Screening for hepatocellular carcinoma (HCC) is controversial.

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5: Screening for Hepatocellular Carcinoma Among Veterans With Hepatitis C on Disease Stage, Treatment Received, and Survival LUCI K. LEYKUM,* HASHEM B. EL SERAG, JOHN CORNELL,* and KYRIAKOS P. PAPADOPOULOS *South Texas Veterans Health Care System, Cancer Therapy and Research Center, University of Texas Health Science Center at San Antonio, San Antonio, Texas; and Baylor College of Medicine, Michael E. DeBakey Veterans Affairs Medical Center, Houston, Texas Background & Aims: The incidence of hepatocellular carcinoma (HCC) is increasing. Despite recommendations for HCC screening from the American Association for the Study of Liver Disease, the National Cancer Institute does not recommend screening. The question of whether screening is of benefit is an important one. The purpose of this study was to examine the determinants of screening, as well as the impact of screening on disease stage, treatment received, and survival in a US veteran population. Methods: Patients with hepatitis C and HCC who receive care in the South Texas Veteran Health Care System were identified using the Veterans Affairs national hepatitis C registry. Screening status was determined by chart review. Potential determinants of screening were assessed. Screened and unscreened patients were compared on the basis of disease stage at diagnosis, treatment received, and survival. Results: Seventy-two patients were identified and included in the analysis, of whom only 16 (22%) were screened. Patients seen by a hepatologist before diagnosis were more likely to be screened. All screened patients were diagnosed with early stage disease, compared with 22% of unscreened patients (P <.001). Screened patients were 10 times more likely to have received potentially curative treatment (95% confidence interval, ). Log-rank test of equality of survivor functions was statistically significant for differences between screened and unscreened groups (P.0005). Conclusions: Our findings support the American Association for the Study of Liver Disease screening recommendations, and suggest that screening is underused. Screening for hepatocellular carcinoma (HCC) is controversial. The 2005 consensus statement from the American Association for the Study of Liver Disease (AASLD) advocates screening. 1 In contrast, the National Cancer Institute does not recommend this practice, 2 stating, based on fair evidence, screening would not result in a decrease in mortality from hepatocellular cancer. However, the emergence of HCC as an important public health issue, and the disparity in outcomes between those diagnosed with early vs intermediate- or latestage disease, have led to an increased focus on screening and have added urgency to the debate. HCC is one of the few cancers whose incidence is increasing in the United States, doubling from 1.4 in 100,000 to more than 3 in 100,000 from the 1970s to the 1990s, 3,4 in large part because of the greater prevalence of hepatitis C. 3 6 Potentially curative treatment options for early disease are mainly surgical (resection or transplantation) or ablative, and the 3-year survival is in the range of 60% 80%. 7 However, HCC is generally resistant to systemic chemotherapy, and the 3-year survival rate for those with intermediate- or late-stage disease is as low as 10% Therefore, screening practices that allow for improved diagnosis at an early stage could have a large impact on survival. The current data regarding screening come largely from Asian populations. Two Asian prospective cohort studies using screening serum -fetoprotein (AFP) levels showed a smaller tumor size at the time of diagnosis, and 1 study reported improved mortality at 1, 2, and 3 years. 12,13 A randomized controlled trial in China using biannual AFP levels showed significantly earlier stages of tumor at the time of diagnosis. 14 In this study, the 1- and 3-year survival was improved in the screened group, but no difference in 5-year survival was found, possibly because aggressive treatments appropriate for early stage disease were not available to many of the patients studied. A second randomized controlled trial using biannual AFP levels and annual ultrasounds showed a decrease in mortality of 37%, 15 but included mainly young patients with hepatitis B who did not have a diagnosis of cirrhosis. The data on HCC screening in non-asian populations are less clear. A prospective French study screened a cohort of patients with cirrhosis (Child Pugh classes A and B) with biannual AFP levels and annual ultrasounds, and did not show a survival benefit for screening. 