Natural History of Untreated Hepatocellular Carcinoma in a US Cohort and the Role of Cancer Surveillance

Accepted Manuscript Natural History of Untreated Hepatocellular Carcinoma in a US Cohort and the Role of Cancer Surveillance Natalia Khalaf, MD, Jun Ying, MS, Sahil Mittal, MD, MS, Sarah Temple, BA, Fasiha Kanwal, MD, MS, Jessica Davila, PhD, Hashem B. El-Serag, MD, MPH

PII: DOI: Reference:

S1542-3565(16)30517-1 10.1016/j.cgh.2016.07.033 YJCGH 54859

To appear in: Clinical Gastroenterology and Hepatology Accepted Date: 28 July 2016 Please cite this article as: Khalaf N, Ying J, Mittal S, Temple S, Kanwal F, Davila J, El-Serag HB, Natural History of Untreated Hepatocellular Carcinoma in a US Cohort and the Role of Cancer Surveillance, Clinical Gastroenterology and Hepatology (2016), doi: 10.1016/j.cgh.2016.07.033. 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.

ACCEPTED MANUSCRIPT

Title: Natural History of Untreated Hepatocellular Carcinoma in a US Cohort and the Role of Cancer Surveillance

RI PT

Short Title: Natural History of Untreated HCC

Natalia Khalaf, MD2,3; Jun Ying, MS1; Sahil Mittal, MD, MS1,3Sarah Temple, BA1; Fasiha

INSTITUTIONAL AFFILIATIONS:

1

SC

Kanwal, MD, MS1,2,3; Jessica Davila, PhD, Hashem B. El-Serag, MD, MPH1,2,3

Center of Innovation, Effectiveness and Quality,

2

M AN U

Sections of Health Services Research, Section of Gastroenterology and Hepatology at the Michael E. DeBakey Veterans Affairs Medical Center and 3Baylor College of Medicine,

Houston, Texas.

TE D

Corresponding Author: Hashem B. El-Serag, MD, MPH. Michael E. DeBakey VA Medical Center, 2002 Holcombe Blvd. (MS152), Houston, TX 77030, Phone (713) 794-8601, Fax (713)

EP

748-7359, e-mail: [email protected]

Electronic Word Count

AC C

Abstract: 263

Manuscript (including abstract, references, tables and figure legends): 6000

Figures/Tables: 8

1

ACCEPTED MANUSCRIPT

Abbreviations: hepatocellular carcinoma, HCC; United States, U.S.; hepatitis C virus, HCV; Barcelona Clinic Liver Cancer, BCLC; alpha-fetoprotein, AFP; electronic medical records, EMR; Veterans Affairs, VA; Compensation and Pension Records Interchange, CAPRI; Veterans

RI PT

Health Information Systems and Technology Architecture, VISTA; Model End-Stage Liver Disease, MELD; Eastern Cooperative Oncology Group, ECOG; ribonucleic acid, RNA; hepatitis B virus, HBV; HBV surface antigen, HBsAg; nonalcoholic fatty liver disease, NAFLD;

SC

computed tomography, CT; magnetic resonance imaging, MRI; overall survival, OS; hazard

M AN U

ratios, HR; confidence intervals, CI.

Conflict of interest: The authors have no conflicts of interest to declare.

Financial Support: This project was supported in part by the National Cancer Institute (R01 CA160738,

TE D

PI: J. Davila); NIDDK K24-04-107 and Texas Digestive Disease Center NIH DK58338 (PI: H El-Serag) the facilities and resources of the Houston Veterans Affairs Health Services Research and Development

EP

Center of Excellence (HFP90-020).

AC C

Author’’s Contributions: HES conceived of the study and participated in its design, acquired the funding, and drafted and revised the manuscript. NK assisted with the data analysis, interpreted the data, and drafted the manuscript. SM assisted with data collection, participated in design and revision of the manuscript. JY and ST cleaned and analyzed data and assisted with data interpretation. FK participated in design and revised the manuscript. JD obtained funding and revised the manuscript. All authors read, revised and approved the final manuscript.

2

ACCEPTED MANUSCRIPT

Abstract: BACKGROUND & AIMS: Determining the natural history and predictors of survival in patients with untreated hepatocellular carcinoma (HCC) in the United States is useful to test existing tumor classifications, identify subgroups of patients likely to benefit from treatment, and estimate lead time related to HCC surveillance.

RI PT

METHODS: We identified a national cohort of 518 veterans diagnosed with HCC from 2004 through 2011, with follow up ending in 2014, who received no palliative or curative treatment. We examined the association between post-diagnosis survival and patient factors, tumor characteristics, and pre-diagnosis surveillance.

M AN U

SC

RESULTS: The mean age at HCC diagnosis was 65.7 years and most patients had hepatitis C (60.6%). Almost all patients (99%) died within the observation period; the median overall survival time was 3.6 months and survival times were 13.4, 9.5, 3.4 and 1.6 months for patients of Barcelona Clinic Liver Cancer (BCLC) stages 0/A, B, C and D, respectively. In addition, model for end-stage liver disease and levels of alpha-fetoprotein were predictive of survival. Nearly 28% received pre-diagnosis HCC surveillance, which was associated with detection of disease at an earlier stage (BCLC 0/A/B, 26.4% vs 14.4%; P=.0006) and slightly longer survival than patients with no surveillance overall (5.2 vs 3.4 months, P=.021); there was no difference in survival times of patients with 0/A stage who did vs did not receive surveillance (10.3 vs 10.5 months).

