Article first published online: 8 MAY 2013
Copyright © 2013 American Association for the Study of Liver Diseases
Volume 57, Issue 6, pages 2103–2105, June 2013
How to Cite
Dufour, J.-F. (2013), C-reactive protein, a prognostic marker in hepatocellular carcinoma. Hepatology, 57: 2103–2105. doi: 10.1002/hep.26250
See Article on Pages 2224.
Potential conflict of interest: Advisory committees: Bayer, BMS, Gilead Science, Janssen Cilag, Jennerex, Merck, Novartis, Roche; Speaking and teaching: Bayer, Novartis, Roche.
- Issue published online: 12 JUN 2013
- Article first published online: 8 MAY 2013
- Accepted manuscript online: 12 JAN 2013 07:12AM EST
- Manuscript Accepted: 27 NOV 2012
- Manuscript Received: 3 OCT 2012
C-reactive protein (CRP), named for its capacity to precipitate C-polysaccharide of Streptococcus pneumonia, is a systemic marker of inflammation. CRP activates the complement pathway. It is a pattern recognition receptor with a pentameric polypeptide structure, which binds to a variety of intrinsic and extrinsic ligands. CRP's highest affinity is towards phosphocholine residues. CRP is produced only in hepatocytes under the transcriptional control of interleukin-6 (IL-6).1 Circulating levels of CRP increase hundred-fold in response to infections and inflammation. Determination of CRP levels is one of the most solicited laboratory tests.
The literature documents an increase in CRP levels in cancer patients. High CRP levels may be of prognostic value since they are associated with poor survival. Among gastrointestinal tumors, esophageal,2 gastric,3 colorectal,4 and pancreatic cancers5 have reported this association. Mechanistically, high CRP levels are either a marker of reactive inflammation to a tumor or a marker of an ongoing inflammatory process that favored tumor development. Tumors frequently show histological evidence of intratumoral and/or peritumoral inflammation. Necrotic cells release proinflammatory signals, which attract inflammatory cells from the surrounding tissue. This inflammatory response may foster the tumoral process rather than contain it. Inflammatory cells help to stimulate and to sustain angiogenesis and promote invasiveness by degrading the extracellular matrix. Inflammation is now recognized as an enabling characteristic of tumors.6
CRP has gained prognostic value in hepatocellular carcinoma (HCC). In a cohort of 90 HCC patients, Nagaoka et al.7 found that CRP levels above 3 mg/dL were predictive of poor prognosis and a decreased survival rate. In patients undergoing resection for HCC, preoperative CRP levels correlated with tumor size and vascular invasion and were predictive of early recurrence.8 In the transplantation setting, high preoperative CRP levels were related to tumor recurrence and poor overall survival; in those specific patients with HCC beyond Milan criteria, high CRP levels were an independent predictor of poor outcome.9 Finally, in a prospective evaluation of a cohort of 133 patients with newly diagnosed HCC, Kinoshita et al.10 reported that CRP levels above 1 mg/dL predicted a shorter survival and were characteristic of high Child-Pugh, CLIP, and JIS scores.
In this issue of Hepatology, Peck-Radosavljevic and co-workers11 investigated the value of serum CRP levels in a large cohort of 615 HCC patients who were not amenable to surgery. The hazard ratio of death increased with CRP values up to 2.5 mg/dL, but not beyond. CRP levels above 1 mg/dL were significantly and independently associated with poor survival upon multivariate analysis in the discovery and in the validation cohorts. Patients with a CRP above 1 mg/dL at diagnosis had a survival of 4 months compared to 20 months for patients with a CRP below 1 mg/dL. The initial value of CRP was a significant negative predictor, independent of other clinical parameters (age, Child Pugh stage, tumor characteristics, and treatment). Interestingly, a second CRP determination obtained ∼6 weeks later in the validation cohort, and therefore post-TACE in most cases, retained the predictive value. The mean CRP levels increased with increasing Barcelona Clinic Liver Cancer (BCLC) stage. CRP levels also stratified patients within the same BCLC stage into long- and short-term survivors. Within BCLC stage B and C, patients with elevated CRP had a shorter survival than patients with low CRP. Within the BCLC C stage, Child B patients with a normal CRP had a survival comparable to Child A patients with an elevated CRP (median survival 15 and 14 months, respectively).
