Volume 14, Issue 2 pp. 228-234
Original Articles
Free Access

Frequency and predictors of de novo hepatocellular carcinoma in patients awaiting orthotopic liver transplantation during the model for end-stage liver disease era

Carla W. Brady

Corresponding Author

Carla W. Brady

Veterans Affairs Medical Center, Durham, NC

Division of Gastroenterology, Duke University Medical Center, Durham, NC

Telephone: 919-684-2052; FAX: 919-684-8264

Division of Gastroenterology, Duke University Medical Center, Box 3913, Durham, NC 27710Search for more papers by this author
Alastair D. Smith

Alastair D. Smith

Division of Gastroenterology, Duke University Medical Center, Durham, NC

Search for more papers by this author
Karen M. Stechuchak

Karen M. Stechuchak

Veterans Affairs Medical Center, Durham, NC

Search for more papers by this author
Cynthia J. Coffman

Cynthia J. Coffman

Veterans Affairs Medical Center, Durham, NC

Department of Bioinformatics and Biostatistics, Duke University, Durham, NC

Search for more papers by this author
Janet E. Tuttle-Newhall

Janet E. Tuttle-Newhall

Division of General Surgery and Critical Care, Duke University Medical Center, Durham, NC

Search for more papers by this author
Dawn Provenzale

Dawn Provenzale

Veterans Affairs Medical Center, Durham, NC

Division of Gastroenterology, Duke University Medical Center, Durham, NC

Search for more papers by this author
Andrew J. Muir

Andrew J. Muir

Division of Gastroenterology, Duke University Medical Center, Durham, NC

Search for more papers by this author
First published: 30 January 2008
Citations: 11

The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs or the Department of Health and Human Services.

Abstract

In the current system of allocation, patients awaiting orthotopic liver transplantation (OLT) remain at risk of developing de novo hepatocellular carcinoma (HCC) and removal from the waiting list. Using the United Network for Organ Sharing database, we calculated the rate and identified predictors of de novo HCC in patients listed for OLT between February 2002 and December 2004. Among 8566 patients, 1167 (13.6%) developed de novo HCC. Predictors of increased odds of de novo HCC were older age, male gender, Asian race, other race, hepatitis C, and hepatitis B. A sensitivity analysis of 2067 patients waiting at least 6 months found that 16.2% developed de novo HCC. Older age [odds ratio (OR) 1.05; 95% confidence interval (CI) 1.03, 1.07], male gender (OR 2.01; 95% CI 1.49, 2.71), Asian race (OR 2.39; 95% CI 1.20, 4.76), other race (OR 1.94; 95% CI 1.40, 2.68), hepatitis C (OR 2.36; 95% CI 1.76, 3.16), and hepatitis B (OR 1.96; 95% CI 1.19, 3.23) remained predictors of increased odds of de novo HCC, and alcoholic liver disease (OR 1.40; 95% CI 1.06, 1.86) emerged as a predictor of increased odds of de novo HCC. A significant proportion of patients listed for OLT develop de novo HCC. Identifying predictors of HCC in these patients may facilitate timely HCC screening and diagnosis. Liver Transpl 14:228–234, 2008. © 2008 AASLD.

Primary liver cancer accounts for approximately 6% of all human cancers worldwide. Hepatocellular carcinoma (HCC) accounts for the majority of primary liver cancers, and its incidence in the United States is rising.1, 2 Patients who are listed for orthotopic liver transplantation (OLT) due to cirrhosis have an increased risk of developing HCC. Use of the Model for End-Stage Liver Disease (MELD) score for prioritization of patients for OLT has increased the number of HCC patients undergoing OLT.3 This has exposed non-HCC patients to an increased likelihood of decompensation, which includes the development of de novo HCC, removal from the transplant waiting list, and death prior to OLT. Quantifying the rate at which de novo HCC develops in patients awaiting OLT and identifying factors that influence de novo HCC development in these patients may be helpful in targeting HCC screening and surveillance practices to specific at-risk populations. In a single-center study of 100 patients, Van Thiel et al.4 showed that 20% of patients listed for OLT develop de novo HCC. This finding has not been validated in a larger patient population, and predictors of the development of de novo HCC in patients awaiting OLT are not known. The present study uses a national database to determine the proportion of patients who develop de novo HCC while listed for OLT and if specific factors influence the development of de novo HCC in such patients.

