Serum fibrosis markers can predict rapid fibrosis progression after liver transplantation for hepatitis C

Authors


  • See Editorial on Page 1237

  • This work was presented in abstract form at the World Transplant Congress, Boston, MA, July 2006.

Abstract

Although recurrent hepatitis C virus (HCV) after liver transplantation (LT) is universal, a minority of patients will develop cirrhosis within 5 years of surgery, which places them at risk for allograft failure. This retrospective study investigated whether 2 serum fibrosis markers, serum hyaluronic acid (HA) and YKL-40, could be used to predict rapid fibrosis progression (RFP) post-LT. These markers were compared with conventional laboratory tests, histological assessment, and hepatic stellate cell activity (HSCA), a key step in fibrogenesis, as assessed by immunohistochemical staining for alpha-smooth muscle actin. Serum and protocol liver biopsy samples were obtained from 46 LT recipients at means of 5 ± 2 (biopsy 1) and 39 ± 6 (biopsy 2) months post-LT, respectively. RFP was defined as an increase in the fibrosis score ≥ 2 from biopsy 1 to biopsy 2 (a mean interval of 33 ± 6 months). The ability of parameters at biopsy 1 to predict RFP was compared with the areas under receiver operating characteristic curves (AUROCs). Of the 46 subjects, 15 developed RFP. Serum HA and YKL-40 performed significantly better than conventional parameters and HSCA in predicting RFP post-LT for HCV at biopsy 1, with AUROCs of 0.89 and 0.92, respectively. The accuracy of serum HA ≥ 90 μg/L and YKL-40 ≥ 200 μg/L in predicting RFP at biopsy 1 was 80% and 96%, respectively. In conclusion, we found that elevated levels of serum HA and YKL-40 within the first 6 months after LT accurately predicted RFP. Larger studies evaluating the role of serum HA and YKL-40 in post-LT management are warranted. Liver Transpl 14:1294–1302, 2008. © 2008 AASLD.

End-stage liver disease from chronic hepatitis C virus (HCV) infection is the leading indication for liver transplantation (LT) in the United States. Although virologic recurrence of HCV is universal, the histologic and clinical course of HCV after LT is highly variable.1 Progression to cirrhosis occurs in 6% to 23% of patients at a median of 3 to 4 years post-LT.1 Several donor, recipient, viral, and transplant-related factors have been associated with more severe recurrent HCV disease after transplantation.2, 3 Efforts to identify individuals at risk for rapid fibrosis progression (RFP) have included histological assessment of allograft biopsies, measurement of liver stiffness by transient elastography (TEG), and measurement of the hepatic venous pressure gradient (HVPG).4–9

A key step in fibrogenesis is the activation of hepatic stellate cells into myofibroblast-like cells that express alpha-smooth muscle actin (α-SMA) and secrete excess matrix proteins into the extracellular matrix (ECM).10 In 2 studies, hepatic stellate cell activity (HSCA), as assessed by the expression of α-SMA in allograft biopsies at 4 months, was found to be a marker for progressive fibrosis in LT recipients with HCV.11, 12 Recently, Cisneros et al.13 reported that levels of HSCA and hepatic transforming growth factor-beta1 (TGF-β1) expression correlated with fibrosis stage and disease progression in patients who had undergone LT.

It has also been suggested that the levels of markers of ECM metabolism reflect the rate of fibrosis progression. Hyaluronic acid (HA) has been shown to be one of the best performing markers for the noninvasive assessment of hepatic fibrosis.14 Serum YKL-40, also known as human cartilage glycoprotein 39 or Chondrex, showed a strong direct linear correlation with the rate of progressive fibrosis in individuals coinfected with HCV and schistosomiasis.15 Furthermore, in the same study, serum YKL-40 levels paralleled serum TGF-β levels as well as the changes in both YKL-40 and TGF-β liver tissue messenger RNA expression. Hepatic messenger RNA expression of YKL-40 confirmed the liver as the likely source of increased serum YKL-40 levels in this setting. We hypothesized that HA and YKL-40 levels could predict progression of liver fibrosis by reflecting the activity of ongoing hepatic fibrogenesis.

