Hepatocellular carcinoma (HCC) is the thirdleading

Transcription

1 Association of Nonalcoholic Fatty Liver Disease (NAFLD) with Hepatocellular Carcinoma (HCC) in the United States From 2004 to 2009 Zobair M. Younossi, 1,2,3 Munkhzul Otgonsuren, 3 Linda Henry, 3 Chapy Venkatesan, 2 Alita Mishra, 1 Madeline Erario, 2 and Sharon Hunt 1 Hepatocellular carcinoma (HCC) is increasingly reported in patients with nonalcoholic fatty liver disease (NAFLD). Our aim was to assess the prevalence and mortality of patients with NAFLD-HCC. We examined Surveillance, Epidemiology and End Results (SEER) registries ( ) with Medicare-linkage files for HCC, which was identified by the International Classification of Diseases for Oncology, third edition codes using topography and morphology codes Medicare-linked data was used to identify NAFLD, hepatitis C virus (HCV), hepatitis B virus (HBV), alcoholic liver disease (ALD), and other liver disease using International Classification of Diseases, Ninth Revision, Clinical Modification codes. NAFLD was also defined by clinical diagnosis (cryptogenic cirrhosis, obese-diabetics with cryptogenic liver disease). A logistic regression model was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for risk of HCC. In addition, adjusted hazard ratios for 1-year mortality were estimated by Cox s proportional hazard regression. A total of 4,929 HCC cases and 14,937 controls without HCC were included. Of the HCC cases, 54.9% were related to HCV, 16.4% to ALD, 14.1% to NAFLD, and 9.5% to HBV. Across the 6-year period (2004 to 2009), the number of NAFLD-HCC showed a 9% annual increase. NAFLD-HCC were older, had shorter survival time, more heart disease, and were more likely to die from their primary liver cancer (all P < ). Of those who received a transplant after HCC (n 5 488), only 5% were related to NAFLD-HCC. In multivariate analysis, NAFLD increased the risk of 1-year mortality (OR, 1.21; 95% CI: ). Additionally, older age, lower income, unstaged HCC increased risk of 1-year mortality while receiving a liver transplant (LT), and having localized tumor stage were protective (all P < 0.05). Conclusions: NAFLD is becoming a major cause of HCC in the United States. NAFLD HCC is associated with shorter survival time, more advanced tumor stage, and lower possibility of receiving a LT. (HEPATOLOGY 2015;62: ) Hepatocellular carcinoma (HCC) is the thirdleading cause of cancer death worldwide. 1 Although hepatitis C virus (HCV) is the most common cause of HCC in the United States, hepatitis B virus (HBV) has been implicated as the most common etiological agent for HCC globally. Recently, there is increasing appreciation that nonalcoholic fatty liver disease (NAFLD) may be also becoming an important cause of HCC. 1-3 There are some indications that NAFLD may soon become a dominant cause for HCC in the Western world. This may be partly related to increasing prevalence of obesity-related NAFLD as well as the declining incidence of new cases of HCV, and the development of Abbreviations: ALDE, alcoholic liver disease; CCI, Charlson comorbidity index; CI, confidence interval; CLD, chronic liver disease; ESRD, end-stage renal disease; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HR, hazard ratio; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; LT, liver transplant; MEDPAR, Medicare Provider Analysis and Review (hospital file); NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NCH, National Claims History (carrier file); OR, odds ratio; PEDSF, Patient Entitlement and Diagnosis Summary File (cancer file); SD, standard deviation; SEER, Surveillance, Epidemiology and End Results; SUMDENOM, The Summarized Denominator (noncancer file). From the 1 Center for Liver Disease, Inova Health System, Falls Church, VA; 2 Department of Medicine, Inova Fairfax Hospital, Falls Church, VA; and 3 Betty and Guy Beatty Center for Integrated Research, Inova Health System, Falls Church, VA. Received May 27, 2015; accepted August 10, 2015.Z 1723