16 In contrast, a recent prospective cohort study of Australian patients screened for HCC with biannual AFP levels and annual ultrasounds, when compared with retrospectively identified, incidentally diagnosed controls, did find a significantly smaller tumor size among screened patients. 17 Based on the published survey of AASLD members, most report performing screening, and many believe that it would be unethical to perform a randomized controlled trial to show its efficacy. 18 However, the lack of clear data in a US population to support these practices, and the divergence between expert opinion and the National Cancer Institute recommendation, make an assessment of HCC screening effectiveness in non- Asian populations imperative. The purpose of this study was to examine the determinants of HCC screening, and the impact of screening on stage, treatment received, and survival in a US veteran cohort. Because the screening practices have evolved, we included biannual AFP levels, annual ultrasound or computer- Abbreviations used in this paper: AASLD, American Association for the Study of Liver Disease; CT, computerized tomography; HCC, hepatocellular carcinoma; VISN, Veterans Integrated Service Network by the AGA Institute /07/$32.00 doi: /j.cgh

2 April 2007 HCC SCREENING IN VETERANS WITH HEP C 509 Table 1. Information Abstracted From Medical Record Variable Definition Date of birth Race White, black, Hispanic, or American Indian based on Veterans administration classification Date of diagnosis Date of confirmation of diagnosis of HCC; if pathologic diagnosis was not made, date that presumptive diagnosis was made based on clinical data End of follow-up evaluation Death, or November 15, 2005 Cirrhosis/Child Pugh class Cirrhosis defined by either a clinical, radiologic, or pathologic diagnosis and scored on the Child Pugh scale Past alcohol use Assessed as yes/no Current alcohol use Number of drinks/wk Psychiatric history Past or current psychiatric diagnosis, including posttraumatic stress disorder, anxiety, depression, dysthymia, or psychosis; does not include substance abuse Serum creatinine level Most recent laboratory value before diagnosis Other chronic medical illnesses At the time of diagnosis, includes diabetes, hypertension, coronary artery disease, congestive heart failure, chronic renal insufficiency, and cerebrovascular disease HCC screening status Based on at least 2 AFP levels or 1 imaging study per year before the time of diagnosis Stage Based on radiologic results: early: 1 lesion 5 cm; 3 lesions 3 cm/each; no portal vein invasion; intermediate/late considered to be anything else Treatment received Potentially curative treatments considered to be transplant, resection, or radiofrequency ablation; other treatments considered palliative Followed up by a hepatologist Based on whether patients were seen by the hepatology clinic before the time of diagnosis Primary care location Based on whether primary care was received at a tertiary academic medical center or outlying clinics ized tomography (CT), or a combination, in our definition of screening. Methods This study was approved by the Internal Review Board of the University of Texas Health Science Center at San Antonio. We used the Veterans Affairs national hepatitis C registry to identify veterans in the South Texas Veterans Health Care System (part of Veterans Integrated Service Network [VISN] 17) who are infected with hepatitis C. This registry automatically identifies and imports data from patients who have laboratory test results indicative of hepatitis C infection (positive hepatitis C antibody or viral RNA) or International Classification of Diseases 9th edition codes related to diagnosis or management of hepatitis C associated with any clinic visit. A total of 4806 hepatitis C positive patients in South Texas were identified as of August The South Texas Veterans Health Care System is composed of 2 main facilities, the Audie L. Murphy Memorial Hospital in San Antonio and Kerrville VA Medical Center, as well as 14 additional out-patient facilities throughout South Texas, in which approximately 74,660 patients were seen in fiscal year We identified the cohort of hepatitis C positive patients who developed HCC in the South Texas Veterans Health Care System from January 1, 2000, through November 15, 2005, using International Classification of Diseases 9th edition codes (hepatocellular carcinoma) and (malignant neoplasm of liver, not specified as primary or secondary). We then performed a chart review to confirm the diagnosis of HCC and to abstract the following data