Outcomes of curative treatment for hepatocellular cancer in nonalcoholic steatohepatitis versus hepatitis C and alcoholic liver disease

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Concomitant increasing incidences of hepatocellular carcinoma (HCC) and nonalcoholic steatohepatitis (NASH) suggest that a substantial proportion of HCC arises as a result of hepatocellular injury from NASH. The aim of this study was to determine differences in severity of liver dysfunction at HCC diagnosis and long-term survival outcomes between patients undergoing curative therapy for HCC in the background of NASH compared to hepatitis C virus (HCV) and/or alcoholic liver disease (ALD). Patient demographics and comorbidities, clinicopathologic data, and long-term outcomes among patients who underwent liver transplantation, hepatic resection, or radiofrequency ablation for HCC were reviewed. From 2000 to 2010, 303 patients underwent curative treatment of HCC; 52 (17.2%) and 162 (53.5%) patients had NASH and HCV and/or alcoholic liver disease. At HCC diagnosis, NASH patients were older (median age 65 versus 58 years), were more often female (48.1% versus 16.7%), more often had the metabolic syndrome (45.1% versus 14.8%), and had lower model for end-stage liver disease scores (median 9 versus 10) (all P < 0.05). NASH patients were less likely to have hepatic bridging fibrosis or cirrhosis (73.1% versus 93.8%; P < 0.001). After a median follow-up of 50 months after curative treatment, the most frequent cause of death was liver failure. Though there were no differences in recurrence-free survival after curative therapy (median, 60 versus 56 months; P = 0.303), NASH patients had longer overall survival (OS) (median not reached versus 52 months; P = 0.009) independent of other clinicopathologic factors and type of curative treatment. Conclusion: Patients with HCC in the setting of NASH have less severe liver dysfunction at HCC diagnosis and better OS after curative treatment compared to counterparts with HCV and/or alcoholic liver disease. (HEPATOLOGY 2012;55:1811–1821)

Concomitant increases in the incidence of hepatocellular carcinoma (HCC) and prevalence of nonalcoholic fatty liver disease (NAFLD) suggest that a substantial proportion of HCC arises as a result of hepatocellular injury from nonalcoholic steatohepatitis (NASH). As a result, HCC is the most rapidly increasing cause of cancer death in the United States.1-6 NAFLD, a spectrum of fatty liver diseases ranging from isolated steatosis, steatohepatitis, and progressive cirrhosis, is the most common chronic liver disease (CLD) in developed societies (paralleling the obesity epidemic) and is likely the leading case of “cryptogenic cirrhosis.”7-13 Up to 50% of patients with NAFLD will develop progressive disease, including NASH, cirrhosis, and/or HCC.1, 2, 8-10, 14-16 Despite a lower incidence of HCC resulting from NASH compared to other CLDs, the high prevalence of NAFLD means that a large percentage of HCC is caused by NASH.11-13, 17-22

Multiple reports describe the natural history of patients with NASH compared to other CLDs, the incidence and risk factors of HCC among those with NASH, and survival outcomes after one type of curative treatment for HCC from NASH compared to other CLDs.1, 4, 12, 13, 16-33 Yet, no previous reports have assessed long-term outcomes between patients with NASH and other CLDs within a framework of multimodal curative therapy, including liver transplantation, resection, and ablation. Thus, the aim of this study was to determine the differences in clinical presentation, histopathology, and survival outcomes among patients undergoing any curative therapy for HCC in the setting of NASH compared to hepatitis C (HCV) and/or alcoholic liver disease (ALD).

Abbreviations

AASLD, American Association for the Study of Liver Diseases; AFP, alpha-fetoprotein; AJCC, American Joint Committee on Cancer; ALD, alcoholic liver disease; BMI, body mass index; DM, diabetes mellitus; HCV, hepatitis C virus; HCC, hepatocellular carcinoma; INR, international normalized ratio; NAFLD, nonalcoholic fatty liver disease; NAS, NAFLD activity score; NASH, nonalcoholic steatohepatitis; MELD, model for end-stage liver disease; OS, overall survival; RFA, radiofrequency ablation; RFS, recurrence-free survival; TACE, transarterial chemoembolization; Y-90, yittrium-90 radioembolization.

Patients and Methods

Patient Selection Criteria, Definitions, and Comparisons.

