Abstract
Background and Aims:
Worldwide, hepatocellular carcinoma (HCC) is a common malignancy. We aimed to prospectively determine the incidence and risk factors of HCC in a United States cohort.
Methods:
The multi-center Hepatocellular Carcinoma Early Detection Strategy study of the National Institutes of Health prospectively enrolled patients with cirrhosis who underwent standard surveillance for HCC. Demographics, medical and family history, etiology of liver disease, and clinical features were evaluated for associations with HCC.
Results:
Between 4/10/2013 and 12/31/2021, 1,723 patients were enrolled and confirmed eligible. During median follow up of 2.2 years (range: 0–8.7 years), there were 109 incident cases of HCC for an incidence rate of 2.4 per 100 person-years; 88 (81%) patients with very early/early BCLC stage (0, A), 20 (18%) intermediate stage (B), 1 (1%) unknown stage. Risk factor analyses were restricted to 1,325 patients, including 95 incident HCC, with at least 6 months of follow up. The majority were men (53.2%), obese or severely obese (median body mass index [BMI] 30.2 kg/m2), and white (86.3%); 42.0% had history of HCV infection, 20.7% had alcoholic liver disease (ALD), and 24.9% had nonalcoholic fatty liver disease (NAFLD). Fourteen risk factors for HCC were significant (p<0.05) in univariate analyses, and a multivariate subset was selected by stepwise logistic regression. The multivariate subset contained gender (p<0.001, male, OR=2.47, 95% C.I. 1.54–4.07), years with cirrhosis (p=0.004, OR=1.06, 95% C.I. 1.02–1.1), family history of liver cancer (p=0.02, yes, OR=2.69, 95% C.I. 1.11–5.86), age (per 5-years, p=0.02, OR=1.17, 95% C.I. 1.03–1.33), obesity (p=0.02, yes, OR=1.7, 95% C.I. 1.08–2.73), aspartate aminotransferase (log(1+AST), p=0.06, OR=1.54 95% C.I. 0.97–2.42), alpha-fetoprotein (log(1+AFP), p=0.07, OR=1.32, 95% C.I. 0.97–1.77), and albumin (p=0.10, OR=0.7, 95% C.I. 0.46–1.07).
Conclusions:
Thus far, this is the largest prospective and geographically diverse study of a United States cohort of patients with cirrhosis that validates known risk factors for HCC (gender, age, obesity, years with cirrhosis, family history of liver cancer, baseline AFP, albumin, and AST). The incidence of HCC was 2.4 % per 100 person years.
Keywords: EDRN, Hepatocellular Carcinoma, Cirrhosis, Risk Factors
Lay Summary
The multi-center hepatocellular carcinoma early detection strategy (HEDS) study utilizes the largest, multi-center, geographically diverse and prospective U.S. patient cohort to validate several risk factors for hepatocellular carcinoma.
Graphical Abstract
INTRODUCTION
The burden of hepatocellular carcinoma (HCC) is an important health care problem and continues to be the most common histologic type of primary liver cancer (1–3). Additionally, HCC is the fastest growing cause of cancer-related deaths in the United States (U.S.) (2). The increase in HCC incidence in the U.S. had been explained in large part by increasing rates of hepatitis C virus (HCV)-related HCC (4), but more recently alcoholic liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) are among the most common chronic liver diseases associated with HCC (5, 6). Annual HCC incidence rates among high-risk patients are around 2% per year (7, 8).
Direct acting antivirals (DAAs) have been highly effective in the treatment of HCV (9). However, patients treated with DAAs still remain at risk for HCC even after achieving sustained viral response (SVR) (10–13). Likewise, hepatitis B virus (HBV) patients with or without cirrhosis treated with oral nucleos(t)ides remain at a significant, though decreased, risk for HCC (14–17). Thus, those with chronic viral hepatitis remain under HCC surveillance even after they begin treatment for HBV regardless of fibrosis/cirrhosis, or after having achieved HCV eradication, while they have residual fibrosis/cirrhosis.
With the change in epidemiology of the etiology of cirrhosis from viral hepatitis to non-infectious causes, there is a need to better determine the HCC risk in a contemporary cohort of patients such as those with NAFLD, ALD and treated viral hepatitis. Having a better understanding of the progression to HCC among these patients would allow better risk-stratification, which can improve and streamline prevention and early detection efforts. Previous studies mostly have been retrospective in design (11, 18). Among the prospective studies that have been conducted, only a few are based on study populations in the U.S. (19–21). Furthermore, these past U.S.-based studies are limited in risk factor evaluation of HCC incidence, recognition of low- and high-risk patient groups, and geographic diversity (3).
We followed a prospective cohort of patients with cirrhosis seen in multiple centers in geographically diverse areas of the U.S. to examine the incidence and risk factors for HCC.
METHODS
The Hepatocellular Carcinoma Early Detection Strategy (HEDS) was created as a multi-center study under the NIH-sponsored Early Detection Research Network (EDRN) to explore the risk factors, incidence, and disease course of HCC in the U.S. and focus on developing novel biomarkers for early detection of HCC (22).
The design of the HEDS study and the patient population has been described in detail in a previous publication. Patients were enrolled from 2013–2021 from 7 centers: Mayo Clinic in Rochester, Minnesota, University of Pennsylvania in Philadelphia, Pennsylvania, Mt. Sinai Medical Center in New York City, New York, University of Michigan in Ann Arbor, Michigan, Saint Louis University in Saint Louis, Missouri, Stanford University in Palo Alto, California, and UT Southwestern in Dallas, Texas. The study has been approved by Institutional Review Board (IRB) at all sites and the data coordinating center. The inclusion and exclusion criteria have been described previously (22). Briefly, the inclusion criteria were of a diagnosis of cirrhosis based on histology, clinical features (imaging indicating portal hypertension, splenomegaly and thrombocytopenia in the setting of a chronic liver disease), or by non-invasive testing (elastography and blood testing). In addition, other inclusion criteria included a Model for End-Stage Liver Disease (MELD) score of < 15, no clinically significant hepatic decompensation (Grade 3–4 encephalopathy, refractory ascites, Child-Turcotte-Pugh class C), not listed for liver transplantation (LT), no cancer history within 5 years, no unexplained liver mass, and no significant comorbid conditions with life expectancy of < 1 year.
Longitudinal follow up and Risk Factors
Patients who met the inclusion and exclusion criteria were enrolled and followed prospectively according to the standard of care for cirrhosis at each site. It was agreed that the standard of care would be for patients to be seen every 6 months, and followed until HCC development, LT, or death. HCC surveillance was done according to each site and provider preference and included US and alpha-fetoprotein (AFP), CT or MRI with or without AFP.
