Vegetative infection of transjugular intrahepatic portosystemic shunts

GASTROENTEROLOGY 1998;115:110–115

LIVER, PANCREAS, AND BILIARY TRACT Vegetative Infection of Transjugular Intrahepatic Portosystemic Shunts ARUN J. SANYAL* and K. RAJENDER REDDY‡ *Division of Gastroenterology, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia; and ‡Department of Hepatology, University of Miami Medical School, Miami, Florida

Background & Aims: The occurrence of and the clinical picture of infection of transjugular intrahepatic portosystemic shunts (TIPS) has not been described previously. We describe the clinical features, associated pathogens, results of treatment of a previously unreported complication of TIPS, and primary infection of TIPS occurring after formation of the neointima. Methods: Patients with TIPS and fever were evaluated to exclude other sources of infection. The diagnosis was based on the occurrence of fever with positive blood cultures and either a thrombus or vegetations on the stent or persistent bacteremia in a patient with a TIPS and no other detectable source of infection despite an extensive search. Results: Eight patients met diagnostic criteria. Two of 8 cases occurred within 10 days of TIPS manipulation despite antibiotic administration before the procedure. The clinical features included fever (8 patients), tender hepatomegaly (5 of 8), hypoxemia (2 of 8), septic pulmonary emboli (1 of 8), septic shock (2 of 8), neutrophilia (5 of 8), and subsequent development of necrotizing fasciitis (1 of 8). Blood cultures were positive in all cases. The organisms included oral and enteric aerobic gramnegative bacteria in 7 of 8 patients and Candida in 1 patient. All 8 responded to administration of antibiotics. Two patients died of myocardial infarction and alcoholic hepatitis, respectively. Conclusions: Infective endotipsitis is an uncommon complication of TIPS. Recognition of its clinical features will facilitate diagnosis. Most patients responded to antibiotic therapy.

ransjugular intrahepatic portosystemic shunts (TIPS) are a recent addition to the therapeutic armamentarium against portal hypertension.1 This involves creation of an intrahepatic tract between the hepatic and portal veins using angiographic methods via a transjugular route. This tract is then dilated and kept patent by deployment of an expandable metal stent across it. Physiologically, TIPS function like side-to-side portacaval shunts but avoid the risks of general anesthesia and major surgery.2 These considerations and their efficacy in

T

achieving hemostasis in patients with refractory variceal hemorrhage3,4 have led to their popularization and widespread use over the last 5 years. As with any new procedure, increasing usage has led to the recognition of both expected5 as well as relatively unexpected6 complications associated with TIPS. Although infections occurring immediately after the procedure because of introduction of skin contaminants or iatrogenic rupture of bile ducts colonized by bacteria are known,7 this article documents a previously unreported complication of TIPS: primary infection of the TIPS weeks to months after its placement. The clinical spectrum of the condition, the microbes associated with such infections, the response to therapy, and the natural history of the condition are described.

Materials and Methods Patient Population and Methods of Evaluation Between 1993 and 1996, a total of 8 cases of TIPS infection were identified by the authors. Although 2 of 165 and 1 of 50 cases of TIPS infection were identified among TIPS performed at the Medical College of Virginia and University of Miami hospitals, respectively, the other 5 cases involved patients who had TIPS placed elsewhere. Patients with TIPS who presented with fever were evaluated clinically, and blood cultures were obtained from two separate venous sites. In patients in whom a common site of infection was not identified, an extensive evaluation including computerized tomography (CT) scan of the abdomen, endoscopic retrograde cholangiopancreatography (ERCP), transesophageal echocardiography, and Doppler sonography of the TIPS were performed. Angiograms were obtained in patients with suspected TIPS occlusion.