2 1724 YOUNOSSI ET AL. HEPATOLOGY, December 2015 highly effective second-generation direct antiviral agents for treatment of HCV. 1,4 Although most cases of HCC occur in the setting of advanced fibrosis or cirrhosis, there are some suggestions that patients with NAFLD without cirrhosis may develop HCC. 5 One study estimated that the yearly cumulative incidence of HCC in NASH-related cirrhosis was 2.6% versus 4.0% in HCV-related cirrhosis during a median follow-up of 3.2 years. 6 The study documented that any alcohol consumption was associated with an increased risk of HCC for both HCV and nonalcoholic steatohepatitis (NASH) patients with cirrhosis. 6 In addition, other data suggest that patients with NASH-related HCC may present with more advanced tumors likely indicating a lack of effective screening for HCC in patients with NASH. 7 It is important to note that most of the data associating NASH and HCC come from tertiary care medical centers. To date, a large, population-based study of NAFLD-related HCC is not available. Therefore, the aim of this study was to determine that association of NAFLD with HCC and related mortality using a population-based database. Materials and Methods The study cohorts were identified from the Surveillance, Epidemiology and End Results (SEER)-Medicare linked database from 2004 to This database links cancer registry data from the National Cancer Institute s SEER program with Medicare enrollment and claims files. 8,9 The data files used included Patient Entitlement and Diagnosis Summary File (PEDSF), inpatient claims known as Medicare Provider Analysis and Review (MEDPAR) files, Physician/Supplier File known as the National Claims History (NCH) files, and claims files for outpatient service (Outpatient). The PEDSF file contains cancer incidence data from the 17 SEER registries, which covers 28% of the U.S. population. In addition to patients demographics and tumor features, it also includes information on date and cause of death, as well as Medicare enrollment. Medicare administrative files contain information on service, date of service, and diagnosis. MEDPAR contains inpatient hospital and skilled nursing facility stay records; NCH contains noninstitutional providers; and Outpatient contains claims filed by hospital outpatient department, ambulatory care, surgical center, outpatient rehabilitation facilities, rural health clinics, and community mental health centers. 8,9 The Summarized Denominator (SUMDE- NOM) file was a random 5% sample of Medicare beneficiaries who resided in one of the SEER areas and did not have cancer. The SUMDENOM file contains information on demographic, enrollment, and entitlement and can be linked to the noncancer (NCH, MED- PAR, and Outpatient) files. Population. We included Medicare beneficiaries who were diagnosed with their first primary liver cancer during in SEER registries. HCC cases were identified by the International Classification of Diseases for Oncology, Third Edition using topography code C22.0 (liver) (SEER cancer site recode 29 5 liver) and morphology codes We excluded cases with tumor histology other than HCC (n 5 10,740). To minimize the possibility of erroneously including metastatic cancer to the liver, persons with simultaneous or previous diagnoses of stomach, colon, lung, pancreatic, breast, prostate, rectal, or metastasis cancers using MEDPAR, NCH, and Outpatient files were also excluded (n 5 2,792; Fig. 1). We also excluded other unrelated causes of liver disease (such as liver fluke) or any other important confounders. Furthermore, using Medicare type of enrollment and date of enrollment data, we excluded patients who had been enrolled at least 1 month in a health maintenance organization during 12 months preceding the diagnosis of HCC as well as patients without at least a 36-month Medicare eligibility window before diagnosis of HCC for ascertaining comorbidity (Fig. 1). In order to select a control group, noncancer cases were identified using the SUMDE- NOM file with similar inclusion/exclusion criteria. A randomly selected sample using a 3:1 frequency was developed from the date of Medicare service use and by SEER registry regions. Identifying Comorbidities. The modified Charlson comorbidity index (CCI) for comorbidities was used to assess comorbidities in this study. 10 Furthermore, we identified medical diagnoses using the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes. We identified heart Address reprint requests to: Zobair M. Younossi, M.D., M.P.H., Betty and Guy Beatty Center for Integrated Research, Claude Moore Health Education and Research Building, 3300 Gallows Road, Falls Church, VA zobair.younossi@inova.org; fax: Copyright VC 2015 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Dr. Younossi consults for Gilead, Bristol-Myers Squibb, AbbVie, Intercept, and GlaxoSmithKline.