from the medical record: age, race, vital status and date of death, cirrhosis, alcohol use, hepatitis B co-infection, other chronic medical conditions, serum creatinine level, location of care, screening status, tumor size at the time of diagnosis, and treatment given. To determine the HCC screening status, notes and test requests were reviewed to ensure that tests were performed for screening purposes and not for work-up of a new finding. Table 1 summarizes the information abstracted, including operational definitions of each variable. Screened patients were compared with unscreened patients using t tests for continuous variables and the Fisher exact tests for dichotomous variables. We used unadjusted and adjusted logistic regression models to examine the potential determinants of screening. We compared the cumulative survival between patients who did and did not receive HCC screening using the Kaplan Meier product limit survival method, and the log-rank test for statistical significance. We used Cox hazard regression analysis to examine the effect of HCC screening on mortality risk while adjusting for other variables. The start date used for survival analysis was the date of first diagnosis, March 7, 2000, with the last date of follow-up evaluation as November 15, All statistical analyses were performed using Stata 8.0 (College Station, TX). Results Of the cohort of hepatitis C positive patients, we identified 92 patients with potential HCC based on International Classification of Diseases 9th edition codes. We excluded 10 patients on the basis of chart review: 2 had benign findings, 3 had metastatic disease from another source, and for 5 patients the etiology of the lesions was unclear. Data were insufficient for abstraction of necessary information for a further 10 patients because the diagnosis was made before the implementation of the electronic medical record in Of the 72 remaining patients, the diagnosis of HCC was based on biopsy examination results in 24 (33%). In 40 patients, the diagnosis was based on an AFP level greater than 400 in the setting of a hypervascular lesion on radiographic imaging (56%). Seven patients had an increasing AFP level in the setting of radiographic imaging, although the absolute number was less than 400 (10%). Finally, in 1 patient with a 20-cm lesion, neither a biopsy examination nor an AFP level was obtained, but the clinical

3 510 LEYKUM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 team caring for the patient, including hepatology and oncology services, believed that the lesion was most consistent with HCC. Sixty-two patients were identified by code (hepatocellular carcinoma) alone, and 6 by code (malignant neoplasm of liver, not specified as primary or secondary) alone. Twenty-four patients had both International Classification of Diseases 9th edition codes attached to their medical records. The positive predictive values of codes 155.0, 155.2, and the combination of the 2 codes were 89.5%, 96.7%, and 100%, respectively. Of the 72 patients with HCC in the final analysis, only 16 (22%) were screened with AFP level and ultrasound or CT. The date of first diagnosis was March 7, 2000, and the last diagnosis was July 11, The date of last follow-up evaluation after the HCC diagnosis was November 15, 2005, at which time 57 patients had died. In screened patients, 6 tumors (38%) initially were identified on the basis of an increase in the AFP level alone. Of these, only 1 AFP level was greater than 400; the others were increased over a previous value. In only 1 of these patients was the follow-up imaging study, an ultrasound, initially reported to be normal. Eight tumors (50%) were found on the basis of an imaging study alone, most frequently triple-phase CT. The AFP levels were normal in all of these instances. In the final 2 patients (12%), both the AFP level and CT were ordered at the same time and both were abnormal, leading to the diagnosis of HCC. Cirrhosis was most likely present in 41 (57%), and probably present in 16 (22%). Cirrhosis was considered most likely on the basis of biopsy results (n 2), or both radiologic and laboratory evidence of cirrhosis (low albumin levels and increased prothrombin time) (n 39). Probable cirrhosis was based on the notation of cirrhosis in clinical notes combined with either laboratory or radiologic evidence of cirrhosis. Fifteen patients (21%) did not have evidence of cirrhosis, 1 of whom had a biopsy examination that revealed bridging fibrosis. Only 1 screened patient did not show clinical or histopathologic signs of cirrhosis at the time of screening. As described earlier, most (79%) had