Because the natural history of recurrent HCV is accelerated in some LT recipients, the post-LT setting is a unique environment for investigating the utility of serum markers of ECM metabolism in predicting RFP. In this retrospective study, we investigated whether serum HA and YKL-40 could predict short- to medium-term progression of fibrosis after LT in comparison with conventional laboratory tests, standard liver histology, and α-SMA expression for HSCA.

Abbreviations

α-SMA, alpha-smooth muscle actin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUROC, area under the receiver operating characteristic curve; BMI, body mass index; CI, confidence interval; CMV, cytomegalovirus; ECM, extracellular matrix; HA, hyaluronic acid; HAI, histologic activity index; HCV, hepatitis C virus; HSCA, hepatic stellate cell activity; HVPG, hepatic venous pressure gradient; IU, international units; LT, liver transplantation; RFP, rapid fibrosis progression; ROC, receiver operating characteristic; SD, standard deviation; TEG, transient elastography; TGF-β, transforming growth factor-beta.

PATIENTS AND METHODS

Subjects

This was a retrospective cohort study approved as a minimal risk protocol by the Mayo Clinic Institutional Review Board. Any HCV-infected individual undergoing LT at Mayo Clinic Jacksonville between March 1998 and December 2001 with a detectable serum HCV RNA level after LT, biopsies and sera available at 4 months and at least 24 months following transplantation, and at least 3 years of clinical follow-up were considered for the study. Exclusion criteria included retransplantation during this time interval as well as evidence of other causes of liver disease, including alcohol abuse and coinfection with hepatitis B virus or human immunodeficiency virus.

Clinical Parameters

The electronic medical records of the study subjects provided the clinical information and laboratory results. Recipient demographic data (age, gender, ethnicity, and body mass index at LT), donor age at LT, LT-related parameters (cold and warm ischemia times and duration of surgery), and donor and recipient cytomegalovirus status, along with pre-LT HCV characteristics (genotype and viral load), were collected. After transplantation, the immunosuppressive therapy, number of allograft rejection episodes, and requirement for intravenous methylprednisone boluses were recorded, along with standard laboratory results, HCV viral levels, allograft biopsy results, and anti-HCV treatment.

Pathological Analyses

All allograft biopsy specimens were stained with hematoxylin and eosin and retrospectively reviewed by a single pathologist blinded to the subject's identification, clinical characteristics, and interval from LT. Each specimen was graded for the histologic activity index (HAI; score 0-18) and staged for the degree of fibrosis (score 0-6) according to Ishak et al.16 In addition, the length and number of portal tracts in each specimen were recorded. RFP was defined as an increase in the fibrosis score by 2 or more from the first biopsy to the second biopsy.

Paraffin-embedded allograft biopsy sections of 4-μm thickness were stained for α-SMA (1A4 clone, prediluted; DakoCytomation, Inc., Carpinteria, CA) with the Dako LSAB2 system according to the manufacturer's recommendations. Antigen retrieval was performed in steam heat for 20 minutes. The tissue was counterstained with hematoxylin. Appropriate positive and negative controls were used. The specimens were reviewed by another pathologist blinded to the subject's identification, clinical characteristics, interval from LT, and conventional histopathology findings. Each specimen was scored for HSCA in the portal tract and fibrous septa (score 0-3) according to the previously described semiquantitative scoring system as follows: 0 = no staining, 1 = less than 10% of mesenchymal cells stained, 2 = 10%-50% of mesenchymal cells stained, and 3 = more than 50% of mesenchymal cells stained.12 Representative images are shown in Fig. 1.

Figure 1.

Representative images of immunohistochemistry staining for alpha-smooth muscle actin expression to determine hepatic stellate cell activity: (A) grade 0 staining, (B) grade 1 staining, (C) grade 2 staining, and (D) grade 3 staining (magnification, ×100).