Abbreviations used in this paper: CT, computerized tomography; ERCP, endoscopic retrograde cholangiopancreatography; TIPS, transjugular intrahepatic portosystemic shunt(s). r 1998 by the American Gastroenterological Association 0016-5085/98/$3.00

July 1998

INFECTION OF TRANSJUGULAR INTRAHEPATIC PORTOSYSTEMIC SHUNTS

Bacteriologic Techniques Five milliliters of venous blood was drawn from each arm, using aseptic precautions, and transferred into separate commercially available blood culture bottles containing 80 mL ESP aerobic medium (Difco Laboratories, Detroit, MI) each. Blood cultures were performed using the ESP semiautomated system (Difco Laboratories).8,9 The methods for species identification varied with the type of bacterial isolate. Streptococci were identified by the API 20 STREP (bioMerieux SA, France)10,11 method, and enterobacteria were identified by the BBL crystal (Becton Dickinson, Meylan, France)12,13 method. Blood cultures were considered to represent clinically significant bacteremia when they were positive from two isolates from separate venipunctures.

111

pressure of ,90 mm Hg in a previously normotensive person and/or a .50 mm Hg decrease in systolic blood pressure in a patient with preexisting hypertension.

Management The initial treatment included a third-generation cephalosporin in 6 of 8 patients. In 1 patient, Timentin (SmithKline Beecham, Philadelphia, PA) was the initial drug used, and in another case penicillin and gentamicin were used as the initial treatment. Once culture results became available, treatment was modified in some cases. Candida sepsis was treated with amphotericin B (0.5 mg · kg21 · day21) for 6 weeks followed by fluconazole orally for 3 months. Streptococcus sanguis was treated with penicillin alone, and S. bovis sepsis was treated with a 6-week course of vancomycin.

Criteria for Diagnosis The criteria for diagnosis were analogous to those used previously for the diagnosis of endocarditis.14–16 These included (1) clinically significant continuous bacteremia (fever and multiple positive blood cultures) along with a vegetation or thrombus in the stent (definite infection); or (2) sustained bacteremia and unremitting fever in a patient with an apparently normal TIPS without an identified site of infection elsewhere in the body (probable infection). Because autopsy material was unavailable, the stents could not be directly studied. Stent thrombosis was defined by near total or total occlusion of stent lumen by a thrombus, and smaller nonocclusive lesions adherent to the stent lining were called vegetations. Septic pulmonary emboli were diagnosed clinically by the presence of fever with multiple positive blood cultures and the appearance of multiple peripheral flameshaped opacities in both lungs compatible with pulmonary emboli. Because tissue samples from the pulmonary lesions were not available, direct bacteriologic examination of these lesions was not possible. Shock was defined as a systolic blood

Results A total of 8 patients with vegetative infection of TIPS were identified between 1993 and 1996. Seven patients were male, and 1 was female (Table 1). The mean age was 45 6 9 years. All patients had either Child–Pugh class B or C cirrhosis.17 Six of 8 patients had underlying alcoholic liver disease, and 1 each had hepatitis C alone and cryptogenic cirrhosis, respectively. Three patients had both hepatitis C and alcoholic liver disease. Diagnosis Six patients met criteria for a diagnosis of definite infection. In each case, multiple blood cultures were positive, indicating the presence of continuous bacteremia (Table 1). A Doppler ultrasound examination of the liver showed the presence of an occlusive thrombus in 4 patients, and vegetation attached to the pseudointimal

Table 1. Demographics, Microbiology, Presentation, Treatment, and Outcome of TIPS Infection No.

Etiology of cirrhosis

Child–Pugh class

1 2

Alcohol Alcohol 1 HCV

A C

S. sanguis E. coli, K. oxytoca

2/2 4/4

3 4

Alcohol Alcohol 1 HCV

B C

E. coli S. aureus

3/4 4/4

5

Cryptogenic

C

S. bovis

2/3

6

Alcohol 1 HCV

C

Acinetobacter calcoaceticus, E. coli, C. albicans

3/4

7

Alcohol

C

K. pneumoniae

5/6

8

HCV

B

E. coli

4/4

Organism(s)

No. positive/ total cultures

Presentation Fever, RUQ pain Fever, RUQ pain, jaundice Fever, jaundice Fevera,b Fever, RUQ pain, jaundice Fever, jaundice

