Departments of Hepatology, Catholic Medical Center, Catholic University of Korea, Seoul, Korea
Division of Hepatology, Department of Internal Medicine, College of Medicine, Catholic University of Korea, 22 Banpo-Daero, Banpo-Dong, Seocho-Gu, Seoul, Korea 137-701. The Catholic University of Korea Incheon St. Mary's Hospital, #56 Dongsu-ro, Bupyeong-gu, Incheon, 403-720, Korea
Although there have been great advances in treatment modalities, hepatocellular carcinoma (HCC) is still a devastating disease. The mortality rate is increasing, and HCC ranks as the third most common cause of cancer deaths worldwide and as the second most common cause of death in Korea.1, 2
Chronic liver disease is one of the main etiologies of HCC, and more than 80% of patients with HCC have been reported to have cirrhosis as well.3 Liver transplantation (LT) is a preferred treatment for selected patients with HCC because it can cure HCC and the primary liver disease at the same time. Since the introduction of the Milan criteria by Mazzaferro et al.,4 survival outcomes after LT have dramatically improved over time, and the 5-year recurrence-free survival (RFS) and overall survival (OS) rates have reached as high as 90% and 70%, respectively.5 However, there are still risks of recurrences (as much as 8%-20%) even when patients meet favorable selection criteria.
There are some limitations to the current selection criteria for transplant candidates. Their major determinants exclusively depend on pretransplant radiological findings. With the Milan criteria, which are the most widely adopted around the world, radiological findings result in a staging error rate of 10% to 46% (either overstaging or understaging).6, 7 In addition, the criteria do not contain strong prognostic parameters for tumor aggressiveness such as tumor differentiation and microvascular invasion.8 However, without biopsy, these factors are not easy to determine before surgery. In addition to tumor number and size, other prognostic markers are needed to predict posttransplant outcomes more accurately.
C-reactive protein (CRP), an acute-phase reactant that is synthesized by hepatocytes in response to systemic inflammation, is helpful for detecting or predicting outcomes of infection, inflammation, and coronary artery disease.9 Recently, it has also been found to be related to adverse outcomes for various malignancies and has been suggested as a prognostic marker. For HCC, a high CRP level has been related to poor survival after the diagnosis as well as poor outcomes after hepatic resection.10, 11 However, the clinical value of CRP remains unclear in the setting of LT for HCC.
Therefore, this study was conducted to determine whether the pretransplant CRP level is predictive of posttransplant tumor recurrence or survival in patients with HCC undergoing LT. The role of CRP was also analyzed according to the Milan criteria.
AFP, alpha-fetoprotein; AUROC, area under the receiver operating characteristic; CI, confidence interval; CRP, C-reactive protein; CT, computed tomography; DDLT, deceased donor liver transplantation; HCC, hepatocellular carcinoma; HR, hazard ratio; IL, interleukin; LDLT, living donor liver transplantation; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; OS, overall survival; PEI, percutaneous ethanol injection; RFA, radiofrequency ablation; RFS, recurrence-free survival; TACL, transarterial chemolipiodolization.
PATIENTS AND METHODS
Between August 2000 and July 2010, 161 patients underwent LT with a preoperative diagnosis of HCC at Seoul Saint Mary's Hospital in South Korea. Three patients who had dysplastic nodules and 1 patient who had a hemangioma according to the final pathology results were excluded from the analysis. Pretransplant serum CRP levels were not available for 72 patients, so those patients were also excluded. Finally, 85 patients whose serum CRP levels at the time of transplantation were available were included in this study. The clinical data for the 85 patients were collected consecutively and were analyzed retrospectively. The median follow-up period was 28.3 months (range = 1.0-118.9 months).