TE D

CONCLUSIONS: Patients with HCCs, including those detected through surveillance, survived for short time periods in the absence of treatment, irrespective of their initial stage at diagnosis. Model for end-stage liver disease scores and levels of alpha-fetoprotein were prognostic factors, independent of BCLC stage. The lead time related to detection by surveillance was modest (<2 months) and therefore unlikely to explain the survival benefit associated with surveillance in previous studies.

AC C

EP

KEY WORDS: epidemiology; outcomes; prognosis; MELD score; AFP

3

ACCEPTED MANUSCRIPT

INTRODUCTION Hepatocellular carcinoma (HCC) is the fastest increasing cause of cancer-related deaths in the United States (U.S.) 1, 2. In addition to the rising incidence rates, and despite advances in

RI PT

diagnosis and treatment, the overall prognosis of HCC remains poor with an estimated 5-year survival rate of only 12% 1. A meta-analysis of observational studies reported that HCC

treatment is underused, with only 22% and 53% of patients receiving curative and non-curative

SC

treatments, respectively 3. However, there are few data on the natural history and determinants of survival in patients with untreated HCC, particularly those with hepatitis C virus (HCV)

M AN U

infection. These data are useful to test and potentially modify existing tumor prognosis classifications, identify subgroups that are likely to benefit in future clinical trials, and quantify the extent of lead time related to HCC surveillance.

TE D

Several non-U.S. studies reported heterogeneous survival rates among patients with untreated HCC, most commonly using the Barcelona Clinic Liver Cancer (BCLC) staging as a prognostic guide 4-7. However, additional variables not included in the BCLC staging, such as alpha-

EP

fetoprotein (AFP) level, have been implicated to predict survival in untreated HCC cohorts 5, 8. In the U.S., where HCV infection is the major cause of HCC 1, HCC screening remains a topic of

AC C

much debate 9, 10. This is because most of the evidence supporting routine surveillance stems from observational cohort and case control studies with no randomized trials on surveillance in patients with HCV 9, 11, 12. The possibility of lead time bias is a main limitation of these studies and the quantification as well as adjustment for this bias in the presence of HCC treatment is difficult. Studying untreated HCC patients who were detected with and without surveillance allows for a better understanding of the extent of lead time bias.

4

ACCEPTED MANUSCRIPT

Using data from a national cohort of 518 veteran patients with untreated HCC, we investigated the natural history of HCC in a mostly HCV-positive cohort, identified prognostic variables

AC C

EP

TE D

M AN U

SC

surveillance on survival in the absence of treatment.

RI PT

associated with survival outside the BCLC classification, and studied the impact of HCC

5

ACCEPTED MANUSCRIPT

MATERIALS AND METHODS Data Sources Study data were obtained from manual review of electronic medical records (EMR) combined

RI PT

with automated Department of Veterans Affairs (VA) administrative data files. EMR reviews were conducted in the VA Compensation and Pension Records Interchange (CAPRI) and

Veterans Health Information Systems and Technology Architecture (VISTA) systems from

SC

nationwide VA facilities. The automated datasets included the Medical SAS (MedSAS)

Outpatient and Inpatient files containing International Classification of Diseases, 9th Revision,

M AN U

Clinical Modification (ICD-9-CM) diagnoses and Common Procedural Terminology (CPT) codes. Date of death, if any, was determined by the Vital Status file, which uses an algorithm from information found in the VA MedSAS Inpatient File, Beneficiary Identification & Records Locator System Death File, Medicare Vital Status file, and Social Security Administration death

Study Population

TE D

file.

EP

We identified 10,695 patients with a suspected HCC diagnosis in all VA hospitals nationwide between October 1, 2004 and September 30, 2011 based on the presence of ICD-9 CM code

AC C

155.0 (malignant neoplasm of liver) and in the absence of ICD-9 CM code 155.1 (intrahepatic cholangiocarcinoma) 13. Based on a desired sample size of 1500 cases with definite HCC, we selected a random computer generated sample of patients with suspected HCC for structured chart review to determine and verify the study eligibility criteria, and included only patients with HCC diagnosis by histopathology or imaging criteria according to the 2005 American Association for the Study of Liver Disease or European Association for the Study of Liver

6

ACCEPTED MANUSCRIPT

Disease guidelines 14-15. We also required the following additional inclusion criteria: recent VA healthcare utilization (at least one inpatient or outpatient encounter at any VA facility within the 1-year prior to the date of HCC diagnosis); no HCC diagnosis made prior to the study period. We

with verified HCC who met the study criteria (Figure 1).

SC

Patient Characteristics

RI PT

reviewed charts of 2,719 patients with suspected HCC to arrive at 1500 unique study subjects

We ascertained demographic features (age, gender, and race), clinical characteristics (etiology

M AN U

and severity of liver disease, BCLC stage at diagnosis), medical and mental conditions from EMR review. We defined confirmed cirrhosis based on the following hierarchy in descending order: liver biopsy results at any time before or at the time of diagnosis of HCC, features suggestive of cirrhosis on abdominal imaging, clinical complications of cirrhosis (ascites, hepatic

TE D

encephalopathy, varices), or laboratory evidence consisting of abnormal values of two of three laboratory tests (albumin <3.0 g/L, platelets <200 000/lL, INR >1.1 between 6 months before and 4 weeks after HCC diagnosis) or an APRI (AST to platelet ratio index) score >2.0 16. Model

EP

End-Stage Liver Disease (MELD) and Child-Pugh scores were calculated from laboratory tests and clinical features obtained from progress notes. BCLC stage (stage 0, very early; A, early; B,

AC C

intermediate; C, advanced; and D, end-stage) was determined by physician documented performance status, Child-Pugh score, tumor size and number, and portal vein involvement. Patients with Eastern Cooperative Oncology Group (ECOG) 0-2 were classified as having “good” performance status, and ECOG ≥3 were classified as having “poor” performance status. HCC characteristics (number of nodules, size of largest nodule, presence of portal vein invasion or thrombosis and metastasis) were determined from cross-sectional imaging reports.