Fever, jaundice, hemoptysis Fever, RUQ pain

Ultrasound findings

Treatment

Outcome

TIPS thrombusa TIPS thrombusa,c

Penicillin 3 4 wk Ceftriaxone 3 4 wk

Alive Dead

TIPS thrombusa,c TIPS patentc

Ceftriaxone 3 4 wk Vancomycin/rifampin 3 6 wk Vancomycin 3 6 wk

Alive Dead

TIPS vegetation

Alive

TIPS patenta,c

Ceftriaxone 3 6 wk Amphotericin (300 mg) followed by fluconazole Ceftriaxone 3 4 wk

TIPS vegetationa,c

Ceftriaxone 3 4 wk

Alive

TIPS thrombusa,c

HCV, hepatitis C virus; RUQ, right upper quadrant. scan did not show evidence of biliary obstruction, gas, or thrombus in portal vein branches or hepatic abscesses. bOccurred within 2 weeks of stent manipulation. cNormal ERCP findings. aCT

Alive

Alive

112

SANYAL AND REDDY

GASTROENTEROLOGY Vol. 115, No. 1

Figure 1. (A) Sonographic evidence of vegetation at the proximal end of a TIPS. (B) The stent extends back into the right atrium, and the vegetation is observed on transesophageal echocardiography at the atrial end of the stent.

lining was identified in 1 patient. Transesophageal echocardiography excluded the possibility of bacterial endocarditis in all cases and showed the presence of vegetation at the atrial end of a stent in a single instance (Figure 1). In 2 patients, no evidence of stent obstruction or vegetation could be identified. However, 1 of these patients had repeated episodes of septic pulmonary emboli, and both had multiple positive blood cultures and no other identifiable source of sepsis. In these patients, a diagnosis of probable shunt infection was made. To be as certain of the diagnosis as possible, an extensive and careful evaluation was performed in each case to look for other potential sources of infection. A CT scan of the abdomen was performed in each instance and did not show any source of infection other than TIPS. There was also no evidence of thrombi or gas in the portal system in any patient. Likewise, a gallium scan was performed in 4 patients and failed to localize the source of sepsis; in these cases nonspecific uptake over the lungs and liver was noted. ERCP examination, performed in 6 patients, did not show evidence of obstruction or calculi in any instance. A colonoscopy showed shallow ulcers in the transverse colon in the patient with S. sanguis bacteremia. In contrast, the patient with S. bovis bacteremia had no abnormalities on colonoscopic examination. Microbiology Although a single organism was identified in 6 cases, 2 patients had polymicrobial sepsis. Gram-negative

aerobes and gram-positive streptococci of oral and enteric origin were the predominant organisms that were identified. However, no single organism stood out as the single most common agent responsible for TIPS infections. One patient had positive blood cultures for both Escherichia coli and Klebsiella. Another patient had positive blood cultures for Acinetobacter, E. coli, and Candida albicans. The microbes were sensitive to a variety of antibiotics including third-generation cephalosporins, ciprofloxacin, and ticarcillin plus clavulanate. The single instance of S. sanguis infection was sensitive to penicillin while the S. bovis identified was sensitive to vancomycin but not penicillin. The Candida species identified was sensitive to both amphotericin and fluconazole. Clinical Features The average duration from the time of TIPS placement to the onset of symptoms was 284 days (range, 110–560 days). Two patients presented within 10 days of follow-up TIPS angiography and dilation of the stent, and 1 patient presented 7 days after colonoscopy. Although a single dose of cefazolin had been administered to those who had undergone angiography, no antibiotics had been administered to the patient who developed symptoms after colonoscopy. Fever was the most common manifestation of TIPS infection and was universally present. The onset was usually insidious, and the temperature peaked 2–3 days after onset. The first episode of fever was associated with shaking chills in 1 patient, and chills occurred in 5 of 8

