Bacterial infections in patients with liver cirrhosis: clinical characteristics and the role of C-reactive protein

Introduction

Patients with cirrhosis are at high risk of developing bacterial infections, sepsis and sepsis-related death [1,2]. There is evidence that about one third of cirrhotic patients present with infection at hospital admission or develop infection during hospitalization [3]. However, the diagnosis of bacterial infection in cirrhotics is often difficult, since these patients usually present with signs of unspecific clinical deterioration and not the classical clinical characteristics of systemic inflammation, such as fever and raised white blood cell (WBC) count. The presence of portal hypertension or hepatic encephalopathy and the effect of β-blockers may modify the clinical picture and the biochemical parameters of systemic inflammatory response syndrome (SIRS). Therefore the diagnosis of bacterial infection and sepsis in patients with cirrhosis remains challenging [4]. On the other hand, one may argue that the detrimental effects of infections in patients with cirrhosis could be, at least in part, the result of delay due to difficulties in the diagnosis, since the prompt initiation of antimicrobial treatment significantly improves the prognosis and clearly reduces mortality rates [5].

The role of C-reactive protein (CRP), a well-known biomarker of inflammation in the general population, has not been clearly defined in patients with advanced liver disease. CRP is an acute-phase reactant that is mainly produced in the liver following stimulation by interleukin (IL) -1 and -6, with increased expression within the first hours after the triggering insult. CRP seems to play a key role in the inflammatory process and provides a link between the innate and adaptive immune systems. In everyday clinical practice, CRP is widely used as an indicator of SIRS, with levels greater than 50 mg/L being associated with sepsis [6]. Several studies in the general population have shown that CRP levels represent a useful marker that not only helps diagnose infection early, but also can predict survival [7].

CRP production might be reduced in patients with impaired liver function and decreased protein synthesis. In contrast, it was shown that patients with cirrhosis without infection had higher levels of IL-6 and increased expression of tumor necrosis factor receptors [8,9]. Therefore, assessing the appropriate CRP levels that would actually help diagnose infection in patients with advanced liver disease remains difficult and there is, to date, a shortage of data dealing with this special issue.

Our aim was to evaluate the role of CRP as a diagnostic tool for bacterial infection in a group of hospitalized patients with cirrhosis. We focused on the value of a single determination of CRP, at the time of admission or on the day after, in identifying patients with infection.

Patients and methods

Results

Two hundred ten consecutive hospitalized patients with liver cirrhosis of different etiologies were included in the analysis. Patient characteristics in the three study groups are shown in Table 1. Viral hepatitis was the cause of chronic liver disease in 87 (41.4%) patients, alcohol consumption in 83 (39.5%), nonalcoholic fatty liver disease in 4 (1.9%), autoimmune liver diseases (autoimmune hepatitis, primary biliary cirrhosis) in 6 (3%) patients, while 30 (14.3%) had cryptogenic cirrhosis.

Table 1 Comparison of the characteristics of the patients included in the three study groups

One hundred patients (71% male, age 63±14.6 years) had documented bacterial infection at admission or during the first 3 days of admission (Group 1) and 110 (78% male, age 61.4±12.7 years) were hospitalized for other reasons (Group 2). The control group (Group 3) included 106 patients (51% male, age 71.4±16 years) who had proven bacterial infections without history or signs of chronic liver disease.

The most common infections in both groups of patients (cirrhotics and non-cirrhotics) were infections of the urogenital tract (pyelonephritis, prostatitis, pelvic inflammatory disease) in 66/206 (32%), respiratory system (pneumonia, bronchopneumonia) in 41/206 (20%), or gastrointestinal tract (cholecystitis, cholangitis, gastroenteritis) in 11/206 (5.3%). Among the cirrhotic patients 45/100 (45%) presented with clinical and laboratory signs of spontaneous bacterial peritonitis. A total of 62/206 (30%) of the patients with infection (20% of the cirrhotics and 30% of the non-cirrhotics) had positive blood cultures. Among the cirrhotics, 66% of the patients had Gram-negative and 33% Gram-positive bacteremia.

Laboratory data

WBC count

There was a significant difference in mean WBC count among the three study groups: 6.92 (range: 4.60-9.38) × 103/mm3 in patients with cirrhosis and bacterial infection vs. 5.75 (range: 4.07-7.70) × 103/mm3 (P=0.02) in cirrhotic patients without infection and 11.28 (range: 8.40-15.20) × 103/mm3 in non-cirrhotic patients with infection (P<0.001). Patients with alcoholic cirrhosis had a higher mean WBC count even without infection 7.23±2.8 × 103/mm3 and no statistically significant difference was found when compared with alcoholic cirrhotics with infection 8.47±4.58 × 103/mm3 (P=0.13).