After obtaining institutional board review approval, demographics, comorbid conditions, clinicopathologic data, radiology reports, curative treatments, and long-term outcomes for patients who underwent definitive curative therapy for pathologically confirmed HCC at the University of Pittsburgh Thomas E. Starzl Transplantation Institute were reviewed. For patients who underwent multiple curative treatments, the date of first definitive therapy was used as the reference for date of curative therapy. Specifically, hepatic radiofrequency ablation (RFA) intended as a “bridge” to liver transplantation was not categorized as definitive curative therapy. Patients who had undergone previous surgical resection, transarterial chemoembolization (TACE), or yittrium-90 radioembolization (Y-90) treatments all had recurrent (in cases of resection) or persistent (in cases of TACE or Y-90) disease noted on radiologic imaging before definitive curative therapy. Patients with HCC arising in a background of NASH were compared to those with HCV and/or ALD-associated HCC. Patients with both HCV and NASH were categorized in the NASH group. Per American Association for the Study of Liver Diseases (AASLD) consensus statements, the alcohol consumption threshold to define NASH included <21 drinks per week for men and <14 drinks per week for women at the height of maximal alcohol intake before curative treatment.34 Using this definition, patients with a previous history of alcohol use that may have predisposed to alcohol-induced liver disease were excluded from the categories of definite NASH and borderline NASH. Criteria for metabolic syndrome were extrapolated from international guidelines35, 36 and included any three of the following: body mass index (BMI) >28.8 kg/m2 (validated as a replacement for elevated waist circumference in men and women)23 and documentation of or medical treatment for dyslipidemia, hypercholesterolemia, hypertension, and/or diabetes mellitus (DM). Active HCV infection was defined by either viral hepatitis noted on histopathologic examination, positive serology, or an elevated viral titer. Preoperative ascites was defined by appearance on radiologic imaging, detection on physical examination, or treatment with diuretics and/or paracentesis. Reported model for end-stage liver disease (MELD) scores do not include upgrades for HCC. Criteria for definitive curative therapy with hepatic ablation, resection, or liver transplantation were not uniform throughout the study period. Every patient was evaluated at a multidisciplinary tumor conference comprising gastroenterologists, hepatologists, transplant surgeons, medical oncologists, and surgical oncologists. For patients who underwent RFA, the size and number of hepatic lesions was determined from last preoperative imaging. Of note, all gross sites of disease (including the few cases of metastatic disease) were resected at definitive curative treatment. In all cases, disease recurrence was noted on postoperative radiologic imaging. For those patients treated with liver transplantation, no donor organs were obtained from executed prisoners or other institutionalized persons.

Histopathology.

Steatosis grade, fibrosis stage, and hepatocyte ballooning were reported as described by Kleiner et al.7 Instead of the precise number of foci per high-power field, lobular inflammation was reported as “none,” “rare/spotty,” or “moderate/heavy.” Each of these terms was then coded in increasing severity from 0 to 2 in calculating the NAFLD activity score (NAS).7 Because the stigmata of NASH may disappear with cirrhosis, the most severe form of each pathologic category (e.g., steatosis, hepatocyte ballooning, and so on) present on examination from the definitive curative treatment or on previous pathology specimens was reported. Pathologist determination of NASH was reported independently of NAS and was categorized as definite, borderline, or none per consensus guidelines.34 Specifically, borderline NASH does not meet classical criteria for steatohepatitis because hepatocellular ballooning and Mallory-Denck bodies may be absent and lobular inflammation may not be prominent. Cases of borderline NASH had no other identifiable causes of CLD. Patients with a pathologic diagnosis of definitive and borderline NASH were grouped together as having HCC from NASH. Patients with definite NASH noted on histopathology and active HCV infection were categorized in the NASH group. T tumor staging was defined according to American Joint Committee on Cancer (AJCC) 7th edition guidelines.37

Statistical Analyses.

PASW software (version 18; SPSS, Inc., Chicago, IL) was used to perform statistical analyses. Baseline characteristics of the sample were characterized by numbers and corresponding percentages and median and interquartile ranges for continuous variables. The normality of continuous variables was examined, and all between-group differences of non-normally distributed continuous variables were tested using nonparametric statistics. Between-group analyses were performed using chi-square and Mann-Whitney U tests. All tests were two-tailed, with a significant P value defined as <0.05. RFS was defined as the duration from date of definitive curative treatment to date of disease recurrence. Patients without disease recurrence were censored at date of last clinical follow-up. Overall survival (OS) was defined as the duration from date of definitive curative treatment to date of last follow-up or death. Continuous variables were categorized based on clinical meaningful differences, so that between-group differences could be examined using Kaplan-Meier survival analyses and the log-rank test. Multivariable stepwise Cox regression analyses were performed to test potential predictors of survival in patients with NASH and HCV/ALD. The predictors were determined using significant between-group differences found using Kaplan-Meier analyses and the log-rank test. Between-group differences in demographic and disease-specific variables that resulted in a value of P < 0.10 were included in the Cox regression models. Cox regression analyses were performed to assess predictors of survival with the use of hazard ratios and 95% confidence intervals. The overall model as well as independent predictors of survival were characterized using a P value of <0.05.

