Hepatocellular cancer (HCC) is the seventh most common cancer and the fourth leading cause of cancer-related death worldwide.1 HCC is the most common type of primary liver cancer and the fastest growing cause of cancer-related deaths in the United States.1–4 Typically occurring in a background of chronic liver disease and cirrhosis, the most common risk factors worldwide have traditionally included hepatitis B virus (HBV), hepatitis C virus (HCV), and alcohol-related liver disease (ALD).5 However, in the past decade, due to increasing rates of obesity and metabolic syndrome, nonalcoholic steatohepatitis (NASH) has emerged as the fastest growing risk factor for the development of HCC.6–12
Despite the increasing global burden of NASH and causative link to HCC, limited data are available comparing outcomes of NASH-related HCC and viral hepatitis (HBV and HCV) and ALD etiologies. Currently, the largest published outcomes-based report used the SEER data set and demonstrated that NASH-related HCC is associated with worse overall survival (OS) than HCC in patients with cirrhosis from viral hepatitis or ALD.6 However, due to the inherent limitations of administrative data sets for studying HCC (eg, lack of granular data regarding tumor stage or underlying liver function), the ability to draw definitive conclusions from this study is limited.13 Previous studies using nonadministrative data have attempted to clarify differences in outcomes between HCC etiologies after adjusting for tumor stage and underlying liver function, but are limited by a paucity of patients and lack of racial/ethnic diverse patient cohorts.7,11,14
Therefore, the aim of our study was to compare the clinical presentation, treatment, and survival of NASH-related HCC with HCC in the setting of viral hepatitis and ALD using a prospectively maintained HCC database from 2 large health systems with a diverse patient population.
Following Institutional Review Board approval, we retrospectively queried a prospectively maintained HCC database identifying all adult patients newly diagnosed with HCC from January 2008 through December 2016 at 2 health systems: Parkland Memorial Health and Hospital System, a safety-net hospital, and UT Southwestern Medical Center, a tertiary referral center. Both health systems have multidisciplinary clinics and conferences, staffed by UT Southwestern faculty from transplant hepatology, surgical oncology, medical oncology, and interventional radiology.15 HCC diagnosis was based on the American Association for the Study of Liver Diseases (AASLD) criteria as previously described.16,17 Patients were excluded if imaging studies were not available to characterize tumor burden or if they had received prior treatment at an outside institution. HBV and HCV etiologies were confirmed by serologies (HBV surface antigen or DNA positivity; HCV antibody or RNA positivity); ALD by heavy alcohol consumption history (>3 drinks per day for men and >2 drinks per day for women, as best determined per medical record review) and absence of other coexisting chronic liver disease; and NASH by demonstration of hepatic steatosis by imaging or biopsy, exclusion of significant alcohol consumption, and absence of coexisting chronic liver disease.18 Patients with HCC and other etiologies of chronic liver disease were excluded from the study.
A retrospective review of each patient’s medical record was performed to obtain patient demographics, clinical history, laboratory data, and imaging results. Laboratory data included alpha-fetoprotein, total bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), albumin, platelet count, and international normalized ratio (INR). Tumor characteristics of interest included the number of lesions, maximum tumor diameter, lymph node involvement, portal vein invasion, presence of extrahepatic metastases, and tumor stage at diagnosis. The Barcelona Clinic Liver Cancer (BCLC) system was used for tumor staging.10
Comorbid conditions such as metabolic syndrome, hyperlipidemia, diabetes, hypertension, and obesity were confirmed by vital reports, laboratory tests, and medication review and were in accordance with the National Cholesterol Education Program Adult Treatment Panel III definition of metabolic syndrome.19 Absence of cirrhosis was defined by Mittal et al20 as: (1) level 1 evidence of noncirrhosis as no evidence of cirrhosis on a resection specimen or liver biopsy within 1 year of HCC diagnosis and no features suggestive of cirrhosis on abdominal imaging within 3 years before HCC diagnosis or (2) level 2 evidence of noncirrhosis as AST to platelet ratio index <1 based on laboratory values within 6 months before and 4 weeks after HCC diagnosis and no features suggestive of cirrhosis on abdominal imaging within 3 years before HCC diagnosis and 2 of 3 test values in the normal range (albumin >3.5 g/L, platelets >200,000/mL, or INR <1.1) within 6 months before and 4 weeks after HCC diagnosis.