At baseline, clinical data that included history of liver disease and its treatment, medical history, family history, medications, and laboratory data for standard clinical care was collected on direct interviews with the patient. Validated questionnaires were given to patients to assess alcohol, tobacco, and coffee/tea exposure and described in detail in a previous publication (22). Physical performance status and body mass index (BMI; calculated from height and weight at enrollment) were collected at the time of enrollment into HEDS. Laboratory and imaging data including US/MRI/CT imaging results, AFP, hepatic biochemical tests, complete blood counts, and MELD scores were obtained. Longitudinally collected data every 6 months at follow up included imaging performed, laboratory data, changes in liver history, medical history, personal cancer history, physical exam (including BMI) and family cancer history. Additionally, patients were asked about changes to substance use and coffee and/or tea consumption at each follow up. Serum and plasma biospecimens were collected at baseline and at each follow up visit longitudinally for research purposes.
Demographics included age, self-reported gender, race and ethnicity. HCV was defined as active (positive for serum HCV RNA) or treated (evidence of SVR in the medical record). Using validated questionnaires, we categorized exposure to alcohol and tobacco as current, former and ever users. Overweight was defined as a BMI between 26–30, obesity was defined as a BMI > 30, and severe obesity as a BMI > 35. Etiology of liver disease was determined by the treating physicians at each site. We defined diabetes based on the use of metformin, hypertension, hyperlipidemia or other diseases based on medical record or as self-reported during enrollment.
Primary Outcome
The primary outcome is the development of HCC after enrollment, and this was defined using the AASLD guidelines (23). The date of HCC confirmation, either by a lesion biopsy or radiologic interpretation as LI-RADS 5 lesion, was used as the HCC diagnosis date. Medical reviews were performed by each site in order to capture incident HCC, LT, and deaths.
Statistical Methods
The annual incidence rate of HCC was estimated as the number of incident HCC cases per 100 person-years observed and reported with 95% Poisson confidence intervals (C.I.) (24, 25). The diagnosed incident cancers were described using the Barcelona Clinic Liver Cancer (BCLC) staging system. To address the possibility that some patients had prevalent HCC cases that were undiagnosed at enrollment, risk factor analyses were restricted to patients with evidence of at least 6 months of study follow-up and no HCC diagnosis within 6 months of enrollment.
Association of clinical characteristics and laboratory markers with later incidence of HCC was examined using Fisher’s exact test and the Wilcoxon test for categorical and continuous variables, respectively. Multivariate associations were examined using forward, backward, and stepwise variable selection followed by logistic regression. Risk factors with p-values less than 0.05 for univariate association with HCC were candidates for selection. Adjusted odds ratios were reported for the multivariate association of selected risk factors. Adjusted odds ratios were also estimated from a dataset with missing variables completed using multiple imputation. The variable selection process was assessed using cross-validation. Sensitivity analyses were conducted for a) including all patients with confirmed HCC and b) excluding patients with HCC confirmed within a year of enrollment.
To account for the risks of LT and death that compete with the risk of HCC, a competing risks analysis was conducted using the method of Fine and Gray (20) for the selected multivariate set of risk factors. This method accounts for observations with small amounts of observed time at risk, and adjusted hazard ratios were estimated from all patients, including any amount of follow-up or timing of HCC diagnosis, with available risk factor data. The competing risk results were used to examine cumulative incidence functions (CIF) of HCC for variables of interest and to plot a distribution of estimated 3-year risks for incident HCC. Concordance statistics (c-statistics) were estimated using the scores from the logistic and competing risk analyses.
RESULTS
A total of 1,723 patients were confirmed to be eligible for analysis, but one patient was excluded due to an unknown date of LT; thus, analysis was restricted to 1,722 patients (Supplemental Figure 1). After a median follow up of 2.20 years (range: 0–8.7 years), a total of 109 patients were determined to have developed HCC. The incident HCC included 88 (81%) patients with very early/early BCLC stage (0, A), 20 (18%) intermediate stage (B), 1 (1%) unknown stage and no known advanced or terminal stage (C, D). The annual incidence rate of HCC was 2.4% (95% C.I. 1.8%−3.1%) during a follow up of 4,510 person years. Among the 1,613 patients without a confirmed diagnosis of HCC, 1,230 had at least 6 months of study follow-up and 95 patients had HCC diagnosis made more than 6 months after enrollment.
Baseline demographic features of the cohort are described in Table 1. In univariate analysis (Table 1), male gender was more common in patients who developed HCC than non-HCC controls (70.5% vs. 51.9%, p <0.001) and those who developed HCC had higher median BMI (32.5 vs. 30.1, p=0.01) and median age (62 vs. 59.5, p=0.03). Obesity was also more common in patients who developed HCC relative to those who did not develop HCC (63.2% vs. 50.3%, p=0.02). History of HCV infection was associated with elevated risk of HCC (56.4% vs 40.9%, p=0.005); there were no significant differences in other etiologies of liver disease between those who developed HCC and those who did not. HCC incidence was the highest among those with etiology of HCV infection, followed by ALD and NAFLD, while infrequent in other conditions such as autoimmune hepatitis which need further study (Supplemental Table 1). We further looked at outcome events based on HCC etiology and the data is shown in Supplemental Table 2. The number of years with cirrhosis (Table 2) also differed significantly between the two groups (median 3.85 vs 2.73 years, p=0.004). Family history of liver cancer was significant (9% vs 3.9%, p=0.05), as was tobacco use (ever vs. never) (p = 0.04).
Table 1.