The diagnosis of HCC was based on consistent findings from at least 2 separate imaging studies [including computed tomography (CT), magnetic resonance imaging, and hepatic angiography]. Otherwise, liver biopsy was performed for radiologically atypical lesions. Chest CT, bone scan, and positron emission tomography CT were performed to exclude distant metastases. Hepatitis B and C viral markers, alpha-fetoprotein (AFP) levels, and liver function tests were also checked. Each patient's liver function was classified according to the Child-Pugh class and the Model for End-Stage Liver Disease (MELD) score.
When HCC was diagnosed, the treatment was based on the tumor characteristics and the patient's liver function. Patients who were eligible for surgical treatment were assigned to undergo either LT if a donor organ was available or partial hepatectomy if the HCC was completely resectable. Otherwise, patients with unresectable tumors were subjected to transarterial chemolipiodolization (TACL), percutaneous ethanol injection (PEI), or radiofrequency ablation (RFA) as appropriate.
Patients underwent deceased donor liver transplantation (DDLT) or living donor liver transplantation (LDLT) according to both the fulfillment of the Milan criteria and the availability of donor livers. For DDLT candidates, the Milan criteria were strictly applied. Because of the extreme scarcity of cadaveric donors in Korea, LDLT with family members was often performed when intrafamilial donation was available. For LDLT, expanded selection criteria were adopted for patients without extrahepatic metastases or macrovascular invasion. For patients with HCC within the Milan criteria, LDLT was performed if intrafamilial donation was possible in the absence of an immediate deceased donor. Patients with HCC beyond the Milan criteria received down-staging treatments for the following reasons: (1) to reduce the tumor burden to allow LT, (2) to exclude unfavorable candidates showing aggressive tumor behavior during the pre-LT treatment, (3) to gain the time needed for intrafamilial donation, and (4) to achieve complete necrosis to improve the prognosis. If tumors were successfully down-staged within the Milan criteria, the patients became candidates for DDLT or LDLT. In addition, LDLT was performed for patients whose tumors were still beyond the Milan criteria but came close to meeting the criteria after down-staging treatments if the patients and their families accepted the risk of recurrence after a full discussion with LT specialists. The details of our current LDLT policy are described in our previous report.12
The serum CRP level was measured with a turbidimetric immunoassay (Wako Chemicals GmbH, Neuss, Germany). A value less than 0.47 mg/dL was defined as normal at our hospital. Histological confirmation was performed after transplantation. The study was approved by the institutional review board of the Catholic University of Korea, and all patients agreed to participate in this study with informed consent.
Posttransplant Management and Follow-Up
Postoperatively, tacrolimus and a corticosteroid were administered as primary immunosuppressive drugs. The corticosteroid was tapered to discontinuation 6 months after LT. For prophylaxis against hepatitis B virus reactivation after transplantation, lamivudine or entecavir and hepatitis B immunoglobulin were administered to patients with hepatitis B virus–related HCC. Patients were followed up aggressively with postoperative CT of the abdomen and chest as well as bone scans at 3-month intervals until 1 year after transplantation and at 6-month intervals from then onward. AFP was measured every 3 months for the first 2 years and every 6 months thereafter.
Management After Tumor Recurrences
Patients whose HCC recurred during the post-LT follow-up period received treatment according to the recurrence site, the number of involved organs, and the patient performance status. Patients who had intrahepatic recurrences were initially treated with TACL, and then locoregional treatments, including, RFA, PEI, and radiotherapy, were considered if they were clinically indicated. If a tumor recurred as a single lesion in 1 extrahepatic organ with resectability, surgical metastasectomy was considered. When multiple sites were involved, systemic chemotherapy (epirubicin, cisplatin, and 5-fluorouracil) was used for patients with a good performance status. For bone metastasis, radiotherapy was performed if it was clinically indicated.