July 1998

INFECTION OF TRANSJUGULAR INTRAHEPATIC PORTOSYSTEMIC SHUNTS

patients within a week of onset of illness. The fever occurred daily, and the pattern varied from intermittent to continuous. When intermittent, the fever lasted several hours and there were no specific diurnal variation in the fever pattern. Two patients had multiple temperature spikes daily; these were associated with shaking chills and profuse diaphoresis (hectic presentation). These 2 patients became hypotensive during at least one febrile episode, requiring volume resuscitation and administration of vasopressors. The fever was associated with tender hepatomegaly (4 of 8 patients) and worsening conjugated hyperbilirubinemia (7 of 8 patients). The degree of bilirubinemia was variable, reaching values as high as 20 mg/dL. Aspartate aminotransferase and alanine aminotransferase values increased by a mean of 20% (range, 25% to 50%; P 5 0.8 by analysis of variance [ANOVA]). In contrast, the alkaline phosphatase level increased by 20%240% (P 5 0.04 by ANOVA). No evidence of biliary obstruction and cholangitis was found in any case by ERCP. Similarly, neither intrahepatic or perihepatic abscesses or pylephlebitis were noted in any case by ultrasound and CT scan examinations of the abdomen. One patient developed repeated temperature spikes associated with blood-tinged sputum and pleuritic chest pain. Radiological studies in this patient showed multiple peripheral flame-shaped opacities compatible with septic pulmonary emboli. No cases of severe hemodynamically significant pulmonary embolism or paradoxical embolism were noted. A second patient developed necrotizing fasciitis approximately 3 months after the diagnosis was established. This patient had polymicrobial sepsis initially. Cultures obtained after tissue debridement showed Staphylococcus aureus and E. coli. No instances of Osler’s nodes, Janeway’s lesions, or Roth’s spots were observed. Subungual hemorrhages were observed in 2 patients. One patient had septic pulmonary emboli, and another had thrombosed TIPS with polymicrobial sepsis. The fever was accompanied by leukocytosis in 5 of 8 patients (range, 12,000–26,000 cells/mm3). In all five instances, this was caused by neutrophilia associated with a ‘‘shift to the left.’’ In 1 patient, the total white blood cell count was normal but there was a mild monocytosis (12%). Only 1 case was associated with a drop in hemoglobin. This was caused by disseminated intravascular coagulation, as documented by increased fibrin split products, increased fibrin dedimers, and increased prothrombin time, as well as partial thromboplastin time. However, no schistocytes were observed in the peripheral smear in this case.

113

Outcome of Infection Five patients became afebrile within 48–72 hours after starting antibiotics. In the other 3 patients, fever associated with sustained bacteremia continued for 3–6 days, despite empiric broad-spectrum antibiotic coverage, until antibiotics could be changed, based on culture and sensitivity results. After appropriate changes in antibiotics, all patients became afebrile within 48 hours. Resolution of fever was accompanied by improvement in associated symptoms within days. After discontinuation of antibiotics, fever did not recur in any case except in the patient with polymicrobial septicemia, which included Candida species. In this patient, fluconazole was administered orally after completion of amphotericin treatment and continued indefinitely thereafter. The median duration of follow-up from the time of diagnosis for these 8 subjects was 18 months (range, 3 months to 3 years). During this time, 2 of 8 patients died. One patient died of a myocardial infarction, and another died of alcoholic hepatitis. A third patient continues to actively abuse alcohol, and has been readmitted six times in 6 months from the time of diagnosis because of alcoholic hepatitis, uncontrolled ascites, and anasarca and intermittent fever, as well as necrotizing fasciitis in one leg. The rest remain relatively stable and ambulatory. None of these patients have been considered for orthotopic liver transplantation.

Discussion Pyogenic infections are an important cause of morbidity and mortality in patients with cirrhosis.18 Although the spectrum of infections after creation of a portacaval shunt, surgically or by TIPS, is generally similar to the spectrum of infections in patients with cirrhosis,18 this article describes a previously unreported infection, i.e., pyogenic infection associated with TIPS occurring weeks to months after its insertion. The presence of infection was clearly documented in these cases by the presence of multiple positive blood cultures obtained from different sites. The association of the infection with the shunt was established by (1) failure to find an alternate source of infection despite an extensive search, (2) presence of continuous bacteremia indicating a focus in continuity with the venous circulation,19 and (3) presence of a potential nidus of infection in the shunt in the form of a thrombus or vegetation. Although the diagnosis could be made with confidence when all the criteria were met, it remained presumptive in the cases in which a nidus for infection in the stent could not be found.