Demographics and clinical characteristics of cohort at baseline.
| Non-HCC controls (n=1,230) | HCC cases (n=95) | Total (n=1,325) | P value | |
|---|---|---|---|---|
| Male, n (%) | 638 (51.9) | 67 (70.5) | 705 (53.2) | <0.001 |
| BMI (kg/m2) | ||||
| Median (25th percentile, 75th percentile) | 30.1 (26.4, 35.3) | 32.5 (27.6, 37.2) | 30.2 (26.5,35.4) | 0.01 |
| Min, max | 16.0, 71.5 | 18.1, 48.3 | 16.0, 71.5 | |
| BMI Category | ||||
| Normal or Underweight | 210 (17.1) | 9 (9.5) | 219 (16.5) | 0.08 |
| Overweight | 401 (32.6) | 26 (27.4) | 427 (32.3) | |
| Obese | 299 (24.3) | 27 (28.4) | 326 (24.6) | |
| Severely Obese | 319 (26) | 33 (34.7) | 352 (26.6) | |
| Obesity | ||||
| Obese or Severely Obese | 618 (50.3) | 60 (63.2) | 678 (51.2) | 0.02 |
| Race, n (%) | ||||
| White or Caucasian | 1052 (86.3) | 82 (86.3) | 1134(86.3) | 0.44 |
| Black or African American | 84 (6.9) | 9 (9.5) | 93 (7.1) | |
| Asian | 27 (2.2) | 0 (0) | 27 (2.1) | |
| Other | 56 (4.6) | 4 (4.2) | 60 (4.6) | |
| Missing | 11 | 0 | 11 | |
| Ethnicity, n (%) | ||||
| Hispanic or Latino | 117 (9.5) | 9 (9.6) | 126 (9.5) | >0.99 |
| Age (y) | ||||
| Median (25th percentile, 75th percentile) | 59.5 (54, 65) | 62 (56, 66) | 60 (54, 65) | 0.03 |
| Min, max | 19.0, 87.0 | 27.0,77.0 | 19.0, 87.0 | |
| Etiology of liver disease, n (%) | ||||
| HCV infection | 498 (40.9) | 53 (56.4) | 551 (42) | 0.005 |
| Treated for HCV | 344 (80.2) | 39 (88.6) | 383 (81) | 0.23 |
| Missing | 69 | 9 | 78 | |
| with DAA | 93 (27.2) | 10 (25.6) | 103 (27) | >0.99 |
| with Interferon | 203 (59.4) | 24 (61.5) | 227 (59.6) | 0.86 |
| SVR attained | 168 (51.7) | 21 (58.3) | 189 (52.4) | 0.49* |
| Missing | 19 | 3 | 22 | |
| HBV infection | 29 (2.4) | 0 (0) | 29 (2.2) | 0.26 |
| HBV therapy exposed | 18 (64.3) | 0 (0) | 18 (64.3) | >0.99 |
| NAFLD | 308 (25.3) | 19 (20.2) | 327 (24.9) | 0.32 |
| Alcoholic Liver Disease | 254 (20.8) | 18 (19.1) | 272 (20.7) | 0.79 |
| Autoimmune hepatitis | 70 (5.7) | 2 (2.1) | 72 (5.5) | 0.16 |
| Cholestatic liver disease (PBC, PSC) | 94 (7.7) | 6 (6.4) | 100 (7.6) | 0.84 |
| Other (Hemochromatosis, Wilson’s, etc.) | 102 (8.4) | 4 (4.3) | 106 (8.1) | 0.23 |
| Missing | 11 | 1 | 12 | |
| Alcohol use, n current (%) | 119 (9.9) | 10 (10.6) | 129 (10) | 0.86 |
| Missing | 28 | 1 | 29 | |
| Tobacco use, n (%) | ||||
| Current | 184 (15) | 15 (15.8) | 199 (15.1) | 0.88 |
| Former | 492 (40.2) | 48 (50.5) | 540 (40.9) | 0.05 |
| Ever | 676 (55.2) | 63 (66.3) | 739 (56) | 0.04 |
| Fully active performance status, n (%) | 833 (67.7) | 65 (68.4) | 898 (67.8) | >0.99 |
| Medications of interest, n ever taken (%) | ||||
| Statins | 326 (27.2) | 25 (26.6) | 351 (27.2) | >0.99 |
| Missing | 32 | 1 | 33 | |
| Metformin | 170 (25.1) | 10 (22.2) | 180 (24.9) | 0.86 |
| Missing | 552 | 550 | 602 | |
| Coffee and tea consumption, n (%) | 884 (81.1) | 66 (81.5) | 950 (81.1) | >0.99 |
| Missing | 140 | 14 | 154 | |
| Cancer history (excl. liver), n (%) | 85 (6.9) | 6 (6.4) | 91 (6.9) | >0.99 |
| Family cancer history, n (%) | ||||
| Liver cancer | 47 (3.9) | 8 (9) | 55 (4.3) | 0.05 |
| Other cancer | 629 (52.2) | 46 (51.7) | 675 (57.6) | >0.99 |
| Missing | 25 | 6 | 31 | |
| Family liver disease history, n (%) | 234 (19.8) | 23 (25.3) | 257 (20.2) | 0.22 |
| Missing | 47 | 4 | 51 | |
Note: Percentages for subvariables are of the subset (e.g. 344 non-HCC controls treated for HCV comprise 80.2% of the 498 with HCV). Percentages (%) for a categorical variable do not include observations with missing values. Variables with 1 (BMI, HBV therapy), 2 (HCV treatment type) 3 (Ethnicity, Cancer history), or 5 (Tobacco use) missing values are not reported above. Considering additional decimal places, the p-value for family history of liver cancer is 0.049 and thus met the univariate variable selection criteria (< 0.05).
p-value is for comparison of attained vs. not attained SVR in those treated for HCV
Table 2.
HCC vs. non-HCC: Clinical and Laboratory data at baseline.