Abbreviations

A1AT, alpha-1-antitrypsin deficiency; AIH, autoimmune hepatitis; ALD, alcoholic liver disease; CI, confidence interval; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; OLT, orthotopic liver transplantation; OPTN, Organ Procurement and Transplantation Network; OR, odds ratio; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; Q1-Q3, first quartile to third quartile; SD, standard deviation.

PATIENTS AND METHODS

Study Population and Data Collection

The Organ Procurement and Transplantation Network (OPTN) database of the United Network for Organ Sharing was used to identify patients for inclusion in this study. The study population consisted of patients 18 years old or older who were listed for OLT between February 2002 and December 2004 and subsequently underwent OLT. Since its creation, the OPTN database has been expanded to include many clinical data, but these variables have been added at different times throughout the history of the database. We limited our study population to patients listed during the MELD era because we were able to obtain more complete data regarding the selected predictor and outcome variables of interest. We excluded patients who were initially listed with a diagnosis of HCC, cholangiocarcinoma, other primary and metastatic liver cancer, fulminant hepatic failure, or a history of previous organ transplantation. We collected the following variables from the OPTN database: age, race/ethnicity, gender, wait time, listing diagnoses, diagnoses at the time of OLT, viral hepatitis serologies, presence or absence of diabetes mellitus, initial calculated MELD score at listing, and presence or absence of incidental tumor found at OLT.

In accordance with guidelines from the Office of Management and Budget, race (white, black or African American, Asian, American Indian and Alaska Native, and Native Hawaiian and other Pacific Islander) and ethnicity (Hispanic or Latino or not Hispanic or Latino) are separately recorded by the US Census.5 Within the OPTN database, separate variables for race and ethnicity are similarly designated. However, the race variable includes a separate category of multiracial, and race and ethnicity variables each have a category that identifies Hispanic ethnicity. In attempting to delineate race from ethnicity, we observed that many patients in our study population were coded as being Hispanic under the race and ethnicity variables. Additionally, many other Hispanic patients were also coded as being multiracial. Therefore, we created a single variable for race, and the levels within our race variable were white, black, Asian, and other.

We divided the listing diagnoses into individual categories for the most common chronic liver diseases. The diagnosis of hepatitis C virus (HCV) liver disease was defined by the presence of a listing diagnosis of HCV, evidence of clinical HCV liver disease, the presence of liver histology consistent with HCV disease, or detectable serum HCV RNA. The diagnosis of hepatitis B virus (HBV) liver disease was defined by the presence of a listing diagnosis of HBV, presence of clinical HBV liver disease, presence of liver histology consistent with HBV liver disease, presence of HBV DNA, having a positive HBV surface antigen, having a positive HBV core antibody with the presence of clinical HBV liver disease, or having a positive HBV core antibody with the presence of liver histology consistent with HBV disease. The presence of other liver diseases was defined by listing diagnoses and diagnoses identified at OLT. Because previous investigations have suggested that HCC may be associated with hemochromatosis and alpha-1-antitrypsin deficiency, we created separate categories for these diagnoses. We combined all other diagnoses into the category of other liver disease. We excluded patients with nondescript diagnoses of end-stage liver disease or unknown disease. However, we included patients with a diagnosis of cryptogenic cirrhosis and combined them with the group of patients carrying the diagnoses of nonalcoholic steatohepatitis (NASH) or nonalcoholic fatty liver disease (NAFLD).

