Spontaneous bacterial peritonitis—diagnosis, treatment and prevention
Dr A. J. Stanley, c/o Ward 9, Glasgow Royal Infirmary, 64 Castle St., Glasgow G4 0SF, UK. E-mail: firstname.lastname@example.org
Spontaneous bacterial peritonitis is a serious complication of cirrhotic ascites, arising most frequently in those with advanced liver disease. Its development leads to a further reduction in the effective arterial blood volume, and it has a mortality rate equivalent to that of a variceal bleed. However, problems remain with regard to the identification and optimal treatment of spontaneous bacterial peritonitis. Several important studies and consensus documents on the condition have recently been published which aid in the identification of patients at risk and help to guide therapy.
In this review, we discuss these publications and address the issues of diagnosis, treatment and both primary and secondary prophylaxis of spontaneous bacterial peritonitis in the light of recent data.
Spontaneous bacterial peritonitis (SBP) is a bacterial infection of ascitic fluid which arises in the absence of any other source of sepsis within the peritoneum or adjacent tissues. Although described relatively recently, it is now recognized as a common and severe complication in patients with cirrhotic ascites. It often develops insidiously, and may remain unrecognized; yet, the mortality rate of a single episode is similar to that of a variceal bleed at 20–40%.1 Thus the survival of a patient with SBP depends on an aggressive approach to diagnosis and treatment and measures should be taken to prevent its occurrence.
Ascites is thought to arise as a result of the marked circulatory and renal abnormalities that are associated with cirrhosis,2–4 and patients who develop this complication have a 2-year survival of approximately 50%.5 The physiological changes leading to its formation have been encompassed in the peripheral vasodilatation hypothesis of Schrier et al.6 This proposes that initial arterial vasodilatation leads to a reduced effective arterial blood volume and subsequent activation of mediators promoting sodium and water retention.7 These include the renin–angiotensin–aldosterone system, the sympathetic nervous system and anti-diuretic hormone. In addition, the renal circulation appears particularly sensitive to angiotensin-II-mediated vasoconstriction,8 which may lead to reduced renal perfusion and glomerular filtration rate.
Thus patients with cirrhosis and ascites exhibit a precarious haemodynamic imbalance. If they are exposed to an additional insult, such as a gastrointestinal bleed, nephrotoxic drugs (e.g. NSAIDs, diuretics, aminoglycosides) or systemic infection, they are at risk of developing renal impairment and the hepatorenal syndrome.
The prevalence of SBP in cirrhotic patients with ascites admitted to hospital has been estimated at 10–30%.9–12 The risk of developing SBP is greater in those with a coexisting gastrointestinal bleed, a previous episode of SBP or low ascitic protein levels (as these patients have lower fluid complement levels and opsonin activity).13–18 Although SBP is usually considered to occur only in patients with cirrhotic ascites, a recent paper from India reported the occurrence of SBP in 18% of patients with ascites secondary to fulminant hepatic failure.19 The prevalence of ascites in this series of patients with fulminant hepatic failure was 29%.
Several important studies and reviews on SBP have been published in recent years. Runyon’s review formed the basis of the American Association for the Study of Liver Diseases (AASLD) Practice Guidelines on SBP.20 Following the publication of further relevant work, the International Ascites Club recently published a consensus document which reiterated many of the AASLD guidelines and added further recommendations.21 Ongoing research in this area has led to the publication of further data on SBP. In the preparation of this review, a search was made using OVID (including MEDLINE and EMBASE) to identify papers published on this subject in the past 5 years, which were then evaluated. We summarize and discuss the recent data and guidelines on SBP below.
Gram-negative aerobic bacteria from the family of Enterobacteriaceae and non-enterococcal Streptococcus spp. are the most common organisms isolated from ascites.9, 10, 22 The three most common isolates are Escherichia coli, Klebsiella pneumoniae and the pneumococci.23 It is proposed that these enteric organisms cross the intestinal mucosal barrier to the mesenteric lymph nodes (a process known as ‘bacterial translocation’) and enter the systemic bloodstream via the thoracic duct.24–26 Animal models have confirmed that bacterial translocation is involved in the pathogenesis of SBP.27, 28 There is no firm evidence that a specific aetiology of liver disease predisposes to SBP or influences the prognosis or response to therapy.