114

SANYAL AND REDDY

A key aspect of the diagnosis is the exclusion of alternate causes of fever and bacteremia in these patients. Acute cholecystitis, subphrenic abscesses, right-sided pneumonia, and pyelonephritis can be easily excluded on clinical grounds and by routine radiological studies. Although pylephlebitis may present with fever, worsening jaundice, and tender hepatomegaly, it may be distinguished from TIPS infection by the presence of gas and echogenic material within the portal vein and its branches.20,21 Furthermore, evidence of hepatic abscesses, as well as intra-abdominal inflammatory processes such as appendicitis or diverticulitis, is often observed by CT scans of the abdomen in patients with pylephlebitis.21 Similarly, TIPS infection may be distinguished from ascending cholangitis by the absence of evidence of biliary obstruction by sonography and ERCP. Little is known about the pathogenesis of TIPS infections. We speculate that local factors, e.g., thrombus formation or sluggish flow, may allow portal bacteremia to seed the pseudointima associated with the stent and establish itself. If true, one might consider this condition analogous to prosthetic valve endocarditis. Clinically, the lesions of prosthetic valve endocarditis cannot be sterilized and usually require valve replacement.22 However, patients with TIPS infection remained free of fever and bacteremia months after completion of a course of antibiotics. This does not preclude the possibility of ‘‘walled-off’’ infection within the thrombus or pseudointima. It is therefore important to consider the possibility of reactivation of the infection when fever recurs in any patient who has had an episode of TIPS infection. In the absence of data on the true incidence of infections associated with thrombi or vegetations in TIPS, it is difficult to make formal recommendations about antibiotic prophylaxis in patients with TIPS undergoing stent manipulation, endoscopy, or other procedures associated with bacteremia. We administered a dose of cefazolin before stent manipulation because two cases occurred within 2 weeks of stent dilation. However, the cost efficacy of such an approach remains to be proven. This is caused, at least in part, by a lack of information about the risks of developing endotipsitis after various interventions. Thus, the risk of bacteremia after colonoscopy is approximately 10%–20%,23 and the risk of endotipsitis is unknown. We do not routinely use antibiotic prophylaxis before endoscopy in patients with TIPS. However, in a patient with thrombosed TIPS, it would appear prudent to use antibiotic prophylaxis directed against enteric flora during procedures that are likely to induce portal bacteremia, e.g., colonoscopy. Similarly, no general recommendation about the treatment of endotipsitis can be made and the treatment must

GASTROENTEROLOGY Vol. 115, No. 1

be individualized. When stent infection cannot be managed effectively with antibiotics, orthotopic liver transplantation with removal of the infected stent may be considered. In conclusion, infective endotipsitis is an uncommon but serious infective complication of TIPS. It is associated with an infected thrombosed stent or with the development of vegetations on the stent. The microbial spectrum responsible for this condition mostly include enteric organism(s). Although no specific recommendations about prophylaxis can be made, the ‘‘septic syndrome’’ can be effectively treated by antibiotics. Although antibiotics can be discontinued after 6–8 weeks in most cases, patients with fungal infections may require longer treatment. The specific regimen of therapy and its duration must be individualized depending on the organisms involved and the patients’ clinical condition.