7

ACCEPTED MANUSCRIPT

Risk Factors for HCC We defined HCV infection as positive anti-HCV antibody or HCV ribonucleic acid (RNA), and

RI PT

hepatitis B virus (HBV) infection as at least one instance of positive HBV surface antigen (HBsAg). Alcohol abuse was defined as history of more than 3 drinks daily, documented alcoholism or alcohol abuse by a health care provider, participation in a substance abuse

SC

treatment program or a history of alcoholic hepatitis. Nonalcoholic fatty liver disease (NAFLD) was defined by histopathology (steatosis on biopsy) or presence of metabolic syndrome in the

M AN U

absence of other causes of chronic liver disease (HCV, HBV, alcohol abuse, primary biliary cirrhosis, primary sclerosing cholangitis, autoimmune hepatitis, hemochromatosis, or Wilson disease). We defined metabolic syndrome using U.S. National Cholesterol Education Program Adult Treatment Panel III (NCEP-ATP III) guidelines 17, except for replacing the elevated waist

HCC Surveillance

TE D

circumference criterion with body mass index (BMI) > 28.8 kg/m2 in both men and women 18.

EP

Patients were classified as having received HCC surveillance if they underwent abdominal imaging and /or AFP surveillance within 2 years of HCC diagnosis. Surveillance by imaging was

AC C

defined as receipt of any liver ultrasound, computed tomography (CT) scan, or magnetic resonance imaging (MRI) in the 2 years prior to HCC diagnosis with surveillance as the listed indication, in the setting of liver transplantation evaluation, follow up of a known non-malignant liver mass, or imaging in a patient with indication listed only as “cirrhosis”. AFP surveillance was defined as receipt of two or more AFP tests at least 6 months apart in the 2 years prior to HCC diagnosis.

8

ACCEPTED MANUSCRIPT

Those who were diagnosed with HCC as a result of symptoms, imaging for an indication other than those listed above, or incidentally during an evaluation for other signs and symptoms were

RI PT

classified in the non-surveillance group. Two hepatologists (HES, SM) manually reviewed the EMR including all imaging reports to determine the presence of pre diagnosis HCC surveillance.

SC

Statistical Analysis

Survival time was defined as time in months from HCC diagnosis to death, last VA encounter or

M AN U

end of study period (October 31, 2014). To identify predictors of overall survival (OS), we compared the demographic, clinical and tumor features between groups of untreated HCC patients who survived < 12 vs. ≥ 12 months using chi-square or Fisher’s exact test for categorical

TE D

variables.

Stepwise Cox regression was performed to identify variables that were associated with 12-month and OS. Potential predictors (age, race, BCLC stage, cirrhosis status, disease etiology variables,

EP

MELD score, AFP level, and surveillance status) that were significantly associated with survival in univariate analyses (P<0.1) were used as input variables. Race, age, surveillance and variables

AC C

with P<0.05 in multivariable analyses were retained in the final models. Hazard ratios (HR) and their 95% confidence intervals (CI) were calculated.

To further examine the possible effect of HCC surveillance on observed survival among patients with untreated HCC, Kaplan-Meier analysis was used to compare OS by BCLC stage in patients with and without HCC surveillance. Mean and median survival times were calculated and log-

9

ACCEPTED MANUSCRIPT

rank tests were performed to compare the distribution among patient groups. Analyses were

AC C

EP

TE D

M AN U

SC

RI PT

performed using SAS version 9.4 (SAS Institute, Cary, NC).

10

ACCEPTED MANUSCRIPT

RESULTS Of 1500 patients with verified HCC, 518 patients did not receive curative or palliative HCC treatment including resection, liver transplantation, ablation, chemoembolization or

RI PT

antineoplastic agents (Figure 1) and the other 982 received at least one treatment specific for HCC. As compared to HCC patients who received treatment, the no treatment group was older, had more advanced BCLC stage at time of diagnosis, higher Child Pugh score and less HCV

SC

(Table 1).

M AN U

The study cohort consisted of 518 patients without any HCC specific treatment. A total of 14 (2.7%) patients had BCLC Stage 0/A, 76 (14.7%) had stage B, 227 (43.8%) had stage C and 166 (32.0%) had stage D disease at the time of diagnosis. (Table 2) The main reasons for absence of HCC treatment in patients with early stage disease (0/A/B) were treatment refusal (37.8%),

TE D

medical ineligibility based on comorbidities (23.3%), disease progression before planned treatment (10.0%) or loss to follow up (10.0%). The mean age at time of HCC diagnosis was 65.7 years, and all but two participants were male (99.6%). Most patients were non-Hispanic

EP

whites (59.5%), followed by African Americans (28.0%) and Hispanics (9.6%). Most patients had HCV (314, 60.6%) or alcohol abuse (411, 79.3%), whereas NAFLD (48, 9.3%), HBV (27,

AC C

5.2%), and other causes (18, 3.5%) of liver disease were less common. Most patients had cirrhosis (426, 82.2%) and were Child-Pugh Class B (227, 43.8%) at time of HCC diagnosis.