Results

A total of 321 patients underwent curative treatment of HCC from 2000 from 2010; 18 had incomplete pathologic data and were excluded from this study. Of the remaining 303 patients, 52 (17.2%) had definitive or borderline NASH and 162 (53.5%) had active HCV and/or ALD. These 214 patients comprised the study cohort. The remaining patients either had no evidence of background liver disease or had other etiologies of CLD not including HCV, ALD, or NASH. Four of fifty-two NASH patients had “borderline” steatohepatitis without any other identifiable cause of CLD. Nine of fifty-two NASH patients had coexistent active HCV infection. Seven of these nine were infected with HCV genotype 1, and each corresponding pathologic specimen demonstrated hepatocyte ballooning and perisinusoidal fibrosis in the background liver and/or large-droplet steatosis, hepatocyte ballooning, or pericellular fibrosis within the tumor.

Demographic, Clinicopathologic, and Curative Treatment Comparisons.

Compared to patients with active HCV/ALD, NASH patients were older, were more often female, had larger BMI at HCC diagnosis, and more frequently had DM, dyslipidemia, and the metabolic syndrome (Table 1). Hepatic synthetic function at HCC presentation (measured by bilirubin, albumin, and international normalized ratio [INR] levels) was worse in patients with HCV/ALD. Ascites was also more common among HCV/ALD patients. MELD scores were slightly higher among HCV/ALD patients. There were no differences in rates of previous TACE or Y-90 treatments or previous surgical procedures. HCV/alcoholic patients less often underwent hepatic resection and more often underwent liver transplantation. Though the number of tumors was greater in HCV/ALD patients, there were no differences in size of largest tumor, frequency of satellite lesions, incidence of T3/4 disease, tumor differentiation, rates of macro-/microvascular invasion, and pathologic nodal or metastatic disease. In the background liver, steatosis, lobular inflammation, and hepatocyte ballooning were all more extensive in NASH specimens compared to HCV/ALD specimens. Correspondingly, the median NAS7 was greater in NASH specimens. Though the majority of patients in each group had bridging fibrosis or cirrhosis on pathological examination, more NASH patients did not have end-stage fibrosis (28.9% versus 6.2%; P < 0.001). Similar differences in demographics, comorbid conditions, active HCV infection, barometers of hepatic synthetic function, MELD score and histologic markers of steatohepatitis, were present in subgroups of patients who underwent hepatic resection and/or ablation and liver transplantation (Supporting Tables 1 and 2). More NASH patients who underwent hepatic resection and/or ablation did not have end-stage fibrosis compared to HCV/ALD counterparts (41.6% versus 12.7%; P = 0.002).

Table 1. Demographics, Clinicopathologic Data, and Curative Treatments Stratified by Underlying Liver Disease
 HCV/ALD (N = 162)NASH (N = 52)P Value
  • *

    Alcohol consumption defined per AASLD consensus statement.34

  • Does not include upgrade for HCC.

  • Each component of surgical therapy separately is listed.