Infection of ascites is facilitated by both the impaired local immune response (poor opsonic capacity of ascitic fluid) and impaired systemic immune response (reduced reticulo-endothelial phagocytic activity).14, 18 Patients may present with fever and abdominal pain, but impaired liver function or renal impairment may be the only findings. Signs of severe peritonitis, such as ileus and shock, are uncommon.29
Although the number of bacteria present in an episode of SBP is very low, they excite an intense inflammatory response. There is a dramatic increase in the concentrations of polymorphonuclear lymphocytes and inflammatory cytokines within the ascitic fluid. Although this is aimed at removing the infection, it may be associated with adverse haemodynamic effects.30, 31 Levels of tumour necrosis factor-α and interleukin-6 increase dramatically within the ascitic fluid and serum of infected individuals. This is thought to lead to a further reduction in effective arterial blood volume, as indicated by a rise in plasma renin activity.32 Recently, SBP has been shown to be associated with significantly higher serum and ascitic fluid levels of chemokines and soluble adhesion molecules, including interleukin-8 and Intercellular Adhesion Molecule-1 (ICAM-1).33 This suggests that they are implicated in the peritoneal infiltration of patients with SBP.
Approximately 30% of patients with SBP will develop renal impairment, and this is the most sensitive predictor of in-hospital mortality.34 At the time of diagnosis of SBP, ascitic interleukin-6 levels, renal failure (blood urea nitrogen > 11 mmol/L or serum creatinine > 133 μmol/L) and mean arterial pressure are independent predictors of the development of renal impairment.32
When to consider SBP
As SBP may pass unrecognized, the International Ascites Club has recommended diagnostic paracentesis in the following situations: (i) all cirrhotics with ascites on admission to hospital; (ii) in-patients with ascites who develop signs of sepsis, hepatic encephalopathy, renal impairment or altered gastrointestinal motility; (iii) all ascitic patients with a gastrointestinal bleed.
Ascitic fluid polymorphonuclear leucocyte count.
Despite using sensitive methods, culture of ascites is negative in 40% of cases with clinical features of SBP and an elevated ascitic polymorphonuclear leucocyte count.35 Furthermore, initiation of therapy cannot be delayed while awaiting culture results. Therefore, ascitic fluid polymorphonuclear leucocyte count is used as an indirect indicator of the presence of SBP. Patients with an ascitic fluid polymorphonuclear leucocyte count of greater than 250/mm3 (0.25 × 109/L) should be considered to have SBP, and this is an indication to prescribe antibiotics (International Ascites Club recommendations; see Table 1). If the patient has haemorrhagic ascites (i.e. ascitic red blood cells > 10 000/mm3), one polymorphonuclear leucocyte should be subtracted for every 250 red blood cells to derive the ascitic fluid polymorphonuclear leucocyte count. A diagnosis of SBP made on symptoms and signs alone is no longer acceptable.
Variants of ascitic fluid infection (adapted from Rimola et al
and Guarner and Soriano70
Ascitic fluid culture.
Ascitic fluid should be obtained for culture at the same time as the diagnostic tap. Ten millilitres should be inoculated into blood culture bottles (aerobic and anaerobic) at the bedside. This increases the yield by 90%.35 Bacteriascites describes a positive ascitic fluid culture with ascitic polymorphonuclear leucocytes of less than 250/mm3. Variants and recommendations on management are shown in Table 1.
Secondary bacterial peritonitis.
This should be suspected if ascitic culture yields more than one organism (especially anaerobes or fungi), or if there is no response to standard empirical antibiotics. It can arise following abdominal surgery or secondary to inflammation or perforation of an intra-abdominal organ. Secondary bacterial peritonitis is likely if at least two of the following are present in ascitic fluid: glucose < 50 mg/dL; protein > 10 g/L; lactate dehydrogenase > normal serum levels.36 The antibiotic regimen should cover anaerobes and enterococci and investigations should be directed at the cause (Table 1).
Antibiotics (Table 2)
Main treatment regimens for spontaneous bacterial peritonitis and their cost
Following the diagnosis of SBP based on an ascitic fluid polymorphonuclear leucocyte count of greater than 250/mm3, empirical therapy should be commenced. The antibiotic chosen should cover the most likely organisms, achieve antibiotic concentrations within ascitic fluid greater than the minimal inhibitory concentration (MIC90) of these organisms and should not impair renal function.