In multivariate analyses, BCLC stage was highly predictive of 12-month mortality risk (stage A vs. D: adjusted HR 0.16, 95% CI 0.07-0.40). Similarly, MELD score (10-19 vs. < 10: adjusted HR 1.60, 95% CI 1.27-2.02; ≥ 20 vs. < 10: adjusted HR 1.83, 95% CI 1.25-2.68) and AFP level

11

ACCEPTED MANUSCRIPT

(≥ 1000 vs. < 10; adjusted HR 1.89; 95% CI 1.44-2.49) were significantly associated with 12month mortality. There was a non-significant trend towards decreased mortality risk in patients

Similar results were found with overall mortality risk. (Table 3)

RI PT

who were enrolled in HCC surveillance programs (adjusted HR 0.84; 95% CI 0.66-1.06).

The vast majority of patients (99.4%) died within the observation period, with a median OS of

SC

3.6 months (IQR=1.4-9.1) and mean OS of 7.6 months (SE=0.49) and a range of 0.1-84.5

months. (Table 4, Figure 2) Only 95 (18.3%) participants were alive at 12 months following

M AN U

HCC diagnosis. The 12-month survival rates for BCLC stage 0/A, B, C and D disease were 64%, 37%, 13% and 7%, respectively. (Supplementary Table 1) Age, race, presence of cirrhosis, etiology of liver disease, performance status, degree of hepatic encephalopathy or enrollment in surveillance programs were not associated with survival at 12 months in univariate analyses. In

TE D

contrast, stage D disease, Child-Pugh Class C, elevated MELD and AFP levels were associated with decreased rates of 12-month survival (P<0.0001). (Table 2)

EP

Of all patients, 144 (27.8%) underwent HCC surveillance, and these patients were more likely to be diagnosed with earlier stage BCLC 0/A/B disease (38/144, 26.4%) compared to those who did

AC C

not receive surveillance (50/348, 14.4%; P=0.0006). (Supplementary Table 2) The difference in median OS between those who were enrolled in surveillance programs and those who were not was only 1.8 months. Within patients in whom BCLC stage was known, this survival difference ranged from 0.2 months for stage 0/A/B to 0.5 months in stage C to 0.8 months in stage D. (Table 5)

12

ACCEPTED MANUSCRIPT

The Kaplan-Meier curves showed the probability of survival stratified by BCLC stage (Figure 2) and surveillance status (Figure 3). As expected, those with lower BCLC stage at diagnosis had significantly longer median OS compared to those with higher stage disease (BCLC 0/A: 13.4

RI PT

months; BCLC B: 9.5 months; BCLC C: 3.4 months; BCLC D: 1.6 months; log-rank P

<0.0001). Furthermore, pairwise comparison showed significant differences in survival between patients with contiguous BCLC stages (B vs. C, P <0.0001; C vs. D, P=0.0001). (Table 4, Figure

SC

2) Patients with pre HCC surveillance also had significantly longer median OS compared to those not in surveillance programs (5.2 vs. 3.4 months; log-rank P=0.021), however, all except 3

M AN U

patients died within the observation period. (Figure 3, Table 5). Of note, there was no difference in median OS in earlier stage disease (0/A/B) when stratified by the presence or absence of surveillance (10.3 vs. 10.5 months, respectively; log-rank P=0.917). Similarly, stratification by surveillance status in higher stage disease showed no difference in mortality (log-rank P=0.375).

AC C

EP

TE D

(Table 5)

13

ACCEPTED MANUSCRIPT

DISCUSSION Studying the natural history of untreated HCC is critical for understanding the prognosis and prognostic factors of HCC and the contribution of surveillance to lead time bias 19. With the

RI PT

exception of few prior studies 5, 7, only one of which focused on HCV-related HCC, our

understanding of the natural history of HCC is largely limited to the reported OS among patients observed in the placebo or non-treatment arms of clinical trials 4. In this study, OS in 518

SC

patients with untreated HCC was, while heterogeneous, very short even among those with early stage disease or those in surveillance programs. Specifically, patients with earlier stage HCC

M AN U

who were diagnosed on the basis of surveillance gained no additional survival benefit in the absence of HCC treatment.

Our study confirmed the importance of BCLC classification as an important predictor of OS.

TE D

The BCLC classification combines tumor features, liver function and functional status to estimate HCC prognosis and guide treatment decisions 20. In our study, BCLC stage was significantly associated with OS with significant differences among patients in contiguous stages

EP

(B vs. C and C vs. D), supporting the utility of this classification system. Additionally, we also found that others factors, like MELD scores and AFP levels, confer additional independent

AC C

prognostic value in untreated patients with HCC. These findings are supported by previous studies describing an association between HCC survival and MELD score components including total bilirubin 8, 21, elevated blood urea nitrogen (BUN) levels 8, and elevated INR 5 as well as AFP levels 5, 8. As BCLC stage correlates with tumor burden, AFP with tumor biology, and MELD with degree of cirrhosis, each of these components represents a unique aspect of the individual HCC patient. In our study, these variables were the most significant predictors of

14

ACCEPTED MANUSCRIPT

survival in untreated HCC, and it is possible that combining one or more in future prospective studies might result in even greater prognostic ability.

RI PT

Most patients in this study (75.8%) had advanced BCLC stage C or D disease at time of cancer diagnosis, with a median OS of 3.4 and 1.6 months, respectively. However, even for patients with BCLC stages 0/A and B, the median survival was only 13.4 and 9.5 months, respectively.

SC

While it may seem surprising that neither curative nor palliative therapy was used in patients with early stage disease, a previous systematic review of HCC patients found that only 22% and

M AN U

53% received curative or palliative therapy, respectively, and only 59% of the subgroup with early HCC received curative treatment 3. We identified and reported documented and inferred reasons for not receiving HCC treatment. Our reported survival rates were generally shorter compared to those of other studies. In Cabibbo et al.’s study of 320 predominately HCV-related

TE D

Italian HCC patients and Gainnini et al.’s study of 600 predominately HBV-related Italian HCC patients, median OS was 6.8 and 9 months, respectively 5, 7. The reasons for the observed shorter survival compared to other studies likely include the unselected nature of our study population,

EP

which represents the spectrum of HCC irrespective of extent of prior care and the high burden of

AC C

comorbid conditions (such as alcohol abuse in 79% of patients).