  • §

    AJCC 7th edition tumor staging.37

At HCC diagnosis   
 Age (years)58 (53-67)65 (57-70)0.023
 BMI (kg/m2)28.7 (25.9-32.0)31.3 (27.6-33.9)0.009
 Female (%)27 (16.7)25 (48.1)<0.001
 Caucasian (%)135 (83.3)47 (94.2)0.215
 DM (%)50 (30.9)28 (53.8)0.003
 Hypertension (%)84 (51.9)31 (59.6)0.329
 Dyslipidemia (%)18 (11.1)17 (32.7)<0.001
 Metabolic syndrome (%)24 (14.8)23 (45.1)<0.001
 History of alcohol use* (%)119 (73.5)0<0.001
 Active HCV infection (%)116 (71.6)9 (17.3)<0.001
 AFP (ng/mL)16.5 (6-106)11.0 (6.0-16.9)0.081
 Albumin (mg/dL)3.4 (3.0-3.9)4.0 (3.3-4.3)0.002
 Bilirubin (mg/dL)1.1 (0.7-1.7)0.6 (0.5-1.4)<0.001
 INR1.2 (1.1-1.3)1.1 (1.0-1.2)0.002
 Creatinine (mg/dL)0.9 (0.8-1.1)0.9 (0.7-1.2)0.735
 MELD score10 (8-12)9 (7-11)0.024
 Aspartate aminotransferase (U/L)73 (42-119)51 (34-70)0.003
 Alanine aminotransferase (U/L)57 (40-96)48 (34-82)0.069
 Ascites (%)54 (33.3)9 (17.3)0.027
 Hepatic encephalopathy (%)11 (6.8)3 (5.8)0.796
Previous treatments   
 TACE or Y-90 (%)38 (23.5)18 (34.6)0.111
 Surgical therapy (%)20 (12.3)3 (5.8)0.411
 Resection7 (4.3)1 (1.9) 
 Ablation13 (8.0)2 (3.8) 
 Resection and ablation2 (1.2)0 
 None142 (87.7)49 (94.2) 
Curative treatment and pathology   
 Transplant (%)83 (51.2)20 (38.5)0.109
 Resection (%)52 (32.1)26 (50.0)0.020
 Ablation (%)32 (19.8)10 (19.2)0.934
 Largest tumor size (cm)3.2 (2.2-4.6)3.0 (2.3-4.6)0.674
 Presence of satellite lesions (%)19 (11.7)5 (9.6)0.870
 Tumor number1 (1-3)1 (1-2)0.049
 T3-4 disease§ (%)29 (17.9)5 (10.6)0.155
 Tumor differentiation (%)  0.241
 Well24 (14.8)13 (26.1) 
 Moderate97 (66)29 (63) 
 Poor26 (17.7)5 (10.9) 
 Other1 (0.6)2 (3.8) 
 Vascular invasion (%)   
 Macro14 (8.6)3 (5.8)0.505
 Micro70 (43.2)21 (40.4)0.720
 Pathologic nodal disease (%)2 (1.2)00.421
 Metastatic disease (%)6 (3.7)1 (1.9)0.530
 Steatosis (%)  <0.001
 <579 (48.8)6 (11.5) 
 5-3350 (30.9)15 (28.8) 
 >33-6622 (13.6)20 (38.5) 
 >668 (4.9)11 (21.2) 
 Lobular inflammation (%)  0.007
 None33 (20.4)4 (7.7) 
 Rare/spotty63 (38.9)14 (26.9) 
 Mild56 (34.6)32 (61.5) 
 Moderate/heavy8 (4.9)2 (3.8) 
Hepatocyte ballooning (%)  <0.001
 None108 (66.7)17 (32.7) 
 Few27 (16.7)14 (26.9) 
 Many25 (15.4)21 (40.4) 
 NAS, range2 (1-3)5 (3-6)<0.001
 NAS ≥4 (%)37 (22.8)35 (67.3)<0.001
 Bridging fibrosis/cirrhosis (%)152 (93.8)38 (73.1)<0.001

Table 2 summarizes comparisons of demographics, clinicopathologic tumor characteristics, and curative treatments between NASH patients with (n = 23) and without (n = 29) metabolic syndrome. Patients with metabolic syndrome had a higher frequency of DM, hypertension, and dyslipidemia. No NASH patients with metabolic syndrome had coexistent HCV infection. There were no significant differences in preoperative alpha-fetoprotein (AFP), albumin, bilirubin, and INR levels, types of curative treatments, number or size of largest HCC tumors, or histopathology of HCC or the background liver between NASH patients with and without metabolic syndrome. Twenty of fifty-two NASH patients (38.5%) had neither HCV nor metabolic syndrome. Median BMI of these patients was 30.1 kg/m2 (range, 26.5-32.6). Sixty percent were female and 30.0%, 45.0%, and 15.0% had DM, hypertension, and dyslipidemia, respectively. There were no differences in any of these parameters when compared to NASH patients with HCV (all P > 0.05).

Table 2. Demographics, Clinicopathologic Data, and Curative Treatments for NASH/HCC Patients Stratified by the Presence of Metabolic Syndrome
 Metabolic Syndrome (N = 23)No Metabolic Syndrome (N = 29)P Value
  • *

    Denotes regular alcohol consumption below the criteria for alcoholic steatohepatitis.34

  • Does not include upgrade for HCC.

  • Each component of surgical therapy separately listed.