evidence of cirrhosis. In those with evidence of cirrhosis, the mean Child Pugh score was 8.2, which was significantly different from that of patients without evidence of cirrhosis (mean Child Pugh score, 2.9) (P.001). Table 2 shows the potential determinants of HCC screening. The median ages of screened and unscreened patients were similar: 54 years (range, y) compared with 53 years (range, y), respectively. Of all patients with HCC, 35 were Hispanic, 28 were white, 6 were black, and 1 was American Indian; the charts of 2 patients did not contain data on race. Screened patients were significantly more likely to be white. No significant differences were found between screened and unscreened patients with respect to Child Pugh score, hepatitis B co-infection, presence or number of other chronic medical illnesses, serum creatinine level, or current or past alcohol use. Although there were no significant differences between groups based on the location where primary care was received, patients who received care in a subspecialty hepatology clinic were more likely to be screened. Twenty-five patients were seen before the diagnosis of HCC in a hepatology clinic; of these, 12 (48%) were screened for HCC. Only 3 (6.4%) patients not seen in a hepatology clinic were screened. Twenty-eight patients were diagnosed with early stage HCC, as defined by 1 tumor less than 5 cm, or 3 tumors less than 3 Table 2. Determinants of Screening in 72 Hepatitis C Positive Patients Diagnosed With HCC in South Texas Between January 2000 and November 2005 Variable Screened (N 16) Unscreened (N 56) P value Average age, y a Race, % white a Average Child Pugh score Hepatitis B co-infection, % Current alcohol use, drinks/wk Past alcohol use, % Serum creatinine level, mg/ml Presence of at least 1 other chronic medical illness, % Number of chronic medical illnesses Psychiatric history, % Followed up by hepatology clinic a before diagnosis, % Followed up by primary care at tertiary center, % a P.05. cm, without portal vein involvement or metastasis. All screened patients were diagnosed with early stage disease based on tumor size, in comparison with only 12 (22%) of those who were not screened (P.0001). Among other potential predictors, only patients who had been seen in a hepatology clinic before the diagnosis were more likely to be diagnosed at an early stage (P.05). Of the entire cohort, 16 (22%) patients received potentially curative treatment as defined by transplant, resection, or radiofrequency ablation. Ten (63%) of these patients had been screened and received the following specific treatments: transplant (n 1), resection (n 1), or radiofrequency ablation (n 8). The 6 remaining patients (40%) had not been screened, and all received radiofrequency ablation. Four additional patients received transarterial chemoembolization; none of these patients had been screened, and none had early stage disease. Patients with early stage disease did not get potentially curative treatments for several reasons. One patient was on the transplant list at the study end date; 2 were evaluated for transplant, but 1 was lost to follow-up evaluation and the other had a massive stroke and was no longer a candidate. One patient was placed in a phase I clinical trial. Progression of disease occurred in 2 patients before curative treatment could be applied. The remaining patients had documented active alcohol use at the time of diagnosis and were lost to follow-up evaluation for varying amounts of time. There was a significant association between HCC screening and receiving potentially curative treatment, with screened patients approximately 10 times more likely to have received such treatment (95% confidence interval, ). This relationship remained significant (P.007) after adjustment for age, race, Child Pugh class, alcohol use, medical and psychiatric comorbidities, and serum creatinine level. However, after adjustment for stage of disease, the association between HCC screening and receipt of potentially curative treatment was no longer significant (P 1.0), suggesting that the association between screening and treatment received is mediated by the