Approximately 28% of the study cohort received pre diagnosis surveillance. These patients were more likely to have earlier stage disease (BCLC 0/A/B), be alive at 12 months, and have longer OS compared to those who did not receive surveillance. Lead time likely explains these findings given that no HCC treatment was received, but it does not seem to add more than one to two months to the observed OS compared to stage matched cases who did not receive HCC

15

ACCEPTED MANUSCRIPT

surveillance. Thus, surveillance did not merely identify HCC patients with less aggressive tumors who were destined to do well as part of their natural history. We believe our data provide support to a real survival benefit associated with HCC surveillance programs – programs that

RI PT

result in diagnosis at an earlier stage followed by stage appropriate therapy as reported by previous studies 22-26.

SC

In our study, the overall 1-, 2- and 3-year survival rates were 64%, 36% and 7%, respectively, for BCLC stage A disease. Little is known about the course of untreated small HCC (a single lesion

M AN U

<5cm in diameter), an increasingly important group given the spread of surveillance programs 26. Curative treatment of early or small HCC is associated with a 50-70% 5-year survival rate 1, 16, 2831

; however, in the absence of randomized prospective controls, it is possible that patients with

early or small HCC have a favorable survival irrespective of treatment 32, 33. Within the confines

TE D

of an overall poor survival with HCC, predicting outcomes in small HCC seems unreliable, and it is possible that some patients with small HCC have relatively long survival irrespective of treatment. Further studies are needed to elucidate prognostic features in this growing subclass of

EP

small HCC.

AC C

This study has few limitations. The retrospective data collection could have led to misclassification of HCC stage and some of the prognostic variables. However, the availability of detailed EMRs including lab results, radiology images and reports, and pathology reports from all VA facilities coupled with the systematic, standardized and thorough data abstraction has minimized the subjectivity of defining the study variables. Our largely male cohort did not allow for differences in gender to be studied and resulted in limited generalizability to women7. We

16

ACCEPTED MANUSCRIPT

employed a wide definition of surveillance to account for change in HCC surveillance guidelines during the study period and therefore an undetermined degree of misclassification could have occurred 34,35. Last, most patients had advanced disease at time of diagnosis and did not have

RI PT

HCC surveillance, which limited the power of our analyses for the subset with early stage

disease and prior HCC surveillance. Although many VA patients get their care exclusively at the VA, some patients with dual insurance coverage (e.g. Medicare for elderly patients) may get

SC

some of their care outside the VA. However, we identified treatment received outside the VA in

M AN U

our review of the medical records.

This is the only U.S. study regarding the natural history of patients with untreated HCC, and the largest on this topic in an HCV-infected cohort. MELD scores and AFP levels predicted mortality independent of BCLC staging. Those who received pre diagnosis HCC surveillance

TE D

had earlier stage disease compared to those not in surveillance programs. However, the survival benefit without HCC treatment was quiet modest (1.8 months overall) and therefore unlikely to explain the effectiveness of surveillance reported by most previous observational studies. These

EP

findings argue surveillance programs in combination with more precise risk stratification are

AC C

needed for appropriate patient management and improved HCC outcomes.

17

ACCEPTED MANUSCRIPT

ACKNOWLEDGEMENTS: This project was supported in part by the National Cancer Institute (R01 CA160738, PI: J. Davila); NIDDK K24-04-107 and Texas Digestive Disease Center NIH DK58338 (PI: H El-Serag) the facilities

RI PT

and resources of the Houston Veterans Affairs Health Services Research and Development Center of

AC C

EP

TE D

M AN U

SC

Excellence (HFP90-020).

18

ACCEPTED MANUSCRIPT

REFERENCES 1. Surveillance, Epidemiology, and End Results (SEER) Program. SEER*Stat database: incidence — SEER 9 Regs research data, Nov 2009 Sub (1973-2007). Bethesda, MD:

RI PT

National Cancer Institute, April 2010.

2. El-Serag HB. Hepatocellular carcinoma. N Engl J Med 2011;365:1118-1127.

3. Tan D, Yopp A, Beg MS, et al. Meta-analysis: underutilisation and disparities of

Pharmacol Ther 2013;38:703-712.

SC

treatment among patients with hepatocellular carcinoma in the United States. Aliment

M AN U

4. Cabibbo G, Enea M, Attanasio M, et al. A metaanalysis of survival rates of untreated patients in randomized clinical trials of hepatocellular carcinoma. Hepatology 2010;51:1274-1283.

5. Cabibbo G, Maida M, Genco C, et al. Natural history of untreatable hepatocellular

TE D

carcinoma: a retrospective cohort study. World J Hepatol 2012;4:256-261. 6. Forner A, Llovet JM, Bruix J. Hepatocellular carcinoma. Lancet 2012;379:1245-1255. 7. Giannini EG, Farinati F, Ciccarese F, et al. Italian Liver Cancer (ITA.LI.CA) group.

EP

Prognosis of untreated hepatocellular carcinoma. Hepatology 2015;61:184-190. 8. Yeung YP, Lo CM, Liu CL, et al. Natural history of untreated nonsurgical hepatocellular

AC C

carcinoma. Am J Gastroenterol 2005;100:1995-2004.

9. Chen JG, Parkin DM, Chen QG, et al. Screening for liver cancer: results of a randomised controlled trial in Qidong, China. J Med Screen 2003;10:204-209.