3 HEPATOLOGY, Vol. 62, No. 6, 2015 YOUNOSSI ET AL Fig. 1. Study flow of the cohort (HCC and control) selection, SEER-Medicare. disease comorbidity by coronary artery disease codes acute myocardial infarction 410, old myocardial infarction 412; and congestive heart failure 428 from information gathered from the diagnosis codes created within 3 years preceding the diagnosis of HCC using the carrier (diagnosed at least twice to be confirmed by the physician), outpatient, and inpatient claims databases. Definition of Different Chronic Liver Diseases. The following chronic liver diseases (CLDs) were identified from the ICD-9 codes: (1) HCV infection , , , , 070.7, and V02.62; (2) HBV infection codes 070.2, 070.3, , , and V02.61; (3) alcoholic liver disease (ALD) codes alcoholic fatty liver disease as 571.0, alcoholic hepatitis as 571.1, alcoholic cirrhosis of the liver as 571.2, alcoholic liver damage as 571.3, and other alcohol-related disorders codes 303, 305.0, V11.3, V79.1, and 291; and (4) diagnosis of NAFLD included codes and Given that NAFLD is typically undercoded, we also defined NAFLD by clinical definition: patients with cryptogenic liver disease/cirrhosis as patients with diabetes (code 250) and obesity (code 278) and/or cirrhosis of liver without mention of alcohol or other liver diseases (code 571.5) and (5) other liver disease includes: autoimmune hepatitis (571.4), biliary cirrhosis (571.6), hemochromatosis (275.0), and Wilson s disease (275.1). No liver disease was defined as individuals without any known liver disease. Treatment After Diagnosis of HCC. Using hospital procedure codes (SRGCDE1-SRGCDE25 by ICD- 9-CM procedure code) as liver transplant (LT) 50.5, we identified LT recipients after and before diagnosis of HCC using the Medicare MEDPAR File. Survival Follow-up Time. Data were censored at December 31, 2011 from the date of HCC diagnosis. Next, within 1-year survival by months was calculated from the date of HCC diagnosis. Also, we identified the five most common mortality causes in the HCC cohort using the cause of death by ICD-10 (COD10V) variable: (1) primary liver cancer (code C22); (2) CLDs where included ALD, toxic liver disease, fibrosis and cirrhosis of liver, other liver disease, and hepatitis (codes K70-K72, K74-K77, B15-B19, and K73); and (3) miscellaneous malignant cancer, such as malignant neoplasm of spleen, mesothelioma, Kaposi s sarcoma,

4 1726 YOUNOSSI ET AL. HEPATOLOGY, December 2015 malignant neoplasms of ill-defined, other secondary and unspecified sites, unspecified malignant neoplasms of lymphoid, hematopoietic and related tissue, and malignant neoplasms of independent multiple sites (codes C26.1, C45, C46, C76-C80, C88, C96.0- C96.2, C96.7, C96.9, and C97); and (4) heart disease, such as hypertensive heart disease, ischemic heart diseases, and other heart diseases (codes I00-I11, I13, and I20-I51). Covariates. Demographic (age and gender), cancer stage (unstaged, localized, and advanced stage as regional or distant), residence region (urban or rural), 2001 Census tract information on percent of persons 25 or more years old with less than 12 years of education, median household income, and further categorized into quartiles (first, second, third, and fourth), modified CCI (categorized as 0, 1, 2, or 31), and date of diagnosis data were available. Statistical Analysis. Covariates were compared between liver disease status by t test trend test for numeric variables and chi-square test for categorical variables. Temporal trends from 2004 to 2009 in proportion liver disease, as well as incidence of HCC, were described. We evaluated risk of HCC by logistic regression model while adjusting