Peck-Radosavljevic and co-workers further considered the presence/absence of clinically evident infection. When unrelated to infection, elevated CRP levels correlated directly with tumor characteristics, in particular with tumor burden. This suggests that elevation of CRP might be tumor-related. The authors also compared the proportion of patients with CRP elevation from unknown origin and from infectious origin, in their cohort as well as in a group of 104 non-HCC patients undergoing a transjugular intrahepatic portosystemic shunt (TIPS) placement. Elevated CRP was more frequently of unknown origin in HCC patients than in non-HCC patients (38% versus 17%). Nevertheless, elevated CRP in both groups was associated with poor prognosis. Similarly, Cervoni et al.12 recently reported a benefit of CRP determination in predicting short-term mortality in patients with advanced cirrhosis.
What pathophysiological mechanisms trigger CRP elevation in HCC? Low-grade inflammation promotes tumor development. Mdr2−/− mice show defective biliary secretion of phospholipids, spontaneous cholangitis, and eventually develop HCC.13 Transgenic mice that express lymphotoxin α:β develop a chronic parenchymal inflammation with the production of cytokines and eventually HCC.14 IL-6, which is the principal regulator of CRP production, is produced by Kupffer cells. In the diethylnitrosamine (DEN) rodent model for HCC, IL-6 rises in response to IL-1a, which is released from necrotic hepatocytes.15 This IL-6 production is gender-specific and may partly explain the male predominance of HCC.16 In fact, IL-6 expression is elevated in patients with cirrhosis and HCC.17 Moreover IL-6 levels were reported to correlate with the development of HCC.18, 19 This suggests that the hepatocellular signaling pathway of IL-6 might be a therapeutic target. The transcription factor STAT3 mediates the IL-6 effects. STAT3 was found to be activated in the majority of HCCs and seems to identify more aggressive tumors.20 Inactivation of the negative feedback loop of the JAK/STAT pathway by methylation of the SCOCS genes is frequent in HCC.20 Hepatocyte-specific STAT3-deficient mice exhibited more than a 6-fold reduction in HCC tumor burden relative to wildtype mice in the DEN model.21 Obesity can also be considered a risk factor for liver tumors through activation of IL-6. Lipid accumulation induces a low-grade inflammatory response. Mice fed a high-fat diet have higher circulating levels of IL-6 and develop HCC more frequently than mice fed a low-fat diet.22 Mice lacking IL-6 display an attenuation of tumorigenesis.22 Experimental and clinical evidence show that low-grade hepatic inflammation provides a permissive environment for malignant transformation and proliferation of hepatocytes.
Currently, sophisticated molecular indices are being investigated for diagnostic and predictive value. Hoshida et al.23 describe a transcriptomic signature in tumor surrounding tissue, which predicted survival after HCC resection. This signature contained a gene set associated with inflammation and downstream targets of IL-6. The cause for this inflammation without overt infection might be the activation of the innate immune system by translocation of bacterial components from the gut, since gut sterilization reduced HCC in an experimental model of hepatocarcinogenesis.24 CRP was not explicitly investigated. Investigations are needed to determine whether CRP levels reflect subclinical bacterial translocation in cirrhosis patients and to compare the predicative value of CRP levels with other transcriptomic signatures in late HCC recurrence.
Does CRP represent an inexpensive, simple prognostic marker for patients with HCC? Peck-Radosavljevic and colleagues' current work appears to indicate that it does, at least in the population of HCCs not amenable for surgery. In particular, CRP was a convincing outcome predictor for BCLC stages B and C, refining the prognosis of Child A and Child B patients. The testing of CRP levels as a variable in the design of trials and in the selection of patients for treatment is warranted.