  • §

    AJCC 7th edition tumor staging.37

At HCC diagnosis   
 Age (years)66 (60-70)63 (54-70)0.363
 BMI (kg/m2)32.0 (29.7-34.3)29.3 (26.9-32.6)0.161
 Female (%)8 (34.8)17 (58.6)0.153
 Caucasian (%)22 (95.6)25 (86.2)0.367
 DM (%)21 (91.3)7 (24.1)<0.001
 Hypertension (%)21 (91.3)10 (34.5)0.001
 Dyslipidemia (%)13 (56.5)4 (13.8)0.003
 Any history of alcohol use* (%)5 (21.7)6 (20.7)0.803
 Active HCV infection (%)09 (31.0)0.010
 AFP (ng/mL)10.0 (4.3-22.0)11.6 (5.0-132.0)0.460
 Albumin (mg/dL)4.0 (3.5-4.2)3.8 (3.3-4.3)0.599
 Bilirubin (mg/dL)0.6 (0.4-1.8)0.6 (0.5-1.2)0.875
 INR1.1 (1.0-1.2)1.1 (1.0-1.2)0.647
 Creatinine (mg/dL)1.1 (0.8-1.2)0.9 (0.7-0.9)0.169
 MELD score9 (7-11)7 (7-10)0.377
 Aspartate aminotransferase (U/L)51 (30-65)51 (35-82)]0.549
 Alanine aminotransferase (U/L)48 (35-79)46 (34-84)0.868
 Ascites (%)4 (17.4)5 (17.2)1.000
 Hepatic encephalopathy (%)03 (10.3)0.245
Previous treatments   
 TACE or Y-90 (%)8 (34.7)10 (34.5)0.787
 Surgical therapy (%)   
 Resection01 (3.4)0.988
 Ablation02 (6.9)0.497
 Resection and ablation001.000
 None23 (100)26 (89.7)0.245
Curative treatment and pathology   
 Transplant (%)8 (34.8)12 (41.4)0.843
 Resection (%)12 (52.2)14 (48.3)0.956
 Ablation (%)6 (26.1)4 (13.8)0.446
 Largest tumor size (cm)3.0 (2.5-4.0)3.3 (2.2-5.0)0.691
 Presence of satellite lesions (%)1 (4.3)4 (13.8)0.368
 Tumor number1 (1-1)1 (1-2)0.352
 T3-4 disease§ (%)2 (9.5)3 (10.3)1.000
 Tumor differentiation (%)  0.200
 Well5 (21.7)8 (27.6) 
 Moderate15 (65.2)14 (48.3) 
 Poor1 (4.3)4 (13.8) 
 Other1 (4.3)1 (3.4) 
 Vascular invasion (%)   
 Macro2 (8.7)1 (3.4)0.577
 Micro8 (34.8)13 (44.8)0.654
 Metastatic disease (%)1 (4.3)00.442
 Steatosis (%)  0.239
 <54 (17.4)2 (6.9) 
 5-336 (26.1)9 (31.0) 
 >33-6610 (43.5)10 (34.5) 
 >663 (13.0)8 (27.6) 
 Lobular inflammation (%)  0.431
 None2 (8.7)2 (6.9) 
 Rare/spotty8 (34.8)6 (20.7) 
 Mild13 (56.5)19 (65.5) 
 Moderate/heavy02 (6.9) 
 Hepatocyte ballooning (%)  0.304
 None6 (26.1)11 (37.9) 
 Few5 (21.7)9 (31.0) 
 Many12 (52.2)9 (31.0) 
 NAS4 (3-6)5 (3-6)0.674
 NAS ≥4 (%)15 (65.2)20 (69.0)0.991
 Bridging fibrosis/cirrhosis (%)16 (69.5)22 (75.9)0.846

Survival Outcomes: All Patients (n = 214).

Postoperative mortality did not differ between NASH patients with metabolic syndrome, compared to HCV/ALD patients (4.3% versus 6.8%; P = 0.912). Median follow-up for all living patients was 50 months. During follow-up, 93 of 214 (43.5%) patients died. Most deaths (51.6%) were caused by hepatic failure, 32.3% were caused by HCC progression without liver failure, and 16.1% were the result of other causes. Median, 1-year, 3-year, and 5-year RFS after curative therapy were 60 months, 74.7%, 60.3%, and 49.0%, respectively (Fig. 1). Median, 1-year, 3-year, and 5-year OS were 60 months, 81.0%, 58.5%, and 49.9%, respectively (Fig. 2). Age at HCC diagnosis greater than 70 years, AFP >100 ng/mL at HCC diagnosis, microvascular tumor invasion, macrovascular tumor invasion, primary T3-4 stage, previous TACE or Y-90 therapy, and liver transplantation (compared to hepatic resection or ablation) were associated with RFS (P < 0.10) on univariable analysis (Table 3). There was no significant difference in RFS between patients with a background NASH versus HCV/ALD (P = 0.303; Fig. 3). Active HCV infection, macrovascular tumor invasion, primary T3-4 stage, AFP >100 ng/mL at HCC diagnosis, albumin <3.5 mg/dL at HCC diagnosis, previous TACE or Y-90 therapy, MELD score, liver transplantation (compared to hepatic resection or ablation), and background NASH were associated with OS (P < 0.10) on univariable analysis (Table 3). NASH patients had longer OS compared to counterparts with HCV and/or ALD (median, not reached versus 52 months; P = 0.009; Fig. 4). Multivariable analyses for RFS and OS are summarized in Table 4. Primary T3-4 stage and liver transplantation were independently associated with RFS. AFP >100 ng/mL at HCC diagnosis, albumin <3.5 mg/dL at HCC diagnosis, liver transplantation, and background NASH were independently associated with OS.

Figure 1.

RFS for the entire cohort (n = 214).

Figure 2.

OS for the entire cohort (n = 214).

Table 3. Univariable Analysis for RFS and OS for the Study Cohort (n = 214)
 RFSOS
 Median (Months)P ValueMedian (Months)P Value
  • *

    Patients who underwent hepatic resection and ablation are categorized as hepatic resection.