The development of de novo HCC was defined as being diagnosed with HCC between initial listing and the time of OLT or having HCC discovered incidentally at OLT. The criteria for HCC have been altered during the MELD era. In the initial MELD allocation system beginning in February 2002, patients with imaging studies revealing stage 1 (1 nodule < 2 cm) and stage 2 (1 nodule of 2.0-5.0 cm; 2-3 nodules, all < 3.0 cm) HCC received exception points equivalent to a 20% probability of pretransplant death within 3 months. In April 2003, modifications were made to the allocation system that included more stringent criteria for the diagnosis of HCC. Patients were also required to have one of the following: tumor greater than 1 cm in size with a vascular blush corresponding to the area of suspicion on the imaging studies, an alpha-fetoprotein greater than 200 ng/mL, an arteriogram confirming a tumor, a biopsy confirming HCC, and chemoembolization or ablation of the lesion. Patients with chronic liver disease and alpha-fetoprotein levels greater than 500 ng/mL also could be listed as stage 1 HCC. In April 2004, further modifications were made to the allocation system. Exception points were no longer given to patients with stage 1 HCC. Patients with alpha-fetoprotein levels greater than 500 ng/mL could still be listed as stage 1 HCC if the alpha-fetoprotein level was rising.6

Statistical Analyses

We estimated the rate of development of de novo HCC in the study population and calculated a 95% confidence interval (CI) using the standard formula for binomial proportions. We examined bivariate relationships between developing de novo HCC and predictor variables that were selected a priori with Pearson chi-square tests for categorical predictors and t tests or Wilcoxon rank sum tests for continuous predictors, depending on their distribution. A Spearman correlation matrix of the predictor variables was examined. With de novo HCC as the outcome variable, we fitted a multiple logistic regression model to generate adjusted odds ratios (ORs) and 95% CIs for the predictor variables. Test statistics with P values less than 0.05 were considered statistically significant. We tested for multicollinearity in the predictor variables and verified the linearity in the log assumption for all continuous variables.7 To address concerns that some de novo HCC cases may have been undetected HCC cases at baseline, we conducted a sensitivity analysis by replicating all aforementioned steps while restricting our sample to those subjects who received a transplant after being on the waiting list for at least 6 months. Analyses were performed with SAS version 9.1 software (SAS Institute, Cary, NC).

RESULTS

We identified 8566 patients who were placed on the United Network for Organ Sharing waiting list for OLT between February 27, 2002 and December 31, 2004 and met criteria for inclusion in our study. Of these patients, 1167 (13.6%) patients developed de novo HCC after being listed for OLT (95% CI, 12.9%, 14.4%). Of those patients who developed de novo HCC, 80.1% (n = 935) were diagnosed prior to OLT, and 10.0% (n = 117) had HCC discovered incidentally at explant. An additional 9.9% (n = 115) of those who developed de novo HCC were identified as being diagnosed with HCC after listing for OLT and as having HCC discovered incidentally at explant. Table 1 shows the various demographic and clinical characteristics of our study population. The mean age was 51.9 ± 9.4 years. The majority of patients were white (76.9%, n = 6571) and of male gender (67.4%, n = 5775). Chronic HCV infection and alcoholic liver disease represented the most common causes of underlying liver disease. The median MELD score at listing was 17.0 (first quartile to third quartile: 13.0-22.0).

Table 1. Demographic and Clinical Characteristics of the Patients (n = 8566)
Characteristic Data
Age at listing (mean ± SD in years) 51.9 ± 9.4
Gender
 Male 5775 (67.4%)
 Female 2791 (32.6%)
Race
 White 6571 (76.9%)
 Black 663 (7.8%)
 Asian 228 (2.7%)
 Other 1085 (12.7%)
Underlying Liver Disease
 HCV 3943 (46.0%)
 ALD 2720 (31.8%)
 NAFLD/NASH 1186 (13.9%)
 PSC 560 (6.5%)
 HBV 548 (6.4%)
 PBC 435 (5.1%)
 AIH 332 (3.9%)
 A1AT 157 (1.8%)
 Hemochromatosis 111 (1.3%)
 Other liver disease 404 (4.7%)
Diabetes mellitus 1752 (20.5%)
Initial calculated MELD at listing [median (Q1-Q3)] 17 (13-22)
Days on waiting list [median (Q1-Q3)] 62 (17-173)
  • Abbreviations: A1AT, alpha-1-antitrypsin deficiency; AIH, autoimmune hepatitis; ALD, alcoholic liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD/NASH, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; Q1-Q3, first quartile to third quartile; SD, standard deviation.
  • * Data on race were missing for 19 subjects; data on initial calculated MELD at listing were missing for 28 subjects, and data for days on the waiting list were missing for 3 subjects.
  • Other liver disease included Wilson's disease, glycogen storage disease, sarcoidosis, amyloidosis, cystic fibrosis, polycystic liver disease, Budd-Chiari syndrome, and cirrhosis secondary to chemical exposures or medications.