liver disease status and other covariates. In addition, adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) for within 1-year mortality were estimated by Cox s proportional hazard regression model in the HCC cohort. We did not examine the following variables owing to a high correlation: coronary artery disease (0.32; P < ) and congestive heart failure (0.49; P < ) with CCI in multivariate-adjusted models. Statistical analyses were conducted using SAS software (version 9.3; SAS Institute Inc., Cary, NC); P value <0.05 was set as statistically significant; data were reported with standard deviations (SDs) or 95% CIs. Sensitivity Analysis. We conducted a sensitivity study to investigate the estimators described above in logistic regression models for predicting risk of HCC. For the sensitivity study, first, we redesigned a randomized cohort from the total noncancer sample (n 5 224,167) for our final cancer cases (n 5 4,979) with different SEER and same frequency ratios (3:1). Next, we computed logistic regression estimators in the above cohort and findings did not change (data not shown). The study received approval from the institutional review board of Inova Fairfax Hospital (Falls Church, VA) with a signed data use agreement with the SEER- Medicare program. Results Description of Study Cohort. After inclusion and exclusion criteria, our final cohorts included 19,916 individuals. Of these, 4,979 had HCC and 14,937 were controls (matched 3:1) without HCC remained for this study (Table 1). In the HCC cohort, patients with NAFLD were older (mean ages were approximately vs years) at the time of diagnosis, more likely to be white, and not disabled with tumors, which were more likely unstaged, as compared to HCC related to viral hepatitis (HCV/HBV; all P < 0.05). Furthermore, comparing the HCC cohort to the control cohort stratified by the type of liver disease, patients with NAFLD-related HCC were older, more likely to be male, other race, and sicker than patients with NAFLD without HCC (all P < 0.05). Patients with HCV-related HCC were more likely to be male, other race, not disabled, and sicker than those with HCV without HCC (all P < 0.05). Temporal Trends in the Proportion of HCC From 2004 to Across the 6-year period, the number of cases of HCC included in our study increased from 670 (2004) to 989 (2009), an 11% annual increase. In a subanalysis assessing the proportion of HCC among patients with different types of liver disease, similar trends were noted in patients with NAFLD with a 9% annual increase from 2004 to 2009 and a 13% annual increase for HCV (all P < ; Fig. 2). Factors Associated With the Diagnosis of HCC. Using multivariate analyses, being male, being nonwhite nonblack race, and having any type of CLD and higher comorbidity scores were independently associated with HCC (Table 2). In fact, being male was associated with almost 4 times the odds of having HCC whereas having NAFLD was associated with almost 3 times increased odds of having the HCC. Mortality Trends Between 2004 and In the HCV/HBV group with HCC, approximately 50% died within 1 year whereas in the NAFLD-related HCC cohort approximately 61% died within 1 year of diagnosis (P < ). In fact, NAFLD patients with HCC had approximately 5 months shorter survival time than HCC related to viral hepatitis (HCV/HBV; all P < 0.05). The survival curve documented that the lowest survival of patients with HCC were in those with NAFLD-related HCC (Fig. 3). Factors Associated With 1-Year Mortality. We determined all factors that were independently associated with dying within the first year after the diagnosis of HCC. Using a multivariate adjusted model, older

