persistent fever prompted the addition of intrathecal gentamicin to her antibiotic regimen after replacement of the external ventricular drain. Although intrathecal antibiotic therapy is generally reserved for the treatment of ventriculitis caused by highly resistant organisms, concern about the well-established risk of morbidity and death associated with persistent ventriculitis caused by S. marcescens infection supports the intrathecal administration of gentamicin.10,15,19 Our patient experienced improvement in her neurological condition and resolution of the infection only after the second replacement of the ventricular catheter and the intrathecal administration of antibiotic agents.

Conclusions

To our knowledge, this case represents the first report of a polymicrobial VP shunt infection due to SBP and the third reported case of an S. marcescens–infected VP shunt. Because the collective series of S. marcescens infections of the CNS has entailed significant rates of morbidity and death, the presence of S. marcescens in a patient with a VP shunt requires surgical intervention and aggressive antibiotic therapy. Intrathecal administration of gentamicin appears to have been instrumental in the resolution of our patient’s infection. In light of the increased morbidity associated with S. marcescens infection of the CNS, intrathecal gentamicin should be considered as part of the first-line therapy in a patient with a shunt infection. Despite the combination of a bowel flora pathogen and a rare opportunistic pathogen isolated from the CSF in this case, the peritoneal cavity remains an acceptable site for CSF diversion in the absence of clinical symptoms and radiographic abnormalities.

References

Article Information

Address reprint requests to: Luis M. Tumialán, M.D., Emory University School of Medicine, Department of Neurosurgery, 1365-B Clifton Road, N.E., Suite 6400, Atlanta, Georgia 30322. email: ltumial@emory.edu.

© AANS, except where prohibited by US copyright law.

Headings

Figures

  • View in gallery

    Fig. 1. Oral and intravenous contrast–enhanced CT of the abdomen obtained immediately after removal of the entire VP shunt system. Arrow indicates the fluid collection near the Douglas cul-de-sac. There was no fluid collection beneath the liver where the distal end of the peritoneal catheter had been.

References

1

Ariel IArad ISoffer D: Autopsy findings of Serratia meningoen-cephalitis in infants. Pediatr Pathol 6:3513581986

2

Baird CO’Connor DPittman T: Late shunt infections. Pediatr Neurosurg 31:2692731999

3

Bayston R: Epidemiology, diagnosis, treatment, and prevention of cerebrospinal fluid shunt infections. Neurosurg Clin N Am 12:703708viii2001

4

Berger ARohrmeister KHaiden NAssadian OKretzer VKohlhauser C: Serratia marcescens in the neonatal intensive care unit: re-emphasis of the potentially devastating sequelae. Wien Klin Wochenschr 114:101710222002

5

Bremer AADarouiche RO: Ventriculoperitoneal shunt infection due to Serratia marcescens. J Infect 50:1381412005

6

Byard RKoszyca BQiao M: Unexpected childhood death due to a rare complication of ventriculoperitoneal shunting. Am J Forensic Med Pathol 22:2072102001

7

Curtis CChock SHenderson THolman MJ: A fatal case of necrotizing fasciitis caused by Serratia marcescens. Am Surg 71:2282302005

8

Garcia-Tsao G: Spontaneous bacterial peritonitis. Gastroenterol Clin North Am 21:2572751992

9

Gaskill SJMarlin AE: Spontaneous bacterial peritonitis in patients with ventriculoperitoneal shunts. Pediatr Neurosurg 26:1151191997

10

Gump WWalsh J: Intrathecal colistin for treatment of highly resistant Pseudomonas ventriculitis. Case report and review of the literature. J Neurosurg 102:9159172005

11

Hejazi AAucken HMFalkiner FR: Epidemiology and susceptibility of serratia marcescens in a large general hospital over an 8-year period. J Hosp Infect 45:42462000

12

Huang CRLu CHChien CCChang WN: Protean infectious types and frequent association with neurosurgical procedures in adult Serratia marcescens CNS infections: report of two cases and review of the literature. Clin Neurol Neurosurg 103:1711742001

13

Jablokow VRKathuria S: Fatal meningitis due to Serratia marcescens after stapedectomy. Arch Otolaryngol 108:34351982

14

Johnson JSCroall JPower JSArmstrong GR: Fatal Serratia marcescens meningitis and myocarditis in a patient with an indwelling urinary catheter. J Clin Pathol 51:7897901998

15

Kourtopoulos HHolm SE: Intraventricular treatment of Serratia marcescens meningitis with gentamicin. Pharmacokinetic studies of gentamicin concentration in one case. Scand J Infect Dis 8:57601976

16

Nourisamie KVyas PSwanson KF: Two unusual complications of ventriculoperitoneal shunts in the same infant. Pediatr Radiol 31:8148162001

17

Piatt JH JrCarlson CV: A search for determinants of cerebrospinal fluid shunt survival: retrospective analysis of a 14-year institutional experience. Pediatr Neurosurg 19:2332421993

18

Pople IKBayston RHayward RD: Infection of cerebrospinal fluid shunts in infants: a study of etiological factors. J Neurosurg 77:29361992

19

Quinn ALParada JPBelmares JO’Keefe JP: Intrathecal Colistin and sterilization of resistant Pseudomonas aeruginosa shunt infection. Ann Pharmacother 39:9499522005

20

Sabbuba NAMahenthiralingam EStickler DJ: Molecular epidemiology of Proteus mirabilis infections of the catheterized urinary tract. J Clin Microbiol 41:496149652003

21

Salomao JFLeibinger RD: Abdominal pseudocysts complicating CSF shunting in infants and children. Report of 18 cases. Pediatr Neurosurg 31:2742781999

22

Sathyanarayana SWylen ELBaskaya MKNanda A: Spontaneous bowel perforation after ventriculoperitoneal shunt surgery: case report and a review of 45 cases. Surg Neurol 54:3883962000

23

Schreffler RTSchreffler AJWittler RR: Treatment of cerebro-spinal fluid shunt infections: a decision analysis. Pediatr Infect Dis J 21:6326362002

24

Stamos JKKaufman BAYogev R: Ventriculoperitoneal shunt infections with gram-negative bacteria. Neurosurgery 33:8588621993

25

Theccanat GHirschfield LIsenberg H: Serratia marcescens meningitis. J Clin Microbiol 29:8228231991

26

Whitehead WEKestle JR: The treatment of cerebrospinal fluid shunt infections. Results from a practice survey of the American Society of Pediatric Neurosurgeons. Pediatr Neurosurg 35:2052102001

27

Yildirim BSari RIsci N: Patients with spontaneous bacterial peritonitis, and malignant and cirrhotic ascites. J Natl Med Assoc 97:2762802005

28

Yu VL: Serratia marcescens: historical perspective and clinical review. N Engl J Med 300:8878931979

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 10 10 10
Full Text Views 67 67 42
PDF Downloads 77 77 56
EPUB Downloads 0 0 0

PubMed

Google Scholar