Cefotaxime is the most widely studied cephalosporin in patients with SBP, and is suitable for empirical therapy for this condition (International Ascites Club recommendations). Prior to 1985, treatment of the condition was suboptimal. With the standard regimen of ampicillin and tobramycin (an aminoglycoside), nephrotoxicity and superinfections were common. A landmark, randomized, comparative study by Felisart et al. then demonstrated that cefotaxime was superior in the treatment of severe infections in patients with cirrhosis.23 In patients with SBP treated with cefotaxime, none developed superinfection or nephrotoxicity, compared to 19% of those given ampicillin–tobramycin.
The dose and duration of therapy were investigated,37 and the AASLD practice guidelines concluded that a regimen of 2 g cefotaxime, eight-hourly for 5 days, was optimum.20 However, another study suggested that 2 g cefotaxime, b.d. for 5 days, is as effective for treating SBP and achieves adequate concentrations of drug within the ascitic fluid.38 Furthermore, cefotaxime is effective in treating SBP which arises in subjects taking oral quinolones as prophylaxis.39 Its spectrum of activity covers the Gram-positive cocci and quinolone-resistant Gram-negative bacilli which are frequently implicated.
Studies have evaluated the efficacy of the new once-daily cephalosporins, such as ceftriaxone, ceftizoxime and ceftazidime, and, although data are limited, they appear to be as effective as cefotaxime.40
Amoxicillin plus clavulinic acid.
In a pilot study 10 years ago, amoxicillin, 1 g, plus clavulinic acid, 200 mg intravenously six-hourly, for 14 days was effective in 85% of cases of SBP.41 This combination was recently compared with cefotaxime.42 Cirrhotic patients suspected of having underlying infection on admission to hospital received either cefotaxime, 1 g intravenously six-hourly, or intravenous amoxicillin–clavulinic acid (1 g/200 mg) eight-hourly then oral amoxicillin–clavulinic acid (500 mg/125 mg) eight-hourly. Both regimens were continued until resolution of infection. The amoxicillin–clavulinic acid combination proved to be as efficacious as cefotaxime with no relevant side-effects. Efficacy was not compromised on switching to oral therapy.
Although intravenous antibiotics have been preferred in the treatment of SBP, recent data suggest that oral quinolones may be equally efficacious. Ofloxacin is a quinolone which (unlike norfloxacin) is well absorbed from the gastrointestinal tract. It has been assessed in a randomized controlled trial of patients with uncomplicated SBP.43 Sixty-four patients received ofloxacin, 400 mg 12-hourly, and 59 patients received intravenous cefotaxime, 2 g six-hourly. Resolution of infection, duration of therapy and survival were similar in each group. Use of this drug is recommended by the International Ascites Club if SBP is uncomplicated and patients have not received prophylactic quinolones.
More recently, a randomized controlled trial has reported that 5 days of oral ciprofloxacin after 2 days of intravenous therapy is as effective as a 1-week intravenous course.44 This provides further evidence for the benefit of oral quinolones. In addition, quinolones are the drug of choice for patients with SBP who are sensitive to beta-lactam antibiotics.
There is now consensus that aminoglycosides have no place in the empirical treatment of SBP due to the unacceptable risk of nephrotoxicity.21
Each of the antibiotic regimens described above achieves resolution of SBP in approximately 90% of cases. As the regimens have similar efficacy, their cost-effectiveness is a major consideration. Table 2 summarizes the main regimens and their costs.
Antibiotics plus albumin
Renal impairment develops in approximately one-third of patients with SBP and is postulated to arise as a result of a further reduction in effective arterial blood volume, mediated by vasoactive cytokines, with a resultant increased renin–angiotensin–aldosterone system activity.31, 33 If this reduction in effective arterial blood volume could be limited, it is possible that the incidence of renal impairment and in-hospital mortality could be reduced.
This issue has recently been addressed by Sort et al.45 and generated great interest.46, 47 In a multicentre randomized study, 126 patients with SBP were assigned to receive either treatment with cefotaxime alone (2 g intravenously six-hourly) or cefotaxime plus intravenous albumin. The albumin was given at a dose of 1.5 g/kg in the first 6 h after diagnosis, followed by a further infusion of 1 g/kg on the third day.