Clinical history of interest included HCC-directed treatments and OS. HCC treatment was categorized as none, curative (transplantation, surgical resection, or ablative techniques), locoregional (transarterial embolization [TACE], yttrium 90, or TARE [transarterial radioembolization], or radiation therapy), or systemic (sorafenib, regorafenib, or clinical trial of systemic therapy). If a patient received multiple treatments, they were classified by the most definitive type of therapy received. For example, if a patient received TACE and liver transplantation, they were classified as having received curative-intent therapy.
Chi-square tests and analysis of variance were used to compare categorical and continuous variables of the study population, stratified by etiology. Logistic regression models were used to evaluate the association between etiology and receipt of curative-intent therapy in patients with BCLC stage 0 or A HCC. Survival analysis was performed by Kaplan-Meier method with log-rank univariate analysis. Patients were censored at time of death, last clinic visit, or end of study period (September 2017). All variables significant in the univariate analysis (log-rank P<.05) were placed in the multivariable model. Hazard ratios (HRs) and P values for the multivariable model were obtained using Cox proportional hazard regression. All tests were 2-sided and performed at the 5% significance level. Statistical analysis was performed using SPSS, version 22 (SPSS Inc).
A total of 1,051 patients were identified as meeting inclusion criteria during the study period: 719 (68.4%) HCV-related HCC, 153 (14.6%) ALD-related HCC, 92 (8.8%) NASH-related HCC, and 87 (8.3%) HBV-related HCC.
No significant change was seen in the proportion of HCC cases related to individual etiologies over time. In the time period from 2008 to 2010, the proportion of NASH-related HCC nearly doubled from 5.5% of all HCC cases to 9.5% in the most recent time period; however, this was not statistically significant when compared with proportions of all etiologies by time period (Figure 1). When comparing absolute incidence of only NASH cases over time, there was a significant increase of new cases over time (15.2% of all NASH cases were diagnosed from 2008–2010, 43.5% from 2011–2013, and 41.3% from 2014–2016; P=.001).
Table 1 describes the differences in clinicopathologic features among included patients. Patients with NASH-related HCC were significantly older and more likely female than those with other etiologies. Patients with ALD and NASH were more likely to be Hispanic white and less likely to be Non-Hispanic black than those with HCV- or HBV-related HCC. As expected, patients with NASH-related HCC had higher rates of metabolic syndrome than those with ALD-, HCV-, and HBV-related etiologies. Patients with NASH and HBV had more preserved liver function as measured by Child-Pugh A score, less portal hypertension as reflected by median platelet count, and higher proportions of noncirrhotic HCC. In a sensitivity analysis only including level 1 evidence for noncirrhosis, NASH-related HCC continued to have the highest proportion of noncirrhosis (16% vs 3% for ALD-, 4% for HCV-, and 11% for HBV-related HCC; P<.001). No difference was seen in HCC surveillance rates within 12 months before HCC diagnosis or BCLC tumor staging among all etiologies; however, patients with NASH and HBV had larger median tumor size.
Clinicopathologic and Treatment Factors
Despite similar tumor stage between HCC etiologies, a higher proportion of patients with NASH-related HCC underwent curative-intent therapy than those with ALD-, HCV-, and HBV-related HCC. Table 2 represents odds of receiving curative-intent therapy. Patients with NASH-related HCC were significantly more likely to receive curative-intent therapy than those with ALD-related HCC (OR, 0.415; 95% CI, 0.21–0.83). Although patients with NASH-related HCC also appeared more likely to receive curative-intent therapy than those with viral hepatitis–related HCC, this difference did not reach statistical significance. As expected, patients with BCLC stage B (OR, 0.109; 95% CI, 0.07–0.18), stage C (OR, 0.027; 95% CI, 0.01–0.06), and stage D HCC (OR, 0.068; 95% CI, 0.04–0.12) were less likely to receive curative-intent therapy compared with those with BCLC stage 0/A HCC. Age, race/ethnicity, insurance status, and surveillance in patients with BCLC stage 0/A HCC were not independently associated with receipt of curative-intent therapy.