| Non-HCC controls At baseline | HCC cases At baseline | p value | |
|---|---|---|---|
| LABORATORY ASSESSMENTS: mean, median (25th, 75th percentile) | |||
|
| |||
| AFP | 7.2, 3.7 (2.4, 6.1) | 9, 5.2 (3.3, 8.9) | <0.001 |
|
| |||
| Albumin | 3.8, 3.9 (3.5, 4.2) | 3.6, 3.6 (3.3, 3.9) | 0.001 |
|
| |||
| ALT | 44, 33 (23, 50) | 54.6, 36 (26.5, 52) | 0.05 |
|
| |||
| AST | 52.3, 40 (29, 60) | 61.7, 50 (31, 74) | 0.01 |
|
| |||
| Alkaline phosphatase | 122, 101 (76, 143) | 121.5, 111 (82, 150) | 0.28 |
|
| |||
| Bilirubin | 1.32, 1 (0.7, 1.7) | 1.18, 1.3 (0.5, 1.65) | 0.86 |
|
| |||
| Direct Bilirubin Missing | 0.47, 0.3 (0.2, 0.6) 603 | 0.46, 0.4 (0.2, 0.6) 46 | 0.64 |
|
| |||
| Bilirubin Total | 1.15, 0.9 (0.6, 1.5) | 1.26, 1.2 (0.8, 1.6) | 0.01 |
|
| |||
| Creatinine | 0.92, 0.83 (0.7, 1) | 0.91, 0.83 (0.74, 1.02) | 0.48 |
|
| |||
| INR | 1.14, 1.1 (1, 1.2) | 1.18, 1.2 (1.1, 1.28) | 0.004 |
|
| |||
| Hematocrit | 39.3, 39.5 (36.1, 42.8) | 38.8 ,38.6 (35.6, 42.1) | 0.23 |
|
| |||
| Hemoglobin | 13.3, 13.3 (12.1, 14.5) | 13.1, 13.1 (12, 14.4) | 0.41 |
|
| |||
| Platelets | 120.1, 109 (75, 154) | 104.7, 94 (55.5, 137) | 0.002 |
|
| |||
| MELD score | 9, 8 (7, 11) | 9.6, 10 (8, 11) | 0.004 |
|
| |||
| Years with Cirrhosis Missing | 4.4, 2.7 (1.1, 6.2) 29 | 6.1, 3.8 (1.6, 8.8) 1 | 0.004 |
|
| |||
|
| |||
| CLINICAL ASSESSMENTS, n (%) | |||
|
| |||
| CTP | |||
| Class A | 919 (74.8) | 63 (67) | 0.11 |
| Class B | 310 (25.2) | 31 (33) | |
|
| |||
| Ascites | |||
| No | 882 (71.8) | 63 (66.3) | 0.30 |
| Controlled | 346 (28.2) | 32 (33.7) | |
| Uncontrolled | 0 (0.0) | 0 (0.0) | |
|
| |||
| Hepatic encephalopathy | |||
| No | 985 (80.1) | 73 (77.7) | 0.60 |
| Controlled | 244 (19.9) | 21 (22.3) | |
| Uncontrolled | 0 (0.0%) | 0 (0.0%) | |
|
| |||
| Varices | |||
| Esophageal | 602 (49.3) | 49 (52.7) | 0.59 |
| Gastric | 99 (8.1) | 8 (8.6) | 0.84 |
| Intra-abdominal | 109 (9.2) | 11 (11.8) | 0.36 |
Note: The numbers of patients without values for Direct bilirubin and years with cirrhosis at enrollment are reported above. Among non-HCC patients, 10,14, and 43 were missing information for esophageal, gastric, and intra-abdominal varices, respectively, and 2 HCC patients for each. All other clinical and laboratory variables above had at most 3 missing values among non-HCC controls and at most one missing value among HCC cases.
Black/African American Race or Hispanic Ethnicity was not a significant difference for risk of HCC (p=0.44 and p>0.99). HCV treatment was not significantly associated with a lower risk of developing HCC (p=0.23). Alcohol use, fully active performance status, other medications taken, coffee and tea consumption, and other family cancer history were not significant factors for the development of HCC.
Baseline clinical and laboratory data were also assessed for this study cohort as shown in Table 2. The median albumin was lower (p=0.001) and median AFP and total bilirubin were higher (p<0.001 and p=0.01, respectively) in those who developed HCC. Additionally, aspartate aminotransferase (AST) was higher in HCC cases than non-HCC controls (p=0.01). Internalized normalized ratio (INR) was higher (p=0.004), and platelets were lower (p=0.002) in the HCC group. The HCC group had higher median MELD score (10.0 vs. 8.0, p=0.004). There was no significant difference in rates of ascites, hepatic encephalopathy, and varices between the two groups. There was heterogeneity in the imaging modality used to confirm absence of HCC at baseline (Table 3). We report censoring and competing events of LT, death, or HCC diagnosis by time since study entry. We found that most LT, death, and HCC events took place within the first 5 years of enrollment (Table 4).
Table 3.
Imaging modality at baseline.
| Time on Study | US** | CT | MRI | Unknown | Total | |
|---|---|---|---|---|---|---|
| No HCC | 1024 (63.5%) 184 (48%) 840 (68.3%) |
279 (17.3%) 103 (26.9%) 176 (14.3%) |
309 (19.2%) 96 (25.1%) 213 (17.3%) |
1 (0.1%) (0%) (0.1%) |
1613 383 1230 |
|
| < 6mos* | ||||||
| ≥ 6mos | ||||||
| HCC | 64 (58.7%) 5 (38.5%) 59 (61.5%) |
19 (17.4%) 5 (38.5%) 14 (14.6%) |
25 (22.9%) 2 (15.4%) 23 (24%) |
1 (0.9%) 1 (7.7%) 0 (0%) |
109 13 96 |
|
| < 6mos* | ||||||
| ≥ 6mos | ||||||
| Total | 1088 (63.2%) | 298 (17.3%) | 334 (19.4%) | 2 (0.1%) | 1722 |
These patients were not included in primary analyses.
Two patients had ultrasound (US) at baseline together with either MRI (1 without HCC) or CT (1 with HCC) and were counted in the MRI or CT category, respectively.
Table 4.
Timing of study exit from censoring, competing events of liver transplant (LT), death, or HCC diagnosis.
| Years | Censor | Liver Transplant | Death | HCC | Total |
|---|---|---|---|---|---|
| Baseline – 0.5 * | 360 | 5 | 18 | 14 | 397 |
| 0.5 – 1 | 131 | 10 | 20 | 9 | 170 |
| 1 – 2 | 169 | 9 | 46 | 26 | 250 |
| 2 – 3 | 135 | 11 | 50 | 16 | 212 |
| 3 – 4 | 140 | 12 | 36 | 12 | 200 |
| 4 – 5 | 140 | 7 | 32 | 9 | 188 |
| 5 – 6 | 109 | 1 | 20 | 15 | 145 |
| 6 – 7 | 63 | 2 | 8 | 4 | 77 |
| 7 – 8 | 60 | 0 | 4 | 2 | 66 |
| 8 – 9 | 15 | 0 | 0 | 2 | 17 |
| Total | 1322 | 57 | 234 | 109 | 1722 |
These patients were not included in primary analyses.
The dichotomization of BMI into obesity (yes/no) is clinically appealing, differentiates HCC at similar levels of significance (p=0.019 vs p=0.015), and, like BMI, accounts for only one degree of freedom in a model. Dichotomizations at clinically familiar levels of overweight and severely obese demonstrated less significance (p=0.06 and 0.07, respectively). Hence, for multivariate analysis, we consider the following 14 clinical characteristics and laboratory markers with univariate association with HCC (p-value < 0.05): gender, age, obesity, smoking history (ever vs never smoked), family history of liver cancer, AFP, albumin, AST, total bilirubin, INR, platelet count, MELD score, history of HCV, and years with cirrhosis. AFP and AST were log transformed. From this list of candidate risk factors, forward, backward, and bi-directional stepwise logistic regressions selected the same set of complementary eight risk factors: gender, age, obesity, family history of liver cancer, years with cirrhosis, AST, AFP, and albumin.