The clinicopathological characteristics of the low- and high-CRP groups were summarized as counts and percentages for categorical variables and as means and standard deviations or medians and ranges for continuous variables. Comparisons between the 2 groups were performed with the t test for continuous variables and with the chi-square test for categorical variables. In addition, correlations between the serum CRP level and other variables were evaluated with the Spearman correlation coefficient. The optimal cutoff value for high and low CRP levels was determined with an area under the receiver operating characteristic (AUROC) analysis. For the primary endpoints, we calculated the RFS and OS rates with the Kaplan-Meier method. The associations between the endpoints and the clinicopathological characteristics were tested with Cox proportional hazards regression methods. In the multivariate regression models, candidate variables were included if the corresponding P value in the fitted univariate regression was less than 0.2. Subgroup analyses by the Milan status (beyond or within the Milan criteria) were also conducted. Proportional hazards assumptions were confirmed through the testing of the Schoenfeld residuals; no relevant violations were found in any models. Furthermore, the final multivariate models were evaluated with the C statistic and the Hosmer-Lemeshow test for their discrimination and calibration abilities. Statistical significance was defined as a P value less than 0.05, and all statistical analyses were performed with R statistical software and SPSS 18.0 for Windows (SPSS, Inc., Chicago, IL). R packages (gof, clinfun, and coxphf) were also used for model diagnostics and for monotone likelihood problems due to rare events; the likelihood converged to a finite value when at least 1 parameter estimate diverged to ±∞.
With a median age of 53 years (range = 34-66 years), there were 64 males (75.3%) and 21 females (24.7%). The majority of the patients (94.1%) had an associated viral infection. Hepatitis B virus was the most frequent type of infection (89.4%), and the frequency of hepatitis C virus was 4.7%. The median MELD score was 14 (range = 6-39), and the AFP level was 16.6 ng/mL (range = 1-2132 ng/mL) at the time of transplantation. The majority of the patients received pretransplant treatments to either control tumor growth or down-stage tumors until LT was performed. TACL was performed either alone or in combination with other therapies in 80.0% of the patients, whereas RFA, PEI, and CyberKnife were applied to a small number. At the diagnosis of HCC, 30 patients (35.3%) had tumors that exceeded the Milan criteria. Nine of these patients were down-staged within the Milan criteria by locoregional therapy. On the other hand, 13 patients whose tumors were originally within the criteria at the time of diagnosis no longer met the criteria over time. Finally, at the time of LT, 51 patients (60.0%) had tumors that met the Milan criteria, and 34 patients (40.0%) had tumors that were beyond the Milan criteria. Only 2 patients underwent DDLT, and the remaining 83 patients underwent LDLT.
Clinicopathological characteristics were compared for the patients whose serum CRP levels were low (<1 mg/dL) and the patients whose serum CRP levels were high (≥1 mg/dL) at the time of transplantation (Table 1). The optimum cutoff value of CRP was established at 0.90 mg/dL for RFS [AUROC = 0.773, 95% confidence interval (CI) = 0.640-0.906, P = 0.001] and 0.64 mg/dL for OS (AUROC = 0.671, 95% CI = 0.570-0.810, P = 0.005). Therefore, the CRP cutoff value of 1 mg/dL was used for further analyses. Twenty-seven patients (31.8%) were found to show high CRP levels. The 2 groups were similar in terms of age, sex, etiology, AFP level, tumor number, Milan criteria proportions, tumor pathology, and pretransplant treatment. However, there were some significant differences between the low- and high-CRP groups. The total bilirubin level, Child-Pugh class, and MELD score were higher for patients with high CRP levels. In addition, the maximal tumor size was increased and intrahepatic metastases were more frequent in the high-CRP group. A trend toward a decreased prothrombin time (%) was observed in the high-CRP group, but it did not reach statistical significance. As a continuous variable, the CRP level was significantly correlated with the tumor size (r = 0.445, P < 0.001), the tumor number (r = 0.375, P = 0.001), the total bilirubin level (r = 0.493, P < 0.001), the albumin level (r = −0.355, P = 0.001), the Child-Pugh score (r = 0.522, P < 0.001), and the MELD score (r = 0.516, P < 0.001).