Multivariable Analysis of Patient, Tumor, and Treatment Variables Associated With Odds of Curative-Intent Therapy Receipt
Figure 2A represents unadjusted OS of the entire cohort. Median OS was 16 months for NASH-, 15 months for ALD-, 14 months for HCV-, and 8 months for HBV-related HCC (P=.030). Patients with HCV-related HCC who had achieved sustained viral response after direct-acting antiviral therapy had improved OS compared with those with untreated HCV (26 vs 12 months; P<.001). Respective 1-, 3-, and 5-year survival rates were 57%, 37%, and 21% for NASH-; 56%, 34%, and 24% for ALD-; 52%, 28%, and 18% for HCV-; and 41%, 18%, and 13% for HBV-related HCC. In subgroup analysis, no significant difference in OS between liver disease etiologies was seen among patients with BCLC stage 0/A or B HCC (P=.352 and P=.413, respectively; Figure 2B and C); however, HCV-related HCC was associated with improved OS compared with ALD-, HBV-, and NASH-related HCC among those with BCLC stage C HCC (5 vs 4, 3, and 3 months, respectively; P=.005; Figure 2D).
Table 3 represents a multivariable analysis of clinicopathologic variables associated with OS of the entire cohort. ALD-related HCC was associated with improved OS compared with NASH-related HCC (HR, 0.522; 95% CI, 0.37–0.78), whereas no significant difference was seen in survival between NASH-related and viral hepatitis–related HCC (HR, 0.872; 95% CI, 0.64–1.19 for HCV; HR, 1.419; 95% CI, 0.94–2.15 for HBV). As expected, advanced BCLC stage was associated with worse survival (HR, 1.449; 95% CI, 1.13–1.89 for BCLC stage B; HR, 2.658; 95% CI, 2.07–3.41 for BCLC stage C; and HR, 3.123; 95% CI, 2.45–3.99 for BCLC stage D), while receipt of surveillance within 12 months (HR, 0.739; 95% CI, 0.60–0.92), and receipt of any therapy (HR, 0.104; 95% CI, 0.08–0.14 for curative; HR 0.382, 95% CI, 0.31–0.48 for locoregional; and HR, 0.529; 95% CI, 0.41–0.68 for systemic therapy) were independently associated with improved survival. Results were unchanged when considering transplant-free survival (HR, 0.58; 95% CI, 0.4–0.8 for ALD; HR, 0.90; 95% CI, 0.7–1.2 for HCV; HR, 1.25; 95% CI, 0.8–1.9 for HBV).
Cox Multivariable Analysis of Patient, Tumor, and Treatment Variables Associated With Death
In our study, patients with NASH-related HCC had significantly higher rates of preserved liver function, including a higher proportion who presented with HCC in the absence of cirrhosis, compared with ALD- and HCV-related HCC. Despite similar receipt of HCC surveillance and BCLC tumor stage at diagnosis, patients with NASH-related HCC had the highest rate of curative-intent therapy. However, NASH-related HCC was associated with worse OS compared with ALD-related HCC but similar survival to viral hepatitis–related HCC.
This is one of the first studies comparing liver disease etiologies in a racially/ethnically diverse cohort. Similar to racial/ethnic disparities in NASH prevalence, we found that patients with NASH were more likely to be Hispanic white, particularly Hispanic white female, and less likely to be Non-Hispanic black compared with other etiologies.21,22 The study by Younossi et al6 grouped race/ethnicity as white versus other and did not subdivide white race into Hispanic and non-Hispanic ethnicities. These data are important when considering the increasing Hispanic white population in the United States.23 Furthermore, racial/ethnic disparities in HCC incidence may worsen in the future given the availability of direct-acting antiviral therapy for HCV and its association with lower HCC incidence.24 In contrast, there are minimal effective treatments for NASH and no preventive measures for progression to HCC.