Bivariate analyses showed that patients who developed de novo HCC were significantly older and were significantly more likely to be male than female and of Asian or other race than of white race (Table 2). The number of days that de novo HCC patients waited on the list prior to OLT was significantly longer than that for patients who did not develop de novo HCC. The development of de novo HCC varied significantly according to the presence of diabetes mellitus, HCV infection, and HBV infection. Patients with NASH/NAFLD, primary sclerosing cholangitis (PSC), primary biliary cirrhosis (PBC), autoimmune hepatitis (AIH), and other liver disease had significantly lower odds of developing de novo HCC. Patients who developed de novo HCC also had significantly lower calculated MELD scores at listing than patients who did not develop de novo HCC.

Table 2. Bivariate Analysis of Predictors of De Novo HCC in Patients Awaiting Orthotopic Liver Transplantation
Variable HCC Present (n = 1167) HCC Not Present (n = 7399) Unadjusted Odds Ratio 95% Confidence Interval P Value
Age at listing (mean ± SD) 54.9 ± 8.0 51.4 ± 9.6 1.04 1.04, 1.05 <0.0001
Gender
 Female 254 (21.8%) 2537 (34.3%) Reference
 Male 913 (78.2%) 4862 (65.7%) 1.88 1.62, 2.17 <0.0001
Race
 White 855 (73.3%) 5716 (77.5%) Reference <0.0001
 Black 84 (7.2%) 579 (7.9%) 0.97 0.76, 1.23 0.8028
 Asian 58 (5.0%) 170 (2.3%) 2.28 1.68, 3.10 <0.0001
 Other 170 (14.6%) 915 (12.4%) 1.24 1.04, 1.49 0.0175
Underlying liver disease
 HCV 797 (68.3%) 3146 (42.5%) 2.91 2.55, 3.32 <0.0001
 ALD 367 (31.5%) 2353 (31.8%) 0.98 0.86, 1.12 0.8099
 NAFLD/NASH 104 (8.9%) 1082 (14.6%) 0.57 0.46, 0.71 <0.0001
 PSC 16 (1.4%) 544 (7.4%) 0.18 0.11, 0.29 <0.0001
 HBV 110 (9.4%) 438 (5.9%) 1.65 1.33, 2.06 <0.0001
 PBC 23 (2.0%) 412 (5.6%) 0.34 0.22, 0.52 <0.0001
 AIH 16 (1.4%) 316 (4.3%) 0.31 0.19, 0.52 <0.0001
 A1AT 14 (1.2%) 143 (1.9%) 0.62 0.36, 1.07 0.0857
 Hemochromatosis 10 (0.9%) 101 (1.4%) 0.63 0.33, 1.20 0.1575
 Other liver disease 16 (1.4%) 388 (5.2%) 0.25 0.15, 0.42 <0.0001
Diabetes mellitus 283 (24.3%) 1469 (19.9%) 1.29 1.12, 1.50 0.0006
Initial calculated MELD at listing [median (Q1-Q3)] 13 (10-16) 17 (14-23) 0.87 0.86, 0.88 <0.0001
Days on waiting list [median (Q1-Q3)] 68 (23-207) 61 (16-168) 1.001 1.00, 1.001 <0.0001
  • Abbreviations: A1AT, alpha-1-antitrypsin deficiency; AIH, autoimmune hepatitis; ALD, alcoholic liver disease; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD/NASH, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; Q1-Q3, first quartile to third quartile; SD, standard deviation.
  • * Other liver disease included Wilson's disease, glycogen storage disease, sarcoidosis, amyloidosis, cystic fibrosis, polycystic liver disease, Budd-Chiari syndrome, and cirrhosis secondary to chemical exposures or medications.