Noninvasive Serum Markers for Fibrosis

Serum YKL-40 and HA concentrations were measured in stored sera obtained from all subjects within 7 days of the date of liver biopsy. Quantification was performed by enzyme-linked immunosorbent assay following the manufacturer's instructions for both YKL-40 (Chondrex, Metra Biosystems, Inc., Mountainview, CA) and HA (Corgenix, Inc., Utah).

Statistical Methods

Subjects were categorized into those with and without RFP. Statistical analyses were performed with Statistical Package for Social Sciences software, version 13.0 (SPSS, Chicago, IL). Continuous variables were summarized with means, standard deviations, medians, and ranges, whereas categorical variables were presented with proportions. An independent Student t test was used to assess the significance of differences between groups for continuous variables. Comparisons of categorical variables between groups were analyzed with Fisher's exact test. Spearman's rank correlation coefficients were calculated to test the association between continuous or ordinal variables. The ability of each parameter to predict RFP was assessed by a comparison of the areas under receiver operating characteristic curves (AUROCs). An AUROC of 1.0 is characteristic of an ideal test, whereas 0.5 indicates a test of no diagnostic value. A pairwise comparison of 2 receiver operating characteristic (ROC) curves was performed by the computation of the variances and covariance of nonparametric AUROC estimates derived from the same sample of cases.17 All differences and associations were considered significant at a 2-sided P value of <0.05.

RESULTS

Between March 1998 and December 2001, 193 patients underwent LT for hepatitis C, of which 46 subjects met entry criteria for inclusion in the study. One hundred forty-seven patients were excluded because there were inadequate follow-up data (59), missing sera or liver biopsy specimens (27), retransplantation (30), death (29), significant alcohol use (1), or concomitant hepatitis B infection (1).

Of 29 patients who died within 3 years after LT, none died from recurrent HCV infection. Malignancy was the cause of death in 7 patients (recurrent hepatocellular carcinoma in 5 and lymphoma and pancreatic cancer in 1 patient each). The remaining 18 patients died from early and late postoperative complications: 11 patients from sepsis, 3 patients from bleeding complications, and 1 patient each from portopulmonary hypertension, pulmonary embolism, cerebrovascular accident, and cardiac arrest. Four patients were lost to follow-up, and the causes of death were unknown.

Of the 30 patients who underwent retransplantation within 3 years of the initial LT, HCV-related graft failure was the indication in 33%: 7 patients from fibrosing cholestatic hepatitis C and 3 patients from recurrent HCV infection. The remaining 20 patients underwent retransplantation for other indications: 7 patients for primary nonfunction of the liver graft, 3 patients for hepatic artery thrombosis, 3 for ischemic cholangiopathy, 4 patients for venous outflow obstruction, 1 patient for rapid progression of de novo autoimmune hepatitis, and 2 patients for chronic rejection.

Table 1 shows that there were no significant differences in recipient demographics, donor age at LT, or LT-related parameters between the 46 subjects included in the study and the remaining 147 excluded patients. There was comparable graft survival between the 46 study subjects and the 86 patients excluded from the study because of inadequate follow-up duration (less than 3 years after LT) and missing sera or liver biopsy specimens according to Kaplan-Meier analysis with a log-rank test (P = 0.4).

Table 1. Comparison of Recipient Demographics and LT-Related Parameters Between Subjects Included in the Study and Patients Excluded from the Study
 Study Subjects (n = 46)Excluded Patients (n = 147)P Value
  • Abbreviations: BMI, body mass index; LT, liver transplantation; SD, standard deviation.

  • *

    Independent Student t test.

  • Fisher's exact test.

Recipient demographics   
 Mean age at LT (years ± SD)49 ± 750 ± 90.5*
 Male gender67%74%0.5
 Caucasian78%75%0.7
 Mean BMI at LT (± SD)28.8 ± 7.128.5 ± 6.00.8*
LT-related parameters   
 Donor age > 60 years old22%20%0.8
 Mean cold ischemia time (hours ± SD)7.5 ± 2.27.7 ± 2.30.6*
 Mean warm ischemia time (minutes ± SD)37 ± 1237 ± 111.0*
 Mean operation duration (minutes ± SD)264 ± 57261 ± 740.8*

Of the 46 subjects included in this study, 15 fulfilled our definition of RFP. The first and second biopsies were performed at means of 5 ± 2 months (a range of 3-8 months) and 39 ± 6 months (a range of 29-49 months) after LT, respectively. The mean elapsed time interval between the 2 biopsies was 33 ± 6 months (a range of 25-40 months). Overall, the mean length of biopsy specimens was 2.4 ± 0.9 cm, with a range of 0.7 to 5.4 cm; 10% of the specimens were less than 1.5 cm long. The mean number of portal tracts per specimen was 13.1 ± 7.8.