Table 1. Clinicopathological Characteristics
Low-CRP Group: <1 mg/dL (n = 58)
High-CRP Group: ≥1 mg/dL (n = 27)
The data are presented as means and standard deviations.
The data are presented as medians and ranges.
Among the 51 patients whose tumors were within the Milan criteria according to radiological findings, 5 patients were beyond the criteria according to pathological findings. Among the 34 patients exceeding the Milan criteria, 5 patients were within the criteria according to the pathological findings. Interestingly, the numbers of patients in the high- and low-CRP groups were also identical by coincidence.
Prognostic Factors for Recurrence and Survival After LT in the Whole Study Population
During the median follow-up of 28.3 months (range = 1.0-118.9 months), 15 patients (17.6%) developed HCC recurrence after LT. Three of these patients (3.5%) were within the Milan criteria, and 12 (14.1%) were beyond the Milan criteria. The recurrence patterns were intrahepatic recurrence alone for 1 patient, extrahepatic recurrence for 13 patients, and concurrent intrahepatic and extrahepatic recurrence for 1 patient. For the relapsing patients, the median time to recurrence was 5.3 months (range = 2.7-35.1 months). One case of intrahepatic recurrence alone was treated with TACL; and two cases, each a single lung or brain recurrence, were treated with surgical resection. Others with multiple extrahepatic metastases received systemic chemotherapy. By the end of the follow-up period, 31 patients (36.5%) had died. The causes of death were HCC progression in 15 (48.4%), hepatic failure in 8 (25.8%), LT complications in 3 (9.7%), and unrelated causes in 5 (16.1%). The 3- and 5-year RFS rates for the whole study population were 81.6% and 78.8%, respectively, and the corresponding rates for OS were 63.6% and 58.7%.
Among the 13 potential variables listed in Table 2, the following were significantly or marginally associated with higher tumor recurrence after LT in univariate analyses: tumors beyond the Milan criteria, high AFP and CRP levels, a large tumor size, microvascular invasion, and moderate/poor tumor differentiation. According to multivariate analyses with the Cox regression model, HCC beyond the Milan criteria [hazard ratio (HR) = 6.29, P = 0.008], a high CRP level (HR = 4.64, P = 0.004), and microvascular invasion (HR = 3.65, P = 0.02) remained significant as independent risk factors for tumor recurrence in the whole population. As for OS after LT, a high CRP level (HR = 2.68, P = 0.01) and microvascular invasion (HR = 2.69, P = 0.01) were identified as independent prognostic factors for poor survival in the multivariate analyses (Table 2).
Table 2. Prognostic Factors for RFS and OS in the Whole Study Population
Univariate Analysis: P Value
Univariate Analysis: P Value
HR (95% CI)
HR (95% CI)
The Milan criteria were based on radiological findings.
Moderate or poor differentiation according to pathological findings.
Prognostic Factors for Recurrence and Survival After LT According to the Milan Criteria
To define the prognostic factors specifically according to the Milan criteria, we performed subgroup analyses for patients with HCC within or beyond the criteria separately (Table 3). For 34 patients beyond the Milan criteria, a high CRP level (HR = 5.96, P = 0.02) and moderate/poor tumor differentiation (HR = 7.31, P = 0.03) were associated with decreased RFS in the multivariate analyses. Furthermore, a high CRP level (HR = 9.27, P < 0.001) and multiple tumors (HR = 3.89, P = 0.02) were associated with decreased OS (Table 4). For patients whose HCC met the Milan criteria, microvascular invasion was associated with decreased RFS with marginal significance (P = 0.06). For a high CRP level, similar trends were found when pathological data were used to determine whether tumors were within or beyond the Milan criteria. A high CRP level was also an independent prognostic factor for poor outcomes in patients whose HCC was beyond the Milan criteria pathologically [HR = 8.16 for RFS (P = 0.003) and HR = 7.76 for OS (P < 0.001)].