Gender 0.251 0.934
 Male (n = 161)Not reached 60 
 Female (n = 52)45 64 
Race 0.320 0.111
 Caucasian (n = 182)60 68 
 Non-Caucasian (n = 32)Not reached 30 
Age 0.048 0.1172
 ≥70 years (n = 50)45 42 
 <70 years (n = 164)65 70 
DM 0.825 0.589
 Present (n = 78)56 59 
 Absent (n = 136)60 70 
Hypertension 0.318 0.329
 Present (n = 115)56 52 
 Absent (n = 99)60 Not reached 
Dyslipidemia 0.647 0.251
 Present (n = 35)Not reached Not reached 
 Absent (n = 179)60 60 
BMI 0.284 0.569
 ≥28.8 (kg/m2) (n = 108)60 70 
 <28.8 (kg/m2) (n = 106)Not reached 60 
Metabolic syndrome 0.708 0.377
 Present (n = 47)Not reached Not reached 
 Absent (n = 167)56 60 
History of alcohol use 0.721 0.448
 Yes (n = 119)56 54 
 No (n = 95)60 64 
Active HCV infection 0.753 0.053
 Yes (n = 125)56 46 
 No (n = 89)Not reached Not reached 
AFP 0.007 0.083
 >100 ng/mL (n = 52)35 34 
 <100 ng/mL (n = 162)65 68 
Albumin 0.584 0.002
 <3.5 mg/dL (n = 100)51 30 
 ≥3.5 mg/dL (n = 114)60 83 
Bilirubin 0.843 0.343
 ≥2 mg/dL (n = 40)Not reached 37 
 <2 mg/dL (n = 174)60 67 
MELD score 0.583 0.073
 ≥13 (n = 46)Not reached 30 
 <13 (n = 168)60 70 
Previous TACE or Y-90 0.012 0.009
 Yes (n = 56)Not reached Not reached 
 No (n = 158)45 46 
Surgical procedure* <0.001 <0.001
 Transplant (n = 103)Not reached Not reached 
 Resection (n = 78)24 35 
 Ablation (n = 33)18 29 
Microvascular invasion 0.0015 0.656
 Yes (n = 91)35 64 
 No (n = 123)Not reached 30 
Macrovascular invasion <0.001 0.003
 Yes (n = 17)5 17 
 No (n = 197)65 67 
T stage <0.001 0.021
 1-2 (n = 180)Not reached 67 
 3-4 (n = 34)6 25 
Fibrosis 0.655 0.185
 Nonbridging fibrosis/no cirrhosis (n = 24)Not reached Not reached 
 Bridging fibrosis/cirrhosis (n = 190)56 59 
Underlying liver disease 0.303 0.009
 NASH (n = 52)60 Not reached 
 HCV/ALD (n = 162)56 52 
Figure 3.

RFS stratified by NASH versus HCV/ALD.

Figure 4.

OS stratified by NASH versus HCV/ALD.

Table 4. Multivariable Analysis for RFS and OS for the Study Cohort (n = 214)
RFSOS
VariableExp B (95% CI)P ValueVariableExp B (95% CI)P Value
  • *

    Reference is transplantation. Patients who underwent hepatic resection and ablation are categorized as hepatic resection.

  • Reference is HCV/ALD.

Age ≥70 years1.011 (0.575-1.775)0.971T3-4 stage1.295 (0.711-2.437)0.398
Previous TACE/Y-900.423 (0.213-1.532)0.398Active HCV infection1.275 (0.775-2.095)0.338
AFP >100 ng/mL1.653 (0.977-2.800)0.062MELD score ≥132.369 (0.792-4.003)0.312
T3-4 stage2.345 (1.300-4.238)0.005Previous TACE/Y-900.713 (0.325-1.413)0.514
Surgical procedure*  AFP >100 ng/mL1.757 (1.110-2.780)0.017
 Resection3.494 (2.049-5.957)<0.001Albumin <3.5 mg/dL2.780 (1.761-4.390)<0.001
 Ablation5.872 (2.868-12.023)<0.001NASH0.509 (0.291-0.888)0.018
   Surgical procedure*  
Microvascular invasion1.762 (0.997-3.114)0.052Resection4.193 (2.378-7.395)<0.001
   Ablation3.435 (1.951-6.050)<0.001

Survival Outcomes: Subgroup Analysis.