Clinical Parameters

No significant differences were noted between those with and without RFP in recipient demographics and clinical characteristics before LT (Table 2). A higher proportion of subjects with RFP received grafts from donors older than 60 years old in comparison with those without RFP (47% versus 10%, P = 0.01). LT-related parameters were otherwise similar between groups (Table 2).

Table 2. Comparison of Recipient Demographics, LT-Related Parameters, and Hepatitis C Viral Parameters Between Subjects With and Without Rapid Fibrosis Progression
 Rapid Fibrosis Progression (n = 15)Slow Fibrosis Progression (n = 31)P Value
  • Abbreviations: BMI, body mass index; CMV, cytomegalovirus; IU, international units; LT, liver transplantation; SD, standard deviation.

  • *

    Independent Student t test.

  • Fisher's exact test.

Recipient demographics   
 Mean recipient age at LT (years) ± SD49 ± 749 ± 81.0*
 Male gender53%71%0.3
 Caucasian73%83%0.7
 Mean BMI at LT ± SD29.7 ± 9.028.6 ± 6.00.7*
LT-related parameters   
 Donor age more than 60 years old47%10%0.01
 Mean cold ischemia time (hours) ± SD7.8 ± 2.47.4 ± 2.20.7*
 Mean warm ischemia time (minutes) ± SD38 ± 1437 ± 110.6*
 Mean duration of operation (minutes) ± SD265 ± 70266 ± 471.0*
 Donor/recipient CMV mismatch13%16%1.0
Post-LT parameters   
 Tacrolimus94%87%0.6
 Cyclosporine6%13%0.6
 Sirolimus3%7%1.0
 Mycophenolate mofetil73%77%1.0
 Allograft rejection before the first biopsy20%16%1.0
 Allograft rejection between biopsies0%0%1.0
Hepatitis C viral parameters   
 Genotype 175%86%0.6
 Mean viral load at first biopsy (log IU) ± SD6.38 ± 0.566.15 ± 0.860.4*
 Mean viral load at second biopsy (log IU) ± SD6.07 ± 0.496.10 ± 0.740.9*
 Undetectable viral load at second biopsy20%32%0.5

No significant difference in immunosuppression was found between groups (Table 2). The number of allograft rejection episodes was comparable, and this resulted in a similar total dosage of intravenous methylprednisone for each group. No subject developed an allograft rejection episode requiring intravenous steroid boluses between the first and second biopsies.

No subject in either group received anti-HCV treatment before the first biopsy, whereas 73% of those with RFP were treated with interferon-based therapy between the first and second biopsies versus 61% of those without RFP (P = 0.52). No significant differences were noted between the groups in the HCV viral load at the first and second biopsies or in the proportion of subjects who had a virological response to treatment (Table 2). The HCV viral load became undetectable at the second biopsy in 11 of the 30 subjects (37%) who received anti-HCV treatment.

Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were not significantly different between subjects with and without RFP at the time of the first allograft biopsy (Table 3). However, the serum aminotransferases were significantly higher in those with RFP when they were measured at the second biopsy (Table 3). Eight subjects (17%) achieved normalization of aminotransferases at the second biopsy. Of these, 7 received anti-HCV treatment, and 5 subjects did not have RFP. No significant associations were found between normalization of aminotransferases and anti-HCV treatment, undetectable HCV viral load at the second biopsy, or rate of fibrosis progression.