Table 3. Prognostic Factors for RFS in Patients With Tumors Within and Beyond the Milan Criteria
Within Milan Criteria (n = 51)
Beyond Milan Criteria (n = 34)
Univariate Analysis: P Value
Univariate Analysis: P Value
HR (95% CI)
HR (95% CI)
NOTE: The Milan criteria were based on radiological findings.
Moderate or poor differentiation according to pathological findings.
Tumor Recurrence and OS According to the Pretransplant CRP Level
The cumulative tumor recurrence and survival rates according to the pretransplant CRP level are shown in Fig. 1. There was a significant difference in RFS between the low- and high-CRP groups (HR = 5.65, P = 0.002; Fig. 1A). The OS rate was also significantly lower for patients with high CRP levels (HR = 2.89, P = 0.005; Fig. 1B).
When the survival of patients was analyzed according to the Milan criteria, no significant survival differences were found in patients with HCC within the criteria [HR = 6.74 for RFS (P = 0.10) and HR = 1.14 for OS (P = 0.86); Fig. 2A,B]. However, for patients whose tumors were beyond the criteria, the survival differences were statistically significant between the low- and high-CRP groups [HR = 9.61 for RFS (P = 0.001) and HR = 10.40 for OS (P < 0.001); Fig. 2C,D].
In the present study, the pretransplant CRP level helped to predict the prognosis after LT for patients with HCC. Both RFS and OS rates after LT were significantly lower in the high-CRP group versus the low-CRP group. In our analysis, the CRP concentration correlated with the size and number of tumors as well as hepatic functional parameters such as the Child-Pugh and MELD scores. These data indicate that an elevated CRP level could be a strong independent prognostic marker of adverse outcomes after LT.
Several reports have documented the prognostic role of the CRP level in patients with various malignancies. The CRP level has been positively correlated with the following: cancer incidence; tumor size, depth, and stage; resistance to chemotherapy; metastatic potential; and mortality.13-17 Since Virchow first mentioned in 1863 that inflammation and cancer are connected to each other,18 this idea has been accepted for several reasons. Inflammation promotes tumor cell proliferation and survival and stimulates angiogenesis and metastasis. It also provides an attractive environment for tumor growth and modulates the immune response in favor of cancer progression.18 Inflammatory cytokines such as interleukin-6 (IL-6) and IL-1β are linked to transcriptional signaling pathways (eg, the Janus kinase/signal transducer and activator of transcription and nuclear factor kappa B pathways) known to be associated with carcinogenesis, tumor growth, and invasion in various malignancies.18, 19 IL-6 is known as one of the main regulators of CRP production.20 Therefore, this connection could be a possible mechanism for the association of a high CRP level with a poor prognosis. Some hematological malignancies are known to directly stimulate CRP production via IL-6 secretion.21, 22 In several studies, the Janus kinase/signal transducer and activator of transcription pathway has been reported to be constitutively activated in HCC.23, 24
Some clinical factors, including the type of HCC (diffuse or massive), tumor size, vascular invasion, extrahepatic metastasis, and poor liver function, have been associated with a high CRP level in patients with HCC. These in turn result in high recurrence and poor survival rates after the diagnosis or surgical resection of HCC.10, 11, 25 In this study, the CRP level correlated well with the tumor burden and hepatic reserve, and the high-CRP group showed elevated total bilirubin levels and Child-Pugh/MELD scores and higher frequencies of intrahepatic metastases in comparison with the low-CRP group. Consistent with previous studies, these findings suggest that CRP reflects tumor aggressiveness and poor liver function. As a result, patients with high pretransplant CRP levels had significantly lower RFS and OS rates than patients with low CRP levels.