Eighty-four patients (9 with HCC) did not have values for all fourteen candidate predictors. Individual predictors were missing for fewer than 0.4% of patients for all candidates except family history of liver cancer (31 patients, 2.3%), years with cirrhosis (30 patients, 2.3%), etiology of liver disease and hence history of HCV (12 patients, 0.4%). Variable selection was performed on the subset complete for all candidate variables and the multivariate models fit on the data set complete for the 8 selected variables (1172 non-HCC and 88 HCC patients). Odds ratios and hazard ratios from the respective multivariate logistic regression and competing risk regressions are shown in Table 5. The logistic model showed that a higher risk of HCC was associated with being male (OR 2.47, 95% C.I. 1.54–4.07), obesity (OR 1.7, 95% C.I. 1.08–2.73), older age (per 5-years OR 1.17, 95% C.I. 1.03–1.33), family history of liver cancer (OR 2.69, 95% C.I. 1.11–5.86), and years with cirrhosis (OR 1.06, 95% C.I. 1.02–1.1). Associations not remaining significant at the 5% level in multivariate analysis suggested a lower risk of HCC to be associated with higher serum albumin (OR 0.7, 95% C.I. 0.46–1.07), and elevated risk to be associated with higher AFP (OR 1.32, 95% C.I. 0.97–1.77) and higher AST (OR 1.54, 95% C.I. 0.97–2.42). The c-statistic, which describes a risk stratification through its interpretation as the probability that the score of an HCC patient is greater than the score of a patient not diagnosed with HCC (over all such pairs), was 0.73 (Table 5). The logistic regression fit to data with missing covariates imputed yielded similar results (Supplemental Table 3). Hazard ratios and the c-statistic from the competing risk analysis, fit on 1,576 patients including 102 HCC cases, were consistent with the results of logistic regression model (Table 5). Estimated coefficients for the two multivariate regressions are in Supplemental Tables 4 and 5.
Table 5.
Multivariate associations with incident HCC in conditional regression models using logistic and competing risk time-to-event methods.
| Logistic Regression 88 HCC 1172 No HCC | Competing Risks 102 HCC 1474 Non-HCC | |
|---|---|---|
| Adjusted OR (95% CI) | Adjusted HR (95% CI) | |
| Male | 2.47 (1.54,4.07) | 2.66 (1.72,4.12) |
| Years with cirrhosis | 1.06 (1.02,1.1) | 1.05 (1,1.09) |
| Family History of Liver Cancer | 2.69 (1.11,5.86) | 1.98 (0.98,4) |
| Age (5yrs) | 1.17 (1.03,1.33) | 1.17 (1.03,1.32) |
| Obese | 1.7 (1.08,2.73) | 1.46 (0.97,2.21) |
| log(1+AST) | 1.54 (0.97,2.42) | 1.38 (0.91,2.08) |
| log(1+AFP) | 1.32 (0.97,1.77) | 1.41 (1.1,1.8) |
| Albumin | 0.7 (0.46,1.07) | 0.55 (0.38,0.8) |
| C-Statistic | 0.73 | 0.75 |
Note: Odds ratios (OR) and hazard ratios (HR) with their 95% confidence intervals for the logistic regression and competing risk regression, respectively. Both models were fit to the data complete for the 8 variables selected by stepwise, backward, and forward selection. Logistic regression was further restricted to patients with at least 6 months of follow-up (which excludes HCC cases diagnosed within 6 months of follow-up). The competing risks for HCC are death and liver transplantation. Point estimates and inferential details are included in Supplemental Tables 3 and 5.
Cross-validation of the variable selection procedure showed that gender, age, obesity, and years with cirrhosis were selected, on average, in 89.2–100% of the 1,000 train-test splits, whereas the unselected 6 of 14 candidate variables were only selected 0.6 – 38% of training sets. AFP, albumin and AST, missing 5% significance in both multivariate models, were respectively selected in 54.8, 50.4, and 71.7% of training sets; family history of liver cancer in 63.5% of training sets (Supplemental Table 6). Sensitivity analyses for including all HCC diagnosed after enrollment or only those more than 6 or 12 months after enrollment show that six risk factors (gender, years with cirrhosis, family history of liver cancer, age, obesity, and AST) are consistently selected with similar odds ratios in all three scenarios (Supplemental Table 7). The risk of HCC associated with being male diminishes somewhat with a longer duration before diagnosis whereas the estimate and significance of AST as a risk factor increases. AFP and albumin were no longer selected when cases diagnosed within 12 months were excluded. Smoking history (ever/never) was selected and marginally significant only when all HCC cases were considered. Both HCV etiology and INR were selected only when all HCC cases diagnosed within 12 months of enrollment were excluded.
The CIF curves illustrate the estimated changes in the hazard for each risk factor relative to a reference patient who is a 59.5 year-old female, without obesity or family history of liver cancer, was diagnosed with cirrhosis 2.7 years ago, and has most recent labs of AFP=4.7, albumin=3.9, and AST=41 (Figure 1). An estimated 1.6% of such patients would be diagnosed with HCC within 3 years. 4.1% of male patients, with all other characteristics the same as the reference group, are expected to be diagnosed within 3-years, whereas 0.9% of patients like the reference group but with albumin=4.81 are expected to have an HCC diagnosis in the same time period.
Figure 1.
Cumulative incidence function curves indicating the probability of HCC over time corresponding to a single characteristic change of a reference patient (black line) who is a 59.5 year-old female, not obese, has no family history of liver cancer, was diagnosed with cirrhosis 2.7 years ago, with most recent labs: AFP=4.7, albumin=3.9 and AST=41 (quantitative variables are median values observed among the non-HCC patients included in primary analysis).
The distribution of 3-year risks of HCC estimated from the competing risk analysis demonstrates that most patients (77.6%) have risks less than that implied by the observed incidence (2.4% for 1 year or 7.3% over 3 years) while 11.6% have 3-year risks between 10% and 47% (Supplemental Figure 2). The cumulative incidence functions for patients stratified by quartiles of risk score demarcate distinct high and low risk groups, and the middle two quartiles comprise an intermediate risk group (Figure 2).
Figure 2.