Table 3. Comparison of Liver Biopsy Characteristics Between Subjects With and Without Rapid Fibrosis Progression
Parameter (Mean ± SD)Rapid Fibrosis Progression (n = 15)Slow Fibrosis Progression (n = 31)P Value
  • Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; HA, hyaluronic acid; HAI, histologic activity index; HSCA, hepatic stellate cell activity; SD, standard deviation.

  • *

    Independent Student t test.

At first allograft biopsy   
 Serum ALT (U/L)112 ± 77102 ± 840.7*
 Serum AST (U/L)130 ± 12379 ± 590.2*
 Biopsy length (cm)2.0 ± 0.82.3 ± 0.90.2*
 HAI3.7 ± 2.33.4 ± 2.50.7*
 Fibrosis score0.6 ± 0.71.1 ± 1.20.1*
 HSCA score1.1 ± 0.90.7 ± 0.90.2*
 Serum HA (μg/L)367 ± 36371 ± 600.007*
 Serum YKL-40 (μg/L)711 ± 542101 ± 490.001*
At second allograft biopsy   
 Serum ALT (U/L)152 ± 15257 ± 390.03*
 Serum AST (U/L)182 ± 23349 ± 420.045*
 Biopsy length (cm)2.7 ± 1.22.5 ± 0.80.4*
 HAI4.8 ± 1.92.6 ± 1.90.001*
 Fibrosis score4.1 ± 1.51.1 ± 1.1<0.001*
 HSCA score1.5 ± 1.40.5 ± 0.60.02*
 Serum HA (μg/L)167 ± 150113 ± 1640.3*
 Serum YKL-40 (μg/L)319 ± 191225 ± 2620.2*

Pathological Analyses

The histological activity index, fibrosis stage, and mean HSCA score at the first biopsy were similar between the 2 groups (Table 3). The HAI was significantly correlated with ALT and AST, as demonstrated by Spearman's rank correlation coefficients of 0.50 and 0.50 at the first biopsy. The HSCA score was significantly correlated with the histological grade and stage, ALT, and AST at the first biopsy, with Spearman's correlation coefficients of 0.62, 0.38, 0.43, and 0.49, respectively (all P ≤ 0.01).

The mean HAI and fibrosis score at the second biopsy were significantly higher for those with RFP than for those without RFP (both P < 0.001; Table 3). The HAI continued to show a significant correlation with ALT and AST, with Spearman's correlation coefficients of 0.43 and 0.41, respectively (both P < 0.001). Figure 2 illustrates the fibrosis scores for each patient at biopsies 1 and 2.

Figure 2.

Fibrosis scores for each patient at first and second liver biopsies.

At the second biopsy, the mean HSCA was significantly higher in subjects who developed RFP than in those without RFP (P = 0.02). However, only histological stage was significantly correlated with the HSCA score, with a Spearman's correlation coefficient of 0.48 (P = 0.001). At that point, HSCA score did not have any significant correlation with histological grade, ALT, or AST (Spearman's correlation coefficients of 0.24, 0.14, and 0.15, respectively; all P > 0.05).

Noninvasive Serum Markers for Fibrosis

Serum concentrations of both HA and YKL-40 were significantly higher at the time of the first biopsy in those who developed RFP in comparison with those who did not (Table 3). Serum YKL-40 levels, but not HA levels, were significantly higher in recipients of donors > 60 years old at the time of the first allograft biopsy (Table 4). There was no significant difference in levels of either marker with regard to donor age by the time of the second biopsy.

Table 4. Comparison of Serum HA and YKL-40 Levels in Liver Transplant Recipients Who Received Organs from Donors Younger and Older than 60 Years of Age at the Time of the First and Second Allograft Biopsies
Serum Concentration (Mean ± SD)Donor Age < 60 Years Old (n = 36)Donor Age > 60 Years Old (n = 10)P Value
  • Abbreviations: HA, hyaluronic acid; SD, standard deviation.

  • *

    Independent Student t test.