In the LT era, there have been some reports about inflammation and posttransplant outcomes. An elevation of IL-6 during or after transplantation has been suggested to be a useful predictor of postoperative infections or graft rejection.26, 27 Recently, Bertuzzo et al.28 analyzed inflammatory markers as prognostic factors in HCC patients undergoing LT. In their study, the pretransplant CRP level was not significantly related to RFS or OS. However, the study had a higher proportion of patients within the Milan criteria and with hepatitis C virus–dominant disease in comparison with ours. In agreement with the previous results, the CRP level was not significant for patients within the Milan criteria in our study either. Further studies are needed to confirm the role of CRP in posttransplant outcomes. In terms of graft rejection, we could not find any significant association between the baseline CRP level and rejection or between the degree of rejection and survival. In the additional analysis, the prognostic role of CRP was not significant for non-HCC patients undergoing LT at our center (data are not shown). We speculate that CRP is not only an inflammatory marker but also a tumor marker, and the predictive role of CRP may be more apparent in HCC patients versus non-HCC patients.
Another major finding of this study was that survival differences between the low- and high-CRP groups were more prominent in patients with HCC beyond the Milan criteria. Because most studies about posttransplant prognosis have included patients within the criteria, little is known about patients beyond the criteria. Patients beyond the criteria are a relatively heterogeneous group in terms of tumor or liver function parameters in comparison with patients with limited tumor numbers and sizes. It is, therefore, necessary to develop distinctive selection criteria for LT in patients beyond the Milan criteria. Sieghart et al.29 reported that osteopontin expression in explanted livers predicted OS after LT in patients beyond the Milan criteria, but this cannot be used as a pretransplant predictor and requires additional immunohistochemical stains. The CRP level can be measured in an easy-to-perform and inexpensive way, and it is a widely available serum marker. Therefore, the pretransplant CRP level is expected to be a simple but powerful biomarker for selecting recipients when LT is being considered for patients whose HCC is beyond the Milan criteria. Because a high CRP level represents tumor aggressiveness, we can speculate that LT should be reserved for patients with high CRP levels on account of the possibility of poor outcomes. Alternatively, neoadjuvant therapy can be chosen with the hope of down-staging; this remains to be confirmed in future studies. A cytokine signal–blocking agent that has been shown to inhibit tumor growth in some HCC cell lines may be a potential neoadjuvant therapy in the future.24, 30
For patients with HCC within the Milan criteria, few prognostic factors were found in our analysis. One reason is the small sample size, and another possible reason is the fact that tumors within the Milan criteria have a low probability of recurrence after LT; this makes other variables less significant in this population. However, the presence of microvascular invasion showed a trend toward decreased RFS with marginal significance. The prognostic significance of microvascular invasion could be confirmed in future studies.
There are some limitations to this study. First, this is a single-institution, retrospective study, and the pretransplant CRP level was not measured routinely until the end of 2006. Almost half of the patients undergoing transplantation for HCC did not have pretransplant CRP levels. This decreased the sample size of our study. Further studies are needed to validate our results prospectively. Second, a relatively high percentage of patients whose HCC was beyond the Milan criteria were included in the analysis. One possible reason is that liver donations from family members were performed frequently because of the absolute shortage of deceased donors in Korea. Lastly, there are many possible conditions affecting the level of CRP, including infections. However, we performed meticulous pretransplant examinations to rule out any other clinical problems (including infections), and LT was not considered for patients who were suspicious for infectious status. Moreover, the CRP level was measured at the time of transplantation. Therefore, the possibility of confounding seems to have been minimized. Whether a pretransplant locoregional treatment could significantly affect CRP levels is not clear from this study because we did not have serial data for CRP before and after therapy. This issue should be taken into account in future studies.
In conclusion, the present study shows that the pretransplant CRP level is a powerful predictor of prognosis in patients with HCC undergoing LT. A high CRP level may represent tumor aggressiveness and poor liver function. In addition, the prognostic role of CRP in predicting outcomes after LT is more evident in patients with HCC beyond the Milan criteria. Therefore, the CRP level can provide valuable information on posttransplant outcomes, especially when LT candidates are being selected from this population.