Another notable finding of our study is the high proportion of NASH-related HCC that presented in the absence of cirrhosis, similar to that seen with HBV-related HCC. This finding has been reported in other studies.20,25 The pathogenesis of HCC developing at higher rates in NASH compared with other etiologies is unclear but may support the hypothesis that HCC pathogenesis is triggered by the same initial insult, but otherwise proceeds on a parallel but separate pathway from fibrosis progression. The high rates of noncirrhotic HCC in patients with HBV have resulted in surveillance recommendations for HCC in noncirrhotic subgroups, such as Asian men aged >40 years and Asian women aged >50 years; however, the role of HCC surveillance in patients with noncirrhotic NASH is less clear.26–28 Given the large burden of NASH in the United States and globally, it would likely not be feasible to perform surveillance in all patients. Furthermore, the incidence of HCC in these patients is unlikely to be sufficiently high for HCC surveillance to be cost-effective. Development of risk stratification tools may help identify subgroups of patients with NASH in whom HCC surveillance would be cost-effective.
Our study adds important prognostic data for the implication of the clinicopathologic differences seen among etiologies by adding valuable information about treatment receipt and survival. Patients with NASH-related HCC had better unadjusted survival, although these differences were partly mitigated after adjusting for tumor burden, liver dysfunction, and receipt of curative-intent therapy. In multivariable models, NASH-related HCC continued to be associated with worse survival compared with ALD-related HCC, but with similar survival compared with HCV-related HCC. Lack of granular data for these factors in large administrative analyses or in studies using small sample sizes with limited statistical power may partly explain the mixed survival outcomes previously reported.6,7 For example, Younossi et al6 found worse survival was associated with NASH-related HCC compared with HCV/HBV; however, their analysis was not able to include type of treatment or underlying liver function. As far as we are aware, our study represents the largest NASH-related HCC study to capture tumor stage, liver dysfunction, HCC-directed therapy, and OS. The large size of our cohort also allowed individual comparison with other etiologies, whereas other studies often combine non-NASH etiologies, which may obscure differences in presentation or prognosis.6,7,12
Study findings were interpreted in light of the following limitations. First, due to a relatively small sample size of patients with NASH only, subset analysis of NASH-related HCC was not adequately powered to determine specific prognostic factors associated with survival in this specific cohort. Second, we were unable to evaluate the prognostic importance of genetic factors, such as the PNPLA3 gene, which may be an individual risk factor for HCC development and poor prognosis.29 Third, our study was conducted in 2 large health systems, representing safety-net and tertiary care centers, but may not be representative of patients seen in other settings. However, the inclusion of 2 diverse types of health systems is important, particularly given differences between safety-net and non–safety-net systems.30 These limitations are outweighed by our study’s strengths, including its large sample size, racially/ethnically diverse cohort, and prospectively collected data.
Our findings show that patients with NASH-related HCC present with more preserved liver function, including 15% presenting without evidence of cirrhosis. Despite similar tumor stage, patients with NASH-related HCC are more likely to undergo curative-intent therapy. However, NASH etiology is associated with worse survival compared with ALD, and similar survival compared with viral hepatitis–related HCC.
Author contributions: Study concept and design: Hester, Yopp. Data acquisition: All authors. Data analysis and interpretation: Hester, Singal, Yopp. Manuscript preparation: Hester, Yopp. Critical revision: All authors.
Disclosures: Dr. Singal has disclosed that he receives honoraria/consulting fees from Bayer, Eisai, Bristol-Myers Squibb, Exelixis, Glycotest, Roche, and Wako Diagnostics. The remaining authors have disclosed that they have not received any financial considerations from any person or organization to support the preparation, analysis, results, or discussion of this article.
Funding: This work was supported by NCI R01MD12565. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding agency had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript.
Global Burden of Disease Cancer CollaborationFitzmaurice CAkinyemiju TF. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 29 cancer groups, 1990 to 2016: a systematic analysis for the Global Burden of Disease study. JAMA Oncol 2018;4:1553–1568.
Vespa JArmstrong DMMedina L. Demographic Turning Points for the United States: Population Projections for 2020 to 2060. United States Census Bureau 2018. Available at: https://www.census.gov/content/dam/Census/library/publications/2018/demo/P25_1144.pdf. Accessed August 8 2018.