Cumulative incidence function curves indicating the probability of HCC over time corresponding to quartiles of risk scores from the competing risk analysis. The first quartile and the fourth quartile demarcate distinct low and high risk groups and the middle two quartiles comprise an intermediate risk group.
DISCUSSION
HCC continues to be a major malignancy worldwide while the incidence has been variable. Several patient characteristics and the rigor of surveillance may influence the detection of HCC and thus the incidence. Chronic HBV and HCV are risk factors for HCC development, with the risk increasing substantially in patients with cirrhosis. According to a systematic review conducted in Asia, HCC incidence rates per 100 person-years were 0.2 in inactive HBV carriers, 0.6 in noncirrhotic patients with chronic HBV, and 3.7 in cirrhotic patients with chronic HBV (26). In cohort studies of HCV, the annual incidence of HCC ranged between 1% and 3%, with the rate increasing to between 1% and 8% after HCV-related cirrhosis developed (21, 27).
Most studies have been retrospective in nature (11, 18), while our study is unique in that it is a large cohort of prospectively followed patients with cirrhosis at risk for HCC in the U.S. Only a few prospective studies on U.S. patient populations have been conducted, focused primarily on age, gender, biomarkers, and cirrhosis etiology as risk factors for HCC incidence (3, 19–21). While these previous studies have developed high-sensitivity HCC detection systems such as GALAD (gender x age x AFP x AFP-L3 x des-gamma-carboxy prothrombin) and the HCC Early Detection Screening algorithm, our large prospectively followed and geographically diverse U.S. cohort of patients with cirrhosis demonstrates a more comprehensive approach to early detection of HCC that includes family history as a risk factor and develops risk stratification measures (19, 20). We found that gender, age, obesity, years with cirrhosis, family history of liver cancer, albumin, AST, and AFP were a complementary set of risk factors of HCC by multivariate analysis. In addition, smoking history (ever vs. never), total bilirubin, INR, platelet count, MELD score, and history of HCV were significant in univariate analyses. We were also able to stratify patients into high- and low-risk groups, which has implications for early detection and prevention of HCC (Figure 2). Our patient population demonstrated an HCC incidence of 2.4% during a follow up of 4,510 person years, which is comparable to HCC incidence rates (1.82%) in recent observations in the U.S. (21).
NAFLD-associated liver disease has also been a major contributor to the pool of patients at risk for HCC. Up to 25% of patients with NAFLD can progress to nonalcoholic steatohepatitis (NASH), up to 20% of patients with NASH have cirrhosis, and NASH-associated cirrhosis has been noted to have a 2.6% incidence of HCC (28–30). Further, in a retrospective cohort study, 34.6% of NAFLD patients with HCC did not have any evidence of cirrhosis, indicating that the risk for NAFLD-associated HCC exists independently of cirrhosis (31). NAFLD is becoming a leading cause of cirrhosis leading to HCC (32).
Several studies have demonstrated an association between obesity and/or metabolic syndrome with HCC (33, 34). A meta-analysis noted the incidence of HCC among persons with NAFLD to be low at 0.44 per 1000 person years (range, 0.29–0.66), relative to those with NASH where the annual incidence has been noted to be over 10-fold higher at 5.29 per 1000 person years (95% CI: 0.75–37.56) (35). These incidence rates are of concern in the context of the rising global burden of obesity. In contrast, HCV can now be effectively cured with resultant interruption in disease progression and decreased risk of progression to HCC. Our study validates in a prospectively followed, large, and geographically diverse cohort that obesity was a strong predictor of development of HCC (p=0.02, OR=1.7, 95% CI: 1.08–2.73). This observation reiterates the importance of the need for structured lifestyle interventions to achieve weight loss in obese patients to mitigate the risk of HCC. A decreased risk of HCC in patients achieving SVR after HCV treatment has been demonstrated in studies evaluating both interferon-based and DAA therapy for HCV (11, 36). However, it is recognized that despite SVR, there remains a late onset risk of HCC in patients with baseline cirrhosis (37). We did not find that treated HCV impacted the risk of HCC, which may be a reflection of a small number of HCV patients successfully treated in this cohort.
The observations of lower risk in women versus higher risk with advancing age and a family history of HCC have been noted in multiple studies while ours is a prospective validation of these observations (38). We did not find Black Race or Hispanic Ethnicity to be a risk factor, as had been observed previously (39) nor in a recent prospective study (40). The lack of demonstration of a relationship between the incidence of HCC and Race/Ethnicity is very likely a reflection of the small numbers of minority patients enrolled in our study. Additional studies with a more balanced distribution of race and ethnicity are required to accurately assess risk of HCC in these populations.
AFP has been consistently demonstrated to be associated with an increased risk of developing HCC among patients with cirrhosis (41, 42). Our study noted AFP to be a predictor of HCC development in patients with cirrhosis across multiple etiologies of liver disease. This is different from its role in surveillance for the detection of early stage HCC and for determining prognosis after the development of HCC. Thus, one could speculate that AFP is also a biomarker of the underlying “biological state” and through several unknown mechanisms predisposes patients to HCC.
There are several strengths of the study, including the prospective data collection in real time from a large U.S. cohort. Importantly, we were able to evaluate demographics, important risk factors (alcohol, tobacco, coffee, while not collecting data on aspirin use), etiology of liver disease, antiviral therapy for viral hepatitis, laboratory data, liver function, and family history. We believe that this is the most thorough evaluation of HCC risk factors done in a prospective manner. The lack of adequate representation of populations representing diverse race/ethnicity backgrounds is a weakness and brings to attention the need to include these groups in future studies.
In conclusion, in a thoroughly conducted prospective cohort of U.S. patients at risk for HCC, the incidence was 2.4 per 100 person-years over follow up of 4,510 person years. The complementary set of predictors for HCC were gender, age, obesity, years with cirrhosis, family history of liver cancer, baseline AFP, albumin, and AST.
Supplementary Material
What You Need to Know Summary.
Background and Context
Hepatocellular carcinoma is a major malignancy worldwide, but its incidence and risk factors have not been investigated with respect to patient geographic diversity in the United States.
New Findings
During a median follow-up of 2.2 years between 2013 and 2021, the incidence of HCC in the largest, multi-center, geographically diverse U.S.- based study was found to be 2.4% per 100 person-years.
Limitations
The conclusions in this study are limited by the small numbers of minority patients enrolled, and a clinical evaluation of our models is beyond the scope of this paper.
Clinical Research Relevance
Most studies on the incidence and risk factors of hepatocellular carcinoma have been retrospective in nature. Thus far, our study is the largest prospective and geographically diverse investigation of a U.S. cohort of patients that validates several known risk factors for hepatocellular carcinoma, including male gender, older age, low albumin, and obesity.
