Abstract and Introduction
Abstract
Background Bacterial infections are common cause of morbidity and mortality in patients with cirrhosis. The early diagnosis of these infections is rather difficult.
Aims To assess the accuracy of acute phase proteins in the identification of bacterial infections.
Methods Concentration of C-reactive protein (CRP), procalcitonin (PCT), lipopolysaccharide-binding protein (LBP), sCD14 and antimicrobial antibodies were measured in serum of 368 well-characterized patients with cirrhosis of whom 139 had documented infection. Clinical data was gathered by reviewing the patients' medical charts.
Results Serum levels of CRP, PCT and LBP were significantly higher in patients with clinically overt infections. Among the markers, CRP – using a 10 mg/L cut-off value– proved to be the most accurate in identifying patients with infection (AUC: 0.93). The accuracy of CRP, however, decreased in advanced stage of the disease, most probably because of the significantly elevated CRP levels in non-infected patients. Combination of CRP and PCT increased the sensitivity and negative predictive value, compared with CRP on its own, by 10 and 5% respectively. During a 3-month follow-up period in patients without overt infections, Kaplan–Meier and proportional Cox-regression analyses showed that a CRP value of >10 mg/L (P = 0.035) was independently associated with a shorter duration to progress to clinically significant bacterial infections. There was no correlation between acute phase protein levels and antimicrobial seroreactivity.
Conclusions C-reactive protein on its own is a sensitive screening test for the presence of bacterial infections in cirrhosis and is also a useful marker to predict the likelihood of clinically significant bacterial infections in patients without overt infections.
Introduction
Bacterial infections are a common cause of morbidity in patients with cirrhosis. A wide range of bacterial infections can decompensate hepatic status and lead to death in patients with cirrhosis. Moreover, bacterial infections have been acknowledged as a potential trigger factor in many complications of cirrhosis, including variceal bleeding, hepatic encephalopathy, renal failure and impairment in clotting factors. Infection significantly increases the mortality rate. Various infections are directly responsible for 30–50% of deaths in patients with cirrhosis. About 50% of the infections present atypical clinical appearance and some disease specific characteristics of this patient population make identification of infectious episodes even more difficult. An appropriate clinical approach would include alertness for predictable infections, and its thorough diagnostic work-up.
Elevation in bacteriaemia related serological markers is an early sign of infection and could assist to identify the infectious episodes more efficiently in this particular patient group. However, the majority of these acute phase proteins (APPs) are exclusively synthesized in the liver, and this may blunt the efficacy of these markers in advanced disease. Furthermore, in cirrhosis several infection independent factors, like hepatocellular carcinoma (HCC), necrosis or the on-going local inflammatory reaction of liver tissue and bacterial translocation (BT) are potentially able to induce the synthesis of these markers limiting their clinical utility. Although bacteriaemia related APP levels in cirrhosis have been measured for quite some years, several unanswered questions remain with regards to their utility as reliable clinical marker. In previous clinical studies with small cohorts, C-reactive protein (CRP) and procalcitonin (PCT) proved to be the most reliable markers in the identification of bacterial infection, but their diagnostic accuracy and the cut-off values were highly variable and sometimes unexpectedly high cut-off values were reported.
Although clinical predictors of bacterial infection [advanced disease, presence of gastrointestinal haemorrhage or severe comorbidity are well characterized, data on serological markers in this clinical context are scarce and not uncontradicted. BT, the passage of bacteria or their products from the gut into the circulation, is a major mechanism in the development of bacterial infections in cirrhosis, as such selective intestinal decontamination with oral antibiotics reduces the overall risk of infections, and improves short-term survival in high-risk patients. Furthermore, sustained chronic BT without overt infection also has remarkable impact on the pathogenetic processes of the disease (elevated proinflammatory cytokine levels and decreased vascular resistance). Elevated serum levels of lipopolysaccharide-binding protein (LBP) and serum or ascitic fluid bacterial deoxyribonucleic acid (DNA) in non-infected patients with cirrhosis are suggested to be related to this process. Previously, our group has suggested that the presence of a serological response to various microbial antigens [e.g. phosphopeptidomannan cell-wall component of Saccharomyces cerevisiae (ASCA), glycans or complex gut bacterial protein lysates (OMP Plus; INOVA Diagnostics, San Diego, CA, USA)] may be considered as a universal marker for the enhanced translocation of gut microflora.
The aims of this study were: (i) to define the best cut-off values for different bacteriaemia related APPs to identify bacterial infections in a large cohort of consecutive patients with cirrhosis; (ii) to investigate whether any combination of these markers could significantly improve the diagnostic accuracy; (iii) to evaluate the interactions between the performance of APPs and the various stages of liver diseases, and presence of disease specific complications; (iv) to determine the predictive power of APP values in patients without overt infection with regards to a future bacterial infection.