In our initial multiple logistic regression model, we included all variables from the bivariate analyses. On the basis of our correlation analysis, the variable representing days waiting on the OLT list was not included in our final model as it was correlated with initial calculated MELD score (ŕ = −0.45).8 A total of 8519 patients were included in the final model, and 47 patients were excluded because of missing data (initial calculated MELD scores were missing in 28 patients, and race was listed as unknown in 19 patients). The final logistic regression model is shown in Table 3. Patients who developed de novo HCC were more likely to be older (OR 1.06; 95% CI, 1.05, 1.07; P < 0.0001)] and of male gender (OR 1.97; 95% CI, 1.66, 2.33; P < 0.0001). Patients of Asian race (OR 2.12; 95% CI, 1.45, 3.10; P = 0.0001) and other race (OR 1.45; 95% CI, 1.19, 1.76; P = 0.0003) also had higher odds of developing de novo HCC compared to those of white race. Underlying liver diseases that were positive predictors of de novo HCC development were HCV infection (OR 2.07; 95% CI, 1.69, 2.53; P < 0.0001) and HBV infection (OR 1.58; 95% CI, 1.21, 2.06; P = 0.0008). Variables associated with decreased odds of developing de novo HCC included NAFLD/NASH (OR 0.62; 95% CI, 0.47, 0.84; P = 0.0016), PSC (OR 0.22; 95% CI, 0.13, 0.38; P < 0.0001), PBC (OR 0.50; 95% CI, 0.31, 0.81; P = 0.0046), AIH (OR 0.52; 95% CI, 0.30, 0.91; P = 0.0216), and other liver disease (OR 0.36; 95% CI, 0.21, 0.62; P = 0.0002). Additionally, higher initial calculated MELD scores were associated with decreased odds of developing de novo HCC (OR 0.86; 95%CI, 0.85, 0.87; P < 0.0001). The c statistic for our model was 0.80.

Table 3. Multiple Logistic Regression Model of Predictors of HCC in Patients Awaiting Orthotopic Liver Transplantation
Variable Adjusted Odds Ratio 95% Confidence Interval P Value
Age at listing 1.06 1.05, 1.07 <0.0001
Male gender 1.97 1.66, 2.33 <0.0001
Race <0.0001
 White race Reference
 Black race 1.23 0.94, 1.60 0.1385
 Asian race 2.12 1.45, 3.10 0.0001
 Other race 1.45 1.19, 1.76 0.0003
HCV 2.07 1.69, 2.53 <0.0001
ALD 0.95 0.80, 1.13 0.5580
NAFLD/NASH 0.62 0.47, 0.84 0.0016
PSC 0.22 0.13, 0.38 <0.0001
HBV 1.58 1.21, 2.06 0.0008
PBC 0.50 0.31, 0.81 0.0046
AIH 0.52 0.30, 0.91 0.0216
A1AT 0.98 0.54, 1.80 0.9512
Hemochromatosis 0.62 0.31, 1.24 0.1738
Other liver disease 0.36 0.21, 0.62 0.0002
Diabetes mellitus 1.13 0.96, 1.33 0.1320
Initial MELD at listing 0.86 0.85, 0.87 <0.0001
  • Abbreviations: A1AT, alpha-1-antitrypsin deficiency; AIH, autoimmune hepatitis; ALD, alcoholic liver disease; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD/NASH, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis.
  • * Other liver disease included Wilson's disease, glycogen storage disease, sarcoidosis, amyloidosis, cystic fibrosis, polycystic liver disease, Budd-Chiari syndrome, and cirrhosis secondary to chemical exposures or medications.