Among those patients who underwent hepatic resection and/or ablation, alcohol use, AFP level, microvascular and macrovascular tumor invasion, T3/4 tumor stage, end-stage fibrosis, and background NASH were associated with RFS on univariable analysis (Supporting Table 3). Dyslipidemia, alcohol use, AFP and albumin levels, MELD score, T3/4 stage, end-stage fibrosis, and background NASH were associated with OS on univariable analysis. Three-year OS among NASH patients was longer (60.9% versus 36.2%; P = 0.029), compared to HCV/ALD counterparts (Supporting Fig. 1). AFP >100 ng/mL at HCC diagnosis (Exp B, 2.745 [1.332-5.658]; P = 0.007) and absence of end-stage fibrosis (Exp B, 0.393 [0.0172-0.896]; P = 0.027) were independently associated with RFS on multivariable analysis. AFP >100 ng/mL (Exp B, 2.175 [1.200-3.952]; P = 0.011), serum albumin <3.5 mg/dL (Exp B, 3.099 [1.802-5.332]; P < 0.001), and background NASH (Exp B, 0.452 [0.243-0.841]; P = 0.013) were independently associated with OS on multivariable analysis. Among patients who underwent liver transplantation, female gender, metabolic syndrome, active HCV infection, AFP >100 ng/mL, albumin <3.5 mg/dL, and T3/4 stage were associated with RFS on univariable analysis (Supporting Table 4). Active HCV infection and albumin <3.5 mg/dL were associated with OS on univariable analysis. There was no significant difference in OS after transplantation between NASH and HCV/ALD patients (3-year 83.3% versus 71.1%; P = 0.204; Supporting Fig. 2). T3/4 stage was independently associated with RFS on multivariable analysis (Exp B, 3.291 [1.116-9.705]; P = 0.032). Though not significant, both active HCV infection (Exp B, 2.252 [0.891-5.688]; P = 0.087) and albumin <3.5 mg/dL (Exp B, 2.316 [0.967-5.544]; P = 0.061) were independently associated with OS on multivariable analysis.

Among those with NASH, differences in RFS (median, 60 months versus not reached; P = 0.364) and OS (median, 64 months versus not reached; P = 0.155) between patients with and without bridging fibrosis or cirrhosis were not significant (Supporting Figs. 3 and 4). Seven of twenty-three NASH patients with metabolic syndrome (30.4%) and 9 of 29 (31.0%) without metabolic syndrome died at last follow-up. Causes of death among patients with metabolic syndrome included HCC progression without liver failure (n = 1), liver failure resulting from HCC progression (n = 2), liver failure without HCC recurrence (n = 2), and coronary artery disease (n = 2). Corresponding causes of death among NASH patients without metabolic syndrome included HCC progression without liver failure (n = 2), liver failure resulting from HCC progression (n = 4), and liver failure without HCC recurrence (n = 3). Causes of death among HCV/ALD patients included HCC progression without liver failure (n = 27), liver failure from HCC progression (n = 10), liver failure without HCC recurrence (n = 27), and other causes (n = 13).

Discussion

In this retrospective study, NASH patients with HCC had less-severe background liver disease at tumor diagnosis compared to HCV and/or ALD counterparts. Among all patients who underwent curative treatment of HCC, 17.2% had NASH—a volume higher than that noted in other series.25, 31 In agreement with other reports, HCC/NASH patients were more often female, were older at HCC diagnosis, had larger BMI, and more often had components of or the diagnosis of metabolic syndrome compared to HCV/ALD patients.1, 9, 18-20, 23, 24, 28, 38-40 Similarly, NASH/HCC specimens also had more-severe steatosis, lobular inflammation, and hepatocyte ballooning (Table 1). Though clinical and individual histopathological findings should not be used as surrogates for an overall pathologic diagnosis of NASH,7, 15, 41-46 these differences reflect the accuracy of pathologist diagnosis of NASH in this patient cohort. Uniquely, this study shows that NASH patients have better synthetic liver function (as measured by serum bilirubin, albumin, and INR), less often had ascites, and had lower MELD scores compared to HCV/ALD counterparts at HCC diagnosis. These discrepancies in demographics, comorbidities, measures of hepatic synthetic function, and background liver histopathology were maintained within subgroup analysis of patients treated with liver transplantation and resection/ablation (Supporting Tables 1 and 2). Moreover, NASH/HCC patients who underwent hepatic resection and/or ablation were less likely to have histopathologic bridging fibrosis or cirrhosis-a finding corroborated by other studies.25, 32, 40 A total of 44.2% of NASH patients in this series had metabolic syndrome. As expected, hypertension, DM, and dyslipidemia were more common in this subgroup relative to other NASH patients. Yet, there were no differences in measures of hepatic synthetic function or tumor or background liver histopathology between NASH patients with and without metabolic syndrome (Table 2).