10. Davila JA, Henderson L, Kramer JR, et al. Utilization of surveillance for hepatocellular carcinoma among hepatitis C virus-infected veterans in the United States. Ann Intern Med 2011;154:85-93.

19

ACCEPTED MANUSCRIPT

11. Trevisani F, De NS, Rapaccini G, et al. Italian Liver Cancer Group. Semiannual and annual surveillance of cirrhotic patients for hepatocellular carcinoma: effects on cancer stage and patient survival (Italian experience). Am J Gastroenterol 2002;97:734-744
.

RI PT

12. Sherman M. Surveillance for hepatocellular carcinoma and early diagnosis. Clin Liver Dis 2007;11:817-837.

13. Davila JA, Weston A, Smalley W, et al. Utilization of screening for hepatocellular

SC

carcinoma in the United States. J Clin Gastroenterol 2007;41:777-782.

2011;53:1020-1022.

M AN U

14. Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. Hepatology

15. EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012;56:908-943.

16. Mittal S, El-Serag HB, Sada YH, et al. Hepatocellular Carcinoma in the Absence of

TE D

Cirrhosis in United States Veterans is Associated With Nonalcoholic Fatty Liver Disease. Clin Gastroenterol Hepatol. 2016;14(1):124-131. 17. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in

EP

Adults. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood

AC C

Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285(19):2486-2497.

18. Ascha MS, Hanouneh IA, Lopez R, et al. The incidence and risk factors of hepatocellular carcinoma in patients with nonalcoholic steatohepatitis. Hepatology 2010;51(6):1972-

1978.

19. Singal AG, El-Serag HB. Hepatocellular Carcinoma From Epidemiology to Prevention: Translating Knowledge into Practice. Clin Gastroenterol Hepatol. 2015;13:2140-2151.

20

ACCEPTED MANUSCRIPT

20. Llovet JM, Bru C, Bruix J. Prognosis of hepatocellular carcinoma: the BCLC staging classification. Semin Liver Dis 1999;19:329–338. 21. Pawarode A, Voravud N, Sriuranpong V, et al. Natural history of untreated primary

1998;21:386-391.

RI PT

hepatocellular carcinoma: a retrospective study of 157 patients. Am J Clin Oncol

22. Wong LL, Limm WM, Severino R, Wong LM. Improved survival with screening for

SC

hepatocellular carcinoma. Liver Transpl 2000;6:320–325.

23. Yuen MF, Cheng CC, Lauder IJ, et al. Early detection of hepatocellular carcinoma

M AN U

increases the chance of treatment: Hong Kong experience. Hepatology 2000;31:330–335. 24. Trevisani F, De NS, Rapaccini G, et al. Semiannual and annual surveillance of cirrhotic patients for hepatocellular carcinoma: effects on cancer stage and patient survival (Italian experience). Am J Gastroenterol 2002;97:734–744.

TE D

25. Trevisani F, Cantarini MC, Labate AM, et al. Surveillance for hepatocellular carcinoma in elderly Italian patients with cirrhosis: effects on cancer staging and patient survival. Am J Gastroenterol 2004;99:1470–1476.

EP

26. El Serag HB, Siegel AB, Davila JA, et al. Treatment and outcomes of treating of hepatocellular carcinoma among Medicare recipients in the United States: a population-

AC C

based study. J Hepatol 2006;44:158–166.

27. Zhang BH1, Yang BH, Tang ZY. Randomized controlled trial of screening for hepatocellular carcinoma. J Cancer Res Clin Oncol 2004;130:417-422.

28. Llovet JM, Bruix J. Early diagnosis and treatment of hepatocellular carcinoma. Baillieres Best Pract Res Clin Gastroenterol 2000;14:991–1008.

21

ACCEPTED MANUSCRIPT

29. Bruix J, Llovet JM. Prognostic prediction and treatment strategy in hepatocellular carcinoma. Hepatology 2002;35:519–524.

1917

RI PT

30. Llovet JM, Burroughs A, Bruix J. Hepatocellular carcinoma. Lancet 2003;362:1907–

31. Ioannou GN, Perkins JD, Carithers RL Jr. Liver transplantation for hepatocellular

Gastroenterology 2008;134:1342–1351.

SC

carcinoma: impact of the MELD allocation system and predictors of survival.

32. Cottone M, Virdone R, Fusco G, et al. Asymptomatic hepatocellular carcinoma in Child's

1989;96:1566-1571.

M AN U

A cirrhosis. A comparison of natural history and surgical treatment. Gastroenterology

33. Barbara L, Benzi G, Gaiani S, et al. Natural history of small untreated hepatocellular carcinoma in cirrhosis: a multivariate analysis of prognostic factors of tumor growth rate

TE D

and patient survival. Hepatology 1992;16:132-137.

34. McGowan CE, Edwards TP, Luong MU, et al. Suboptimal surveillance for and knowledge of hepatocellular carcinoma among primary care providers. Clin

EP

Gastroenterol Hepatol. 2015;13:799-804. 35. Dalton-Fitzgerald E, Tiro J, Kandunoori P, et al. Practice patterns and attitudes of

AC C

primary care providers and barriers to surveillance of hepatocellular carcinoma in patients with cirrhosis. Clin Gastroenterol Hepatol. 2015;13:791-798.