For the sensitivity analysis, 2067 patients were identified by the restriction of the sample to those on the waiting list for at least 6 months. In this reduced sample, a slightly higher proportion of patients (16.2%, n = 334) developed de novo HCC while awaiting OLT. In comparison with the full sample, similar trends were observed in the sensitivity analysis in the proportion of patients who were diagnosed with de novo HCC prior to OLT (79.3%, n = 265), the proportion of patients who had HCC discovered incidentally at explant (10.8%, n = 36), and the proportion of patients identified as being diagnosed with HCC after being listed for OLT and as having HCC discovered incidentally at explant (9.9%, n = 33).

The variables included in our final logistic regression model for the sensitivity analysis were different than the model generated on the full sample because of data and statistical issues. We found that in the reduced sample certain predictors had very low prevalence rates [for example, hemochromatosis (0.87%) and alpha-1-antitrypsin deficiency (1.45%)]; therefore, the variable other liver disease was expanded to include the presence of any of the following diagnoses: PSC, PBC, AIH, alpha-1-antitrypsin deficiency, or hemochromatosis (Table 4). Furthermore, we detected multicollinearity in our initial logistic regression model.9 On the basis of our correlation analysis, the variable representing other liver disease was not included in our final sensitivity model as it correlated with HCV (ŕ = −0.50).8 We did not have the same issues with correlation between days on the waiting list and initial calculated MELD score in the sensitivity analysis as we did in the full analysis and were able to include both variables in the model for the sensitivity analysis. In comparing the results of the sensitivity analysis model to those of the full analysis, we noted that ORs were of similar magnitude and statistical significance with the following exceptions: NAFLD/NASH and alcoholic liver disease. Older age (OR 1.05; 95% CI 1.03, 1.07; P < 0.0001), male gender (OR 2.01; 95% CI 1.49, 2.71; P < 0.0001), Asian race (OR 2.39; 95% CI 1.20, 4.76; P = 0.0131) and other race (OR 1.94; 95% CI 1.40, 2.68; P < 0.0001) compared to white race, HCV (OR 2.36; 95% CI 1.76, 3.16; P < 0.0001), and HBV (OR 1.96; 95% CI 1.19, 3.23; P = 0.0084) remained predictors of increased odds of developing de novo HCC (Table 5). A higher initial calculated MELD score at listing remained a predictor of decreased odds of developing de novo HCC (OR 0.89; 95% CI 0.86, 0.92; P < 0.0001). Although NAFLD/NASH was associated with a decrease in the odds of developing de novo HCC in the full sample, this was no longer the case in the smaller sample of the sensitivity analysis. Alcoholic liver disease was not associated with de novo HCC in the full sample, but in the sensitivity analysis, alcoholic liver disease was associated with increased odds of developing de novo HCC (OR 1.40; 95% CI 1.06, 1.86; P = 0.0183). Also, in the sensitivity analysis, we observed that the variable days on the waiting list was not statistically associated with de novo HCC (OR = 1.00, 95% CI = 1.00, 1.001, P = 0.1407).