At first allograft biopsy   
 Serum HA (μg/L)139 ± 230276 ± 3130.2*
 Serum YKL-40 (μg/L)180 ± 209737 ± 6670.027*
At second allograft biopsy   
 Serum HA (μg/L)125 ± 163149 ± 1560.7*
 Serum YKL-40 (μg/L)228 ± 258343 ± 1670.2*

The ability of each parameter measured at the time of first biopsy to predict RFP in the second biopsy was analyzed with ROC curves with the corresponding AUROC (Table 5). Serum HA and YKL-40 were comparable and better at predicting RFP than conventional histological and biochemical markers as well as HSCA (all P < 0.05). In addition, HSCA along with conventional histological and biochemical markers were not predictive of RFP (all P > 0.05).

Table 5. Comparison of Receiver Operating Characteristic Curves with Corresponding AUROCs Between Parameters
Parameter at First BiopsyAUROC (95% Confidence Interval)
  1. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUROC, area under the receiver operating characteristic curve; HA, hyaluronic acid; HAI, histologic activity index; HSCA, hepatic stellate cell activity.

HAI0.56 (0.38-0.74)
Fibrosis score0.42 (0.24-0.59)
HSCA score0.63 (0.45-0.81)
Serum ALT (U/L)0.59 (0.40-0.83)
Serum AST (U/L)0.64 (0.46-0.83)
Serum HA (μg/L)0.89 (0.80-0.98)
Serum YKL-40 (μg/L)0.92 (0.79-1.00)

With the ROC curves, different serum levels of HA and YKL-40 were evaluated for optimal predictive ability. A serum HA concentration of ≥90 μg/L at initial biopsy had a sensitivity and specificity of 80%, a positive predictive value of 67%, a negative predictive value of 89%, and an accuracy of 80% in predicting subjects who would develop RFP by the second biopsy. A serum YKL-40 concentration of ≥200 μg/L measured at the first biopsy had a sensitivity and specificity of 87% and 100%, respectively, with positive and negative predictive values of 100% and 87%, and an accuracy of 96% in predicting the subsequent development of RFP.

Subgroup analyses were performed separately on the 30 and 16 subjects who did and did not receive anti-HCV treatment, respectively. The ability of serum HA and YKL-40 measured at the first biopsy to predict RFP was comparable for both groups with AUROCs of 0.90 [95% confidence interval (CI) of 0.79-1.00] and 0.91 (95% CI of 0.73-1.00) for serum HA and 0.89 (95% CI of 0.72-1.00) and 1.00 for serum YKL-40.

Further subgroup analyses were performed by the exclusion of 13 patients who developed an undetectable viral load at the second biopsy and 8 patients who achieved normalization of aminotransferases. In 33 patients who remained viremic at the second biopsy, the ability of serum HA and YKL-40 measured at the first biopsy to predict RFP was comparable to the ability for the whole cohort with AUROCs of 0.87 (95% CI of 0.74-0.99) and 0.92 (95% CI of 0.76-1.00), respectively. Similarly, the ability of serum HA and YKL-40 to predict RFP in 38 patients who had abnormal aminotransferases at the second biopsy was comparable to the ability for the whole cohort with AUROCs of 0.87 (95% CI of 0.75-0.98) and 0.98 (95% CI of 0.95-1.00), respectively.

The correlation between each of the measured parameters was evaluated. Of note, regardless of the timing of the biopsy, the serum levels of both markers were significantly correlated, with a Spearman's rank correlation coefficient of 0.36 (P = 0.001). At the first biopsy, serum YKL-40 levels, but not HA levels, were significantly correlated with HSCA, with a Spearman's rank correlation coefficient of 0.33 (P = 0.03). At the second biopsy, there was no correlation between levels of either serum marker and HSCA.

Long-Term Follow-Up

The mean length of clinical follow-up for the 46 study subjects was 70 ± 17 months. Of the 15 subjects identified as having RFP, all showed evidence of progressive liver disease. Four subjects died, 3 subjects underwent retransplantation, 2 subjects developed decompensated cirrhosis, 3 subjects progressed to cirrhosis on protocol liver biopsy without clinical decompensation, and 3 subjects progressed to advanced fibrosis (Ishak's stage 3-4) at last follow-up. Of the 31 subjects without RFP, only 4 subjects had defined progression of their liver disease (3 subjects progressed to stage 3-4 fibrosis and 1 subject developed compensated cirrhosis). One subject died from septicemia and renal failure, and another underwent retransplantation for hepatic artery thrombosis. The other 25 subjects were stable, with protocol liver biopsies confirming no to moderate fibrosis (stage 0-2) at last follow-up.

DISCUSSION

This is the first study to demonstrate that 2 serum markers of ECM metabolism, serum HA and YKL-40, measured within the first 8 months after LT could identify patients with short- to medium-term risk of hepatic fibrosis progression following LT for chronic HCV infection. The importance of our findings is supported by the longer term follow-up of the study cohort. Those subjects who were defined as having RFP developed important clinical endpoints: death, hepatic decompensation, and cirrhosis of the liver allograft. Previous studies have shown that histological features such as increased inflammatory infiltrate as well as staining for HSCA were predictive of early disease progression.4, 5, 8, 11, 12 However, in the current study, we found that these markers showed only a weak correlation with the 2 serum markers of ECM metabolism and were relatively poor predictors of the subsequent development of hepatic fibrosis.

At the outset of this study, we hypothesized that serum levels of HA and YKL-40 might reflect not only the stage of liver disease but also the activity of ECM metabolism, so that those individuals with the highest levels of these markers would be at greatest risk of disease progression. Consistent with this hypothesis, Kamal et al.15 showed that high levels of YKL-40 were strongly correlated with the development of progressive fibrosis in a cohort of subjects infected with HCV and schistosomiasis.

There is also indirect evidence from other studies that the levels of HA and YKL-40 reflect both the rate of ECM deposition or metabolism and the stage of fibrosis. These markers are often very high in active inflammatory states of the liver, such as alcoholic hepatitis, and are often more elevated than in patients with less active cirrhosis.18 Similarly, in some studies, their levels have been shown to fall with successful interferon therapy prior to any expected reduction in the hepatic fibrosis stage.19–22 Although we did not find high levels of inflammatory activity in the first biopsy specimens, we did find that the markers were highly predictive of early fibrosis progression. However, the liver allograft is a unique environment, and direct comparisons with other states may not be appropriate. Kamal et al.15 found that serum TGF-β increased in parallel with the severity of liver damage and fibrosis progression in patients coinfected with HCV and schistosomiasis. Other investigators have emphasized the importance of HSCA as a marker of disease recurrence.11, 12 In contrast, Cisneros et al.13 recently concluded that mechanisms other than overactivation of hepatic stellate cells and an overexpression of TGF-β were probably involved in the accelerated hepatic fibrosis in LT patients with HCV recurrence. There are factors in the early postoperative period that have been associated with disease progression. A prime example of this is the association between older donor allografts and RFP, which has been observed by our group and others.23–25 We found that serum YKL-40 levels were significantly higher at the time of the first biopsy in recipients of organs from donors > 60 years of age. Thus, we speculate that our findings suggest a specific role for YKL-40 and possibly HA in reflecting the activity of new ECM turnover or deposition in the liver following transplantation, rather than simply confirming the presence of fibrosis or quantity of liver ECM.

An unexpected finding of our study was that although we observed higher serum levels of HA and YKL-40 at the time of the first liver biopsy, the levels of these markers appeared to fall by the time of the second biopsy. One explanation is that the immediate postoperative period is a complex time during which the insult to the liver is multifactorial, being related to HCV graft infection, graft perfusion, and biliary function as well as a host of immunological effects. Hence, the levels of these markers may be reflecting a more severe immediate injury and wound response that is not observed at the time of the second liver biopsy. Despite this, both early fibrosis progression and the ability of these markers to predict this appear to be highly predictive of subsequent disease progression as reflected in the longer term clinical outcomes.

Previous investigators have reported that the degree of hepatic inflammation predicts progressive recurrent disease.4, 5, 8 However, the site of inflammatory activity (lobular versus portal) and timing of index biopsy (3 versus 12 months post-LT) differed in these studies. In our cohort, there was some degree of necroinflammation in both groups at the initial biopsy, but we did not find that the HAI per se was predictive of RFP. In the study by Blasco et al.,7 mild acute hepatitis was the most common finding at the 3-month biopsy post-LT, but there was no reported association between the extent of inflammatory activity and subsequent development of liver fibrosis.

Activation of hepatic stellate cells, as assessed by α-SMA expression, has been shown to be a marker for progressive fibrosis in patients undergoing LT for HCV.11, 12 However, it was recently shown that hepatic α-SMA expression was similar in LT patients with recurrent HCV and HCV patients who did not undergo LT, and this suggests that overactivation of hepatic stellate cells is not the only explanation for accelerated hepatic fibrosis post-LT.13 In our study, we found that there was a reasonable correlation between the HSCA score at the first biopsy and RFP, with a c-statistic of 0.63. We suspect that the α-SMA stain did not perform as well as the serum markers in part because of sampling error as well as the imprecise nature of grading immunohistological slides. Gawrieh et al.12 emphasized the importance of distinguishing parenchymal α-SMA expression (zones 1, 2, and 3) from mesenchymal α-SMA expression (portal tracts and fibrous septa). They speculated that the hepatic stellate cell/myofibroblast population was heterogeneous within the liver. We did not assess α-SMA expression in that manner, but our results support the previous observations that in some individuals, HSCA can predict RFP.

Serum markers of fibrosis form part of an expanding armamentarium in evaluating the state of the liver allograft. Like TEG, they are truly noninvasive and can be repeated frequently, with minimal patient discomfort. In a study of 169 paired liver biopsies and TEG measurements, TEG was shown to diagnose moderate or higher stages of liver allograft fibrosis accurately.6 There was a direct correlation between liver stiffness as assessed by TEG and HVPG. Blasco et al.7 demonstrated a significant positive association between HVPG and fibrosis stage. They found that an HVPG of 6 mm Hg or greater measured 12 months after LT was more accurate than the fibrosis stage at predicting clinical decompensation. Our preliminary data suggest that serum HA and YKL-40 may have a role even earlier in recurrent disease stratification than these other tests. Although their true prognostic potential remains to be defined, in our study, all of the 15 patients identified as having RFP by HA and YKL-40 died, underwent retransplantation, or had at least stage 3 fibrosis, with or without hepatic decompensation on longer term follow-up. This compares with 1 death, 1 patient requiring retransplantation, and 6 other patients progressing to at least stage 3 fibrosis in the 31 subjects without RFP. Furthermore, anti-HCV treatment between the 2 biopsies did not have any impact on the ability of serum HA and YKL-40 to predict RFP. We anticipate that tests other than liver biopsy will provide important additional insights into the risk of progressive disease and long-term outcome in patients transplanted for HCV. This information can then be used to determine who may be a suitable candidate for anti-HCV therapy, given the challenging nature of this treatment.26–32

Clearly, this study has a number of limitations, the largest of which is the ability to generalize the results to a larger cohort of HCV individuals undergoing LT. Problems include the small sample size and the relatively large number of subjects excluded from the study. However, we showed that the demographic and clinical characteristics of the excluded subjects were very similar to those who were included in the sample subset. Furthermore, we purposefully excluded individuals who had rapid disease progression and in whom the presence of rapidly progressive liver disease was manifested clinically. Indeed, exclusion of those with more severe disease may have had the effect of reducing the strength of our findings. Further criticisms of studies that use needle liver biopsy as the gold standard are the issues of sampling and interpreter error.33–36 However, the large number of adverse outcomes reported in those patients with RFP strongly suggests that the needle biopsy results did indeed identify the majority of individuals at risk of disease progression and adverse clinical outcome. We are therefore satisfied that despite the small sample size and the limitations of needle biopsy, the methods used in this study were valid.

In summary, we found that serum levels of HA and YKL-40 could accurately predict those subjects at risk for RFP after LT for chronic HCV infection, performing significantly better than conventional histology assessments of inflammation and fibrosis, standard biochemical testing, or assessment of HSCA. Serum markers of fibrosis should be evaluated further in LT recipients with HCV.

Ancillary