CRP

The mean CRP value in patients with cirrhosis and infection was 34±38.8 mg/L, compared to 147±95 mg/L in non-cirrhotic patients with infection (P<0.001) and 5.7±7.3 mg/L in the group of patients who had cirrhosis without infection (P<0.001, Table 1). Moreover, a statistically significant number, 19/100 (19%), of the cirrhotic patients had normal CRP levels (<5 mg/L), despite documented bacterial infection, compared with 2/106 (1.9%) of the non-cirrhotic patients (P<0.001). When patients who presented with BT>38°C were taken into account, the difference between the CRP levels among cirrhotic and non-cirrhotic patients became non-significant (P=0.29).

Regarding the etiology of cirrhosis, patients with bacterial infections and autoimmune liver disease presented with a lower mean CRP level (1.75±0.07 mg/L) compared to patients with cirrhosis due to viral (6.3±1.3 mg/L) or alcoholic (5.35±0.9 mg/L) etiology (P=0.26). Among the cirrhotic patients who had no signs of infection, 33.3% (all with alcoholic liver disease) presented with CRP>5 mg/L (max 40 mg/L). In the univariate analysis, serum CRP concentrations, WBC and MELD score were significantly associated with bacterial infection in cirrhotics (Table 2). Multivariate analysis, however, showed that only CRP levels and MELD score were associated significantly with infection (Table 3).

Table 2 Univariate analysis: CRP levels, WBC count and MELD score were significantly associated with bacterial infection in cirrhotic patients

Table 3 Multivariate analysis: CRP levels and MELD score were significantly associated with the presence of bacterial infection in cirrhotic patients

In cirrhotic patients with the same MELD score, the probability of infection increased by 8% for each one-unit increase in CRP concentrations. Likewise, for each one-unit increase in MELD score the probability of infection increased by 40% in patients with the same CRP value.

Using ROC curves, the probability of infection in cirrhotics was estimated according to the CRP levels. Specifically, a cutoff CRP level >10 mg/L indicated the presence of infection with a sensitivity of 68%, a specificity of 84.55% and an area under the ROC curve (AUROC) of 0.8197.

The best cutoff level for CRP according to the severity of the liver disease, determined by the CPS, showed a slight variation, with lower cutoff levels in more advanced liver disease. In patients with advanced liver disease (CPS-C) a CRP cutoff level of 5.78 was found to be significant in identifying patients who might have bacterial infection (Table 4, Fig. 1).

Table 4 Cutoff levels for CRP according to the severity of the liver disease

Figure 1 ROC curve representing the diagnostic value of CRP according to the three different disease stages as represented by the CPS CRP, C-reactive protein; ROC, receiver operating characteristic; CPS, Child-Pugh score

Discussion

Bacterial infections in patients with cirrhosis represent a major cause of hospitalization and a common reason for clinical deterioration or death [13-15]. Immediate treatment initiation is required, but clear clinical and laboratory data supporting the diagnosis are lacking. Our study aimed to evaluate the characteristics of these patients to set a clinical context that could help improve clinicians’ skills in detecting early bacterial infection in patients with liver function impairment.

We performed a retrospective analysis of prospectively collected data from 210 consecutive hospitalized patients with cirrhosis. We compared clinical and laboratory parameters between cirrhotics admitted to the hospital with and without infection, and we also included a control group of 106 patients who had evidence of infection without chronic liver disease.

Regarding the clinical data, several reviews underline the lack of typical signs of infection and inflammatory response in patients with advanced liver disease; however, they are based on a limited number of studies, most of which were retrospective and published some decades ago [16]. In a more recent study [1], the prevalence, incidence and clinical relevance of bacterial infection were documented for a total of 405 admissions (361 cirrhotic patients). It was underlined that infections are common and may also occur during hospitalization for other reasons. Interestingly, 46% of the patients with infections were asymptomatic.