Table 4. Demographic and Clinical Characteristics of Patients in the Sensitivity Analysis (n = 2067)
Characteristic Data
Age at listing (mean ± SD in years) 51.6 ± 9.0
Gender
 Male 1344 (65.0%)
 Female 723 (35.0%)
Race
 White 1569 (76.2%)
 Black 142 (6.9%)
 Asian 51 (2.5%)
 Other 298 (14.5%)
Underlying liver disease
 HCV 1023 (49.5%)
 ALD 605 (29.3%)
 NAFLD/NASH 280 (13.6%)
 PSC 145 (7.0%)
 HBV 112 (5.4%)
 PBC 147 (7.1%)
 AIH 70 (3.4%)
 A1AT 30 (1.5%)
 Hemochromatosis 18 (0.9%)
 Other liver disease 90 (4.4%)
Diabetes mellitus 431 (20.9%)
Initial calculated MELD at listing [median (Q1-Q3)] 14 (11-17)
Days on waiting list [median (Q1-Q3)] 322 (234-477)
  • Abbreviations: A1AT, alpha-1-antitrypsin deficiency; AIH, autoimmune hepatitis; ALD, alcoholic liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD/NASH, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; Q1-Q3, first quartile to third quartile; SD, standard deviation.
  • * Data on race were missing for 7 subjects; data on initial calculated MELD at listing were missing for 4 subjects.
  • This is representative of other liver disease as defined in the full sample, incorporating Wilson's disease, glycogen storage disease, sarcoidosis, amyloidosis, cystic fibrosis, polycystic liver disease, Budd-Chiari syndrome, and cirrhosis secondary to chemical exposures or medications. In the sensitivity analysis, other liver disease was expanded to include the following variables: PSC, PBC, AIH, alpha-1-antitrypsin deficiency, hemochromatosis, and other liver disease. This new creation of the variable other liver disease represented 480 patients (23.2%).
Table 5. Multiple Logistic Regression Model of Predictors of HCC in Patients Awaiting Orthotopic Liver Transplantation for the Sensitivity Analysis
Variable Adjusted Odds Ratio 95% Confidence Interval P Value
Age at listing 1.05 1.03, 1.07 <0.0001
Male gender 2.01 1.49, 2.71 <0.0001
Race <0.0001
 White race Reference
 Black race 1.22 0.72, 2.07 0.4508
 Asian race 2.39 1.20, 4.76 0.0131
 Other race 1.94 1.40, 2.68 <0.0001
HCV 2.36 1.76, 3.16 <0.0001
ALD 1.40 1.06, 1.86 0.0183
NAFLD/NASH 1.26 0.79, 2.00 0.3359
HBV 1.96 1.19, 3.23 0.0084
Diabetes mellitus 1.21 0.89, 1.63 0.2205
Initial MELD at listing 0.89 0.86, 0.92 <0.0001
Days on waiting list 1.000 1.000, 1.001 0.1407
  • NOTE: c = 0.725.
  • Abbreviations: ALD, alcoholic liver disease; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; NAFLD/NASH, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis.

DISCUSSION

To our knowledge, this is the first study to use national data to quantify the rate of de novo HCC in patients awaiting OLT. We found that a clinically significant proportion of patients developed de novo HCC while awaiting OLT, and it appears that existing screening practices identified most of these cases. However, approximately 20% of our study population was identified as having HCC diagnosed at explant. In identifying predictors of de novo HCC in patients awaiting OLT, we made a number of interesting observations.

Previous studies have indicated that Asian race is a predictor of HCC, and the association of Asian race with HCC has been thought to be influenced largely by the increased prevalence of HBV infection in the Asian population.2, 10 In order to examine specifically the impact of Asian race on the development of de novo HCC, we entered Asian race and HBV infection as separate factors in our regression model. As a result, we observed that patients of Asian race had approximately twice the odds of developing de novo HCC as patients of white race, and this was independent of the presence of HBV infection. The reasons for this observation are unclear but may reflect genetic differences in the predisposition to HCC development. Previous studies have also indicated that blacks are also more likely to develop HCC than whites.2, 9 The lack of association of black race with de novo HCC in our study population may reflect previously reported racial disparities in the utilization of OLT.11 Additionally, the inability to measure the impact of Hispanic ethnicity as an independent factor on de novo HCC development in our cohort resulted from the inability to distinguish accurately Hispanic ethnicity from one's racial identity within the database.

Older age and male gender were also associated with increased odds of developing de novo HCC, and this confirms findings of previous studies. Data from the Surveillance Epidemiology and End Results program have indicated that the mean age of patients diagnosed with HCC is 65 years.12 However, the overall mean age of patients who developed de novo HCC in our study was 54.9 years, which represents a younger population of HCC patients. This likely reflects the selection of younger patients for OLT listing on the basis of a decreased likelihood of having fewer and/or less severe comorbid illnesses, for it has been shown that 3-year and 5-year patient survival is lower for patients 65 years old and older compared to younger patients.13