Basic Research Relevance
While our study focuses on the clinical detection and manifestations of hepatocellular carcinoma, it calls to attention the need to further investigate the physiological and biochemical mechanisms behind the unique roles of newer biomarkers in early detection and development of hepatocellular carcinoma. Understanding these mechanisms will bring healthcare providers closer to earlier diagnosis of HCC in those who are at high risk for this malignancy.
Acknowledgments
Opinions expressed by the authors are their own and this material should not be interpreted as representing the official viewpoint of the U.S. Department of Health and Human Services, the National Institutes of Health, or the National Cancer Institute. The HEDS cohort has an associated biorepository for blinded validation of novel biomarkers, and as such we will not be sharing data or other study materials. We would like to acknowledge the support for this work: NIH R01 CA237659-01 and U01 CA086402. We would also like to thank Suditi Rahematpura and Bethany Nahri So for their assistance with manuscript formatting.
Grant Support:
NIH R01 CA237659-01, U01 CA086402
Abbreviations:
- ALT
alanine aminotransferase
- ALD
alcoholic liver disease
- AFP
alpha-fetoprotein
- AST
aspartate aminotransferase
- BMI
body mass index
- DAAs
direct acting antivirals
- CIF
cumulative incidence function
- U.S.
United States
- EDRN
Early Detection Research Network
- HBV
hepatitis B virus
- HCV
hepatitis C virus
- HCC
hepatocellular carcinoma
- HEDS
Hepatocellular Carcinoma Early Detection Strategy
- MELD
Model for End-Stage Liver Disease
- BCLC
Barcelona Clinic Liver Cancer
- LT
liver transplantation
- INR
internalized normalized ratio
- NAFLD
nonalcoholic fatty liver disease
- NASH
nonalcoholic steatohepatitis
- SVR
sustained viral response
Footnotes
Data Statement: The HEDS cohort has an associated biorepository for blinded validation of novel biomarkers, and as such we will not be sharing data or other study materials.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Massarweh NN, El-Serag HB. Epidemiology of Hepatocellular Carcinoma and Intrahepatic Cholangiocarcinoma. Cancer Control 2017;24:1073274817729245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ryerson AB, Eheman CR, Altekruse SF, Ward JW, Jemal A, Sherman RL, Henley SJ, et al. Annual Report to the Nation on the Status of Cancer, 1975–2012, featuring the increasing incidence of liver cancer. Cancer 2016;122:1312–1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Singal AG, El-Serag HB. Hepatocellular Carcinoma From Epidemiology to Prevention: Translating Knowledge into Practice. Clin Gastroenterol Hepatol 2015;13:2140–2151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Davila JA, Morgan RO, Shaib Y, McGlynn KA, El-Serag HB. Hepatitis C infection and the increasing incidence of hepatocellular carcinoma: a population-based study. Gastroenterology 2004;127:1372–1380. [DOI] [PubMed] [Google Scholar]
- 5.Cabrera R, Singal AG, Colombo M, Kelley RK, Lee H, Mospan AR, Meyer T, et al. A Real-World Observational Cohort of Patients with Hepatocellular Carcinoma: Design and Rationale for TARGET-HCC. Hepatol Commun 2021;5:538–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Huang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol 2021;18:223–238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Howe HL, Wu X, Ries LA, Cokkinides V, Ahmed F, Jemal A, Miller B, et al. Annual report to the nation on the status of cancer, 1975–2003, featuring cancer among U.S. Hispanic/Latino populations. Cancer 2006;107:1711–1742. [DOI] [PubMed] [Google Scholar]
- 8.Villanueva A Hepatocellular Carcinoma. N Engl J Med 2019;380:1450–1462. [DOI] [PubMed] [Google Scholar]
- 9.Maan R, van Tilborg M, Deterding K, Ramji A, van der Meer AJ, Wong F, Fung S, et al. Safety and Effectiveness of Direct-Acting Antiviral Agents for Treatment of Patients With Chronic Hepatitis C Virus Infection and Cirrhosis. Clin Gastroenterol Hepatol 2016;14:1821–1830 e1826. [DOI] [PubMed] [Google Scholar]
- 10.Backus LI, Belperio PS, Shahoumian TA, Mole LA. Impact of Sustained Virologic Response with Direct-Acting Antiviral Treatment on Mortality in Patients with Advanced Liver Disease. Hepatology 2019;69:487–497. [DOI] [PubMed] [Google Scholar]
- 11.Kanwal F, Kramer J, Asch SM, Chayanupatkul M, Cao Y, El-Serag HB. Risk of Hepatocellular Cancer in HCV Patients Treated With Direct-Acting Antiviral Agents. Gastroenterology 2017;153:996–1005 e1001. [DOI] [PubMed] [Google Scholar]
- 12.Kutala B, Valla D, Marcellin P. Things fall apart with hepatocellular carcinoma and direct-acting antivirals. Expert Opin Drug Saf 2018;17:107–109. [DOI] [PubMed] [Google Scholar]
- 13.Reig M, Marino Z, Perello C, Inarrairaegui M, Ribeiro A, Lens S, Diaz A, et al. Unexpected high rate of early tumor recurrence in patients with HCV-related HCC undergoing interferon-free therapy. J Hepatol 2016;65:719–726. [DOI] [PubMed] [Google Scholar]
- 14.Ahn J, Lee HM, Lim JK, Pan CQ, Nguyen MH, Ray Kim W, Mannalithara A, et al. Entecavir safety and effectiveness in a national cohort of treatment-naive chronic hepatitis B patients in the US - the ENUMERATE study. Aliment Pharmacol Ther 2016;43:134–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hoang JK, Yang HI, Le A, Nguyen NH, Lin D, Vu VD, Chaung K, et al. Lower liver cancer risk with antiviral therapy in chronic hepatitis B patients with normal to minimally elevated ALT and no cirrhosis. Medicine (Baltimore) 2016;95:e4433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lin D, Yang HI, Nguyen N, Hoang J, Kim Y, Vu V, Le A, et al. Reduction of chronic hepatitis B-related hepatocellular carcinoma with anti-viral therapy, including low risk patients. Aliment Pharmacol Ther 2016;44:846–855. [DOI] [PubMed] [Google Scholar]
- 17.Lok AS, McMahon BJ, Brown RS Jr., Wong JB, Ahmed AT, Farah W, Almasri J, et al. Antiviral therapy for chronic hepatitis B viral infection in adults: A systematic review and meta-analysis. Hepatology 2016;63:284–306. [DOI] [PubMed] [Google Scholar]