5 HEPATOLOGY, Vol. 62, No. 6, 2015 YOUNOSSI ET AL Table 1. Characteristics of Study Cohort, SEER-Medicare, HCC Variable Total (Overall) N HCV N HBV N Autoimmune Hepatitis/ Biliary Cirrhosis N ALD N NAFLD N Age at cancer diagnosis, (10.21) (10.65) (10.25) (8.65) (8.05) (8.11) mean (SD) Survival months, (18.33) (18.63) (20.85) (17.22) (15.73) (17.14) mean (SD) Died within 1 year (%) 2,726 (54.7) 1,404 (51.3) 206 (43.7) 147 (57.9) 540 (66.1) 429 (61.2) Cause of death within 1 year (%) Primary liver cancer 2,507 (75.6) 1,304 (73.5) 230 (86.1) 128 (75.3) 446 (73.0) 399 (80.9) CLD 559 (16.8) 348 (19.6) 18 (6.7) 30 (17.6) 110 (18.0) 53 (10.7) Heart disease 120 (3.6) 58 (3.3) NA NA 29 (4.7) 18 (3.7) Male (%) 3,522 (70.7) 1,894 (69.2) 346 (73.5) 139 (54.7) 713 (87.3) 430 (61.3) Race (%) White 3,031 (60.9) 1,560 (57.0) 111 (23.6) 198 (78.0) 630 (77.1) 532 (75.9) Other 1,348 (39.0) 1982 (43) 360 (76.4) 56 (22) 187 (23.9) 169 (24.1) CCI (%) (5.6) 179 (6.5) 60 (12.7) 9 (3.5) 26 (3.2) 4 (0.6) (16.2) 476 (17.4) 115 (24.4) 29 (11.4) 92 (11.3) 95 (13.6) (14.7) 386 (14.1) 91 (19.3) 38 (15.0) 99 (12.1) 117 (16.7) 31 3,163 (63.5) 1,695 (62.0) 205 (43.5) 178 (70.1) 600 (73.4) 485 (69.2) Medicare eligibility status (%) Age based 3,380 (67.9) 1,529 (55.9) 378 (80.3) 220 (86.6) 633 (77.5) 620 (88.4) Disabled 1,543 (31.0) 1,176 (43.0) 87 (18.5) 32 (12.6) 178 (21.8) 70 (10.0) Tumor stage (%) Unstaged 582 (11.7) 295 (10.8) 41 (8.7) 37 (14.6) 121 (14.8) 88 (12.6) Localized 2,782 (55.9) 1,559 (57.0) 276 (58.6) 141 (55.5) 411 (50.3) 395 (56.3) Advanced 1,615 (32.4) 882 (32.2) 154 (32.7) 76 (29.9) 285 (34.9) 218 (31.1) Comorbidities (%) Heart Disease 1,104 (22.2%) 527 (19.3%) 71 (15.1%) 61 (24.0%) 226 (27.7%) 219 (31.2%) LT, 448 (9.0) 308 (11.3) 30 (6.4) 21 (8.3) 49 (6.0) 40 (5.7) year of diagnosis (%) (13.5) 343 (12.5) 67 (14.2) 40 (15.7) 119 (14.6) 101 (14.4) (14.8) 428 (15.6) 62 (13.2) 48 (18.9) 108 (13.2) 93 (13.3) (15.8) 442 (16.2) 72 (15.3) 28 (11.0) 146 (17.9) 99 (14.1) (18.0) 480 (17.5) 93 (19.7) 40 (15.7) 156 (19.1) 128 (18.3) (18.0) 498 (18.2) 81 (17.2) 42 (16.5) 134 (16.4) 142 (20.3) (19.9) 545 (19.9) 96 (20.4) 56 (22.0) 154 (18.8) 138 (19.7) No HCC Variable Total (Overall) HCV N 5 14,937 P Value N HBV N Autoimmune Hepatitis/ Biliary Cirrhosis N ALD N NAFLD N 5 1,243 No Liver Disease N 5 12,436 Male (%) 6,546 (43.8) < (58.8) 55 (47.0) 46 (40.0) 414 (68.5) 425 (34.2) 5,358 (43.1) Race (%) White 11,356 (76.0) < (62.1) 34 (29.1) 91 (79.1) 431 (71.4) 870 (70.0) 9,668 (77.7) Black 1,542 (10.3) (20.1) NA NA 124 (20.5) 199 (16.0) 1,116 (9.0) Other 2,038 (13.6) < (17.8) 73 (62.4) 16 (13.9) 49 (8.1) 174 (14.0) 1,651 (13.3) CCI (%) 0 9,723 (65.1) < (25.6) 30 (25.6) 24 (20.9) 186 (30.8) 32 (2.6) 9,343 (75.1) 1 2,433 (16.3) (22.0) 28 (23.9) 27 (23.5) 144 (23.8) 287 (23.1) 1,854 (14.9) 2 1,135 (7.6) < (16.1) 17 (14.5) 17 (14.8) 87 (14.4) 275 (22.1) 671 (5.4) 31 1,646 (11.0) < (36.3) 42 (35.9) 47 (40.9) 187 (31.0) 649 (52.2) 568 (4.6) Medicare eligibility status (%) Age based 11,379 (76.2) < (34.6) 83 (70.9) 88 (76.5) 344 (57.0) 873 (70.2) 9,845 (79.2) Disabled 3,523 (23.6) < (64.7) 31 (26.5) 26 (22.6) 259 (42.9) 366 (29.4) 2,568 (20.6) Comorbidities (%) Heart Disease 1599 (10.7%) < (18.7%) 25 (21.4%) 27 (23.5%) 162 (26.8%) 436 (35.1%) 870 (7.0%) Mean (SD) for numeric variables and N (%) for categorical variables. NA, not applicable. Data does not sum up to 100% due to a small sample size of ESRD (< 1% in the total final cohort). P values reported by t test or Chi-square test for comparison between the total HCC and the total No HCC. Significant (P < 0.05) by t test or Chi-square test for stratified comparison each liver disease between HCC and No HCC.