4 April 2007 HCC SCREENING IN VETERANS WITH HEP C 511 Figure 1. Kaplan Meier survival curves for hepatitis C positive patients in South Texas who were diagnosed with HCC, based on screening status. Gray line, screened patients; black line, unscreened patients. fact that screened patients were more likely to be diagnosed with early stage disease. Finally, the relationship between race and treatment received was not significant. Approximately 18% of unscreened patients were alive at 1 year, compared with 75% of screened patients (P.001). Kaplan Meier survival curves for screened and unscreened patients are shown in Figure 1. Log-rank test of equality of survivor functions was statistically significant for differences between screened and unscreened groups (P.0005). The average survival among unscreened patients was 8.5 months (95% confidence interval, ), whereas that of screened patients was 19.8 months (95% confidence interval, ). Besides the patient who suffered a massive stroke, all other patients died of complications related to HCC. Unadjusted and adjusted hazard ratios are shown in Table 3. In the unadjusted model, screening, stage of disease, treatment received, hepatology evaluation before diagnosis, primary care physician location, and past psychiatric history were associated significantly with a decreased risk of death. In a model that examined the joint effect of screening and treatment, screening did not have a significant association with survival, but treatment received did, suggesting that this mediates the relationship between the screening and survival. The global test of proportional hazards assumption was fulfilled in all models. Discussion This study describes the impact of screening for HCC on outcomes in a US population with hepatitis C and HCC. Screened patients were more likely to be diagnosed with early stage disease, were more likely to receive potentially curative treatment, and had improved survival. This study had several limitations. First, the sample size was small. Despite this, our results were statistically significant. Second, our population was relatively younger at the age of diagnosis with HCC than that described in other studies, which partly may have explained the observed screening benefit. The younger age may be the result of the high prevalence of alcohol use, which may have influenced the course of disease in these patients. In addition, 40% were co-infected with hepatitis B. Third, our population was limited to veterans with hepatitis C, and all patients were male, limiting our ability to generalize these results. Fourth, it is possible that survival data could be misleading because patients may not have died of HCC. However, we did not use alternative end points such as transplant or resection because the only patients who received these interventions were screened. Finally, during the period of time of this study, our transplant candidates were referred to and followed up by the transplant program at University Hospital. Because of this arrangement, patients may have been referred more easily and transplanted in a more timely fashion than those who receive care at a place that is geographically distant from a transplant center, improving their chances of survival. The observational retrospective cohort design raises the issue of selection bias. It is conceivable that screened patients were different from, or healthier than, unscreened patients. We adjusted our results for factors indicative of general health, including age, Child Pugh class, serum creatinine level, alcohol use, and psychiatric and medical comorbidities. We did not find differences in current alcohol use, Child Pugh class, or serum creatinine level, but our sample size may have been too small to detect these differences. Lead time bias also could play a role in the observed survival gain (11.3 mo) seen among screened patients. However, screened patients were more likely to receive potentially curative treatment, which could account for the survival benefit. Patients followed up by hepatologists were more likely to be screened than those who were seen only by a primary provider. Table 3. Risk of Death Examined for 72 Hepatitis C Positive Patients in South Texas Diagnosed With HCC Between January 2000 and November 2005 Hazard ratio Unadjusted analysis 95% CI Adjusted analysis Hazard ratio 95% CI HCC screening 0.27 a Age, y N/A N/A Child Pugh class N/A N/A Past alcohol use, y/n N/A N/A Current alcohol use, N/A N/A drinks/wk Hepatitis B co-infection N/A N/A Presence of other chronic N/A N/A illness b Number of comorbidities N/A N/A Serum creatinine level, N/A N/A mg/dl Race, white N/A N/A Psychiatric disease 0.47 a a PCP at tertiary center 1.47 a Hepatology assessment 0.44 a before diagnosis Early stage HCC 0.17 a Receipt of potentially curative treatment 0.15 a a NOTE. Unadjusted and adjusted Cox hazard regression model. CI, confidence interval; N/A, not applicable; PCP, primary care physician. a P.05. b Includes diabetes, hypertension, coronary artery disease, congestive heart failure, chronic renal insufficiency, and cerebrovascular disease.

5 512 LEYKUM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 This suggests that primary care providers are either unaware of AASLD screening recommendations, or do not believe in their validity. However, 43% of patients seen by a hepatologist were not screened. These patients did not have clinical evidence of cirrhosis based on chart review, and thus presumably were considered low risk for HCC. This screening strategy is consistent with current AASLD recommendations, which recommend screening hepatitis C positive patients with cirrhosis, or who have other risk factors for HCC such as alcohol use or hepatitis B co-infection. However, even these seemingly low-risk patients still developed HCC, highlighting the difficulty of assessing cirrhosis clinically, and the need for reliable ways to assess the possibility of evolving cirrhosis or bridging fibrosis. Our findings are supportive of the current AASLD recommendation that AFP level alone should not be used for screening because it was normal in 8 of 16 screened patients (50%) at the time of diagnosis. However, it is difficult to assess the efficacy of the recommendation to screen with AFP level combined with ultrasound based on our population because 1 patient had a normal ultrasound in the setting of an increased AFP level, and most were screened with triple-phase CT. In conclusion, this study showed that screening patients with hepatitis C for HCC is effective in a US veteran population, and suggests the need for further study in an expanded US patient population. However, screening appears to be underused, even among patients evaluated by hepatologists. It is difficult to predict who receives screening based on Child Pugh class, despite the recommendation of screening for most groups of patients with cirrhosis. These data highlight the need for further study regarding the reasons that patients are not screened, the most effective way to screen, and how best to assess HCC risk in the absence of clinical cirrhosis. References 1. Bruix J, Sherman M. Management of hepatocellular carcinoma, AASLD practice guideline. Hepatology 2005;42: National Cancer Institute types/liver/. Accessed May 23, El-Serag HB, Davila JA, Peterson NJ, et al. The continuing increase in the incidence of hepatocellular carcinoma in the United States: an update. Ann Intern Med 2003;139: El-Serag HB, Mason AC. Risk factors for the rising rates of primary liver cancer in the United States. Arch Intern Med 2000;160: Strader DB, Wright T, Thomas DL, et al. Diagnosis, management, and treatment of hepatitis C. Hepatology 2004;39: Lauer GM, Walker BD. Hepatitis C virus infection. N Engl J Med 2001;345: Bruix J, Llovet JM. Prognostic prediction and treatment strategy in hepatocellular carcinoma. Hepatology 2002;35: Llovet JM, Burroughs A, Bruix J. Hepatocellular carcinoma. Lancet 2003;362: Arii S, Yamaoka Y, Futagawa S, et al. Results of surgical and nonsurgical treatment for small sized hepatocellular carcinomas: a retrospective and nationwide survey in Japan. Hepatology 2000;32: Llovet JM, Bustamante J, Castells A. Natural history of untreated nonsurgical hepatocellular carcinoma: rationale for the design and evaluation of therapeutic trials. Hepatology 1999;29: Di Maio M, De Maio E, Perrone F, et al. Hepatocellular carcinoma systemic treatments. J Clin Gastroenterol 2002;35(Suppl 2): S109 S Tang ZY. Screening and early treatment of primary liver cancer with special reference to the east part of China. Ann Acad Med Singapore 1980;2: Wu JC, Lee SD, Hsiao KJ, et al. Mass screening of primary hepatocellular carcinoma by alpha-fetoprotein in a rural area of Taiwan a dried blood spot method. Liver 1988;8: Chen JG, Parkin DM, Chen QG, et al. Screening for liver cancer: results of a randomized controlled trial in Qidong, China. J Med Screening 2003;10: Zhang BH, Yang BH, Tang ZY. Randomized controlled trial of screening for hepatocellular carcinoma. J Cancer Res Clin Oncol 2004;130: Pateron D, Ganne N, Tinchet JC, et al. Prospective study of screening for hepatocellular carcinoma in Caucasian patients with cirrhosis. J Hepatol 1994;20: Rusli F, Knight V, Patella S, et al. Surveillance for hepatocellular carcinoma: effect on tumour size and survival. Poster presentation at the American Association for the Study of Liver Disease, Boston, MA; October 29 November 2, Chasalani N, Said A, Ness R, et al. Screening for hepatocellular carcinoma in patients with cirrhosis in the United States: results of a national survey. Am J Gastroenterol 1999;94: U.S. Department of Veterans Affairs. va.gov/performance/fy05%20vital%20signs%20reports.htm. Accessed June 23, Address requests for reprints to: Luci K. Leykum, MD, MBA, Assistant Professor of Medicine, University of Texas Health Science Center at San Antonio, South Texas Veterans Health Care System, 7400 Merton Minter Boulevard, Ambulatory Care 11C6, San Antonio, Texas Leykum@uthscsa.edu. Supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research, and Development Service, South Texas Veterans Health Care System, San Antonio, Texas. The authors would like to thank Dr Steven Schenker for his insightful comments, and Dr Judy Patterson for her invaluable help with database abstraction. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs.