HCV reinfection and superinfection

HCV reinfection and superinfection are common among actively injecting drug users. This might further complicate the development of an effective HCV vaccine."



"In a large longitudinal cohort of DU in Amsterdam, we demonstrated that 39% of DU with a documented HCV seroconversion during follow-up, experience multiple HCV infections over time"



"At least some individuals despite having the immunological capacity to clear initial infection, fail to elicit an immune response that provides sufficient protection against HCV reinfection. Partial protective immunity, however, might result in lower peak viremia, increased rates of spontaneous viral clearance following reinfection, or protection against strains of the same HCV subtypes. Nevertheless, the results presented in this study further complicates vaccine development. Therefore, HCV harm reduction will remain dependent on precautionary measures preventing the further spread of HCV, and treatment of those who are chronically infected."

Jnl of Hepatology
June 2009
Articles in Press

Frequent HCV reinfection and superinfection in a cohort of injecting drug users in Amsterdam



Thijs J.W. van de Laar1Corresponding Author Informationemail address, Richard Molenkamp2, Charlotte van den Berg13, Janke Schinkel2, Marcel G.H.M. Beld2, Maria Prins13, Roel A. Coutinho134, Sylvia M. Bruisten13



Received 9 September 2008; received in revised form 23 April 2009; accepted 5 May 2009. published online 18 June 2009.

Uncorrected Proof



ABSTRACT


Background/Aims



This study investigates the occurrence of HCV reinfection and superinfection among HCV seroconverters participating in the Amsterdam Cohort Studies among drug users from 1985 through 2005.



This study comprises all DU participants (n=59) who seroconverted for HCV during follow-up in the Amsterdam Cohort Studies (ACS) between December 1985 and November 2005



Methods



HCV seroconverters (n=59) were tested for HCV RNA at five different time points: the last visit before seroconversion (t=−1), the first visit after seroconversion (t=1), six months after (t=2) and one year after (t=3) seroconversion, and the last visit prior to November 2005 (t=4). If HCV RNA was present, part of the NS5B region was amplified and sequenced. Additional phylogenetic analysis and cloning was performed to establish HCV reinfection and superinfection.



Results



Multiple HCV infections were detected in 23/59 (39%) seroconverters; 7 had HCV reinfections, 14 were superinfected, and 2 had reinfection followed by superinfection. At the moment of HCV reinfection, 7/9 seroconverters were HIV-negative: persistent HCV reinfection developed in both HIV-positive cases but also in 4/7 HIV-negative cases. In total, we identified 93 different HCV infections, varying from 1 to 4 infections per seroconverter. Multiple HCV infections were observed in 10/24 seroconverters with spontaneous HCV clearance (11 reinfections, 3 superinfections) and in 13/35 seroconverters without viral clearance (20 superinfections).



Conclusions



HCV reinfection and superinfection are common among actively injecting drug users. This might further complicate the development of an effective HCV vaccine.



Introduction



Injecting drug users (DU) are at high risk for hepatitis C virus (HCV) infection through the shared use of needles and injection equipment. The reported HCV seroprevalence among injecting DU ranges from 30 to 90% in Europe, North-America, and Australia [1], [2], [3]. Over decades, persistent HCV viremia can cause chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma [4]. Spontaneous resolution of the virus occurs in 15–40% of those infected [5]. However, both HCV reinfection and HCV superinfection have been documented among individuals with ongoing risk behaviour [6], [7], suggesting that neither viral clearance nor ongoing HCV infection consistently protect against new HCV infection. Evidence for partial protective immunity against HCV infection is derived largely from chimpanzee studies. Chimpanzees that previously cleared HCV and were rechallenged with homologous HCV strains, generally showed lower levels of HCV viremia and self-limited infection [8], [9]. Whether humans previously infected with HCV gain protection against a second HCV infection remains controversial [6], [10], [11], [12].



The incidence rate ratio of naıve HCV infection versus HCV reinfection in a high-risk population, is an indirect but frequently used tool to gain insight into the presence or absence of partial HCV immunity [6], [10], [11], [12]. Clearly, if HCV infection occurs at a lower rate in previously exposed individuals compared to naı¨ve individuals it is reasonable to assume some level of immunity is being induced from past exposure to HCV. Alternatively, if HCV reinfection occurs at similar or higher rates, this means that at least some individuals despite having the immunological capacity to clear initial HCV infection, fail to elicit an immune response that provides sufficient protection against HCV reinfection [12]. Insight into HCV protective immunity after previous exposure contributes to vaccine development and also provides health guidance for preventive strategies in populations at risk. Therefore, we studied the occurrence of HCV reinfection and superinfection in a well-defined cohort of injecting DU from the moment of HCV seroconversion until the end of follow-up. This study uniquely combines longitudinal data from prospectively identified HCV seroconverters with an epidemiologic and phylogenetic approach.



Discussion



In a large longitudinal cohort of DU in Amsterdam, we demonstrated that 39% of DU with a documented HCV seroconversion during follow-up, experience multiple HCV infections over time. Both HCV reinfection and superinfection were common in this high-risk population. Traditional HCV incidence calculations based on HCV antibody seroconversion, with the assumption that HCV seropositive individuals no longer are susceptible, therefore underestimate the true HCV incidence and do not properly reflect HCV transmission dynamics within a network of high-risk individuals [22].



As the occurrence of HCV reinfection after initial viral clearance has been previously described [6], [10], [11], [12], [23], there is no question that in humans HCV protective immunity is not even close to being complete. The existence of partial protective HCV immunity in humans, however, remains controversial. Previous studies among anti-HCV positive DU showed that HCV infection occurred at lower rate in previously exposed individuals than in naı¨ve individuals [11], [12], [24]. Despite reinfection rates of 3–12%, this suggests that some level of immunity is being induced from past exposure. However, these studies were conducted in DU with prevalent HCV infection; the period of high-risk behaviour which led to anti-HCV seroconversion most likely preceded the study period by several years. As a result, previously HCV-infected DU were generally older (median age 41–47 years) and had less active drug use compared to HCV-uninfected DU [24]. In fact, 35–58% of participants in those studies had quit injecting before the actual start of the study period, being therefore at minimal risk of HCV reinfection. In contrast, studies performed in HCV seroconverter cohorts of young (<30 years) actively injecting DU implied absence of partial HCV protective immunity based on the fact that the incidence of HCV reinfection in these cohorts was similar [6] or even higher [10] than the incidence of initial infection, causing 46–50% of DU who previously cleared the virus to experience a second viremic episode.



The data in our study are concordant with those of other seroconverter cohorts of young actively injecting DU [6], [10]; 42% of previously exposed DU experienced HCV reinfection. Also in our cohort, preliminary incidence calculations suggest that the incidence of HCV reinfection were at least similar to the incidence of naı¨ve HCV infection. The incidence of HCV reinfection declined from 20.4/100 PY in the period 1985–1995 to 4.17/100 PY in the period 1995–2005, whereas the incidence of initial HCV infection in our cohort dropped from 27.5/100 PY in the late 1980s to approximately 2.0/100 PY in recent years [13]. Based on this incidence rate ratio it is tempting to imply that partial HCV protective immunity does not exist, but is this justified? As HCV reinfection might be associated with lower peak viremia and higher rates of spontaneous viral clearance, the real data that address this question are those regarding the outcome and nature of HCV reinfection. In contrast to long-lasting antibody seroconversion, HCV reinfection requires detection of new HCV viremia [24]. The probability of detecting transient HCV reinfection strongly depends on the frequency and interval of RNA testing [24]. In our study design the relatively long RNA testing intervals will support detection of all persistent HCV reinfections, but short episodes of resolved viremia are easily missed. Indeed 14/24 (58%) seroconverters resolved initial HCV viremia but lacked evidence for HCV reinfection, implying either rapid spontaneous viral clearance, protection against HCV reinfection or cessation of HCV risk behaviour. Of them, 7/14 (50%) were at minimal risk for HCV reinfection as they quit injection within 6 months after HCV seroconversion, the remaining 7 DU had no detectable HCV viremia despite ongoing risk behaviour. Additionally, 3/9 seroconverters with documented HCV reinfection resolved a second or even third viremic episode. Besides infrequent sampling, data on persistence of HCV reinfection are often skewed by HIV/HCV coinfection. HIV-coinfection is associated with increased rates of HCV persistence (up to 95%); lack of robust CD4+ T cell responses may contribute to the failure of early viral control of HCV [25], [26]. HCV reinfection persisted in the two DU coinfected with HIV, but also in 4/7 DU without HIV.



Our study has several limitations. Despite our clear-cut definitions of HCV reinfection, coinfection and superinfection, the unpredictable course of HCV infection makes it difficult to distinguish them as such. Mosley et al. [27] showed that intercalating HCV RNA positivity among those who eventually clear infection and intercalating HCV RNA negativity among those who eventually develop persistent infection, are common especially during the initial phase of HCV infection [27]. It cannot be excluded that in our study DU with ‘probable reinfections’ actually have fluctuating levels of low viremia, which would have led us to overestimate the rate of HCV reinfection. To minimise the risk of such misclassification, we defined viral clearance as two consecutive HCV RNA-negative visits at least 4 months apart. Underestimation of the rate of HCV reinfection was also very likely for three reasons. First, HCV reinfection in humans and chimpanzees has been associated with short episodes of low viremia [9], [12], therefore reinfections might have been missed as a result of too low frequency of sampling and RNA levels below detection limit. Second, in the context of a steady injection partner, reinfection with a HCV strain almost identical to the initial strain, cannot be excluded. Third, superinfection requires detection of at least two divergent viral strains at the same time point, and our method did not exclude the possibility of rapid viral clearance of the initial strain followed by reinfection with a second strain. Although the latter would explain the difficulties we experienced with detecting minority variants in DU with superinfection, it seems more likely that the high viral turnover of HCV causes even the smallest fitness differences between two competing strains to result in rapid disappearance of the less fit strain over time [28]. Other laboratory methods involving strain-specific primers are currently being developed to specifically amplify minority variants.



In conclusion, both HCV reinfection and superinfection are common among actively injecting DU. At least some individuals despite having the immunological capacity to clear initial infection, fail to elicit an immune response that provides sufficient protection against HCV reinfection. Partial protective immunity, however, might result in lower peak viremia, increased rates of spontaneous viral clearance following reinfection, or protection against strains of the same HCV subtypes. Nevertheless, the results presented in this study further complicates vaccine development. Therefore, HCV harm reduction will remain dependent on precautionary measures preventing the further spread of HCV, and treatment of those who are chronically infected.



Results




Study population



In the ACS, 456/1259 (36.2%) participating DU were HCV-antibody negative at entry. Their median age was 31 years (IQR, 28–36 years), 70% was male and 11/456 (2.4%) were HIV-positive. Only 132/456 (28.9%) HCV-antibody negative DU had ever injected, of whom 64 (14.0%) in the last 6 months [13], [21]. We identified 59 DU who seroconverted for HCV during ACS follow-up: 58% was male and their median age at HCV seroconversion was 29 years (IQR, 25–34 years). The median year of HCV seroconversion in the ACS was 1991 [IQR: 1989–1994], or 2.23 years (IQR: 0.93–6.49 years) after initiation of drug injection. The median ACS follow-up time since HCV seroconversion was 7.06 years [IQR, 2.81–12.1]. Only one HCV seroconverter denied ever injecting. HCV/HIV coinfection occurred in 13/59 (22%) HCV seroconverters; HCV infection preceded HIV infection in 6/13, 5 contracted both viruses during the same brief period, and two DU were HIV-positive before they acquired HCV.



Dominant HCV variant detection



The 59 HCV seroconverters were tested for HCV RNA on a total of 326 visits, varying from 2 to 10 visits per DU. At 211/326 (65%) of these visits, HCV RNA was detected using a real- time PCR based on the 5′-UTR; log HCV viral loads varied from 2 to 6.14IU/ml. Amplification and sequencing of the HCV NS5B region succeeded in 170/177 (96%) samples with a viral load exceeding 1000IU/ml and in 15/34 (44%) samples with a viral load below 1000IU/ml. Sequencing analysis of HCV RNA-positive samples revealed that at least 23/59 (39%) HCV seroconverters had evidence for multiple HCV infections over time. According to our definitions, 7 HCV seroconverters had HCV reinfections, 13 were superinfected, 2 were reinfected and subsequently superinfected, and 1 was coinfected and subsequently superinfected. In total, 93 different HCV infections were identified: 59 initial infections, 11 reinfections, 22 superinfections and 1 coinfection. Genotyping and sequencing succeeded for 74/93 (80%) of HCV infections, finding a genotype distribution of 1a (47%), 3a (32%), 4d (8%), 1b (6%), 2a/b (6%) and 4a (1%).



HCV seroconverters with spontaneous HCV clearance



Spontaneous HCV viral clearance occurred in 24/59 (41%) HCV seroconverters including 2 DU who were HIV-positive at the moment of HCV seroconversion. Within this group of resolvers, 10/24 (42%) DU had evidence for multiple HCV infections over time; 7 DU had HCV reinfections, 2 DU had HCV reinfection followed by HCV superinfection, and 1 DU had HCV superinfection but spontaneously resolved both viral strains (Fig. 1a). In 2/9 DU with evidence for HCV reinfection we could confirm the presence of two distinct HCV strains over time (confirmed reinfection). In the other 7 DU, however, viral characterisation of either the primary (n=6) or the reinfecting (n=1) HCV strain failed due to rapid viral clearance of especially primary HCV viremia (probable reinfection).



In 6/9 (67%) seroconverters with confirmed or probable HCV reinfection, HCV reinfection persisted (Fig. 1a). It must be noted that 2/6 seroconverters were coinfected with HIV when they developed persistent HCV reinfection: DU 18917 already was HIV-positive before HCV seroconversion and DU 12905 developed persistent HCV re- and superinfection around HIV-seroconversion. The other four DU with persistent HCV reinfection were HIV-negative at the time; 1 DU (DU 19927) eventually acquired HIV years after the establishment of persistent HCV reinfection. None of the three HCV seroconverters that cleared a second or even third viremic episode were coinfected with HIV.



HCV seroconverters without spontaneous HCV clearance



Chronic HCV infection without evidence of previous viral clearance occurred in 35/59 (59%) HCV seroconverters. According to our definitions, 13/35 (37%) seroconverters had HCV superinfections over time (Fig. 1b); at least two clearly distinct HCV strains were detected in all 13 of them. From 6/13, we isolated three or even four distinct HCV viral strains. In DU 18934, the dominant viral strain switched from HCV genotype 4a to genotype 1a, subsequently to genotype 3a and eventually to a heterologous strain of genotype 1a (denoted as 1a∗). DU 12962 had clear evidence of HCV coinfection. In the last visit predating HCV antibody presence, two distinct HCV strains were detected using the NS5B PCR, one was HCV subtype 1b and one was HCV subtype 4d. HCV subtype 4d became the dominant viral strain, however it seems to have submerged after superinfection with HCV genotype 1a. No particular pattern in the switch of HCV strains was observed; for example the dominant HCV strains switched from genotype 1a to 3a as often as from genotype 3a to 1a.



Sequencing and phylogenetic analysis



A phylogenetic tree was constructed of all 184 HCV sequences obtained from the 59 HCV seroconverters (Fig. 2). Minor genetic variations between HCV isolates obtained from one host, were classified as intrahost HCV evolution. However, 5 seroconverters had clear evidence of phylogenetically distinct HCV infections with strains of the same genotype (Fig. 2). HCV strains obtained from 2 other seroconverters were separated by bootstrap value >70, and genetic variation exceeded intrahost evolution observed in other participants (Fig. 2). Although reinfection with a very similar HCV strain for these 2 seroconverters is plausible, intrahost genetic evolution cannot be excluded.



Minority HCV variant detection



Two patients, DU 18898 and DU 18917, were selected for minority HCV variant detection. DU 18898 was selected because of two genotype switches (1b→4d→1a) within a period of 15 months. We selected, amplified, and sequenced 40–56 clones representing the four HCV RNA-positive visits during this interval (Fig. 1b). For each time point, all clones were quasispecies of one HCV strain consistent with the dominant genotype detected at that time point. Hence, no dual infections were detected (data not shown).



DU 18917 was selected because, after resolution of the primary HCV 1a infection, we observed reinfection with a different strain of HCV genotype 1a (1a∗), subsequent superinfection with HCV 3a, and eventually a back-switch to the HCV 1a∗ strain, suggesting 3a/1a∗ superinfection (Fig. 1a). We selected, amplified, and sequenced 35 clones from the first and last HCV RNA sample. Except for variations in quasispecies, all clones belonged to the dominant subtype 1a and heterologous 1a∗. For the three intermediate time points, 80–100 clones were sequenced. Evidence of superinfection was found only in the third intermediate sample; 98/100 clones were of subtype 1a∗ and 2/100 clones were subtype 3a (data not shown).



Materials and methods



Study population and sample selection



This study comprises all DU participants (n=59) who seroconverted for HCV during follow-up in the Amsterdam Cohort Studies (ACS) between December 1985 and November 2005 [13]. The ACS is an open, prospective cohort study of both injecting and non-injecting DU, initiated in December 1985 [14]. Recruitment is still ongoing and takes place via local methadone posts, sexually transmitted diseases clinics, and by word of mouth. At ACS visits every 4–6 months, participants complete a standardised questionnaire about personal health, risk behaviour and socio-economic situation; blood is drawn for HIV-testing and storage at −80°C. The moment of HCV seroconversion was calculated as the midpoint between the last HCV-seronegative visit and the first HCV-seropositive visit.



For each HCV seroconverter, samples taken at five time points were selected, if available: the last visit before HCV antibody seroconversion (t=−1), the first visit with a positive HCV antibody test (t=1), visits approximately six months later (t=2) and one year later (t=3), and the last ACS visit prior to November 2005 (t=4). A subset of 19 HCV seroconverters from the ACS identified in an earlier study had more than five samples available [15].



Definitions: reinfection, coinfection and superinfection



Spontaneous viral clearance was assumed if two or more consecutive visits, at least 4 months apart, showed HCV antibodies without detectable HCV RNA. During our study period, none of the participating DU had received HCV antiviral treatment.



HCV reinfection refers to a situation in which a primary HCV infection is spontaneously cleared prior to subsequent infection with a HCV strain of a homologous or heterologous lineage. All subjects who had two or more consecutive anti-HCV positive visits without detectable HCV RNA, at least 4 months apart, prior to an HCV RNA-positive visit were defined as reinfected. This includes (i) HCV seroconverters that had HCV RNA positive visits, separated by at least two RNA negative visits, but also (ii) HCV seroconverters that remained HCV RNA negative at their first anti-HCV positive test (t=1), then returned at least one more consecutive RNA negative visit prior to an RNA-positive visit.



HCV coinfection and HCV superinfection both refer to HCV dual infections. Coinfection is defined as infection with two or more heterologous HCV strains simultaneously or within a window period too narrow for the first HCV infection to have resulted in detectable HCV antibodies. Superinfection is defined as a subsequent infection with a heterologous HCV strain in the presence of a previous HCV strain and the antibodies that it has generated. All changes in HCV strains over time without a demonstrated spontaneous clearance in between were considered as superinfections.



HCV RNA quantification



HCV serum RNA was quantified by an in-house real-time PCR assay based on the 5′-UTR region of the HCV genome [15]. In brief, RNA extraction was performed using the Boom method [16], in which 200μl of serum and 15μl of internal control were added to 900μl of lysis buffer and 20μl of size-fractioned coarse silica particles. RNA was eluted in a volume of 100μl and transcribed to cDNA as detailed elsewhere [17]. Real-time PCR mixes (25μl total volume) contained 12.5μl of 2× LC480 probes master, 0.6μM of forward and reverse primers (HCV47F: 5′-GTGAGGAACTACTGTCTTCACG-3′, HCV312R: 5′-ACTCGCAAGCACCCTATCAGG-3′), and 0.2μM of labeled HCV and IC taqman probes (HCV-P129: 5′-FAM-CTCCCGGGAGAGCCATAGTGGTCTGCG-MGB-NFQ-3′, HCV-IC: 5′-VIC-ATGGCCACAGCGCCGCGGTGTTAGTGC-MGB-NFQ-3′). Real-time PCR was performed on a Roche LC480 using the following cycling conditions: 2min at 50°C and 10min at 95°C followed by 50 cycles of: 20s at 95°C, 20s at 55°C, and 1min at 72°C. Quantification of viral RNA was performed by using standard curves which were produced by linear regression analysis of dilution series of plasmid DNA.



Dominant HCV variant detection: the NS5B PCR



HCV RNA-positive samples for each DU were selected to document changes in the dominant HCV viral variant in individuals over time. From each sample, 3μl cDNA was used as input for a nested multiplex PCR which amplifies 449 nucleotides of the HCV NS5B region (nt 8546–8994). Except for some modifications made to use cDNA instead of RNA as PCR input, conditions and primers of the NS5B PCR were those we described earlier [18].



Minority HCV variant detection: the cloning PCR



The cloning PCR is a single round PCR targeting the NS5B region (337 bp, nt 8279–8615). In brief, 25μl of cDNA was added to 25μl reaction mixture containing PCR II Buffer (Applied Biosystems), 200μmol/L of each dATP, dCTP and dGTP, plus 400μmol/L dUTP (Applied Biosystems), 0.1μg/μL bovine serum albumin (Roche Diagnostics), 0.9μM sense primer 5′-TATGAYACCCGCTGYTTTGACTC-3′, 0.9μM antisense primer 5′-TAYCTVGTCATAGCCTCCGTGAA-3′, 0.5U Uracil N Glycolase (Applied Biosystems) and 2.5U Amplitaq Gold (Applied Biosystems). The final MgCl2 concentration was 2.5mmol/L. PCR cycling conditions were 2min at 50°C and 10min at 95°C followed by 45 cycles of: 20s at 95;°C, 20s at 55°C and 60s at 72°C, with a final incubation of 5min at 72°C. The cloning PCR product was cut from the agarose gel, resolved in 900μl lysis buffer and purified using a shortened Boom-isolation protocol [16]. Purified amplicons were ligated into PCR II vector using the TOPO TA Cloning kit (Invitrogen). The plasmid was used to transform competent Escherichia coli cells, which were plated on LB agar plates containing ampicillin (100μg/ml) and Xgal (5-bromo-4-chloro-3-indolyl-B-galactoside). Based on Xgal selection, 50–100 white clones, containing the PCR insert, were picked for each time point and grown overnight at 37°C on LB agar plates. For each clone, the PCR insert was amplified using M13 primers, according to the TOPO TA Cloning kit protocol, and then sequenced.

2.6. Sequencing and phylogenetic analysis



PCR products were sequenced using previously described methods [18]. Viral genotype was determined after phylogenetic analysis of the NS5B sequences obtained (GenBank Accession Nos. FJ024088 to FJ024273) along with established GenBank reference sequences [19]. A HCV phylogenetic tree was constructed by the neighbour-joining method in Mega version 4.0 [20], using the Tamura–Nei substitution model with γ-distribution (α=0.40). Bootstrap values (n=1000) were calculated to analyze the robustness of tree topology.

OraSure to conduct more testing on hepatitis C product

OraSure to conduct more testing on hepatitis C product
Thu Jun 25, 2009 8:59am EDT


* Says to conduct additional testing and trial of HCV test

* Sees $3 mln charge

* Says facing problems in manufacturing rapid HIV test

* Says advisory committee to review OTC HIV test study

June 25 (Reuters) - OraSure Technologies Inc (OSUR.O) said it would need to conduct an additional study for its hepatitis C virus (HCV) test as health regulators raised concerns of potential interpretation bias, delaying its marketing approval.

The U.S. Food and Drug Administration indicated that the clinical data could have been biased because the same operators performed the test and interpreted the results on multiple specimen types derived from the same patient, the company said.

In October, OraSure filed a premarket approval (PMA) application for its OraQuick HCV test for use wi th multiple specimen types, including venous whole blood, fingerstic k whole blood, oral fluid and other sample types.

OraSure had expected to address the FDA's concerns without material impact to the clinical program.

However, the FDA recently concluded that additional clinical testing will be required to obtain approval of the PMA for a venous whole blood claim, the company said.

A new clinical study will also be required for approval of claims for oral fluid and other sample types, it added.

The exact timing and costs related to the work will not be fully determined until after protocols are submitted and reviewed by the FDA, which should occur in the next several weeks, the company said.

The company expects to record a non-cash impairment charge of $3 million related to a portion of the milestone payments previously made under a license for certa in HCV patents.

MANUFACTURING ISSUES

OraSure also said it has been experiencing an intermittent difficulty in manufacturing a component required for its rapid HIV test, adding that second-quarter revenue would be hurt if it was unable to resume full-scale production of the test this week.

The OraQuick Advance Rapid test, which detects antibod ies to HIV-1 and HIV-2 in 20 minutes, is approved by the FDA for use with oral fluid, fingerstick or venous whole blood and plasma specimens.

The company also expects gross margin for the second quarter to be negatively impacted primarily due to higher unabsorbed production costs and scrap expenses.

COMMITTEE TO REVIEW OTC TEST

OraSure said the FDA indicated that an advisory committee should review and approve the results of an observed use study and the remaining clinical activities related to the company's over-the-counter (OTC) HIV test.

In August, the company had submitted the results of its observed use study to the FDA to obtain approval for an OTC version of its OraQuick test.

The study was designed to assess an individual's ability to interact with the product packaging, comprehend the instructions for us e, take the test and interpret the results.

Shares of the company closed at $3.28 Wednesday on Nasdaq.

"Maintenance Therapy Did Not Reduce Incidence of HCC

"Maintenance Therapy Did Not Reduce
Incidence of HCC in the HALT-C Study"



"Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial."

RESULTS:
"maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort."

"The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis."

"The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls."

Incidence of Hepatocellular Carcinoma and Associated Risk Factors in Hepatitis C-Related Advanced Liver Disease

Gastroenterology Jan 2009 Anna S. LokCorresponding Author Informationemail address, Leonard B. Seeff, Timothy R. Morgan, Adrian M. di Bisceglie, Richard K. Sterling, Teresa M. Curto, Gregory T. Everson, Karen L. Lindsay, William M. Lee, Herbert L. Bonkovsky, Jules L. Dienstag, Marc G. Ghany##, Chihiro Morishima, Zachary D. Goodman, HALT-C Trial Group Received 27 May 2008; accepted 11 September 2008. published online 19 September 2008.

Background & Aims

Although the incidence of hepatocellular carcinoma (HCC) is increasing in the United States, data from large prospective studies are limited. We evaluated the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) cohort for the incidence of HCC and associated risk factors.

Methods

Hepatitis C virus-positive patients with bridging fibrosis or cirrhosis who did not respond to peginterferon and ribavirin were randomized to groups that were given maintenance peginterferon for 3.5 years or no treatment. HCC incidence was determined by Kaplan-Meier analysis, and baseline factors associated with HCC were analyzed by Cox regression.

Results

1,005 patients (mean age, 50.2 years; 71% male; 72% white race) were studied; 59% had bridging fibrosis, and 41% had cirrhosis. During a median follow-up of 4.6 years (maximum, 6.7 years), HCC developed in 48 patients (4.8%). The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis. A multivariate analysis model comprising older age, black race, lower platelet count, higher alkaline phosphatase, esophageal varices, and smoking was developed to predict the risk of HCC.

Conclusions

We found that maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort. Baseline clinical and laboratory features predicted risk for HCC. Additional studies are required to confirm our finding of HCC in patients with chronic hepatitis C and bridging fibrosis.

Chronic hepatitis C is a potentially progressive condition, characterized by slowly evolving fibrosis that can culminate in cirrhosis.1, 2 Over time, compensated cirrhosis can advance to decompensated cirrhosis terminating in hepatic failure and death. A common cause for liver-related death among persons with chronic hepatitis C is hepatocellular carcinoma (HCC), occurring predominantly in those with cirrhosis,3, 4, 5 but reports have appeared of HCC occurring, although far less commonly, in persons with bridging fibrosis without definite cirrhosis.6, 7, 8, 9, 10 In studies from Asia, the annual incidence of HCC in subjects with hepatitis C-related cirrhosis is reported to be 4%-10%, whereas in European studies, the reported annual incidence is 0.5%-5% in comparable subjects.11, 12, 13 Data on the incidence of HCC in subjects with advanced hepatitis C followed prospectively in the United States are limited. Furthermore, the factors that predict a high likelihood of future HCC in such subjects are not clearly defined.

Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial. A total of 1050 persons with chronic hepatitis C and advanced fibrosis or cirrhosis who were previous nonresponders to antiviral therapy were randomized to receive either a reduced dose of pegylated interferon (peginterferon) or no treatment for 3.5 years.18 Because HCC was one of the planned trial outcomes, the HALT-C trial provided an opportunity to determine in a prospective controlled manner whether long-term peginterferon treatment reduced the incidence of HCC and to identify risk factors associated with the development of HCC in a US cohort with chronic hepatitis C and advanced fibrosis or cirrhosis.

Discussion

The major objective of the HALT-C Trial was to determine whether long-term treatment with peginterferon would reduce progression of hepatitis C, including the development of HCC. Support for the possibility that maintenance therapy might inhibit disease progression had come from a 2-year study of interferon maintenance therapy in a US population.23 Additionally, several early reports suggested that interferon therapy reduced the development of HCC in Japanese and Italian patients with chronic hepatitis C.14, 15, 16, 17 Following initiation of the HALT-C Trial in 2000, numerous additional studies were published that addressed the issue of antiviral treatment and development of HCC, leading Craxi and Camma to perform a meta-analysis in 2005 of the then existing studies.24 The analysis included 20 trials, most of which were not randomized studies and 19 of which showed (13 significantly) that antiviral therapy reduced the rate of development of HCC. The validity of the meta-analysis, however, was limited by the marked heterogeneity among the studies. The authors concluded that, although interferon treatment appeared to have a beneficial impact on reducing the development of HCC, the effect was slight and was most evident among the relatively few patients with cirrhosis who achieved a sustained virologic response. More recent studies from Japan and Taiwan, however, have continued to support a positive impact of antiviral therapy on the occurrence of HCC,25, 26, 27, 28 but patients in these trials were not randomized, and, again, the benefit was observed mainly in patients who had achieved a sustained virologic response.

In the present study, we found no significant difference in the incidence of HCC between patients randomized to receive peginterferon maintenance therapy (23/495; 4.7%) and those randomized to no treatment (25/510; 4.9%). The cumulative incidence of HCC 5 years after randomization in the 2 cohorts was almost identical (treated, 5.4% and controls, 5.0%). These data, collected prospectively from a large number of subjects followed closely for a median of 4.6 years, suggest strongly that maintenance therapy with peginterferon, 90 µg/week, does not reduce the incidence of HCC in patients with chronic hepatitis C and advanced fibrosis or cirrhosis.

Because treatment had no effect on the rate of development of HCC in this trial, data from the 2 cohorts (treated and untreated) could be combined to provide information on the annual incidence of HCC among patients with advanced HCV-related liver disease in the United States followed prospectively. Thus, for the combined cohort, the 3-year incidence of HCC from study entry was 1.9%, increasing to 5.9% at 5 years, yielding an annual incidence of HCC of 1.1%. This figure is consistent with but at the lower range of previous estimates of the annual incidence of HCC both in the United States,29 now increasing,30 and in Europe3, 4, 12 but clearly lower than the rates reported among HCV-infected persons in Japan.31, 32, 33

The most likely explanation for the disparity between this trial and others in the efficacy of interferon-based therapy in reducing the incidence of HCC is the difference in study design between the HALT-C Trial and many of the previous studies. The HALT-C Trial was a large, well-powered, prospective, randomized, multicenter trial, with clear, predefined criteria for the diagnosis of HCC. Many prior studies of interferon treatment were relatively small, nonrandomized, and/or retrospective, with all the potential pitfalls of such observations. Furthermore, in several previous reports, the decreased incidence of HCC was observed mainly in patients who achieved a sustained virologic response, subjects who were excluded from the HALT-C Trial. Potentially, the HALT-C Trial might have demonstrated efficacy of antiviral therapy in preventing HCC had the maintenance treatment consisted of full-dose peginterferon and ribavirin instead of half-dose peginterferon only. However, it is unlikely that patients with advanced fibrosis or cirrhosis could tolerate full-dose combination therapy for 3.5 years; only 59% of patients in the HALT-C Trial randomized to receive half-dose peginterferon therapy were able to tolerate this lower dose for the entire 3.5 years. Finally, the lower incidence of HCC observed in the HALT-C Trial compared with previous studies could be related to differences in radiologic and histologic diagnostic criteria for HCC. In the current study, cases of HCC identified within 12 months of enrollment were excluded because the tumors may have been present prior to entry into the trial. Although 23% (11/48) of the HCC cases in our trial were categorized as presumed, rather than definite, HCC, almost all of these cases fulfilled currently accepted diagnostic criteria for HCC.34, 35, 36 Indeed, the diagnostic criteria utilized in this study were very stringent, and the outcome was the same whether we included presumed cases or not.

The rate of HCC was higher in those whose baseline liver biopsy displayed cirrhosis than in those with bridging fibrosis (6.1% vs 3.9%, respectively; P = .08). Although HCC has been reported rarely in patients with chronic hepatitis C in the absence of cirrhosis, the high frequency of HCC among subjects without cirrhosis in the present study was unexpected.6, 7, 8, 9, 37 One possible explanation is understaging of the baseline liver biopsy resulting from sampling error; however, 8 of the 23 patients with HCC who had no cirrhosis at enrollment also had no cirrhosis on follow-up liver biopsy (all 8 patients) or on explant (1 patient). Alternatively, hepatic fibrosis may have progressed between the time of the last study biopsy and the subsequent diagnosis of HCC because the interval between the last biopsy with no cirrhosis and HCC diagnosis varied from 0 to 34 months (median, 16). Whatever the explanation for our observation, the development of HCC in persons with noncirrhotic fibrosis raises the question of whether such patients should undergo HCC surveillance as is recommended for persons with established cirrhosis.33

In seeking to identify features that might predict the future development of HCC, we undertook an extensive analysis of baseline demographic characteristics; metabolic and viral factors; laboratory values (including HCC biomarkers); disease severity; histologic findings; markers of portal hypertension; treatment assignment; physical activity; and history of alcohol use, smoking, and hormone use in women. On univariate analysis, treatment group assignment was not associated with risk of HCC, and we could not confirm associations reported by others between diabetes or alcohol use and the development of HCC.38, 39, 40, 41, 42, 43 In the current study, subjects with HCC had a lower body mass index, not only at enrollment, but also, based on historical information, at ages 20 and 40 years, than subjects in whom HCC did not develop. Prior reports of an association between obesity and HCC may have included patients with nonalcoholic fatty liver disease.44, 45, 46 In this study, a strong association was observed between the risk of HCC and a low platelet count, a high AST level, and the presence of esophageal varices. This finding is consistent with an association in several studies from Asia between a low platelet count and the development of HCC47, 48 and an association in a study from Italy between the presence of esophageal varices and HCC.49

Numerous efforts have been made to develop predictive models for the development of HCC in persons with advanced chronic liver disease.50, 51, 52, 53 In multivariate analyses, these investigators have identified varying factors, including age, gender, platelet count, prothrombin time, albumin level, and AFP as predictive variables. In the current study, we found that gender, cirrhosis, and AFP were not significant in all models tested, whereas DCP was. Smoking was not a significant factor on univariate analysis but was significant in several regression models tested and showed a trend in our final model. Several simplified models that included factors easily accessible in routine clinical settings were developed to aid in predicting the risk of HCC among patients with hepatitis C and advanced fibrosis or cirrhosis. These models may be useful for selecting high-risk patients with bridging fibrosis and cirrhosis for HCC surveillance.

In conclusion, in the current analysis of hepatitis C patients with advanced fibrosis enrolled in the HALT-C Trial, 3.5 years of maintenance therapy with peginterferon did not reduce the incidence of HCC. The 1% annual incidence of HCC in this US cohort was lower than those reported in studies of similar patients in Japan and Europe. Modeling based on clinical and laboratory features can identify patients at higher risk who might be candidates for more intense HCC surveillance. Additional studies are required to confirm the occurrence of HCC among patients with bridging fibrosis but no cirrhosis and to determine whether such patients should undergo HCC surveillance.

Results

A total of 1050 patients were randomized between January 2001 and August 2004. Forty-five patients were excluded for the following reasons: (1) prevalent HCC, defined as HCC diagnosed within 12 months after enrollment (n = 5); (2) patients who had less than 12 months follow-up after enrollment (n = 38); and (3) patients with a diagnosis of presumed HCC who were followed for at least 24 months and did not show radiologic or clinical progression of their liver masses (n = 2). The characteristics of the remaining 1005 patients included in this analysis are shown in Table 1. The mean age of the patients was 50.2 years, 71% were men, and 72% were of white ethnicity. Baseline biopsy specimens showed bridging fibrosis (Ishak score 3 or 4) in 59% and cirrhosis (Ishak score 5 or 6) in the remaining 41% of patients. The vast majority of the patients (93%) had genotype 1 infection.

Incidence of HCC

During a median follow-up of 4.6 years after randomization (maximum of 6.7 years), 48 (4.8%) patients met criteria for HCC, 37 (77%) for definite HCC, and 11 (23%) for presumed HCC (Table 2). Thirty-four (71%) of the 48 patients with HCC were identified during the randomized phase and the remaining 14 during the extended follow-up period. Thirty-six (75%) patients had early stage (T1/T2) HCC at the time of diagnosis. Twenty-six (54.2%) patients were eligible for potentially curative treatment: 3 had surgical resection, 17 underwent liver transplantation, and 6 had radiofrequency ablation ± transarterial chemoembolization (TACE). An additional 11 patients had TACE. Thus, in total, 37 (77.1%) patients had tumors diagnosed at an early enough stage that allowed them to receive effective therapies.

HCC developed in a similar proportion of peginterferon-treated patients, 4.5% (23/495), and controls, 4.9% (25/510). The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls. The cumulative incidence of HCC 3 and 5 years after randomization was 1.9% and 5.4% for peginterferon-treated patients and 1.9% and 5.0% for controls, respectively (P = .78) (Figure 1A). The incidence of HCC in the peginterferon-treated and control groups was similar regardless of the inclusion or exclusion of cases of presumed HCC.

HCC was diagnosed in a higher percent of patients with cirrhosis, 6.1% (25/408), than in those with bridging fibrosis, 3.9% (23/597). The estimated annual incidence of HCC was 0.8% among patients with bridging fibrosis and 1.4% among those with cirrhosis. The cumulative incidence of HCC 3 and 5 years after randomization was 2.6% and 7.0% for patients with cirrhosis and 1.4% and 4.1% for those with bridging fibrosis, respectively (P = .08) (Figure 1B).

HCC in Patients Without Cirrhosis at Enrollment

Of the 23 patients with HCC and an Ishak fibrosis score of 3 or 4 at enrollment, 13 advanced to a score of 5 or 6, 6 remained with a score of 3 or 4, and 2 had a score of 2 on subsequent biopsies, liver explants, or surgically resected specimen; 2 patients did not have repeat biopsies (Table 3). Thus, 17% (8/48) of the patients in whom HCC developed did not have cirrhosis on at least 2 biopsies more than 2 years apart. HCC was diagnosed after a median of 46.5 months (range, 26-72) from enrollment in these 8 patients. At enrollment, 6 of these 8 patients had platelet counts >150,000/mm3, and none had esophageal varices; however, at the time of the diagnosis of HCC, only 1 patient had a platelet count >150,000/mm3 (Table 4).

Risk Factors for HCC

Baseline clinical and laboratory characteristics of the patients with and without HCC are shown in Table 1. Based on univariate analysis, older age; lower body mass index; lower white cell and platelet counts; a lower albumin level; higher levels of alkaline phosphatase, AST, alanine aminotransferase (ALT), AFP, and DCP; and the presence of esophageal varices were significantly associated with HCC (each P < .05). Black race, higher international normalized ratio of prothrombin time (INR), and histologic evidence of cirrhosis showed a trend toward significance (P < .1), whereas gender, diabetes, alcohol consumption, smoking, treatment assignment, HCV genotype, and bilirubin were not significantly associated with HCC.

Models to Predict HCC Incidence

Several Cox proportional hazards models were developed to predict the risk of HCC (Table 5). The initial model (model I), which included significant variables from our univariate analysis or from previously published analyses, was modified to eliminate variables that were not significant on multivariate analysis. Three simplified models (models II-IV) are shown because they include variables that are available in most patients with chronic hepatitis C. The final model (model IV) included age in years (continuous), black race (black = 1, nonblack = 0), alkaline phosphatase in units per liter (continuous), the presence of esophageal varices (present = 1; absent = 0), smoking (ever smoked = 1, never smoked = 0), and platelet count x 1000/mm3 (continuous). The regression formula for model IV is as follows: age*0.049 + black race*0.712 + alkaline phosphatase*0.006 + esophageal varices*0.777 + ever smoked*0.749 + platelets*-0.011. A score was calculated for each subject included in this study, and cases were ranked in order of their calculated scores in this model. Three cut points (0 to log10 [1.50], log10 [1.50] to log10 [3.25], and >log10 [3.25]) were selected to reflect low, intermediate, and high risk, as distinguished by 5-year cumulative incidence of HCC of <1%, 1%-5%, and >5%. The cumulative incidence of HCC 3 and 5 years after randomization was 0% and 0.4% for patients in the low-risk group, 1.5% and 4.2% for patients in the intermediate-risk group, and 6.1% and 17.8% for patients in the high-risk group, respectively (Figure 2).

Low platelet count and presence of esophageal varices were highly significant in the models tested. The impact of esophageal varices and platelet count on the incidence of HCC is shown in Figure 3A and B. The cumulative incidence of HCC 3 and 5 years after randomization was 5.3% and 13.1%, 2.4% and 5.7%, and 0.6% and 2.6% for patients with baseline platelet count <100, 100-149, and >150 x 1000/mm3, respectively. The estimated annual incidence of HCC for patients in these 3 strata of platelet counts was 2.6%, 1.1%, and 0.5%, respectively.

Patients and Methods

The design of the HALT-C Trial has been described previously.18, 19 Briefly, patients with detectable HCV RNA at 10 clinical centers had to meet the following criteria for enrollment: failure to have achieved a sustained virologic response after previous interferon treatment with or without ribavirin, the presence of advanced hepatic fibrosis on liver biopsy (Ishak fibrosis score, ≥3), no history of hepatic decompensation or HCC, and the absence of defined exclusion criteria (eg, liver disease other than hepatitis C, uncontrolled medical or psychiatric conditions, or contraindications to use of interferon or ribavirin).

Patients were treated with peginterferon alfa-2a (180 µg weekly) and ribavirin (1-1.2 g daily) for at least 24 weeks ("Lead-in" phase). Patients who had detectable serum HCV RNA at treatment week 20 (protocol definition of nonresponse) were randomized at week 24 to maintenance therapy (peginterferon alfa-2a, 90 µg weekly) or to no treatment for the next 3.5 years. Patients with undetectable serum HCV RNA at week 20 were considered responders and continued combination therapy for a total of 48 weeks. Patients in the responder group in whom HCV RNA became detectable later during treatment (breakthrough) or after stopping treatment (relapse) were offered enrollment into the randomized trial. After approval of peginterferons, the protocol was amended to allow randomization of patients who had failed treatment with peginterferon and ribavirin outside this trial (Express patients). Following completion of the 3.5 years of the randomized trial, all patients were invited to continue follow-up without treatment.

All patients were required to have an ultrasound, computed tomography, or magnetic resonance imaging with no evidence of hepatic mass lesions suspicious for HCC. Subjects enrolled in the Lead-in phase were required to have serum α-fetoprotein (AFP) <200 ng/mL, whereas those enrolled as Express patients had to have AFP <1000 ng/mL at the screening visit. Of the 237 Express patients enrolled, only 3 had AFP >200 ng/mL at entry, and the highest AFP value was 315 ng/mL.

All patients had a liver biopsy performed prior to enrollment. For Lead-in patients, the baseline liver biopsy was done at least 2 months following the last course of interferon therapy and within 12 months prior to enrollment. For Express patients, a liver biopsy must have been performed within 18 months prior to randomization, or a baseline liver biopsy must have been performed at least 8 weeks after the end of the prior course of peginterferon and ribavirin and within 24 weeks prior to randomization. Liver biopsies were repeated 1.5 and 3.5 years after randomization. All biopsy specimens were reviewed in conference by a panel of 12 hepatic pathologists, who used the Ishak scoring system to grade inflammation (0-18) and to stage fibrosis (0-6).20 Liver biopsies performed for the diagnosis of HCC and explant livers of patients who underwent liver transplantation for HCC were also submitted for review by the group of hepatic pathologists.

HCC Surveillance and Diagnostic Criteria

Patients were seen every 3 months during the 3.5 years of the randomized trial and every 6 months thereafter. Patients who stopped treatment continued to be followed in the study unless consent was withdrawn or a liver transplantation had occurred. A complete blood count, a liver panel, and AFP level were obtained at the local clinical center at each visit. Des-γ-carboxy prothrombin (DCP) levels at enrollment were tested in batches with an enzyme immunoassay (Eisai Company, Tokyo, Japan) in a central laboratory at the University of Michigan; the results were not reported to the investigators at the clinical sites. Lead-in patients underwent an ultrasound examination of the liver 20 weeks and 12 months after enrollment and then every 6-12 months. Express patients underwent ultrasound examinations 6 months after enrollment and then every 6-12 months. Patients with an elevated or rising AFP and those with new lesions on ultrasound were evaluated further with a computed tomography or magnetic resonance imaging. Diagnostic liver biopsy and HCC treatment were conducted at the discretion of investigators at each site. All patients underwent an endoscopy at the time of randomization and after 3.5 years (ie, at the end of the randomized trial). For the current analysis, patients were categorized as not having varices (absent) or as having varices, regardless of their size.

Two definitions of HCC were adopted: one for "definite" HCC and one for "presumed" HCC. Definite HCC was defined by histologic confirmation or a new mass lesion on imaging with AFP levels increasing to >1000 ng/mL. Presumed HCC was defined as a new mass lesion on ultrasound in the absence of histology and AFP was <1000 ng/mL in conjunction with one of the following characteristics: (1) 2 liver imaging studies showing a mass lesion with characteristics of HCC (vascular enhancement, wash out), (2) progressively enlarging lesion on ultrasound leading to death, or (3) 1 additional imaging study showing a mass lesion with characteristics of HCC that either increased in size over time or was accompanied by increasing AFP levels. Tumors were staged based on the modified United Network of Organ Sharing TNM system.21 Early HCC was defined as tumor stage T1 (single lesion <2 cm in diameter) or T2 (single lesion between 2 and 5 cm or no more than 3 lesions each <3 cm in diameter).

All cases of HCC (presumed and definite) were reviewed by an Outcomes Review Panel composed of rotating panels of 3 investigators. The outcomes review panel did not include any radiologist, and radiographic studies were not reviewed. However, the outcome panel did review serial radiology reports from each HCC case to ascertain that the predefined criteria were met, the date when these criteria were first met, and the tumor staging. In cases in which the outcomes review panel was unable to determine whether the patient had met criteria for HCC, the investigators were requested to provide additional information including follow-up imaging or histology, and the case was re-reviewed. All patients who met criteria for presumed or definite HCC continued to be followed, and results of subsequent imaging or histology (including liver explant and surgically resected specimen) were submitted for review. An earlier report on all trial outcomes (clinical and histologic) included randomized patients with definite HCC (incident and prevalent) identified during the first 3.5 years after randomization (up to April 30, 2007).19 The current analysis includes all randomized patients in the HALT-C Trial who had a diagnosis of definite or presumed HCC as judged by the Outcomes Review Panel prior to October 15, 2007.

Statistical Analyses

Statistical analyses were performed at the Data Coordinating Center with SAS release 9.1 (SAS Institute, Cary, NC). Cumulative incidence of HCC was determined by Kaplan-Meier analysis, and differences were compared with the log-rank test. Relation of baseline factors to risk of HCC was assessed by t test, χ2 test, or univariate Cox proportional hazards regression. Several multivariate Cox proportional hazards models to estimate the risk of HCC were tested. Variables with a P value < .1 on univariate analysis and variables reported previously to be associated with HCC (eg, gender; race/ethnicity; presence of cirrhosis, alcohol consumption, and smoking) were entered into a multivariate analysis (model I). Treatment assignment was also included because the primary goal of the HALT-C Trial was to determine whether maintenance therapy with peginterferon would prevent clinical outcomes, including HCC, in patients with chronic hepatitis C and advanced fibrosis or cirrhosis. Alternative multivariate analyses that included AFP and DCP separately and in combination as well as aspartate aminotransferase (AST) and platelet count separately and combined as AST-platelet ratio index (APRI)22 were tested. Variables not readily available in clinical practice were deleted from model I to create models with greater clinical utility (models II, III, and IV). The cohort was divided into 3 risk groups based on the resulting individual estimates of model IV. The cumulative incidence of HCC among these groups was determined by Kaplan-Meier analysis and compared with the log-rank test. A 2-sided significance level of 5% was used for all analyses.

"Maintenance Therapy Did Not Reduce Incidence of HCC

"Maintenance Therapy Did Not Reduce
Incidence of HCC in the HALT-C Study"



"Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial."

RESULTS:
"maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort."

"The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis."

"The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls."

Incidence of Hepatocellular Carcinoma and Associated Risk Factors in Hepatitis C-Related Advanced Liver Disease

Gastroenterology Jan 2009 Anna S. LokCorresponding Author Informationemail address, Leonard B. Seeff, Timothy R. Morgan, Adrian M. di Bisceglie, Richard K. Sterling, Teresa M. Curto, Gregory T. Everson, Karen L. Lindsay, William M. Lee, Herbert L. Bonkovsky, Jules L. Dienstag, Marc G. Ghany##, Chihiro Morishima, Zachary D. Goodman, HALT-C Trial Group Received 27 May 2008; accepted 11 September 2008. published online 19 September 2008.

Background & Aims

Although the incidence of hepatocellular carcinoma (HCC) is increasing in the United States, data from large prospective studies are limited. We evaluated the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) cohort for the incidence of HCC and associated risk factors.

Methods

Hepatitis C virus-positive patients with bridging fibrosis or cirrhosis who did not respond to peginterferon and ribavirin were randomized to groups that were given maintenance peginterferon for 3.5 years or no treatment. HCC incidence was determined by Kaplan-Meier analysis, and baseline factors associated with HCC were analyzed by Cox regression.

Results

1,005 patients (mean age, 50.2 years; 71% male; 72% white race) were studied; 59% had bridging fibrosis, and 41% had cirrhosis. During a median follow-up of 4.6 years (maximum, 6.7 years), HCC developed in 48 patients (4.8%). The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis. A multivariate analysis model comprising older age, black race, lower platelet count, higher alkaline phosphatase, esophageal varices, and smoking was developed to predict the risk of HCC.

Conclusions

We found that maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort. Baseline clinical and laboratory features predicted risk for HCC. Additional studies are required to confirm our finding of HCC in patients with chronic hepatitis C and bridging fibrosis.

Chronic hepatitis C is a potentially progressive condition, characterized by slowly evolving fibrosis that can culminate in cirrhosis.1, 2 Over time, compensated cirrhosis can advance to decompensated cirrhosis terminating in hepatic failure and death. A common cause for liver-related death among persons with chronic hepatitis C is hepatocellular carcinoma (HCC), occurring predominantly in those with cirrhosis,3, 4, 5 but reports have appeared of HCC occurring, although far less commonly, in persons with bridging fibrosis without definite cirrhosis.6, 7, 8, 9, 10 In studies from Asia, the annual incidence of HCC in subjects with hepatitis C-related cirrhosis is reported to be 4%-10%, whereas in European studies, the reported annual incidence is 0.5%-5% in comparable subjects.11, 12, 13 Data on the incidence of HCC in subjects with advanced hepatitis C followed prospectively in the United States are limited. Furthermore, the factors that predict a high likelihood of future HCC in such subjects are not clearly defined.

Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial. A total of 1050 persons with chronic hepatitis C and advanced fibrosis or cirrhosis who were previous nonresponders to antiviral therapy were randomized to receive either a reduced dose of pegylated interferon (peginterferon) or no treatment for 3.5 years.18 Because HCC was one of the planned trial outcomes, the HALT-C trial provided an opportunity to determine in a prospective controlled manner whether long-term peginterferon treatment reduced the incidence of HCC and to identify risk factors associated with the development of HCC in a US cohort with chronic hepatitis C and advanced fibrosis or cirrhosis.

Discussion

The major objective of the HALT-C Trial was to determine whether long-term treatment with peginterferon would reduce progression of hepatitis C, including the development of HCC. Support for the possibility that maintenance therapy might inhibit disease progression had come from a 2-year study of interferon maintenance therapy in a US population.23 Additionally, several early reports suggested that interferon therapy reduced the development of HCC in Japanese and Italian patients with chronic hepatitis C.14, 15, 16, 17 Following initiation of the HALT-C Trial in 2000, numerous additional studies were published that addressed the issue of antiviral treatment and development of HCC, leading Craxi and Camma to perform a meta-analysis in 2005 of the then existing studies.24 The analysis included 20 trials, most of which were not randomized studies and 19 of which showed (13 significantly) that antiviral therapy reduced the rate of development of HCC. The validity of the meta-analysis, however, was limited by the marked heterogeneity among the studies. The authors concluded that, although interferon treatment appeared to have a beneficial impact on reducing the development of HCC, the effect was slight and was most evident among the relatively few patients with cirrhosis who achieved a sustained virologic response. More recent studies from Japan and Taiwan, however, have continued to support a positive impact of antiviral therapy on the occurrence of HCC,25, 26, 27, 28 but patients in these trials were not randomized, and, again, the benefit was observed mainly in patients who had achieved a sustained virologic response.

In the present study, we found no significant difference in the incidence of HCC between patients randomized to receive peginterferon maintenance therapy (23/495; 4.7%) and those randomized to no treatment (25/510; 4.9%). The cumulative incidence of HCC 5 years after randomization in the 2 cohorts was almost identical (treated, 5.4% and controls, 5.0%). These data, collected prospectively from a large number of subjects followed closely for a median of 4.6 years, suggest strongly that maintenance therapy with peginterferon, 90 µg/week, does not reduce the incidence of HCC in patients with chronic hepatitis C and advanced fibrosis or cirrhosis.

Because treatment had no effect on the rate of development of HCC in this trial, data from the 2 cohorts (treated and untreated) could be combined to provide information on the annual incidence of HCC among patients with advanced HCV-related liver disease in the United States followed prospectively. Thus, for the combined cohort, the 3-year incidence of HCC from study entry was 1.9%, increasing to 5.9% at 5 years, yielding an annual incidence of HCC of 1.1%. This figure is consistent with but at the lower range of previous estimates of the annual incidence of HCC both in the United States,29 now increasing,30 and in Europe3, 4, 12 but clearly lower than the rates reported among HCV-infected persons in Japan.31, 32, 33

The most likely explanation for the disparity between this trial and others in the efficacy of interferon-based therapy in reducing the incidence of HCC is the difference in study design between the HALT-C Trial and many of the previous studies. The HALT-C Trial was a large, well-powered, prospective, randomized, multicenter trial, with clear, predefined criteria for the diagnosis of HCC. Many prior studies of interferon treatment were relatively small, nonrandomized, and/or retrospective, with all the potential pitfalls of such observations. Furthermore, in several previous reports, the decreased incidence of HCC was observed mainly in patients who achieved a sustained virologic response, subjects who were excluded from the HALT-C Trial. Potentially, the HALT-C Trial might have demonstrated efficacy of antiviral therapy in preventing HCC had the maintenance treatment consisted of full-dose peginterferon and ribavirin instead of half-dose peginterferon only. However, it is unlikely that patients with advanced fibrosis or cirrhosis could tolerate full-dose combination therapy for 3.5 years; only 59% of patients in the HALT-C Trial randomized to receive half-dose peginterferon therapy were able to tolerate this lower dose for the entire 3.5 years. Finally, the lower incidence of HCC observed in the HALT-C Trial compared with previous studies could be related to differences in radiologic and histologic diagnostic criteria for HCC. In the current study, cases of HCC identified within 12 months of enrollment were excluded because the tumors may have been present prior to entry into the trial. Although 23% (11/48) of the HCC cases in our trial were categorized as presumed, rather than definite, HCC, almost all of these cases fulfilled currently accepted diagnostic criteria for HCC.34, 35, 36 Indeed, the diagnostic criteria utilized in this study were very stringent, and the outcome was the same whether we included presumed cases or not.

The rate of HCC was higher in those whose baseline liver biopsy displayed cirrhosis than in those with bridging fibrosis (6.1% vs 3.9%, respectively; P = .08). Although HCC has been reported rarely in patients with chronic hepatitis C in the absence of cirrhosis, the high frequency of HCC among subjects without cirrhosis in the present study was unexpected.6, 7, 8, 9, 37 One possible explanation is understaging of the baseline liver biopsy resulting from sampling error; however, 8 of the 23 patients with HCC who had no cirrhosis at enrollment also had no cirrhosis on follow-up liver biopsy (all 8 patients) or on explant (1 patient). Alternatively, hepatic fibrosis may have progressed between the time of the last study biopsy and the subsequent diagnosis of HCC because the interval between the last biopsy with no cirrhosis and HCC diagnosis varied from 0 to 34 months (median, 16). Whatever the explanation for our observation, the development of HCC in persons with noncirrhotic fibrosis raises the question of whether such patients should undergo HCC surveillance as is recommended for persons with established cirrhosis.33

In seeking to identify features that might predict the future development of HCC, we undertook an extensive analysis of baseline demographic characteristics; metabolic and viral factors; laboratory values (including HCC biomarkers); disease severity; histologic findings; markers of portal hypertension; treatment assignment; physical activity; and history of alcohol use, smoking, and hormone use in women. On univariate analysis, treatment group assignment was not associated with risk of HCC, and we could not confirm associations reported by others between diabetes or alcohol use and the development of HCC.38, 39, 40, 41, 42, 43 In the current study, subjects with HCC had a lower body mass index, not only at enrollment, but also, based on historical information, at ages 20 and 40 years, than subjects in whom HCC did not develop. Prior reports of an association between obesity and HCC may have included patients with nonalcoholic fatty liver disease.44, 45, 46 In this study, a strong association was observed between the risk of HCC and a low platelet count, a high AST level, and the presence of esophageal varices. This finding is consistent with an association in several studies from Asia between a low platelet count and the development of HCC47, 48 and an association in a study from Italy between the presence of esophageal varices and HCC.49

Numerous efforts have been made to develop predictive models for the development of HCC in persons with advanced chronic liver disease.50, 51, 52, 53 In multivariate analyses, these investigators have identified varying factors, including age, gender, platelet count, prothrombin time, albumin level, and AFP as predictive variables. In the current study, we found that gender, cirrhosis, and AFP were not significant in all models tested, whereas DCP was. Smoking was not a significant factor on univariate analysis but was significant in several regression models tested and showed a trend in our final model. Several simplified models that included factors easily accessible in routine clinical settings were developed to aid in predicting the risk of HCC among patients with hepatitis C and advanced fibrosis or cirrhosis. These models may be useful for selecting high-risk patients with bridging fibrosis and cirrhosis for HCC surveillance.

In conclusion, in the current analysis of hepatitis C patients with advanced fibrosis enrolled in the HALT-C Trial, 3.5 years of maintenance therapy with peginterferon did not reduce the incidence of HCC. The 1% annual incidence of HCC in this US cohort was lower than those reported in studies of similar patients in Japan and Europe. Modeling based on clinical and laboratory features can identify patients at higher risk who might be candidates for more intense HCC surveillance. Additional studies are required to confirm the occurrence of HCC among patients with bridging fibrosis but no cirrhosis and to determine whether such patients should undergo HCC surveillance.

Results

A total of 1050 patients were randomized between January 2001 and August 2004. Forty-five patients were excluded for the following reasons: (1) prevalent HCC, defined as HCC diagnosed within 12 months after enrollment (n = 5); (2) patients who had less than 12 months follow-up after enrollment (n = 38); and (3) patients with a diagnosis of presumed HCC who were followed for at least 24 months and did not show radiologic or clinical progression of their liver masses (n = 2). The characteristics of the remaining 1005 patients included in this analysis are shown in Table 1. The mean age of the patients was 50.2 years, 71% were men, and 72% were of white ethnicity. Baseline biopsy specimens showed bridging fibrosis (Ishak score 3 or 4) in 59% and cirrhosis (Ishak score 5 or 6) in the remaining 41% of patients. The vast majority of the patients (93%) had genotype 1 infection.

Incidence of HCC

During a median follow-up of 4.6 years after randomization (maximum of 6.7 years), 48 (4.8%) patients met criteria for HCC, 37 (77%) for definite HCC, and 11 (23%) for presumed HCC (Table 2). Thirty-four (71%) of the 48 patients with HCC were identified during the randomized phase and the remaining 14 during the extended follow-up period. Thirty-six (75%) patients had early stage (T1/T2) HCC at the time of diagnosis. Twenty-six (54.2%) patients were eligible for potentially curative treatment: 3 had surgical resection, 17 underwent liver transplantation, and 6 had radiofrequency ablation ± transarterial chemoembolization (TACE). An additional 11 patients had TACE. Thus, in total, 37 (77.1%) patients had tumors diagnosed at an early enough stage that allowed them to receive effective therapies.

HCC developed in a similar proportion of peginterferon-treated patients, 4.5% (23/495), and controls, 4.9% (25/510). The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls. The cumulative incidence of HCC 3 and 5 years after randomization was 1.9% and 5.4% for peginterferon-treated patients and 1.9% and 5.0% for controls, respectively (P = .78) (Figure 1A). The incidence of HCC in the peginterferon-treated and control groups was similar regardless of the inclusion or exclusion of cases of presumed HCC.

HCC was diagnosed in a higher percent of patients with cirrhosis, 6.1% (25/408), than in those with bridging fibrosis, 3.9% (23/597). The estimated annual incidence of HCC was 0.8% among patients with bridging fibrosis and 1.4% among those with cirrhosis. The cumulative incidence of HCC 3 and 5 years after randomization was 2.6% and 7.0% for patients with cirrhosis and 1.4% and 4.1% for those with bridging fibrosis, respectively (P = .08) (Figure 1B).

HCC in Patients Without Cirrhosis at Enrollment

Of the 23 patients with HCC and an Ishak fibrosis score of 3 or 4 at enrollment, 13 advanced to a score of 5 or 6, 6 remained with a score of 3 or 4, and 2 had a score of 2 on subsequent biopsies, liver explants, or surgically resected specimen; 2 patients did not have repeat biopsies (Table 3). Thus, 17% (8/48) of the patients in whom HCC developed did not have cirrhosis on at least 2 biopsies more than 2 years apart. HCC was diagnosed after a median of 46.5 months (range, 26-72) from enrollment in these 8 patients. At enrollment, 6 of these 8 patients had platelet counts >150,000/mm3, and none had esophageal varices; however, at the time of the diagnosis of HCC, only 1 patient had a platelet count >150,000/mm3 (Table 4).

Risk Factors for HCC

Baseline clinical and laboratory characteristics of the patients with and without HCC are shown in Table 1. Based on univariate analysis, older age; lower body mass index; lower white cell and platelet counts; a lower albumin level; higher levels of alkaline phosphatase, AST, alanine aminotransferase (ALT), AFP, and DCP; and the presence of esophageal varices were significantly associated with HCC (each P < .05). Black race, higher international normalized ratio of prothrombin time (INR), and histologic evidence of cirrhosis showed a trend toward significance (P < .1), whereas gender, diabetes, alcohol consumption, smoking, treatment assignment, HCV genotype, and bilirubin were not significantly associated with HCC.

Models to Predict HCC Incidence

Several Cox proportional hazards models were developed to predict the risk of HCC (Table 5). The initial model (model I), which included significant variables from our univariate analysis or from previously published analyses, was modified to eliminate variables that were not significant on multivariate analysis. Three simplified models (models II-IV) are shown because they include variables that are available in most patients with chronic hepatitis C. The final model (model IV) included age in years (continuous), black race (black = 1, nonblack = 0), alkaline phosphatase in units per liter (continuous), the presence of esophageal varices (present = 1; absent = 0), smoking (ever smoked = 1, never smoked = 0), and platelet count x 1000/mm3 (continuous). The regression formula for model IV is as follows: age*0.049 + black race*0.712 + alkaline phosphatase*0.006 + esophageal varices*0.777 + ever smoked*0.749 + platelets*-0.011. A score was calculated for each subject included in this study, and cases were ranked in order of their calculated scores in this model. Three cut points (0 to log10 [1.50], log10 [1.50] to log10 [3.25], and >log10 [3.25]) were selected to reflect low, intermediate, and high risk, as distinguished by 5-year cumulative incidence of HCC of <1%, 1%-5%, and >5%. The cumulative incidence of HCC 3 and 5 years after randomization was 0% and 0.4% for patients in the low-risk group, 1.5% and 4.2% for patients in the intermediate-risk group, and 6.1% and 17.8% for patients in the high-risk group, respectively (Figure 2).

Low platelet count and presence of esophageal varices were highly significant in the models tested. The impact of esophageal varices and platelet count on the incidence of HCC is shown in Figure 3A and B. The cumulative incidence of HCC 3 and 5 years after randomization was 5.3% and 13.1%, 2.4% and 5.7%, and 0.6% and 2.6% for patients with baseline platelet count <100, 100-149, and >150 x 1000/mm3, respectively. The estimated annual incidence of HCC for patients in these 3 strata of platelet counts was 2.6%, 1.1%, and 0.5%, respectively.

Patients and Methods

The design of the HALT-C Trial has been described previously.18, 19 Briefly, patients with detectable HCV RNA at 10 clinical centers had to meet the following criteria for enrollment: failure to have achieved a sustained virologic response after previous interferon treatment with or without ribavirin, the presence of advanced hepatic fibrosis on liver biopsy (Ishak fibrosis score, ≥3), no history of hepatic decompensation or HCC, and the absence of defined exclusion criteria (eg, liver disease other than hepatitis C, uncontrolled medical or psychiatric conditions, or contraindications to use of interferon or ribavirin).

Patients were treated with peginterferon alfa-2a (180 µg weekly) and ribavirin (1-1.2 g daily) for at least 24 weeks ("Lead-in" phase). Patients who had detectable serum HCV RNA at treatment week 20 (protocol definition of nonresponse) were randomized at week 24 to maintenance therapy (peginterferon alfa-2a, 90 µg weekly) or to no treatment for the next 3.5 years. Patients with undetectable serum HCV RNA at week 20 were considered responders and continued combination therapy for a total of 48 weeks. Patients in the responder group in whom HCV RNA became detectable later during treatment (breakthrough) or after stopping treatment (relapse) were offered enrollment into the randomized trial. After approval of peginterferons, the protocol was amended to allow randomization of patients who had failed treatment with peginterferon and ribavirin outside this trial (Express patients). Following completion of the 3.5 years of the randomized trial, all patients were invited to continue follow-up without treatment.

All patients were required to have an ultrasound, computed tomography, or magnetic resonance imaging with no evidence of hepatic mass lesions suspicious for HCC. Subjects enrolled in the Lead-in phase were required to have serum α-fetoprotein (AFP) <200 ng/mL, whereas those enrolled as Express patients had to have AFP <1000 ng/mL at the screening visit. Of the 237 Express patients enrolled, only 3 had AFP >200 ng/mL at entry, and the highest AFP value was 315 ng/mL.

All patients had a liver biopsy performed prior to enrollment. For Lead-in patients, the baseline liver biopsy was done at least 2 months following the last course of interferon therapy and within 12 months prior to enrollment. For Express patients, a liver biopsy must have been performed within 18 months prior to randomization, or a baseline liver biopsy must have been performed at least 8 weeks after the end of the prior course of peginterferon and ribavirin and within 24 weeks prior to randomization. Liver biopsies were repeated 1.5 and 3.5 years after randomization. All biopsy specimens were reviewed in conference by a panel of 12 hepatic pathologists, who used the Ishak scoring system to grade inflammation (0-18) and to stage fibrosis (0-6).20 Liver biopsies performed for the diagnosis of HCC and explant livers of patients who underwent liver transplantation for HCC were also submitted for review by the group of hepatic pathologists.

HCC Surveillance and Diagnostic Criteria

Patients were seen every 3 months during the 3.5 years of the randomized trial and every 6 months thereafter. Patients who stopped treatment continued to be followed in the study unless consent was withdrawn or a liver transplantation had occurred. A complete blood count, a liver panel, and AFP level were obtained at the local clinical center at each visit. Des-γ-carboxy prothrombin (DCP) levels at enrollment were tested in batches with an enzyme immunoassay (Eisai Company, Tokyo, Japan) in a central laboratory at the University of Michigan; the results were not reported to the investigators at the clinical sites. Lead-in patients underwent an ultrasound examination of the liver 20 weeks and 12 months after enrollment and then every 6-12 months. Express patients underwent ultrasound examinations 6 months after enrollment and then every 6-12 months. Patients with an elevated or rising AFP and those with new lesions on ultrasound were evaluated further with a computed tomography or magnetic resonance imaging. Diagnostic liver biopsy and HCC treatment were conducted at the discretion of investigators at each site. All patients underwent an endoscopy at the time of randomization and after 3.5 years (ie, at the end of the randomized trial). For the current analysis, patients were categorized as not having varices (absent) or as having varices, regardless of their size.

Two definitions of HCC were adopted: one for "definite" HCC and one for "presumed" HCC. Definite HCC was defined by histologic confirmation or a new mass lesion on imaging with AFP levels increasing to >1000 ng/mL. Presumed HCC was defined as a new mass lesion on ultrasound in the absence of histology and AFP was <1000 ng/mL in conjunction with one of the following characteristics: (1) 2 liver imaging studies showing a mass lesion with characteristics of HCC (vascular enhancement, wash out), (2) progressively enlarging lesion on ultrasound leading to death, or (3) 1 additional imaging study showing a mass lesion with characteristics of HCC that either increased in size over time or was accompanied by increasing AFP levels. Tumors were staged based on the modified United Network of Organ Sharing TNM system.21 Early HCC was defined as tumor stage T1 (single lesion <2 cm in diameter) or T2 (single lesion between 2 and 5 cm or no more than 3 lesions each <3 cm in diameter).

All cases of HCC (presumed and definite) were reviewed by an Outcomes Review Panel composed of rotating panels of 3 investigators. The outcomes review panel did not include any radiologist, and radiographic studies were not reviewed. However, the outcome panel did review serial radiology reports from each HCC case to ascertain that the predefined criteria were met, the date when these criteria were first met, and the tumor staging. In cases in which the outcomes review panel was unable to determine whether the patient had met criteria for HCC, the investigators were requested to provide additional information including follow-up imaging or histology, and the case was re-reviewed. All patients who met criteria for presumed or definite HCC continued to be followed, and results of subsequent imaging or histology (including liver explant and surgically resected specimen) were submitted for review. An earlier report on all trial outcomes (clinical and histologic) included randomized patients with definite HCC (incident and prevalent) identified during the first 3.5 years after randomization (up to April 30, 2007).19 The current analysis includes all randomized patients in the HALT-C Trial who had a diagnosis of definite or presumed HCC as judged by the Outcomes Review Panel prior to October 15, 2007.

Statistical Analyses

Statistical analyses were performed at the Data Coordinating Center with SAS release 9.1 (SAS Institute, Cary, NC). Cumulative incidence of HCC was determined by Kaplan-Meier analysis, and differences were compared with the log-rank test. Relation of baseline factors to risk of HCC was assessed by t test, χ2 test, or univariate Cox proportional hazards regression. Several multivariate Cox proportional hazards models to estimate the risk of HCC were tested. Variables with a P value < .1 on univariate analysis and variables reported previously to be associated with HCC (eg, gender; race/ethnicity; presence of cirrhosis, alcohol consumption, and smoking) were entered into a multivariate analysis (model I). Treatment assignment was also included because the primary goal of the HALT-C Trial was to determine whether maintenance therapy with peginterferon would prevent clinical outcomes, including HCC, in patients with chronic hepatitis C and advanced fibrosis or cirrhosis. Alternative multivariate analyses that included AFP and DCP separately and in combination as well as aspartate aminotransferase (AST) and platelet count separately and combined as AST-platelet ratio index (APRI)22 were tested. Variables not readily available in clinical practice were deleted from model I to create models with greater clinical utility (models II, III, and IV). The cohort was divided into 3 risk groups based on the resulting individual estimates of model IV. The cumulative incidence of HCC among these groups was determined by Kaplan-Meier analysis and compared with the log-rank test. A 2-sided significance level of 5% was used for all analyses.

HCC Rates of Not Reduced by Maintenance Therapy in HALT-C Study

HCC Rates of Not Reduced by Maintenance Therapy in HALT-C Study


"Incidence Rates and Risk Factors Associated with Hepatocellular Carcinoma (HCC) in Patients with Advanced Liver Disease Due to Hepatitis C: Results of the HALTC Trial"



Incidence Rates and Risk Factors Associated with Hepatocellular ...A.S. Lok presented this data at EASL. ... Data regarding HCC imcidence rates and risk factors among patients with HCV-related cirrhosis in the USA are scant ...
www.natap.org/2008/EASL/EASL_51.htm


STUDY AIMS

To determine in the HALT-C Cohort
--the incidence of HCC
--the effect of maintenance interferon therapy on HCC development
--the risk factors associated with HCC development


Cumulative 5 yr incidence of HCC among the HALT-C cohort

-- in the interferon-treated patients and controls was similar: 5.4% vs 5.0% (p=0.78)

-- In patients with cirrhosis was higher than those with bridging fibrosis: 7.0% vs 4.1% (p=0.08)

19% (9/48) of patients with HCC did not show cirrhosis either on baseline or on follow-up biopsies or explant, indicating that the absence of cirrhosis in these patients was not likely to be due to sampling error.

HCC development was significantly associated with older age; lower BMI; platelets and albumn; higher AST, ALT, alkaline phosphate, AFP and DCP; and the presence of esophageal varices.

AUTHOR CONCLUSIONS

Maintenance therapy with low dose peginterferon is ineffective in reducing the development of HCC in patients with advanced liver disease due to HCV.

Modeling of risks can identify patients at highest risk of HCC.

Additional studies are required to confirm our finding of a high risk of HCC in patients with chronic hepatitis C and bridging fibrosis to determine if HCC surveillance is warranted in these patients.

long-term treatment with interferon-based regimens slowed progression of fibrosis

"Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial."



RESULTS:
"maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort."
"The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis."

"The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls."





Incidence of Hepatocellular Carcinoma and Associated Risk Factors in Hepatitis C-Related Advanced Liver Disease



Gastroenterology Jan 2009

Anna S. LokCorresponding Author Informationemail address, Leonard B. Seeff‡, Timothy R. Morgan§, Adrian M. di Bisceglie∥, Richard K. Sterling¶, Teresa M. Curto#, Gregory T. Everson, Karen L. Lindsay‡‡, William M. Lee§§, Herbert L. Bonkovsky∥∥, Jules L. Dienstag¶¶, Marc G. Ghany##, Chihiro Morishima, Zachary D. Goodman‡‡‡, HALT-C Trial Group

Received 27 May 2008; accepted 11 September 2008. published online 19 September 2008.



Background & Aims



Although the incidence of hepatocellular carcinoma (HCC) is increasing in the United States, data from large prospective studies are limited. We evaluated the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) cohort for the incidence of HCC and associated risk factors.



Methods



Hepatitis C virus-positive patients with bridging fibrosis or cirrhosis who did not respond to peginterferon and ribavirin were randomized to groups that were given maintenance peginterferon for 3.5 years or no treatment. HCC incidence was determined by Kaplan–Meier analysis, and baseline factors associated with HCC were analyzed by Cox regression.



Results



1,005 patients (mean age, 50.2 years; 71% male; 72% white race) were studied; 59% had bridging fibrosis, and 41% had cirrhosis. During a median follow-up of 4.6 years (maximum, 6.7 years), HCC developed in 48 patients (4.8%). The cumulative 5-year HCC incidence was similar for peginterferon-treated patients and controls, 5.4% vs 5.0%, respectively (P = .78), and was higher among patients with cirrhosis than those with bridging fibrosis, 7.0% vs 4.1%, respectively (P = .08). HCC developed in 8 (17%) patients whose serial biopsy specimens showed only fibrosis. A multivariate analysis model comprising older age, black race, lower platelet count, higher alkaline phosphatase, esophageal varices, and smoking was developed to predict the risk of HCC.



Conclusions



We found that maintenance peginterferon did not reduce the incidence of HCC in the HALT-C cohort. Baseline clinical and laboratory features predicted risk for HCC. Additional studies are required to confirm our finding of HCC in patients with chronic hepatitis C and bridging fibrosis.





Chronic hepatitis C is a potentially progressive condition, characterized by slowly evolving fibrosis that can culminate in cirrhosis.1, 2 Over time, compensated cirrhosis can advance to decompensated cirrhosis terminating in hepatic failure and death. A common cause for liver-related death among persons with chronic hepatitis C is hepatocellular carcinoma (HCC), occurring predominantly in those with cirrhosis,3, 4, 5 but reports have appeared of HCC occurring, although far less commonly, in persons with bridging fibrosis without definite cirrhosis.6, 7, 8, 9, 10 In studies from Asia, the annual incidence of HCC in subjects with hepatitis C-related cirrhosis is reported to be 4%–10%, whereas in European studies, the reported annual incidence is 0.5%–5% in comparable subjects.11, 12, 13 Data on the incidence of HCC in subjects with advanced hepatitis C followed prospectively in the United States are limited. Furthermore, the factors that predict a high likelihood of future HCC in such subjects are not clearly defined.



Data from several previous studies, mostly from Japan, suggested that long-term treatment with interferon-based regimens slowed progression of fibrosis and reduced the occurrence of HCC in persons with chronic hepatitis C virus (HCV) infection.14, 15, 16, 17 To determine whether these results apply also in the United States, we undertook the Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis (HALT-C) Trial. A total of 1050 persons with chronic hepatitis C and advanced fibrosis or cirrhosis who were previous nonresponders to antiviral therapy were randomized to receive either a reduced dose of pegylated interferon (peginterferon) or no treatment for 3.5 years.18 Because HCC was one of the planned trial outcomes, the HALT-C trial provided an opportunity to determine in a prospective controlled manner whether long-term peginterferon treatment reduced the incidence of HCC and to identify risk factors associated with the development of HCC in a US cohort with chronic hepatitis C and advanced fibrosis or cirrhosis.



Discussion



The major objective of the HALT-C Trial was to determine whether long-term treatment with peginterferon would reduce progression of hepatitis C, including the development of HCC. Support for the possibility that maintenance therapy might inhibit disease progression had come from a 2-year study of interferon maintenance therapy in a US population.23 Additionally, several early reports suggested that interferon therapy reduced the development of HCC in Japanese and Italian patients with chronic hepatitis C.14, 15, 16, 17 Following initiation of the HALT-C Trial in 2000, numerous additional studies were published that addressed the issue of antiviral treatment and development of HCC, leading Craxi and Camma to perform a meta-analysis in 2005 of the then existing studies.24 The analysis included 20 trials, most of which were not randomized studies and 19 of which showed (13 significantly) that antiviral therapy reduced the rate of development of HCC. The validity of the meta-analysis, however, was limited by the marked heterogeneity among the studies. The authors concluded that, although interferon treatment appeared to have a beneficial impact on reducing the development of HCC, the effect was slight and was most evident among the relatively few patients with cirrhosis who achieved a sustained virologic response. More recent studies from Japan and Taiwan, however, have continued to support a positive impact of antiviral therapy on the occurrence of HCC,25, 26, 27, 28 but patients in these trials were not randomized, and, again, the benefit was observed mainly in patients who had achieved a sustained virologic response.



In the present study, we found no significant difference in the incidence of HCC between patients randomized to receive peginterferon maintenance therapy (23/495; 4.7%) and those randomized to no treatment (25/510; 4.9%). The cumulative incidence of HCC 5 years after randomization in the 2 cohorts was almost identical (treated, 5.4% and controls, 5.0%). These data, collected prospectively from a large number of subjects followed closely for a median of 4.6 years, suggest strongly that maintenance therapy with peginterferon, 90 μg/week, does not reduce the incidence of HCC in patients with chronic hepatitis C and advanced fibrosis or cirrhosis.



Because treatment had no effect on the rate of development of HCC in this trial, data from the 2 cohorts (treated and untreated) could be combined to provide information on the annual incidence of HCC among patients with advanced HCV-related liver disease in the United States followed prospectively. Thus, for the combined cohort, the 3-year incidence of HCC from study entry was 1.9%, increasing to 5.9% at 5 years, yielding an annual incidence of HCC of 1.1%. This figure is consistent with but at the lower range of previous estimates of the annual incidence of HCC both in the United States,29 now increasing,30 and in Europe3, 4, 12 but clearly lower than the rates reported among HCV-infected persons in Japan.31, 32, 33



The most likely explanation for the disparity between this trial and others in the efficacy of interferon-based therapy in reducing the incidence of HCC is the difference in study design between the HALT-C Trial and many of the previous studies. The HALT-C Trial was a large, well-powered, prospective, randomized, multicenter trial, with clear, predefined criteria for the diagnosis of HCC. Many prior studies of interferon treatment were relatively small, nonrandomized, and/or retrospective, with all the potential pitfalls of such observations. Furthermore, in several previous reports, the decreased incidence of HCC was observed mainly in patients who achieved a sustained virologic response, subjects who were excluded from the HALT-C Trial. Potentially, the HALT-C Trial might have demonstrated efficacy of antiviral therapy in preventing HCC had the maintenance treatment consisted of full-dose peginterferon and ribavirin instead of half-dose peginterferon only. However, it is unlikely that patients with advanced fibrosis or cirrhosis could tolerate full-dose combination therapy for 3.5 years; only 59% of patients in the HALT-C Trial randomized to receive half-dose peginterferon therapy were able to tolerate this lower dose for the entire 3.5 years. Finally, the lower incidence of HCC observed in the HALT-C Trial compared with previous studies could be related to differences in radiologic and histologic diagnostic criteria for HCC. In the current study, cases of HCC identified within 12 months of enrollment were excluded because the tumors may have been present prior to entry into the trial. Although 23% (11/48) of the HCC cases in our trial were categorized as presumed, rather than definite, HCC, almost all of these cases fulfilled currently accepted diagnostic criteria for HCC.34, 35, 36 Indeed, the diagnostic criteria utilized in this study were very stringent, and the outcome was the same whether we included presumed cases or not.



The rate of HCC was higher in those whose baseline liver biopsy displayed cirrhosis than in those with bridging fibrosis (6.1% vs 3.9%, respectively; P = .08). Although HCC has been reported rarely in patients with chronic hepatitis C in the absence of cirrhosis, the high frequency of HCC among subjects without cirrhosis in the present study was unexpected.6, 7, 8, 9, 37 One possible explanation is understaging of the baseline liver biopsy resulting from sampling error; however, 8 of the 23 patients with HCC who had no cirrhosis at enrollment also had no cirrhosis on follow-up liver biopsy (all 8 patients) or on explant (1 patient). Alternatively, hepatic fibrosis may have progressed between the time of the last study biopsy and the subsequent diagnosis of HCC because the interval between the last biopsy with no cirrhosis and HCC diagnosis varied from 0 to 34 months (median, 16). Whatever the explanation for our observation, the development of HCC in persons with noncirrhotic fibrosis raises the question of whether such patients should undergo HCC surveillance as is recommended for persons with established cirrhosis.33



In seeking to identify features that might predict the future development of HCC, we undertook an extensive analysis of baseline demographic characteristics; metabolic and viral factors; laboratory values (including HCC biomarkers); disease severity; histologic findings; markers of portal hypertension; treatment assignment; physical activity; and history of alcohol use, smoking, and hormone use in women. On univariate analysis, treatment group assignment was not associated with risk of HCC, and we could not confirm associations reported by others between diabetes or alcohol use and the development of HCC.38, 39, 40, 41, 42, 43 In the current study, subjects with HCC had a lower body mass index, not only at enrollment, but also, based on historical information, at ages 20 and 40 years, than subjects in whom HCC did not develop. Prior reports of an association between obesity and HCC may have included patients with nonalcoholic fatty liver disease.44, 45, 46 In this study, a strong association was observed between the risk of HCC and a low platelet count, a high AST level, and the presence of esophageal varices. This finding is consistent with an association in several studies from Asia between a low platelet count and the development of HCC47, 48 and an association in a study from Italy between the presence of esophageal varices and HCC.49



Numerous efforts have been made to develop predictive models for the development of HCC in persons with advanced chronic liver disease.50, 51, 52, 53 In multivariate analyses, these investigators have identified varying factors, including age, gender, platelet count, prothrombin time, albumin level, and AFP as predictive variables. In the current study, we found that gender, cirrhosis, and AFP were not significant in all models tested, whereas DCP was. Smoking was not a significant factor on univariate analysis but was significant in several regression models tested and showed a trend in our final model. Several simplified models that included factors easily accessible in routine clinical settings were developed to aid in predicting the risk of HCC among patients with hepatitis C and advanced fibrosis or cirrhosis. These models may be useful for selecting high-risk patients with bridging fibrosis and cirrhosis for HCC surveillance.



In conclusion, in the current analysis of hepatitis C patients with advanced fibrosis enrolled in the HALT-C Trial, 3.5 years of maintenance therapy with peginterferon did not reduce the incidence of HCC. The 1% annual incidence of HCC in this US cohort was lower than those reported in studies of similar patients in Japan and Europe. Modeling based on clinical and laboratory features can identify patients at higher risk who might be candidates for more intense HCC surveillance. Additional studies are required to confirm the occurrence of HCC among patients with bridging fibrosis but no cirrhosis and to determine whether such patients should undergo HCC surveillance.



Results



A total of 1050 patients were randomized between January 2001 and August 2004. Forty-five patients were excluded for the following reasons: (1) prevalent HCC, defined as HCC diagnosed within 12 months after enrollment (n = 5); (2) patients who had less than 12 months follow-up after enrollment (n = 38); and (3) patients with a diagnosis of presumed HCC who were followed for at least 24 months and did not show radiologic or clinical progression of their liver masses (n = 2). The characteristics of the remaining 1005 patients included in this analysis are shown in Table 1. The mean age of the patients was 50.2 years, 71% were men, and 72% were of white ethnicity. Baseline biopsy specimens showed bridging fibrosis (Ishak score 3 or 4) in 59% and cirrhosis (Ishak score 5 or 6) in the remaining 41% of patients. The vast majority of the patients (93%) had genotype 1 infection.



Incidence of HCC



During a median follow-up of 4.6 years after randomization (maximum of 6.7 years), 48 (4.8%) patients met criteria for HCC, 37 (77%) for definite HCC, and 11 (23%) for presumed HCC (Table 2). Thirty-four (71%) of the 48 patients with HCC were identified during the randomized phase and the remaining 14 during the extended follow-up period. Thirty-six (75%) patients had early stage (T1/T2) HCC at the time of diagnosis. Twenty-six (54.2%) patients were eligible for potentially curative treatment: 3 had surgical resection, 17 underwent liver transplantation, and 6 had radiofrequency ablation ± transarterial chemoembolization (TACE). An additional 11 patients had TACE. Thus, in total, 37 (77.1%) patients had tumors diagnosed at an early enough stage that allowed them to receive effective therapies.



HCC developed in a similar proportion of peginterferon-treated patients, 4.5% (23/495), and controls, 4.9% (25/510). The estimated annual incidence of HCC was 1.1% in peginterferon-treated patients and 1.0% in controls. The cumulative incidence of HCC 3 and 5 years after randomization was 1.9% and 5.4% for peginterferon-treated patients and 1.9% and 5.0% for controls, respectively (P = .78) (Figure 1A). The incidence of HCC in the peginterferon-treated and control groups was similar regardless of the inclusion or exclusion of cases of presumed HCC.



HCC was diagnosed in a higher percent of patients with cirrhosis, 6.1% (25/408), than in those with bridging fibrosis, 3.9% (23/597). The estimated annual incidence of HCC was 0.8% among patients with bridging fibrosis and 1.4% among those with cirrhosis. The cumulative incidence of HCC 3 and 5 years after randomization was 2.6% and 7.0% for patients with cirrhosis and 1.4% and 4.1% for those with bridging fibrosis, respectively (P = .08) (Figure 1B).



HCC in Patients Without Cirrhosis at Enrollment



Of the 23 patients with HCC and an Ishak fibrosis score of 3 or 4 at enrollment, 13 advanced to a score of 5 or 6, 6 remained with a score of 3 or 4, and 2 had a score of 2 on subsequent biopsies, liver explants, or surgically resected specimen; 2 patients did not have repeat biopsies (Table 3). Thus, 17% (8/48) of the patients in whom HCC developed did not have cirrhosis on at least 2 biopsies more than 2 years apart. HCC was diagnosed after a median of 46.5 months (range, 26–72) from enrollment in these 8 patients. At enrollment, 6 of these 8 patients had platelet counts >150,000/mm3, and none had esophageal varices; however, at the time of the diagnosis of HCC, only 1 patient had a platelet count >150,000/mm3 (Table 4).



Risk Factors for HCC



Baseline clinical and laboratory characteristics of the patients with and without HCC are shown in Table 1. Based on univariate analysis, older age; lower body mass index; lower white cell and platelet counts; a lower albumin level; higher levels of alkaline phosphatase, AST, alanine aminotransferase (ALT), AFP, and DCP; and the presence of esophageal varices were significantly associated with HCC (each P < .05). Black race, higher international normalized ratio of prothrombin time (INR), and histologic evidence of cirrhosis showed a trend toward significance (P < .1), whereas gender, diabetes, alcohol consumption, smoking, treatment assignment, HCV genotype, and bilirubin were not significantly associated with HCC.



Models to Predict HCC Incidence



Several Cox proportional hazards models were developed to predict the risk of HCC (Table 5). The initial model (model I), which included significant variables from our univariate analysis or from previously published analyses, was modified to eliminate variables that were not significant on multivariate analysis. Three simplified models (models II–IV) are shown because they include variables that are available in most patients with chronic hepatitis C. The final model (model IV) included age in years (continuous), black race (black = 1, nonblack = 0), alkaline phosphatase in units per liter (continuous), the presence of esophageal varices (present = 1; absent = 0), smoking (ever smoked = 1, never smoked = 0), and platelet count × 1000/mm3 (continuous). The regression formula for model IV is as follows: age*0.049 + black race*0.712 + alkaline phosphatase*0.006 + esophageal varices*0.777 + ever smoked*0.749 + platelets*−0.011. A score was calculated for each subject included in this study, and cases were ranked in order of their calculated scores in this model. Three cut points (0 to log10 [1.50], log10 [1.50] to log10 [3.25], and >log10 [3.25]) were selected to reflect low, intermediate, and high risk, as distinguished by 5-year cumulative incidence of HCC of <1%, 1%–5%, and >5%. The cumulative incidence of HCC 3 and 5 years after randomization was 0% and 0.4% for patients in the low-risk group, 1.5% and 4.2% for patients in the intermediate-risk group, and 6.1% and 17.8% for patients in the high-risk group, respectively (Figure 2).



Low platelet count and presence of esophageal varices were highly significant in the models tested. The impact of esophageal varices and platelet count on the incidence of HCC is shown in Figure 3A and B. The cumulative incidence of HCC 3 and 5 years after randomization was 5.3% and 13.1%, 2.4% and 5.7%, and 0.6% and 2.6% for patients with baseline platelet count <100, 100–149, and >150 × 1000/mm3, respectively. The estimated annual incidence of HCC for patients in these 3 strata of platelet counts was 2.6%, 1.1%, and 0.5%, respectively.



Patients and Methods



The design of the HALT-C Trial has been described previously.18, 19 Briefly, patients with detectable HCV RNA at 10 clinical centers had to meet the following criteria for enrollment: failure to have achieved a sustained virologic response after previous interferon treatment with or without ribavirin, the presence of advanced hepatic fibrosis on liver biopsy (Ishak fibrosis score, ≥3), no history of hepatic decompensation or HCC, and the absence of defined exclusion criteria (eg, liver disease other than hepatitis C, uncontrolled medical or psychiatric conditions, or contraindications to use of interferon or ribavirin).



Patients were treated with peginterferon alfa-2a (180 μg weekly) and ribavirin (1–1.2 g daily) for at least 24 weeks (“Lead-in” phase). Patients who had detectable serum HCV RNA at treatment week 20 (protocol definition of nonresponse) were randomized at week 24 to maintenance therapy (peginterferon alfa-2a, 90 μg weekly) or to no treatment for the next 3.5 years. Patients with undetectable serum HCV RNA at week 20 were considered responders and continued combination therapy for a total of 48 weeks. Patients in the responder group in whom HCV RNA became detectable later during treatment (breakthrough) or after stopping treatment (relapse) were offered enrollment into the randomized trial. After approval of peginterferons, the protocol was amended to allow randomization of patients who had failed treatment with peginterferon and ribavirin outside this trial (Express patients). Following completion of the 3.5 years of the randomized trial, all patients were invited to continue follow-up without treatment.



All patients were required to have an ultrasound, computed tomography, or magnetic resonance imaging with no evidence of hepatic mass lesions suspicious for HCC. Subjects enrolled in the Lead-in phase were required to have serum α-fetoprotein (AFP) <200 ng/mL, whereas those enrolled as Express patients had to have AFP <1000 ng/mL at the screening visit. Of the 237 Express patients enrolled, only 3 had AFP >200 ng/mL at entry, and the highest AFP value was 315 ng/mL.



All patients had a liver biopsy performed prior to enrollment. For Lead-in patients, the baseline liver biopsy was done at least 2 months following the last course of interferon therapy and within 12 months prior to enrollment. For Express patients, a liver biopsy must have been performed within 18 months prior to randomization, or a baseline liver biopsy must have been performed at least 8 weeks after the end of the prior course of peginterferon and ribavirin and within 24 weeks prior to randomization. Liver biopsies were repeated 1.5 and 3.5 years after randomization. All biopsy specimens were reviewed in conference by a panel of 12 hepatic pathologists, who used the Ishak scoring system to grade inflammation (0–18) and to stage fibrosis (0–6).20 Liver biopsies performed for the diagnosis of HCC and explant livers of patients who underwent liver transplantation for HCC were also submitted for review by the group of hepatic pathologists.



HCC Surveillance and Diagnostic Criteria



Patients were seen every 3 months during the 3.5 years of the randomized trial and every 6 months thereafter. Patients who stopped treatment continued to be followed in the study unless consent was withdrawn or a liver transplantation had occurred. A complete blood count, a liver panel, and AFP level were obtained at the local clinical center at each visit. Des-γ-carboxy prothrombin (DCP) levels at enrollment were tested in batches with an enzyme immunoassay (Eisai Company, Tokyo, Japan) in a central laboratory at the University of Michigan; the results were not reported to the investigators at the clinical sites. Lead-in patients underwent an ultrasound examination of the liver 20 weeks and 12 months after enrollment and then every 6–12 months. Express patients underwent ultrasound examinations 6 months after enrollment and then every 6–12 months. Patients with an elevated or rising AFP and those with new lesions on ultrasound were evaluated further with a computed tomography or magnetic resonance imaging. Diagnostic liver biopsy and HCC treatment were conducted at the discretion of investigators at each site. All patients underwent an endoscopy at the time of randomization and after 3.5 years (ie, at the end of the randomized trial). For the current analysis, patients were categorized as not having varices (absent) or as having varices, regardless of their size.



Two definitions of HCC were adopted: one for “definite” HCC and one for “presumed” HCC. Definite HCC was defined by histologic confirmation or a new mass lesion on imaging with AFP levels increasing to >1000 ng/mL. Presumed HCC was defined as a new mass lesion on ultrasound in the absence of histology and AFP was <1000 ng/mL in conjunction with one of the following characteristics: (1) 2 liver imaging studies showing a mass lesion with characteristics of HCC (vascular enhancement, wash out), (2) progressively enlarging lesion on ultrasound leading to death, or (3) 1 additional imaging study showing a mass lesion with characteristics of HCC that either increased in size over time or was accompanied by increasing AFP levels. Tumors were staged based on the modified United Network of Organ Sharing TNM system.21 Early HCC was defined as tumor stage T1 (single lesion <2 cm in diameter) or T2 (single lesion between 2 and 5 cm or no more than 3 lesions each <3 cm in diameter).



All cases of HCC (presumed and definite) were reviewed by an Outcomes Review Panel composed of rotating panels of 3 investigators. The outcomes review panel did not include any radiologist, and radiographic studies were not reviewed. However, the outcome panel did review serial radiology reports from each HCC case to ascertain that the predefined criteria were met, the date when these criteria were first met, and the tumor staging. In cases in which the outcomes review panel was unable to determine whether the patient had met criteria for HCC, the investigators were requested to provide additional information including follow-up imaging or histology, and the case was re-reviewed. All patients who met criteria for presumed or definite HCC continued to be followed, and results of subsequent imaging or histology (including liver explant and surgically resected specimen) were submitted for review. An earlier report on all trial outcomes (clinical and histologic) included randomized patients with definite HCC (incident and prevalent) identified during the first 3.5 years after randomization (up to April 30, 2007).19 The current analysis includes all randomized patients in the HALT-C Trial who had a diagnosis of definite or presumed HCC as judged by the Outcomes Review Panel prior to October 15, 2007.



Statistical Analyses



Statistical analyses were performed at the Data Coordinating Center with SAS release 9.1 (SAS Institute, Cary, NC). Cumulative incidence of HCC was determined by Kaplan-Meier analysis, and differences were compared with the log-rank test. Relation of baseline factors to risk of HCC was assessed by t test, χ2 test, or univariate Cox proportional hazards regression. Several multivariate Cox proportional hazards models to estimate the risk of HCC were tested. Variables with a P value < .1 on univariate analysis and variables reported previously to be associated with HCC (eg, gender; race/ethnicity; presence of cirrhosis, alcohol consumption, and smoking) were entered into a multivariate analysis (model I). Treatment assignment was also included because the primary goal of the HALT-C Trial was to determine whether maintenance therapy with peginterferon would prevent clinical outcomes, including HCC, in patients with chronic hepatitis C and advanced fibrosis or cirrhosis. Alternative multivariate analyses that included AFP and DCP separately and in combination as well as aspartate aminotransferase (AST) and platelet count separately and combined as AST-platelet ratio index (APRI)22 were tested. Variables not readily available in clinical practice were deleted from model I to create models with greater clinical utility (models II, III, and IV). The cohort was divided into 3 risk groups based on the resulting individual estimates of model IV. The cumulative incidence of HCC among these groups was determined by Kaplan–Meier analysis and compared with the log-rank test. A 2-sided significance level of 5% was used for all analyses.