6 1728 YOUNOSSI ET AL. HEPATOLOGY, December 2015 age, having an unstaged tumor, being eligible for Medicare based on disability or end-stage renal disease (ESRD), and having NAFLD were all independently associated with risk of death within the first year after the diagnosis of HCC (Table 3). In contrast, receiving an LT and having localized tumor stage were protective against risk of mortality within 1 year. Discussion The aim of our study was to assess the association of NAFLD with HCC and the associated mortality using a population database. Our results indicate that there is an increase in the rate of HCC reported to SEER. We suspect this increase is not only related to an increase in the prevalence of liver diseases such as NAFLD (9% annual increase in the rate of NAFLD-related HCC), but also potentially owing to increased awareness of HCC screening in clinical practice. Furthermore, our data confirmed that HCV remains the most common diagnosis associated with HCC, followed by ALD and NAFLD. These data are in line with a recent report from Altekruse et al., who reported a significant increase in the crude rate of HCC (6% annually) between 2000 and It is important to note that there are some interesting differences between NAFLD and viral-hepatitis-related HCC. Patients with NAFLD and HCC were older at the time of diagnosis and more likely to be white, with 5 months shorter survival time than patients with viralhepatitis-related HCC. In addition, more patients (62%) with NAFLD-related HCC died within 1 year than those with HCC related to viral hepatitis. In fact, the majority (84.3%) of patients with NAFLD-related HCC died of their primary liver cancer. These results are interesting and potentially suggest a poorer prognosis of NAFLD-related HCC. There are probably a number of potential explanations for this poor prognosis. First, as reported previously, patients with NAFLD do not seem to get screening for HCC as compared to those with HCV. 12 Additionally, ultrasound (the current HCC screening tool) may fail to pick up small tumors in NAFLD patients owing to visceral obesity and intrahepatic fat. 13,14 Third, there are reports that NAFLD patients without cirrhosis may be at some risk for developing HCC. Given the current recommendations for HCC screening, these patients without cirrhosis with NAFLD will not be considered for screening per the current screening guidelines Owing to the lack of pathological information about non-hcc liver parenchyma, our study was unable to provide accurate information about the rate of HCC in NAFLD patients without cirrhosis. Nevertheless, given the large number of patients with NAFLD worldwide (both with and without cirrhosis) and the higher mortality rates for NAFLD-related HCC, this issue is of utmost importance and needs to be addressed. It is also important that the underlying mechanisms that connect development of HCC in non-cirrhotic NAFLD must be further investigated. 20,21 Another reason for poor prognosis of NAFLD patients with HCC may be related to the presence of metabolic comorbidities, such as cardiovascular disease. In this study, we found cardiovascular disease to be higher in our patients with NAFLD-related HCC as compared to the other HCC cohorts. Presence of cardiac comorbidities may reduce the chance of these patients to receive aggressive treatment, such as surgical resection or LT. 22,23 In fact, our data show that fewer patients with NAFLD-related HCC receive an LT as compared to other HCC patients. Furthermore, the Table 2. Multivariate Adjusted ORs With 95% CIs for HCC, SEER-Medicare, Effect OR (95% CI) Fig. 2. Trends in proportion of HCC, by liver disease status: , SEER-Medicare. Male 3.80 ( ) Race White Black 0.59 ( ) Other 1.66 ( ) CCI ( ) ( ) ( ) Liver disease HCV ( ) HBV ( ) Autoimmune hepatitis/biliary cirrhosis ( ) NAFLD 2.62 ( )