Our data confirms these previous findings, showing that the typical clinical symptoms or signs of bacterial infection were present in less than half of our cirrhotic patients. More specifically, despite bacterial infection, only one third of our patients had fever >38°C at admission, while most of them had WBC within normal ranges. This was eventually attributed to a “relative” increase in WBC count to normal values in patients with known leucopenia due to the presence of hypersplenism. In this context, patients with alcoholic liver disease form a special group. Although we did not include patients with acute alcoholic hepatitis, patients with alcoholic cirrhosis also present with relative higher WBC due to neutrophil priming [17], even in the case of hypersplenism. Indeed, we found no statistically significant difference in WBC between alcoholic cirrhotics with or without infections, indicating that WBC are unable to respond to further bacterial challenge in these patients. However, regardless of the WBC, it is well known that alcoholic patients present with defective neutrophil function and are especially prone to develop infections [18]; thus, other biomarkers are needed in these patients.

It is noteworthy that the severity of liver disease was found to inversely correlate with the presence of typical clinical presentation; patients with infection and fever had significantly lower MELD scores compared with those without fever: the more severe the liver disease, the less typical the clinical symptoms and signs. Therefore, in everyday clinical practice, high suspicion is needed when dealing with patients who present with signs of unexplained worsening of their liver function (increase in MELD score), despite the absence of fever or leukocytosis.

How can the quantitative assessment of CRP be helpful in this context? CRP measurement is a cheap and practical test, widely used as a biomarker of infection but also noninfectious inflammation in many disease entities [19]. However, it has been proven quite nonspecific, having an established diagnostic and prognostic value especially in patients with sepsis.

Indeed, the role of CRP in diagnosing bacterial infections in patients with impaired liver function is in conflict with studies debating whether it really represents a reliable marker [20]. Other studies have correlated CRP as a factor independently associated with mortality in patients with liver disease [21]. The different design and methodology of these studies might be an explanation for the contradictory results.

In our cohort, there was a significant difference between CRP values at presentation between cirrhotic and non-cirrhotic patients with infections, with cirrhotics having lower values even in severe cases of bacterial infection with bacteremia; 19/100 patients with cirrhosis and infection had CRP values within normal values. These results may lead to the apparent conclusion that CRP cannot represent a reliable marker of early diagnosis of infection in patients with cirrhosis.

But what about values of CRP that are “near normal”, and what does “normal” mean in patients who have impaired protein synthesis? In a further analysis, we tried to define a useful “low” cutoff level for CRP, through the generation of ROC curves. A level of 10 mg/L was found to indicate infection with a sensitivity of 68%, a specificity of 84.55% and an AUROC of 0.8197. This cutoff level is comparable with the results of one study from Hungary [22] (9.2 mg/L AUC 0.93) but is much lower than those in other reports (6.5-80 mg/L) [23,24]. The Hungarian study, like ours, involved a large number of patients from everyday clinical practice, while the other studies included smaller cohorts of patients with severe infectious episodes (patients in the intensive care unit or with sepsis).

Furthermore, analyzing the diagnostic accuracy of CRP levels according to the severity of the liver disease, we found an indirect proportional relationship between the CRP cutoff level and CPS: the higher the CPS, the lower the diagnostic CRP cutoff level (CPS-A: 21.3 mg/L, CPS-B: 17 mg/L, CPS-C: 5.8 mg/L). This finding has important clinical implications, showing that in patients with advanced liver disease the smallest CRP elevation (5.8 mg/L with normal value <5 mg/L) could be a sign of bacterial infection requiring antibiotic initiation.

Other studies also reported that for more advanced liver disease the diagnostic capacity of CRP was lower for the same CRP level, with AUROC values decreasing in direct relation to the severity of the liver disease [15]. In this context, and for practical reasons, we took a step forward in suggesting a possible cutoff level for each CPS stage.

Another biomarker that was found to correlate with bacterial infections was procalcitonin [25]. When studied in patients with liver disease and compared with CRP an acceptable accuracy has been reported [26]. However, procalcitonin is not an easy tool for routine use in everyday clinical practice.

In conclusion, patients with cirrhosis and bacterial infections usually present with a nonspecific acute phase response, making classical findings such as fever and high WBC of little help in diagnosis. On the other hand, CRP with a cutoff value of 10 mg/dL may be a useful clinical marker of infection. Moreover, CRP cutoff levels were found to differ according to the severity of liver disease. Although these findings could be helpful in everyday practice, the decision to start antibiotics must be individualized and not be based only on a certain CRP level, but on the right clinical suspicion, the stability of the patient’s clinical condition, the presence of comorbidity and the risk of death or complications, issues that might still remain difficult for the treating physician.