NASH patients with HCC have better OS after curative treatment compared to corresponding patients with HCV and/or ALD. Dramatic differences in demographics, comorbidities, severity of liver dysfunction at HCC diagnosis, and types of curative therapy precluded a “case-control” matching between groups. To account for these differences, we performed a multivariable analysis to determine those factors independently associated with postoperative outcomes. NASH patients had longer OS after curative treatment (median, not reached versus 52 months; P = 0.009; Fig. 4), compared to HCV/ALD counterparts that was independent of clinicopathologic factors, MELD score, and type of curative treatment (Table 3). Given (1) similarities in T stage, frequency of satellite lesions, tumor differentiation (Table 1), and RFS (Fig. 3) between groups, (2) the majority of patients had early-stage HCC, (3) most patients had well-compensated liver disease at HCC diagnosis, and (4) cause of death in the most patients was liver failure, this difference in OS was likely the result of differences in the type of background liver disease and not tumor aggressiveness. Among transplant patients, albumin <3.5 mg/dL and active HCV infection were associated with OS, suggesting that recurrent HCV infection is likely a key reason for the shorter survival of HCV/ALD transplanted patients compared to NASH counterparts. In resected and/or ablated patients, background NASH was associated with OS independent of severity of fibrosis, MELD score, and tumor stage. Importantly, discrepancies in OS between NASH and HCV/ALD patients were not the result of differences in postoperative mortality (e.g., death within 90 days of surgery).

Increasingly recognized is the synergistic role of NASH with HCV infection in exacerbating liver disease and promoting HCC development.1, 9, 22, 43, 47 The incidence of NASH among all patients with active HCV infection who underwent curative treatment of HCC in our study was 7.2%—similar to that found in other series.47, 48 Recognized histopathologic findings of superimposed NASH on HCC were found in these specimens-including background hepatocyte ballooning and perisinusoidal fibrosis and malignant hepatocyte ballooning, large-droplet steatosis, and pericellular fibrosis.47, 49 Because the majority of these patients were infected with HCV genotype 1, it is likely that the histopathologic evidence of NASH was more likely the result of metabolic factors associated with NASH in non-HCV patients, as opposed to the steatotic effects of HCV which are more often observed with genotype 3.49, 50 Despite categorizing patients with coexistent definitive histopathologic NASH and active HCV infection in the NASH group, measures of synthetic liver function, MELD scores, and histopatholgic fibrosis were all less severe and OS after curative therapy was prolonged among NASH patients relative to HCV/ALD counterparts. Interestingly, none of the NASH patients with metabolic syndrome had coexistent HCV infection. Clearly, more studies are needed to determine the synergistic role of these two common CLDs in promoting hepatic fibrosis and hepatocellular carcinogenesis.

Several limitations to our study should be considered. Imperfect interrater agreement on the presence and magnitude of certain histologic features and lack of consensus on features distinguishing NASH from steatosis and inflammation mean that the assignment of NASH is not absolute.7, 51 Sampling variability and adequacy and tumor viability (particularly in cases of previous TACE or Y-90 treatment) may have influenced histologic interpretations.6, 7, 43, 50 Because only cases of definitive or borderline NASH were included in the NASH group, we may have underestimated the incidence of HCC arising from NASH. It is increasingly recognized that a large percentage of patients with HCC arising from cryptogenic cirrhosis in fact may have “burnt-out NASH” because characteristic steatosis, lobular inflammation, and ballooning degeneration may disappear with fibrosis progression.1, 2, 6, 8, 12, 20, 30, 38, 40, 43, 50, 52 Occult alcohol use may have clouded the differentiation between alcoholic and nonalcoholic steatohepatitis.50 Because preoperative serum triglyceride, high-density lipoprotein, and/or fasting glucose levels, waist circumference, and blood-pressure measurements were not available for most patients, we used surrogates for each parameter, including medication treatment and BMI. Because there were likely some patients with unrecognized elements of metabolic syndrome, we may have underestimated its presence among NASH patients-accounting for the lower prevalence of this condition relative to other series. Despite prolonged follow-up after curative treatment for HCC, median RFS and OS has not been reached. Though more extensive follow-up may alter the significance of other clinicopathologic variables on long-term outcome, it is unlikely that conclusions regarding the improved survival of the NASH cohort relative to HCV/ALD patients would be affected given the distribution of deaths over the follow-up period (Fig. 4). Importantly, only patients with early HCC stages, compensated liver disease (in cases of hepatic resection and ablation), and ability to tolerate operative intervention were included in this study. Thus, results of our study may not apply to the majority of patients with advanced HCC precluding curative therapy. Moreover, comparisons of NASH patients were limited to counterparts with HCV/ALD and may not apply to those with other CLDs. Though our study is one of the largest evaluating long-term outcomes after curative therapy of HCC in NASH, the numbers of patients in each subgroup were relatively small. These size limitations may have masked further differences in outcomes among patients within each subgroup, particularly the influence of end-stage fibrosis on long-term survival among NASH patients. Thus, a similar multi-institutional evaluation would be beneficial in confirming our findings of prolonged survival among NASH patients.

In summary, NASH patients with HCC have less-severe liver dysfunction at HCC diagnosis and better OS after curative treatment compared to counterparts with HCV and/or ALD.

Ancillary