7 HEPATOLOGY, Vol. 62, No. 6, 2015 YOUNOSSI ET AL Fig. 3. Adjusted survival curves by liver disease. diagnosis of NAFLD in HCC patients was independently associated with higher 1-year mortality (Tables 1 and 3). These data are consistent with those reported from tertiary care centers, indicating a poorer prognosis for NAFLD-related HCC when compared with viralhepatitis-related HCC. 24 Table 3. Multivariate-Adjusted HRs With 95% CIs for Within 1-Year Mortality in the Cancer Cohort, SEER-Medicare, Effect HR (95% CI) Age at diagnosis of HCC 1.01 ( ) Race White Black 1.11 ( ) Other 0.81 ( ) CCI ( ) ( ) ( ) Medicare eligibility status Age-based ESRD 1.64 ( ) Disabled 1.28 ( ) Tumor stage Advanced Unstaged 1.31 ( ) Localized 0.53 ( ) Liver disease Autoimmune hepatitis/biliary cirrhosis HCV 0.95 ( ) HBV 0.96 ( ) ALD 1.27 ( ) NAFLD 1.21 ( ) LT 0.13 ( ) Year of diagnosis ( ) ( ) ( ) ( ) ( ) Providing these data from a population-based database strengthens the evidence associating NAFLD HCC and its poor prognosis. Given that NAFLD (prevalence rates: 20%-25%) and NASH (prevalence rates: 3%-5%) have a very high prevalence in the general population, these data have tremendous implications regarding the future burden of HCC in the United States. 25,26 In fact, data reported from tertiary care centers have already indicated cryptogenic cirrhosis (presumed NAFLD cirrhosis) as an important cause of HCC. 27 Our finding builds on the data from tertiary care medical centers and highlights the importance of screening these patients for HCC. Another important area of investigation is to determine the exact pathogenic mechanism that predisposes NAFLD patients to HCC. It is possible that the chronic active inflammatory state of obesity, metabolic conditions, and NAFLD may lead to hepatocarcinogenesis. 20,23 There are possibly a number of limitations to this study. First, the Medicare administrative data files provide an underestimation of the true prevalence of NAFLD. Also, some patients with cryptogenic cirrhosis included in the NAFLD cohort may have had a history of alcohol use in the past. Although we limited our cohort with full 3-year data about alcohol use, we may have not accounted for all such patients. Furthermore, we were not able to study the impact of HCC screening on the increasing rate of HCC reported to SEER. However, the data on the increasing rate of HCC in the United States are consistent with those reported by others. 1,2,19 Furthermore, we found that survival was better in patients with HCC who had a staged tumor versus those who were unstaged or advanced stage tumors. This may provide indirect evidence that these patients with staged tumors may have been picked up through a surveillance program and undergone a staging process and treatment. Another limitation of the study

8 1730 YOUNOSSI ET AL. HEPATOLOGY, December 2015 has to do with the linkage between SEER data and Medicare. Usually all cases reported by the SEER registries are matched to the Medicare database. For each matched patient, a record is created in PEDSF, and if available, their entire Medicare claims before and after diagnoses are extracted. However, depending on the year of diagnosis, this linked data may not show up until a later time point. Nevertheless, strength of the SEER data linkage with Medicare is the method used to include subjects from the same region with cancer, which can serve as controls. In summary, our analysis of this large, populationbased study suggests that NAFLD is becoming an important cause of HCC, and these rates are increasing by approximately 10% per year. Our data also showed that patients with NAFLD-related HCC suffer a worst prognosis than other causes of HCC. In fact, NAFLD is an independent predictor of mortality 1 year after the diagnosis. Furthermore, patients with NAFLD-related HCC are less likely to receive an LT. These data may help inform clinicians, clinical investigators, and policy makers to increase the awareness of the increased risk of HCC in patients with NAFLD. In this context, a screening strategy for NAFLD patients with advanced fibrosis (stage 3 or 4) is appropriate. Given the lack of strong evidence from large-scale, population-based data, it is premature to recommend screening all patients with NAFLD regardless of stage of liver disease. Nevertheless, future studies on the efficacy, effectiveness, and cost effectiveness of HCC screening programs in NAFLD patients should be undertaken. s 1. El-Serag HB. Epidemiology of viral hepatitis and hepatocellular carcinoma. Gastroenterology 