- 18.Kanwal F, Kramer JR, Mapakshi S, Natarajan Y, Chayanupatkul M, Richardson PA, Li L, et al. Risk of Hepatocellular Cancer in Patients With Non-Alcoholic Fatty Liver Disease. Gastroenterology 2018;155:1828–1837 e1822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Singal AG, Tayob N, Mehta A, Marrero JA, El-Serag H, Jin Q, Saenz de Viteri C, et al. GALAD demonstrates high sensitivity for HCC surveillance in a cohort of patients with cirrhosis. Hepatology 2022;75:541–549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tayob N, Kanwal F, Alsarraj A, Hernaez R, El-Serag HB. The Performance of AFP, AFP-3, DCP as Biomarkers for Detection of Hepatocellular Carcinoma (HCC): A Phase 3 Biomarker Study in the United States. Clin Gastroenterol Hepatol 2023;21:415–423 e414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kanwal F, Khaderi S, Singal AG, Marrero JA, Loo N, Asrani SK, Amos CI, et al. Risk factors for HCC in contemporary cohorts of patients with cirrhosis. Hepatology 2023;77:997–1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Borges KA, Dai J, Parikh ND, Schwartz M, Nguyen MH, Roberts LR, Befeler AS, et al. Rationale and design of the Hepatocellular carcinoma Early Detection Strategy study: A multi-center longitudinal initiative of the National Cancer Institute’s Early Detection Research Network. Contemp Clin Trials 2019;76:49–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Marrero JA, Kulik LM, Sirlin CB, Zhu AX, Finn RS, Abecassis MM, Roberts LR, et al. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology 2018;68:723–750. [DOI] [PubMed] [Google Scholar]
- 24.Ulm K A simple method to calculate the confidence interval of a standardized mortality ratio (SMR). Am J Epidemiol 1990;131:373–375. [DOI] [PubMed] [Google Scholar]
- 25.Johnson NL, Kitz S. Discrete distributions. New York: John Wiley and Sons, 1969. [Google Scholar]
- 26.Fattovich G, Bortolotti F, Donato F. Natural history of chronic hepatitis B: special emphasis on disease progression and prognostic factors. J Hepatol 2008;48:335–352. [DOI] [PubMed] [Google Scholar]
- 27.El-Serag HB, Kanwal F. Epidemiology of hepatocellular carcinoma in the United States: where are we? Where do we go? Hepatology 2014;60:1767–1775. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Khan FZ, Perumpail RB, Wong RJ, Ahmed A. Advances in hepatocellular carcinoma: Nonalcoholic steatohepatitis-related hepatocellular carcinoma. World J Hepatol 2015;7:2155–2161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.McCullough AJ. The clinical features, diagnosis and natural history of nonalcoholic fatty liver disease. Clin Liver Dis 2004;8:521–533, viii. [DOI] [PubMed] [Google Scholar]
- 30.Ascha MS, Hanouneh IA, Lopez R, Tamimi TA, Feldstein AF, Zein NN. The incidence and risk factors of hepatocellular carcinoma in patients with nonalcoholic steatohepatitis. Hepatology 2010;51:1972–1978. [DOI] [PubMed] [Google Scholar]
- 31.Mittal S, El-Serag HB, Sada YH, Kanwal F, Duan Z, Temple S, May SB, et al. Hepatocellular Carcinoma in the Absence of Cirrhosis in United States Veterans is Associated With Nonalcoholic Fatty Liver Disease. Clin Gastroenterol Hepatol 2016;14:124–131 e121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Geh D, Manas DM, Reeves HL. Hepatocellular carcinoma in non-alcoholic fatty liver disease-a review of an emerging challenge facing clinicians. Hepatobiliary Surg Nutr 2021;10:59–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Jinjuvadia R, Patel S, Liangpunsakul S. The association between metabolic syndrome and hepatocellular carcinoma: systemic review and meta-analysis. J Clin Gastroenterol 2014;48:172–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Saunders D, Seidel D, Allison M, Lyratzopoulos G. Systematic review: the association between obesity and hepatocellular carcinoma - epidemiological evidence. Aliment Pharmacol Ther 2010;31:1051–1063. [DOI] [PubMed] [Google Scholar]
- 35.Younossi ZM, Koenig AB, Abdelatif D, Fazel Y, Henry L, Wymer M. Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology 2016;64:73–84. [DOI] [PubMed] [Google Scholar]
- 36.van der Meer AJ, Veldt BJ, Feld JJ, Wedemeyer H, Dufour JF, Lammert F, Duarte-Rojo A, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA 2012;308:2584–2593. [DOI] [PubMed] [Google Scholar]
- 37.Ioannou GN, Beste LA, Green PK, Singal AG, Tapper EB, Waljee AK, Sterling RK, et al. Increased Risk for Hepatocellular Carcinoma Persists Up to 10 Years After HCV Eradication in Patients With Baseline Cirrhosis or High FIB-4 Scores. Gastroenterology 2019;157:1264–1278 e1264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Liu P, Xie SH, Hu S, Cheng X, Gao T, Zhang C, Song Z. Age-specific sex difference in the incidence of hepatocellular carcinoma in the United States. Oncotarget 2017;8:68131–68137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Younossi ZM, Stepanova M, Negro F, Hallaji S, Younossi Y, Lam B, Srishord M. Nonalcoholic fatty liver disease in lean individuals in the United States. Medicine (Baltimore) 2012;91:319–327. [DOI] [PubMed] [Google Scholar]
- 40.Kanwal F, Khaderi S, Singal AG, Marrero JA, Loo N, Asrani SK, Amos CI, et al. Risk Factors for Hepatocellular Cancer in Contemporary Cohorts of Patients with Cirrhosis. Hepatology 2023;77(3):997–1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Marrero JA, Feng Z, Wang Y, Nguyen MH, Befeler AS, Roberts LR, Reddy KR, et al. Alpha-fetoprotein, des-gamma carboxyprothrombin, and lectin-bound alpha-fetoprotein in early hepatocellular carcinoma. Gastroenterology 2009;137:110–118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Singal AG, Mukherjee A, Elmunzer BJ, Higgins PD, Lok AS, Zhu J, Marrero JA, et al. Machine learning algorithms outperform conventional regression models in predicting development of hepatocellular carcinoma. Am J Gastroenterol 2013;108:1723–1730. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.