References 1. Sanyal AJ, Shiffman ML. Transjugular intrahepatic portosystemic shunt: a medical perspective. Dig Dis 1995;13:153–162. 2. Grace N. The side-to-side portacaval shunt revisited. N Engl J Med 1994; 330:208–209. 3. Sanyal AJ, Freedman AM, Luketic VA, Purdum PP, Shiffman ML, Tisnado J, Cole PE. Transjugular intrahepatic portosystemic shunts for patients with active variceal hemorrhage unresponsive to sclerotherapy. Gastroenterology 1996;111:138–146. 4. Ring EJ, Lake JR, Roberts JP, Gordon RL, LaBerge JM, Read AE, Sterneck MR, Ascher NL. Using transjugular intrahepatic portosystemic shunts to control variceal bleeding before liver transplantation. Ann Intern Med 1992;116:304–309. 5. Sanyal AJ, Freedman AM, Shiffman ML, Purdum PP, Luketic VA, Cheatham AK. Portosystemic encephalopathy after transjugular intrahepatic portosystemic shunt: results of a prospective controlled study. Hepatology 1994; 20:46–55. 6. Sanyal AJ, Freedman AM, Purdum PP, Shiffman ML, Luketic VA. The hematologic consequences of transjugular intrahepatic portosystemic shunts. Hepatology 1996; 23:32–39. 7. Freedman AM, Sanyal AJ. The spectrum of complications following transjugular intrahepatic portasystemic shunts (TIPS). Semin Intervent Radiol 1994;11:161–177. 8. Washington JA II. Blood cultures: principles and new approaches. J Med Technol 1984;1:604–609. 9. Anonymous. Technical information: ESPtm blood culture system. Detroit, MI: Difco, 1994. 10. Tillotson GS. An evaluation of the API 20 strep system. J Clin Pathol 1982;35:468–471. 11. Colman G. The application of computers to the classification of streptococci. J Gen Microbiol 1968;50:149–158. 12. Sanders AC, Faber JE, Cook TM. A rapid method for the characterization of enteric pathogens using paper discs. Appl Microbiol 1957;5:36–40. 13. Kampfer P, Rauhoff O, Dott W. Glycosidase profiles of members of the family enterobacteriaceae. J Clin Microbiol 1991;29:2877– 2879. 14. Maki DG, Agger WA. Enterococcal bacteremia: clinical features, the risk of endocarditis and management. Medicine 1988;67: 248–269.

July 1998

INFECTION OF TRANSJUGULAR INTRAHEPATIC PORTOSYSTEMIC SHUNTS

15. Von Reyn CF, Levy BS, Arbeit RD, Friedland G, Crumpacker CS. Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med 1981;94:504–518. 16. McKinsey DS, Ratts TE, Bisno AL. Underlying cardiac lesions in adults with infective endocarditis. Am J Med 1987;82:681–688. 17. Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni ML, Williams R. Transection of the esophagus for bleeding varices. Br J Surg 1960;60:646–649. 18. Wyke RJ. Bacterial infections complicating liver disease. Baillieres Clin Gastroenterol 1989;3:187–210. 19. Werner AS, Cobbs CG, Kaye D, Hook EW. Studies on the bacteremia of bacterial endocarditis. JAMA 1967;202:199–203. 20. Drabick JJ, Landry FJ. Suppurative pylephlebitis. South Med J 1991;84:1396–1398. 21. Bolt RJ. Diseases of hepatic blood vessels. In: Berk JE, ed. Bockus gastroenterology. 4th ed. Philadelphia: Saunders, 1985: 3259–3277.

115

22. Cowgill D, Addonizio P, Hopeman AR, Harken PH. Prosthetic valve endocarditis. Curr Probl Cardiol 1986;11:623–664. 23. Shorvon PJ, Eykyn SJ, Cotton PB. Gastrointestinal instrumentation, bacteremia and endocarditis. Gut 1983;24:1078–1093.

Received June 20, 1997. Accepted March 10, 1998. Address requests for reprints to: Arun J. Sanyal, M.D., Division of Gastroenterology, MCV Station 980341, Medical College of Virginia, Richmond, Virgnia 23298-0341. Fax: (804) 828-4945. Supported in part by an award from the American College of Gastroenterology and by National Institutes of Health grants DK 53213-01 and DK 00065 to the Clinical Research Center of the Medical College of Virginia. The authors thank Dr. James Arrowood, who performed the transesophageal echocardiography on some of the patients described in this study.