It is not surprising that we observed that PSC, PBC, AIH, and other liver diseases were associated with decreased odds of developing de novo HCC. These diseases have not been traditionally associated with an increased risk of HCC compared to other forms of end-stage liver disease. In the larger patient sample, alcohol was not a predictor of HCC. The sensitivity analysis of patients listed for at least 6 months did demonstrate that alcoholic liver disease was associated with increased risk of development of HCC. The sensitivity analysis likely represents a more uniform patient group that resulted in the statistically significant finding, and it also supports earlier observations of the association of alcoholic liver disease with HCC and data on the role of alcohol in the initiation, promotion, and progression of liver cancer.14, 15 The results for NAFLD/NASH were also not consistent between the two analyses. In our full sample, we observed a decrease in the odds of de novo HCC development among patients with NAFLD/NASH. The sensitivity analysis did not confirm this finding, but this is likely related to the reduction in power with the markedly smaller number of patients with NAFLD/NASH in the sensitivity analysis. Our initial findings are consistent with more recently published data by Sanyal et al.,16 who observed that the rate of HCC development in patients with NASH-related cirrhosis is significantly lower than the rate of HCC development in patients with HCV-related cirrhosis.

Another observation of interest is that higher calculated MELD scores at listing are associated with significantly decreased odds of developing de novo HCC. In fact, patients with lower initial calculated MELD scores had higher odds of developing de novo HCC; thus, it appears that the MELD score should not influence HCC screening and surveillance in patients with end-stage liver disease. This was one of the more intriguing findings of our analysis. Although the ability of the MELD score to predict survival has been validated for various forms of liver disease, the MELD score does not capture all sequelae of liver disease. Given that the median initial MELD in the HCC patients was 13 with an interquartile range of 10 to 16, many patients had a low MELD and likely were listed for transplantation because of more advanced portal hypertension with complications not captured by MELD such as encephalopathy and ascites. Portal hypertension deprives the liver of portal blood flow and promotes liver atrophy. The latter might stimulate the progenitor populations to expand, and HCC derived from those cells would probably grow more quickly and come to clinical attention earlier than tumors that arise from more mature hepatocytes. In support of this hypothesis, it has been shown that the presence of a transjugular portosystemic shunt increases the risk for developing HCC.17 This is an area that clearly deserves further study to confirm and better understand these findings.

There are clear limitations to our study. First, we cannot confirm the diagnosis of HCC for the patients in the analysis. As discussed in the Patients and Methods section, the criteria for listing for HCC have been modified during the MELD era. A recent evaluation of liver transplant recipients who had not received pretransplant ablative treatment revealed that the overall accuracy of radiologic staging was only 44.1%.18 The accuracy is presumably much higher for patients undergoing ablation or embolization, but this finding further highlights the need for more effective screening strategies. Second, we could not accurately determine if the development of HCC was a reason for the removal of patients from the waiting list. Therefore, we were not able to include in our study patients who were removed from the waiting list before undergoing OLT, and it is likely that this limitation may have led to an underestimation of the rate of de novo HCC among patients listed for OLT.

The use of an administrative database also prevented an assessment of the impact of HCC screening and surveillance practices on the development of de novo HCC. Many studies, including the current study, have identified patients that are at increased risk of developing HCC and are therefore candidates for HCC screening. Although patients with cirrhosis due to underlying liver diseases such as PSC, PBC, and AIH have a decreased likelihood of de novo HCC, these patients should still undergo HCC screening. However, patients with factors that are associated with an increased likelihood of developing de novo HCC may benefit from more aggressive HCC screening practices than other patients with end-stage liver disease who do not have factors that further increase their risk for HCC. Unfortunately, the ideal interval and modalities for HCC screening and surveillance in patients with liver disease remain unknown. Knowledge of the intervals and modalities used for HCC screening and surveillance in patients awaiting OLT will be helpful in ensuring more timely diagnosis of HCC, which could reduce the rate of incidental HCC found at explant and prevent patients from being removed from the waiting list because of advanced HCC.

    The full text of this article hosted at iucr.org is unavailable due to technical difficulties.