2012;142: El-Serag HB. Hepatocellular carcinoma. N Engl J Med 2011;365: Kikuchi L, Oliveira CP, Carrilho FJ. Nonalcoholic fatty liver disease and hepatocellular carcinoma. Biomed Res Int 2014;2014: Anstee QM, Targher G, Day CP. Progression of NAFLD to diabetes mellitus, cardiovascular disease or cirrhosis. Nat Rev Gastroenterol Hepatol 2013;10: Ertle J, Dech^ene A, Sowa JP, Penndorf V, Herzer K, Kaiser G, et al. Non-alcoholic fatty liver disease progresses to hepatocellular carcinoma in the absence of apparent cirrhosis. Int J Cancer 2011;128: 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: White DL, Kanwal F, El-Serag HB. Association between nonalcoholic fatty liver disease and risk for hepatocellular cancer, based on systematic review. Clin Gastroenterol Hepatol 2012;10: e2. 8. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care 2002;40(8 Suppl):IV Welzel TM, Graubard BI, Zeuzem S, El-Serag HB, Davila JA, McGlynn KA. Metabolic syndrome increases the risk of primary liver cancer in the United States: a study in the SEER-Medicare database. HEPATOLOGY 2011;54: Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data. J Clin Epidemiol 2000; 53: Altekruse SF, Henley SJ, Cucinelli JE, McGlynn KA. Changing hepatocellular carcinoma incidence and liver cancer mortality rates in the United States. Am J Gastroenterol 2014;109: Piscaglia F, Baroni GS, Barchetti A, Pecorelli A, Marinelli S, Tiribelli C, et al. Clinical patterns of hepatocellular carcinoma (HCC) in non alcoholic fatty liver disease (NAFLD): a multicenter case-control study. J Hepatol 2015;62:S437. doi: /S (15) Singal AG, Yopp AC, Gupta S, Skinner CS, Halm EA, Okolo E, et al. Failure rates in the hepatocellular carcinoma surveillance process. Cancer Prev Res 2012;5: Singal AG, Yopp A, Skinner CS, Packer M, Lee WM, Tiro JA. Utilization of hepatocellular carcinoma surveillance among American patients: a systematic review. J Gen Intern Med 2012;27: Singal AG, Pillai A, Tiro J. Early detection, curative treatment, and survival rates for hepatocellular carcinoma surveillance in patients with cirrhosis: a meta-analysis. PLoS Med 2014;1:e EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012;56: Singal AG, Tiro JA, Gupta S. Improving hepatocellular carcinoma screening: applying lessons from colorectal cancer screening. Clin Gastroenterol Hepatol 2013;11: Hassett M, Yopp AC, Singal AG. Surveillance for hepatocellular carcinoma: how can we do better? Am J Med Sci 2013;346: El-Serag HB, Alsarraj A, Richardson P, Davila JA, Kramer JR, Durfee J, et al. Hepatocellular carcinoma screening practices in the Department of Veterans Affairs: findings from a national facility survey. Dig Dis Sci 2013;58: Streba LA, Vere CC, Rogoveanu I, Streba CT. Nonalcoholic fatty liver disease, metabolic risk factors, and hepatocellular carcinoma: an open question. World J Gastroenterol ;21: Noureddin M, Rinella ME. Nonalcoholic fatty liver disease, diabetes, obesity, and hepatocellular carcinoma. Clin Liver Dis 2015;19: Ortiz-Lopez C, Lomonaco R, Orsak B, Finch J, Chang Z, Kochunov VG, et al. Prevalence of prediabetes and diabetes and metabolic profile of patients with nonalcoholic fatty liver disease (NAFLD). Diabetes Care 2012;35: Lonardo A, Ballestri S, Marchesini G, Angulo P, Loria P. Nonalcoholic fatty liver disease: a precursor of the metabolic syndrome. Digest Liver Dis 2015;47: Singal AG, Conjeevaram HS, Volk ML, Fu S, Fontana RJ, Askari F, et al. Effectiveness of hepatocellular carcinoma surveillance in patients with cirrhosis. Cancer Epidemiol Biomarkers Prev 2012;21: Younossi ZM, Reyes MJ, Mishra A, Mehta R, Henry L. Systematic review with meta-analysis: non-alcoholic steatohepatitis a case for personalised treatment based on pathogenic targets. Aliment Pharmacol Ther 2013;39: Chalasani N, Younossi Z, Lavine JE, Diehl AM, Brunt EM, Cusi K, et al. The diagnosis and management of non-alcoholic fatty liver disease: practice Guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association. HEPATOLOGY 2012;55: Marrero JA, Fontana RJ, Su GL, Conjeevaram HS, Emick DM, Lok AS. NAFLD may be a common underlying liver disease in patients with hepatocellular carcinoma in the United States. HEPATOLOGY 2002; 36: