Progression of liver fibrosis among injection drug users with chronic hepatitis C


  • Potential conflict of interest: Nothing to report.


Although most hepatitis C virus (HCV) infections are acquired by injection drug use, prospective data on the progression of liver fibrosis are sparse. Baseline liver biopsies were obtained (1996–1998) on a random sample of 210 out of 1667 HCV-positive injection drug users (IDUs). Subjects were followed biannually, with a second biopsy offered to those eligible. Paired biopsies were scored 0 to 6 (modified Ishak score), significant fibrosis was defined as score 3 or greater, and progression of fibrosis was defined as an increase 2 or more units or clinical evidence of end-stage liver disease. Predictive values of blood markers [FibroSURE, aspartate aminotransferase-to-platelet-ratio index (APRI) and alanine aminotransferase (ALT)] were assessed for detection of contemporaneous and future liver fibrosis. Among 119 prospectively followed IDUs, 96% were African American; 97% HCV genotype 1a/b; 27% HIV-infected, and median age was 42 years. Most (90.7%) did not have significant liver fibrosis at first biopsy. Although predictive value for detecting insignificant fibrosis at first biopsy was greater than 95% for FibroSURE, APRI, and ALT, specificities were 88.9%, 72.7%, and 72.7%, respectively. After 4.2 years median follow-up, 21% had progression of fibrosis, which was significantly associated with serum level of HCV RNA and ALT. No serological test had predictive value greater than 40% for contemporaneous or future significant fibrosis. Even initial biopsy result had only a 30.4% value for predicting future significant fibrosis. In conclusion, significant liver fibrosis and progression were detected in some, but not most, IDUs in this cohort. In this setting with low fibrosis prevalence, FibroSURE, ALT, and APRI tests predict insignificant fibrosis; however, further work is needed to find noninvasive markers of significant liver fibrosis. (HEPATOLOGY 2006;43:788–795.)

Injection drug use is the chief source of hepatitis C virus (HCV) infection in Western nations, and most injection drug users (IDUs) have HCV infection.1, 2 The natural history of HCV infection in IDUs must be understood, not just because they constitute the majority of HCV-infected persons, but also because of emerging data regarding treatment. HCV infection can be eradicated in approximately half of persons with currently available medications.3–5 Conversely, treatment of HCV infection is complex, associated with adverse reactions, and usually requires 48 weeks of weekly interferon alfa injections and daily oral ribavirin. Consequently, HCV treatment is most advisable for persons at highest risk of fibrosis progression,3, 6 a guideline that becomes even more important in populations such as IDUs in which there is a high incidence of other (competing) medical issues and substantial barriers to traditional medical care.7, 8

Despite the importance of understanding the natural history of liver disease in IDUs and of identifying those in need of treatment, remarkably little published information is available on either topic. One reason is that IDUs do not commonly receive regular care for HCV infection in academic centers, where most natural history studies have been conducted. In addition, liver biopsy is the only widely accepted marker of the stages of liver fibrosis that precede liver failure (end-stage liver disease), and IDUs uncommonly undergo liver biopsy and can be difficult to longitudinally monitor.9

We have overcome these challenges. Initially, the incidence of end-stage liver disease, the final clinical expression of HCV infection, was characterized in 1,667 HCV-infected IDUs.9 To further evaluate the magnitude of clinically silent liver disease, we performed a liver biopsy on randomly selected cohort members between 1996 and 1998.10 The chief purposes of the current investigation are: (1) to assess prospectively the rate of the progression of liver fibrosis among HCV-infected IDUs by repeating liver biopsies 3 to 5 years later and (2) to identify persons at risk for progression (in need of treatment) by using noninvasive testing.


HCV, hepatitis C virus; IDU, intravenous drug user; ALT, alanine aminotransferase; MHAI, modified hepatic activity index; AST, aspartate aminotransferase; ROC, receiver operating curve; APRI, aspartate aminotransferase-to-platelet-ratio.

Materials and Methods

Population and Study Design

Initial Biopsy Group.

Between 1988 and 1989, 2,946 persons who acknowledged illicit injection drug use within 10 years were enrolled in a natural history study, as previously described.11 All participants were more than 17 years of age, free of acquired immunodeficiency syndrome, and consented to semiannual follow-up that included administration of a questionnaire and blood collection. After serological testing described later, all participants with at least 1 year of follow-up and HCV antibodies at enrollment (n = 1,625) or during follow-up (n = 42) were considered members of the HCV-infected IDU cohort.9

Between May 1996 and June 1998, a cross-sectional study was performed to determine the severity and correlates of clinically silent liver disease among members of the HCV-infected IDU cohort.10 Two hundred ten subjects were randomly selected from eligible participants after stratification for serum alanine aminotransferase (ALT) level, as previously described.10

Candidates were offered enrollment in the biopsy substudy by a single study coordinator and examined by a hepatologist. A questionnaire was administered a median of 8 days before each biopsy to assess medical information and drug/alcohol-using behavior. Subjects were excluded from the initial biopsy study if they already had clinical evidence of end-stage liver disease (defined as ascites, encephalopathy, unexplained prothrombin time ≥14 seconds or esophageal varices) or had a serious non–liver-related medical condition, as previously described.10 Subjects were not excluded if they had hepatitis B; however, HBV DNA was only detected in one subject using the Roche COBAS PCR (Roche Molecular Systems, Basel, Switzerland).12

Prospective Biopsy Group.

After the initial biopsy, subjects were followed every 6 months. Exposures were reevaluated by questionnaire, blood was drawn, and evidence of liver-related disease or treatment was assessed by physical examination and review of medical records for subjects who responded affirmatively to questions designed to detect these outcomes. Repeat liver biopsies were offered between March 2001 and May 2003 to all who were still eligible. Three persons who already had cirrhosis at the initial biopsy were excluded because a second biopsy was not clinically indicated, as were 25 subjects in whom HCV RNA was not detected in blood. Thus, a total of 182 persons were considered for the second biopsy protocol.

Repeat transcutaneous liver biopsy was performed and processed using the same protocol as the initial biopsy except specimens were obtained under ultrasound guidance using a Monopty core biopsy device with an 18-gauge needle. Slides from the two biopsies were read concurrently by a single, experienced hepatopathologist, who graded the degree of inflammation and the stage of fibrosis by using the Ishak modified hepatic activity index (MHAI) and Ishak fibrosis index13 as well as the METAVIR system.14 Hepatic fat (steatosis) was assessed using a semiquantitative method based on the average percentage of fat (0, 1–30, 31–60, >60%) on hematoxylin-eosin section. The hepatopathologist was blinded to clinical markers other than HCV antibody status. The hepatopathologist judged the adequacy of each specimen with regard to the study aim of assessing the progression of fibrosis, and persons whose specimens were inadequate were excluded from analysis. The median length of biopsy specimens was 11 mm.

Laboratory Methods

HCV antibody, HCV RNA, HCV genotype, HIV antibody, and liver enzyme testing were done as described elsewhere.11, 15–17

FibroSURE testing was done on serum specimens stored at −70°C and not thawed until testing, which was done by Laboratory Corporation of America (Burlington, NC). FibroSURE is based on a patented algorithm that incorporates results of six individual serum biochemicals, including alpha-2 macroglobulin, haptoglobulin, apoplipoprotein A-1, total bilirubin, gamma glutamyl transferase, and ALT. Fibrosis and necroinflammatory activity scores are calculated as individual numeric values between zero and 1.0, with higher numbers indicating a greater amount of fibrosis and/or necroinflammatory activity, as previously published under the name FibroTest (Biopredictive, Paris, France).18, 19

Statistical Analysis

Fibrosis progression was defined categorically as a change in Ishak fibrosis score of 2 or greater or development of end-stage liver disease because a change of 2 or more Ishak units was judged to exceed sampling error20 and thus be clinically significant. A baseline fibrosis progression rate was calculated by dividing the fibrosis score on the first biopsy by the duration of HCV infection, estimated from the time from first injection drug use. (The validity of this estimation was previously assessed in this cohort.2, 21) In addition, an observed fibrosis progression rate was calculated by subtracting the fibrosis score of the first biopsy from the second and dividing by the years between.

To identify persons who had progression of fibrosis, chi-square, Mantel Haenszel, and Fisher's exact testing were used to compare categorical data, and Wilcoxon rank sum was used to compare continuous data non-parametrically. Age was categorized into the highest quartile and below.

Because the distributions of the three markers of liver inflammation were highly concordant, total MHAI inflammatory scores were analyzed. Total MHAI inflammatory scores were classified as high if they were 5 or greater, whereas the remainder were classified as low. Because no convention has been established, this grouping was based solely on the distribution of the scores in this dataset.

For analyses of the predictive value of FibroSURE and aspartate aminotransferase-to-platelet-ratio index (APRI) test results of contemporaneous and future significant fibrosis, liver fibrosis was defined as significant if the Ishak fibrosis score was 3 or higher, according to published guidelines for HCV treatment.3, 6 Because FibroSURE results are reported typically by using the METAVIR system, all biopsy specimens were also scored by that system in which significance is defined as 2 or more units; there was 100% concordance in classification of significant fibrosis by METAVIR and Ishak systems. Because all analyses of predictive values, sensitivity and specificity were based on detection of significant fibrosis, Ishak data are shown for consistency with all other analyses. The comparison of FibroSURE-predicted fibrosis results with what was actually observed was accomplished by using a METAVIR–Ishak conversion provided by the manufacturer.


This research was approved by the institutional review boards of the Johns Hopkins School of Medicine (liver biopsy/fibrosis marker studies) and Bloomberg School of Public Health (IDU cohort, HCV testing).


Prospective Study Population

Between March 2001 and May 2003, liver biopsies were performed on 121 (66.5%) of the 182 eligible persons. The reasons that biopsies were not obtained on others were as follows: lost to follow-up (n = 25), refused (n = 5), protocol violation (n = 5), died of non–liver-related causes (n = 14), and serious non–liver-related medical condition (n = 9), and developed end-stage liver disease before the second biopsy was offered (n = 3). Though not eligible for repeat biopsy, the latter three were classified as “progressors” in this analysis. Of the 121 with a second biopsy, five had inadequate biopsy sample size for evaluation. Thus, 116 with paired biopsies were analyzed along with three subjects who developed end-stage liver disease after their initial biopsy (total n = 119).

The baseline (first biopsy) characteristics of the 119 persons prospectively evaluated for progression of fibrosis are shown in Table 1. The median age was 41.8 years (range, 27.6–58.3 years), and the estimated duration of HCV infection (time from first injection use) was a median of 21.8 years (range, 8.2–35 years). Ninety-six percent of participants were African American, 82% were male, 27% were HIV infected, 97% were HCV genotype 1a or 1b, 4% were hepatitis B surface antigen or HBV DNA positive, whereas 54% reported injection drug use and 58% reported alcohol use in the 6-month period before the first biopsy. The 119 persons prospectively evaluated for progression of fibrosis were younger but did not significantly differ from the 63 eligible subjects who were not studied in terms of race, gender, or estimated duration of HCV infection.

Table 1. Baseline Characteristics of Eligible Study Subjects
 Study group (N = 119)Not studied (N = 63)
  • Abbreviations: IQR, interquartile range; ALT, alanine aminotransferase; HCV, hepatitis C virus. Data shown represent characteristics at baseline (first biopsy). Study group includes 116 persons with a repeat liver biopsy that was analyzable as well as three persons who developed end-stage liver disease. The “not studied” includes 5 persons whose second biopsy was not analyzable, 14 who died of non–liver-related conditions, 25 persons who were lost to follow-up, 5 who refused repeat biopsy, 5 with study protocol violations, and 9 for whom repeat biopsy was not indicated because of medical comorbidity. Alcohol intake refers to those who reported any alcohol intake within 6 months of the initial biopsy. ALT evaluations represent results of two determinations made over a 6-month period before the initial biopsy.

  • *

    P < .05.

  • Estimated from the time from first injection drug use, which was assessed at enrollment into the cohort in 1988–89.

Median age (years)*41.843.9
Sex (% male)82.477.8
Race (% Black)95.896.8
HIV-infected (%)26.938.1
Estimated years of HCV infection21.825.1
Alcohol intake (%)58.170.5
Median ALT IU/L (IQR)39 (28, 67)43 (25, 66)

Study Outcomes

The median interval of follow-up after the initial biopsy was 4.2 years (range, 2.8–6.0 years). Second biopsy fibrosis scores ranged from 0 to 6. Eight (6.7%) had cirrhosis (Ishak modified fibrosis score of 5 or 6, including three subjects with end-stage liver disease), 15 (12.6%) had septate fibrosis (Ishak modified fibrosis score of 3 or 4), and 96 (80.6%) had mild or no detectable fibrosis (Ishak modified fibrosis score of 0–2)(Table 2). Total MHAI inflammatory scores ranged from 1 to 8 (of a total of 18), and 71.6% had a total MHAI inflammatory score of less than 5, which was considered low.

Table 2. Fibrosis Scores at First and Second Liver Biopsy and Among 119 HCV-Infected Persons
inline image

Progression of Fibrosis

Progression of fibrosis, defined as an increase in fibrosis score of at least 2 Ishak points or end-stage liver disease, occurred in 25 (21%) of 119 subjects. When considered as a function of the time between biopsies, the median observed fibrosis progression rate was 0.11 Ishak modified fibrosis unit/year. However, there were marked differences among subjects: range, −0.68 to 1.42. To evaluate whether fibrosis scores at the second biopsy could be predicted by scores at the first biopsy and the estimated duration of HCV infection, a baseline fibrosis progression rate was calculated: median 0.04 Ishak modified fibrosis unit/year (range, 0.00–0.21 units/year). For each person, the median difference in the observed minus the estimated baseline progression rate of fibrosis was 0.00 (range, −0.87–1.34). However, those with the most discordance in observed versus predicted rates, that is, those with differences in progression rates of fibrosis outside the middle quartiles (those in the lowest and highest quartiles), did not substantially differ in risk factors for progression from those showing progression rate of fibrosis concordance (data not shown).

Correlates of Progression.

Progression of fibrosis was observed more often in persons with higher HCV RNA levels and higher ALT, aspartate aminotransferase (AST), and gamma glutamyltransferase levels at the initial biopsy (Table 3). Progression of fibrosis also was observed more often if interval ALT or AST levels were elevated. For example, progression of fibrosis was observed in 10.5% of those with normal AST levels at each follow-up visit, 21.1% of those whose AST was elevated at less than half of visits, and 29.3% of those whose AST was elevated at more than half of visits (P = .04). However, 28% of those with progression of fibrosis had normal aminotransferases and lower HCV RNA levels at baseline. No associations were detected between progression of fibrosis and gender, age, lifetime alcohol use, alcohol use during study interval, injection drug use frequency or duration, steatosis, or body mass index.

Table 3. Correlates of Progression of Liver Fibrosis Among 119 HCV-Infected Injection Drug Users
CharacteristicProgressors (n = 25)Non-progressors (n = 94)P
  • Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma glutamyl transpeptidase; NS, not significant (P > .10). Liver fibrosis progression was defined as second biopsy fibrosis scores ≥ 2 points more than the initial (n = 22) or development of end-stage liver disease (n = 3) among 119 injection drug users followed a median of 4.2 years. Characteristics were defined at baseline (first biopsy) unless indicated. ALT, AST, and GGT were stratified at 1.5 times the upper limit of normal for the assay.

  • *

    Characteristics were defined using data collected during semiannual interval study visits as follows: alcohol use was defined as any alcohol within 6 months of initial biopsy; injection drug use was defined as any injection drug use within 6 months of initial biopsy; interferon alfa use was defined as any treatment with interferon alfa at initial biopsy; steatosis was defined as any evidence of fatty change on biopsy.

  • Results of FibroSURE and aspartate aminotransferase to platelet ratio index (APRI) contemporaneous with initial biopsy. FibroSURE results were missing for 1 progressor and 3 non-progressors.

 Percent unless indicated 
Univariate analysis   
% Male8482NS
% Black9296.8NS
% Age > 46 years2419.2NS
Alcohol use*5658.7NS
Injection drug use*6850NS
Interferon alfa use*4.63.2NS
Median HCV RNA level (log copies/mL)7.116.63<.04
 ALT ≥ 60 IU/L4424.2.05
 AST ≥ 52 IU/L4827.5.05
 GGT ≥ 140 IU/L4020.9.05
First-biopsy FibroSURE fibrosis (median)0.450.31NS
First-biopsy FibroSURE inflammatory (median)0.150.09NS
First-biopsy APRI (median)0.640.45NS
Baseline total MHAI ≥ 53233NS
Steatosis at baseline*3223.4NS
Body mass index (median)24.224.3NS
Steatosis at second biopsy*27.323.1NS
HIV infected2427.7NS

Persons with progression of fibrosis were not more likely to be HIV infected. However, those with CD4+ lymphocyte counts less than 200/mm3 at or before the first biopsy were excluded from the initial protocol, and there were too few HIV-infected subjects to stratify analyses by use of antiretroviral drugs, HIV RNA level, or CD4+ lymphocyte count.

After multivariate logistic regression, variables significantly associated with progression of fibrosis from univariate analysis remained associated: elevated HCV RNA level (≥7.2 log copies/mL) at baseline (adjusted OR = 3.04, P = .02) and elevated ALT (≥60 IU/mL) at baseline (adjusted OR = 2.70, P = .04). In a subset analysis of patients with biopsies at least 1 cm in length, odds ratios were essentially unchanged (data not shown).

Few subjects received interferon alfa–based treatment for HCV infection. At baseline, 4.6% of progressors versus 3.2% of non-progressors had prior treatment with interferon alfa. During the study interval, only three were treated with interferon alfa (with/without ribavirin), all three with failure to respond to treatment (2 discontinued because of side effects); MHAI fibrosis scores at second biopsy were 1, 4, and 5.

Progression Algorithms

To examine whether significant liver fibrosis could be detected more accurately by algorithms that incorporate multiple noninvasive fibrosis markers, we assessed two recently characterized panels by blood testing near the first biopsy. The FibroSURE testing was done by the commercial laboratory on previously unthawed sera collected a median of 1 month after (interquartile range [IQR], 1.5 months before to 2.8 months after) the initial biopsy for 115 of 119 subjects. The median FibroSURE fibrosis score was 0.32 (IQR, 0.18–0.57), and the median FibroSURE inflammatory score was 0.10 (IQR, 0.05–0.20). For 53 subjects whose FibroSURE result predicted a biopsy score of insignificant fibrosis (<0.31), all but one had insignificant fibrosis (and that one had steatosis) (predictive value for insignificant fibrosis = 98.1%) (Tables 4 and 5). The area under the receiver operating curve (ROC) was 0.74. However, among 37 whose FibroSURE result predicted a biopsy score with significant fibrosis (>0.48), only five had significant fibrosis on biopsy (predictive value for significant fibrosis = 11.9%). Biopsy results that were greater than 1 Ishak fibrosis stage different than FibroSURE prediction were not different (P > .10) from those within one stage with regard to biopsy length, time from serum collection to biopsy, or HIV status (data not shown).

Table 4. Comparison of FibroSURE Score and Actual Fibrosis Score for 115 Individuals Undergoing First Liver Biopsy
FibroSURE ResultsIshak Liver Biopsy Result
  • NOTE. Because FibroSURE is typically characterized by using METAVIR liver biopsy results, biopsies were also scored by this system. There was 100% concordance in scoring “significant” liver disease: all biopsies scored ≥3 by Ishak were scored ≥2 by METAVIR. For this table, predicted fibrosis scores are reported in the Ishak system for consistency with other analyses by using a METAVIR–Ishak conversion provided by the manufacturer.

  • *

    Subject had >50% steatosis.

Table 5. Relative Performance of Blood Screening Test Algorithms for Detection of Contemporaneous Liver Fibrosis
  • NOTE. Liver fibrosis was defined as significant if the score was ≥3 Ishak or ≥2 METAVIR, according to published guidelines for HCV treatment.3, 6 There was 100% concordance in pathologist classification of significant disease by the two systems.

  • PPV, positive predictive value; NPV, negative predictive value; ROC, receiver operator curve.

  • *

    Note that for FibroSURE and aspartate aminotransferase to platelet ratio index (APRI), there are persons with intermediate results between the thresholds for significant and insignificant fibrosis. Thus, their performance is reported for each outcome, whereas for ALT and AST, the results for insignificant fibrosis are the reciprocal of those shown for significant fibrosis: sensitivity of significant fibrosis equals specificity of insignificant fibrosis; PPV of significant fibrosis equals NPV of insignificant fibrosis.

Significant fibrosis     
FibroSURE > 0.4855.664.811.994.40.74
 APRI > 1.518.294.425.091.80.70
 ALT > ULN72.772.921.696.3 
 AST > ULN81.863.618.897.1 
Insignificant fibrosis*     
 FibroSURE < 0.3148.688.998.112.9 
 APRI < 0.558.972.795.415.4 

APRI results were calculated from blood collected within 45 days of the first liver biopsy on all 119 subjects. The median APRI was 0.46 (IQR, 0.32–0.79). For 66 subjects whose APRI result predicted a biopsy score of insignificant fibrosis (<0.5), 63 (95.4%) had insignificant fibrosis (predictive value for insignificant fibrosis = 95.4%). For prediction of significant fibrosis, the specificity of an APRI result greater than 1.5 was 94.4%. However, the sensitivity was only 18.2% at this APRI threshold. The area under the ROC for detection of significant fibrosis was 0.70.

The highest agreement among liver biopsy, FibroSURE, and APRI for classification of liver fibrosis was in excluding significant liver fibrosis, a designation made by all three tests for 66 (57.9%) of subjects. Of 39 additional subjects who did not have significant fibrosis on initial biopsy, 34 were misclassified as having significant fibrosis by FibroSURE only, two by APRI, and three by both algorithms. Of the nine classified as having significant disease on the initial biopsy, one was similarly classified by FibroSURE and APRI, four by FibroSURE only, one by APRI only, and three by neither APRI or FibroSURE.

Because the ultimate goal of liver fibrosis evaluation is to anticipate future disease course, we also evaluated the degree to which FibroSURE and APRI results from near the time of the initial biopsy predicted the second biopsy fibrosis score. Of 53 persons with initial FibroSURE fibrosis scores less than 0.31, 49 (92.5%) had insignificant fibrosis on the second biopsy (Table 6). Of 42 persons who had FibroSURE scores greater than 0.48, 13 (31%) had significant fibrosis on the second biopsy. (Only 5 of the 13 had significant fibrosis on the initial biopsy.) Of 67 persons with initial APRI scores less than 0.5, 59 (88.1%) had insignificant fibrosis on the second biopsy. For predicting significant fibrosis on the second biopsy, even the sensitivity of detection of significant fibrosis on the first biopsy was low (30.4%). The first biopsy, FibroSURE and APRI results were similar with regard to predicting future insignificant liver fibrosis (predictive values of 85.0%, 92.5%, and 88.1%, respectively). In contrast, detection of significant liver fibrosis at a first biopsy is a better predictor of second biopsy significant fibrosis (63.6%) compared with FibroSURE (31%) and APRI (37.5%).

Table 6. Performance of FibroSURE as a Screening Test for Significant Future Fibrosis in a Cohort of 119 Individuals
  • NOTE. Same blood test results as used to assess fibrosis with biopsy 1 were used to assess the fibrosis detected a median of 4.2 years later.

  • *

    Note that for FibroSURE and aspartate aminotransferase (APRI), there are persons with intermediate results between the thresholds for significant and insignificant fibrosis. Thus, their performance is reported for each outcome, whereas for ALT and AST, the results for insignificant fibrosis are the reciprocal of those shown for significant fibrosis: sensitivity of significant fibrosis equals specificity of insignificant; PPV of significant fibrosis equals NPV of insignificant fibrosis.

Significant fibrosis     
FibroSURE >0.4861.969.
APRI > 1.513.094.837.582.00.68
ALT > ULN47.872.929.785.4 
AST > ULN65.265.631.388.7 
Biopsy 1 significant fibrosis30.495.863.685.2 
Insignificant fibrosis*     
FibroSURE <0.3152.180.992.527.4 
APRI < 0.561.565.288.128.8 


Identification of IDUs with serious HCV-related liver disease is one of the most important challenges to reducing HCV-related morbidity and mortality in the United States (and most Western countries). This is true not only because IDUs constitute the majority of infected persons, but also because treatment requires a substantial investment of public health resources that is rarely received by IDUs. Given the future likelihood of more effective, less complicated HCV treatment and of individuals transitioning to less chaotic life cycles, the critical short-term objective must be identification of the few IDUs for whom existing therapy is urgently needed.

With resources provided by the National Institute of Health, we performed serial liver biopsies on a cohort of community-based IDUs. Because this will rarely be feasible outside of a research setting, we also evaluated commercially available blood markers of liver fibrosis. High predictive values (>95%) for detection of insignificant liver fibrosis were achieved by using the FibroSURE, as well as the APRI and liver enzyme measures. If confirmed in other settings, these data suggest that existing blood tests could be used to identify a subset of IDUs for whom liver biopsy could be deferred. However, this high “negative” predictive value was achieved in large part because of the low prevalence of significant fibrosis in our cohort, and these blood tests were not sensitive methods of detecting significant liver fibrosis.

The inability to reliably detect significant fibrosis was especially disappointing given our overall goal of finding the subset of IDUs for whom treatment of HCV infection is most urgent. Whereas the low prevalence of persons with significant liver fibrosis was encouraging, it diminished the positive predictive value of our surrogate markers and reduced the precision of our estimates of tests' performance for screening for significant fibrosis.

The results of FibroSURE and APRI testing differ from what was previously reported.18, 22 Differences in the accuracy of the tests in our cohort could be explained by the low prevalence of significant fibrosis or variation in the distributions of individual components of the FibroSURE or APRI panels caused by injection drug use itself, or perhaps by race. The performance of other serological tests is altered in IDUs in whom there is an increased risk of biological false-positive syphilis tests, hypergammaglobulinemia, and abnormal hepatitis B serologies.23, 24 Additional research is needed to investigate whether modifications in the FibroSURE and APRI algorithms can improve their performance among IDUs. Racial differences may occur in the performance of tests to screen for HCV-related liver fibrosis, just as racial differences occur in spontaneous clearance of HCV infection9, 25 and response to interferon alfa therapy for chronic hepatitis C.26, 27 Slower natural history of HCV-related fibrosis has also been reported in African Americans.28, 29 These data underscore the importance of validating screening tests in a variety of clinical settings.

The principal reason for assessing the current stage of liver fibrosis is to provide information about the likelihood of future progression, and hence the relative urgency of treatment. Thus, our analysis of how well testing between 1996 and 1998 predicted liver fibrosis results a median of 4 years later is particularly clinically relevant. Unfortunately, no test reliably predicted future fibrosis. Even the initial liver biopsy result was only 30% sensitive as a method of screening for future significant liver fibrosis. These findings differ from those reported by Yano and coworkers, in which the initial biopsy was strongly predictive of progression, perhaps because the proportion progressing to cirrhosis was much higher in that study.30

Measurement error in the biopsy itself would diminish its own predictive value, as well as that of the blood “surrogate” markers. This is an especially important concern in this study because the median size of the liver biopsies was only 1.1 cm. Improved detection of significant fibrosis and better correlation of serum markers with liver histology have been reported with larger tissue samples.19, 31 However, no evidence was seen of systematic misclassification. For example, the size of the biopsies that changed by 2 or more points was not different than those with less change. In addition, the proportion of persons who had a 2 unit or more progression of fibrosis was not significantly greater among those who were predicted to have significant fibrosis by FibroSURE but did not on initial biopsy compared with others. The persons with discrepancies between biopsy and FibroSURE also were largely different from those with biopsy–APRI discrepancies. Finally, a subset analysis of correlates of progression of fibrosis among those with biopsies 1 cm in length or longer did not differ from results of all biopsy lengths combined. Overall, these data underscore the importance of additional work to improve methods to detect liver fibrosis among HCV-infected persons.

In this study of HCV-infected IDUs, the rate of fibrosis progression and the association with elevated serum liver transaminase results are similar to what was previously reported by Ghany and coworkers.32 Although it can be difficult to compare between studies because of differences in histological scoring systems, others have also reported relatively low rates of progression, whereas higher rates were noted in yet other settings.19, 33, 34 In our cohort, factors are present that would be expected to reduce the apparent rate of progression and others that might increase it. Examples of the former are the relatively young age of infection and predominance of African Americans.28, 34, 35 Conversely, other factors might have been expected to increase the risk of the progression of fibrosis in this cohort, including predominance of men, low utilization of interferon-alfa anti-HCV treatment, and relatively high rates of HIV infection and alcohol use, compared with other natural history studies.

The cofactors associated with the progression of fibrosis also differ among studies.19, 34–36 However, this difference is not surprising given the number of factors that could contribute to the progression of fibrosis and the difference in their occurrence in various settings. This study exclusively analyzed fibrosis progression among IDUs, the largest group worldwide of HCV-infected individuals and those least likely to undergo histologic evaluation of liver disease severity.

Several attributes of the study design and cohort members might alter the findings. As mentioned previously, if we required all liver biopsy specimens to be larger than 2 cm, the predictive value of noninvasive tests (and the first liver biopsy) would have improved. In addition, we could have altered the algorithms for FibroSURE and APRI to improve their performance in our setting. However, the primary objective of our research was to assess methods for detection of liver fibrosis that would be clinically relevant and reproducible. Consequently, the biopsy protocols corresponded with the standard of care at our institution, and the noninvasive testing was done by commercial laboratories and by published algorithms. Another potential limitation of the study is the possibility of exaggerating fibrosis progression by studying a population with low initial disease prevalence, so-called “deviation to the mean.” In this regard, it is important that our overall rate of progression was not greater than that reported by others.

In summary, these data show that significant liver fibrosis and progression occur in some, but not most, members of this IDU cohort. Although serum markers for fibrosis of ALT, Fibrosure, and APRI can predict insignificant fibrosis, additional research will be needed to improve their utility to detect significant liver fibrosis.