With the standard treatment, renal impairment developed in 33% of patients, whereas with the combination therapy it occurred in only 10%. The in-hospital mortality rates were 28% and 10%, respectively. Although the study was not blind and thus is open to criticism regarding treatment bias, the latter figure represents the lowest reported mortality rate for SBP. The difference in mortality rates remained significant after 3 months of follow-up at 41% and 22%, respectively.
The study reported that those patients who developed renal impairment had evidence of further stimulation of the renin–angiotensin–aldosterone system with a dramatic rise in plasma renin activity. This is in keeping with the hypothesis of renal impairment in SBP. In patients treated with albumin in addition to antibiotics, plasma renin activity was suppressed to levels below baseline. In those given antibiotics alone, plasma renin activity increased and remained significantly raised during the period of observation. For the series of patients overall, independent predictors of development of renal impairment were baseline serum bilirubin > 68 μmol/L, creatinine > 91 μmol/L and treatment with cefotaxime alone.
This suggests that the greatest benefit of additional albumin infusion may be in those with more advanced liver disease or impaired renal function. Identification of such a high-risk subgroup is important when considering the cost implications of prescribing large volumes of albumin. Before recommending this expensive therapy, further studies should be undertaken to confirm these results and, in particular, to determine whether smaller volumes of albumin or cheaper plasma expanders would suffice.
Assessing treatment response
The resolution of SBP is commonly associated with a rapid improvement in the patient’s general condition. If there is no such rapid improvement, a follow-up paracentesis is recommended 48 h after commencement of antibiotics.48 A fall in the ascitic polymorphonuclear lymphocyte count of > 25% suggests that the choice of antibiotic is appropriate (International Ascites Club recommendations). If there is no fall in the polymorphonuclear lymphocyte count, alternative antibiotics should be given, either empirically or according to available sensitivities, and the possibility of secondary bacterial peritonitis should be reconsidered.
Consider referral for liver transplantation
If a patient survives an episode of SBP, the prognosis remains poor. The 1-year and 2-year survival rates following a first episode of SBP are estimated to be 30–50% and 25–30%, respectively, and can be attributed to the severity of the underlying liver disease.17, 49, 50 Survival after liver transplantation is higher; therefore, it is generally recommended that such patients should be considered for this procedure.
PREVENTION OF SBP
In view of the high mortality following an episode of SBP, prevention is clearly important. Unfortunately, most studies of prophylactic antibiotics in this condition have examined heterogeneous patient populations with both primary and secondary prophylaxis being assessed within the same study. Interestingly, recent animal studies have suggested that bacterial overgrowth and translocation in cirrhosis can be decreased using propranolol or cisapride to increase intestinal transit.51, 52 Studies on cirrhotic patients are awaited to evaluate this further.
Prevention of recurrent SBP
Patients who have survived an episode of SBP have a 40–70% 1-year probability of a further episode.13, 17 One randomized placebo-controlled trial has examined the efficacy of antibiotics purely for secondary prophylaxis of SBP.53 Long-term norfloxacin, 400 mg/day, reduced the recurrence of SBP at 1 year from 68% to 20%, with a particular reduction in SBP secondary to Gram-negative bacilli. The study was not designed to assess survival. On the basis of these results, long-term oral norfloxacin is advised for all patients recovering from an episode of SBP until resolution of ascites, transplantation or death (International Ascites Club recommendations).48
Specific subgroups of cirrhotic patients have been demonstrated to be at high risk of a first episode of SBP, namely those with low ascitic fluid total protein (< 10 g/L) and those hospitalized with a gastrointestinal haemorrhage.
Low ascitic fluid total protein.
Ascitic fluid total protein has been shown to be an independent predictor of SBP. Runyon prospectively studied cirrhotic patients during hospitalization and found that 15% of patients with ascitic protein ≤ 10 g/L developed SBP compared to 2% of those with ascitic protein > 10 g/L.15 Longer term prospective studies confirmed that patients with ascitic protein < 10 g/L had a 1-year probability of developing SBP of 20–43%.18, 54 The incidence was greatest in those with Child C liver disease and in those who did not receive short-term prophylaxis if admitted with a gastrointestinal bleed. In those patients with ascitic fluid protein > 10 g/L, the incidence of SBP with up to 3 years of follow-up was negligible. Therefore, primary prophylaxis is inappropriate in this group.
Studies have examined the role of primary antibiotic prophylaxis in patients with low ascitic total protein. Novella et al. randomized patients to receive antibiotics either continuously or only during hospitalization.55 The 1-year incidence of SBP was 9% and 33%, respectively, in favour of continuous antibiotic prophylaxis, but this study had no control arm. In the light of these findings, the AASLD guidelines suggest that patients with ascitic protein < 10 g/L ought to receive prophylactic antibiotics during hospitalization.20
Grange et al. subsequently published the only placebo-controlled trial of primary prophylaxis of SBP in patients with ascitic protein < 15 g/L, and the results were less impressive.56 Patients were prescribed either placebo or norfloxacin, 400 mg daily. Over a 6-month follow-up, SBP developed in only five patients (9%) in the placebo arm and none in the treatment arm. Prophylactic antibiotics were shown to be of benefit only when patients with SBP were grouped with those who developed an isolated bacteraemia. On this basis, the International Ascites Club was unable to reach a consensus on the need for prophylactic antibiotics for patients with low ascitic protein levels in the absence of previous SBP.21
However, a recent study has suggested that selected patients with low ascitic total protein may benefit from primary prophylaxis.57 One hundred and nine patients with ascitic total protein ≤ 10 g/L were followed prospectively for a mean of 45 weeks. SBP developed in 26%. Subgroup analysis revealed that patients with a high bilirubin (> 54.7 μmol/L) and low platelets (< 98 000/mm3) had a higher 1-year risk of SBP and a lower 1-year survival. However, further studies are required to confirm these findings.
Cirrhotic patients with gastrointestinal haemorrhage.
All cirrhotic patients who develop an upper gastrointestinal bleed are at risk of a variety of bacterial infections, including SBP, within the first few days following the bleed. Bacteria of enteric origin are most commonly implicated58 and the development of infection is associated with a poor prognosis.59, 60
A recent meta-analysis of antibiotic prophylaxis in cirrhotic patients with a gastrointestinal bleed reported a reduction in SBP for patients given prophylactic treatment,13 although only one of the four individual trials showed a significant reduction in SBP.61–64 The antibiotics used included norfloxacin, ofloxacin and ciprofloxacin with or without amoxicillin–clavulinic acid. The benefit was greatest in those patients with more advanced liver disease. Importantly, the meta-analysis reported a significant improvement in short-term survival for any cirrhotic patient with a gastrointestinal bleed given prophylactic antibiotics. Therefore, antibiotic prophylaxis should be administered to all cirrhotic patients with a gastrointestinal bleed whether ascites is present or not. Oral norfloxacin, 400 mg b.d. for at least 7 days, was recommended by the International Ascites Club,21 and oral ciprofloxacin, 500 mg b.d. for 7 days, by the recent British Society of Gastroenterology (BSG) guidelines.65
Drug resistance and costs
There are obvious concerns regarding the long-term use of antibiotics in such patients. These include the emergence of antibiotic-resistant bacteria, particularly within hospitals, and the costs of therapy. There have been several recent reports of quinolone-resistant bacteria in cirrhotic patients receiving prolonged norfloxacin prophylaxis.66, 67 This suggests that quinolone prophylaxis should be used with caution, particularly in areas with a high prevalence of quinolone-resistant bacteria, and prophylaxis should be limited to those at greatest risk of SBP.
Two studies have assessed the economics of long-term antibiotic prophylaxis in patients with cirrhotic ascites.68, 69 They reported a cost saving when high-risk patients were given prophylactic therapy, compared with the treatment of confirmed cases only. However, the decision to use prophylactic antibiotics for patients with cirrhotic ascites must be made on the basis of the rates of local bacterial antibiotic resistance and costs.
Spontaneous bacterial peritonitis remains a serious complication of cirrhotic ascites and has a high mortality. Presentation is often insidious, and therefore clinicians must have a high index of suspicion and a low threshold for a diagnostic paracentesis. If SBP is confirmed, antibiotic treatment should be immediately instituted, and concomitant albumin infusion may improve the outcome. The prognosis is poor in those who survive, and liver transplantation should be considered in suitable patients. Long-term secondary prophylaxis with norfloxacin is required for all patients with proven SBP. Primary prophylaxis is recommended for patients presenting with a gastrointestinal bleed and may be helpful in those with advanced liver disease and low ascitic protein levels.