22

ACCEPTED MANUSCRIPT

Tables Table 1. Characteristics of HCC patients who received and did not receive any treatment

68 (13.1%) 138 (26.6%) 91 (17.6%) 48 (9.3%) 173 (33.4%)

159 (16.2%) 275 (28.0%) 238 (24.2%) 118 (12.0%) 192 (19.6%)

308 (59.5%) 145 (28.0%) 50 (9.7%) 15 (2.9%)

588 (59.9%) 248 (25.3%) 129 (13.1%) 17 (1.7%)

0.09

14 (2.7%) 76 (14.7%) 227 (43.8%) 166 (32.1%) 426 (82.2%) 314 (60.6%) 27 (5.2%) 411 (79.3%) 48 (9.3%)

182 (18.5%) 275 (28.0%) 337 (34.3%) 87 (8.9%) 775 (78.9%) 701 (71.4%) 42 (4.3%) 798 (81.3%) 72 (7.3%)

<0.00001

169 (34.6%) 227 (46.4%) 93 (19.0%)

511 (55.7%) 349 (38.1%) 57 (6.2%)

<0.00001

205 (39.6%) 313 (60.4%)

385 (39.2%) 597 (60.8%)

0.91

TE D

AC C

EP

RI PT

P-value

<0.00001

SC

>70 Race White African American Hispanic Other BCLC Stage* 0/A (early) B (intermediate) C (advanced) D (end stage) Cirrhosis* HCV HBV Alcohol Use NAFLD Child-Pugh Class* A B C ECOG Performance Status Good Poor

Received Treatment (n=982)

M AN U

Age (years) <55 56-60 61-65 66-70

Received No treatment (n=518)

0.14 <0.0001 0.36 0.41 0.19

*Missing data not included

23

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 2: Potential determinants of survival among 518 patients with untreated hepatocellular carcinoma. Total Survival Survival Survival ≥12 Months P-value¥ <12 Months Months N=518 N=423 (%) N=95 (%) (mean) Age (years) <55 68 62 (91.2) 6 (8.8) 5.55 .0927 7.38 56-60 138 113 (81.9) 25 (18.1) 6.34 61-65 91 77 (84.6) 14 (15.4) 6.63 66-70 48 39 (81.2) 9 (18.8) 9.10 >70 173 132 (76.3) 41 (23.7) Race 7.84 White 308 249 (80.8) 59 (19.2) 0.1426 6.02 African American 145 125 (86.2) 20 (13.8) 11.30 Hispanic 50 36 (72.0) 14 (28.0) 5.66 Other 15 13 (86.7) 2 (13.3) BCLC Stage* 19.32 0/A (early) 14 5 (35.7) 9 (64.3) <0.0001 13.43 B (intermediate) 76 48 (63.2) 28 (36.8) C (advanced) 227 198 (87.2) 29 (12.8) 6.01 3.95 D (end stage) 166 154 (92.8) 12 (7.2) 7.24 426 354 (83.1) 72 (16.9) .0687 Cirrhosis* 314 259 (82.5) 55 (17.5) 7.14 0.5478 HCV 27 22 (81.5) 5 (18.5) 6.37 1.000 HBV 411 334 (81.3) 77 (18.7) 7.65 .6489 Alcohol Use 8.44 48 38 (79.2) 10 (20.8) 0.6393 NAFLD Child-Pugh Class* 10.49 A 169 124 (73.4) 45 (26.6) <0.0001 6.32 B 227 194 (85.5) 33 (14.5) 3.60 C 93 88 (94.6) 5 (5.4) MELD Score* 9.21 <10 155 120 (77.4) 35 (22.6) <0.0001 6.49 10-19 259 223 (86.1) 36 (13.9) 3.07 ≥20 45 43 (95.6) 2 (4.4) AFP level* 8.70 <10 99 83 (83.8) 16 (16.2) <0.0001 10.11 10-100 95 65 (68.4) 30 (31.6) 6.85 100-1000 103 85 (82.5) 18 (17.5) >1000 168 156 (92.9) 12 (7.1) 3.74 9.29 144 111 (77.1) 33 (22.9) 0.2125 In a Surveillance Program* ECOG Performance Status 7.44 Good 205 171 (83.4) 34 (16.6) .4037 7.68 Poor 313 252 (80.5) 61 (19.5)

24

ACCEPTED MANUSCRIPT

Total N=518

Survival Survival <12 Months ≥12 Months N=423 (%) N=95 (%)

Survival Months (mean)

P-value¥

AC C

EP

TE D

M AN U

SC

RI PT

Encephalopathy* None 438 353 (80.6) 85 (19.4) 7.93 0.5237 6.31 Mild 46 39 (84.8) 7 (15.2) 5.20 Severe 6 6 (100.0) 0 (0) Ascites 10.93 None 244 173 (70.9) 71 (29.1) <0.0001 4.83 Mild 202 183 (90.6) 19 (9.4) 4.07 Severe 72 67 (93.1) 5 (6.9) Size of Largest Lesion* 11.64 <5cm 193 129 (66.8) 64 (33.2) <0.0001 6.46 5-10cm 154 133 (86.4) 21 (13.6) 4.33 ≥10 cm 152 142 (93.4) 10 (6.6) Lesion Number* 8.95 0.5164 Uninodular 186 148 (79.6) 38 (20.4) 6.85 Multinodular 331 274 (82.8) 57 (17.2) 3.60 Portal Vein Invasion or 176 168 (95.5) 8 (4.5) <0.0001 Thrombosis* 2.81 86 83 (96.5) 3 (3.5) 0.0001 Metastatic Lesions* BCLC, Barcelona Clinic Liver Cancer; HCV, hepatitis C virus; HBV, hepatitis B virus; NAFLD, nonalcoholic fatty liver disease; MELD, Model End-Stage Liver Disease score; AFP, alphafetoprotein; ECOG, Eastern Cooperative Oncology Group. ¥ Chi-square or Fisher’s exact test used for values <5. *Missing data not included.

25

ACCEPTED MANUSCRIPT

Table 3: Multivariate Cox regression models of factors associated with mortality in 518 patients with untreated hepatocellular carcinoma.

P-value

BCLC Stage 0/A B C D

0.29 (0.16-0.52) 0.36 (0.26-0.48) 0.69 (0.55-0.85) Ref.

<0.0001

MELD Score < 10 10-19 ≥ 20

Ref. 1.55 (1.25-1.91) 1.86 (1.29-2.68)

0.0002

AFP Level <10 10-100 100-1000 >1000

Ref. 0.83 (0.62-1.11) 1.26 (0.95-1.68) 1.86 (1.44-2.41)

<0.0001

SC

HR (95% CI)

M AN U

Variable (N=)

RI PT

Overall Mortality*

AC C

EP

TE D

Surveillance Program Yes 0.83 (0.67-1.03) .0521 No Ref. BCLC, Barcelona Clinic Liver Cancer; MELD, Model End-Stage Liver Disease score; AFP, alpha-fetoprotein. ± 95 censored. * 3 censored.

26

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 4: Overall survival in 518 patients with untreated hepatocellular carcinoma based on Kaplan-Meier analysis. Log-rank Log-rank PMean Survival Median Survival Range P-value of value of Median Months (SE) Months (IQR) Months Mean Survival Survival All Patients 7.60 (0.49) 3.62 (1.38-9.05) 0.1-84.5 BCLC Stage 0/A (n=14) 19.32 (4.72) 13.37 (8.19-27.14) 2.9-71.3 <0.0001 A vs. B 0.24 B (n=76) 13.43 (1.54) 9.54 (5.13-16.30) 0.9-84.5 B vs. C <0.0001 C (n=227) 6.01 (0.56) 3.36 (1.61-6.25) 0.2-60.0 C vs. D 0.0001 D (n=166) 3.95 (0.46) 1.56 (0.66-4.28) 0.1-29.8 BCLC, Barcelona Clinic Liver Cancer; SE, Standard error; IQR, Interquartile range.

27

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 5: Overall survival in 492 patients with untreated hepatocellular carcinoma and known surveillance status based on Kaplan-Meier analysis. Log-rank Mean Survival Median Survival P-value of Months (SE) Months (IQR) Median Survival Enrolled in Surveillance Program Yes (n=144) 9.29 (1.02) 5.21 (2.02-11.04) 0.021 No (n=348) 7.08 (0.57) 3.39 (1.18-8.21) Stage 0/A/B (n=88) Yes (n=38) 14.95 (2.39) 10.33 (5.16-17.07) 0.917 No (n=50) 14.27 (1.99) 10.51(5.66-18.65) Stage C (n=212) Yes (n=45) 6.93 (1.10) 3.91(2.14-8.65) 0.310 No (n=167) 5.97 (0.69) 3.42 (1.57-5.59) Stage D (n=161) Yes (n=49) 4.21 (0.79) 2.07 (0.79-5.56) 0.375 No (n=112) 3.71 (0.57) 1.23 (0.59-3.42) SE, standard error; IQR, Interquartile range.

28

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Supplementary Table 1: Survival Data: Time from HCC diagnosis to death in 518 patients with untreated hepatocellular carcinoma by BCLC staging.* Survival Rate % (n/N) 6 Month Survival Rate All Participants 33.0% (171/518) Stage 0/A 85.7% (12/14) Stage B 64.5% (49/76) Stage C 25.1% (57/227) Stage D 16.9% (28/166) Unknown Stage 71.4% (25/35) 12 Month Survival Rate All Participants 18.3% (95/518) Stage 0/A 64.3% (9/14) Stage B 36.8% (28/76) Stage C 12.8% (29/227) Stage D 7.2% (12/166) Unknown Stage 48.6% (17/35) 24 Month Survival Rate All Participants 7.9% (41/518) Stage 0/A 35.7% (5/14) Stage B 15.8% (12/76) Stage C 5.3% (12/227) Stage D 3.0% (5/166) Unknown Stage 20.0% (7/35) 36 Month Survival Rate All Participants 2.9% (15/518) Stage 0/A 7.1% (1/14) Stage B 9.2% (7/76) Stage C 1.3% (3/227) Stage D 0% (0/166) Unknown Stage 11.4% (4/35) *Death or last documented activity in the VAMC EMR up to 10/31/2014.

29

ACCEPTED MANUSCRIPT

Supplementary Table 2: Frequency of HCC diagnosis made on surveillance in 518 patients with untreated HCC by BCLC staging. Enrolled in Surveillance Program Unknown (%) 0 (0.0%) 2 (2.6%) 15 (6.6%) 5 (3.0%) 4 (11.4%)

P-value 0.0006

AC C

EP

TE D

M AN U

SC

Yes (%) No (%) 0/A (n=14) 9 (64.3%) 5 (35.7%) B (n=76) 29 (38.2%) 45 (59.2%) C (n=227) 45 (19.8%) 167 (73.6%) D (n=166) 49 (29.5%) 112 (67.5%) Unknown (n=35) 12 (34.3%) 19 (54.3%) BCLC, Barcelona Clinic Liver Cancer.

RI PT

BCLC Stage

30

ACCEPTED MANUSCRIPT

FIGURE LEGENDS Figure 1: Flowchart showing derivation of study cohort

hepatocellular carcinoma stratified by BCLC staging.

RI PT

Figure 2: Kaplan-Meier curve showing overall mortality in 518 patients with untreated

Figure 3: Kaplan-Meier curve showing overall mortality in 492 patients with untreated

hepatocellular carcinoma stratified by HCC surveillance status. Unknown BCLC stage included

AC C

EP

TE D

M AN U

SC

in analysis.

31

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT