Histopathology and clinical correlates of end-stage hepatitis B cirrhosis: A possible mechanism to explain the response to antiviral therapy

Authors


Abstract

In chronic liver disease associated with histological necroinflammation, clinical severity is frequently greater in those with higher grades of activity. Conventional wisdom assumes that necroinflammation is mild or absent in patients with end-stage hepatitis B virus (HBV) cirrhosis due to the frequent presence of mildly elevated aminotransferase levels, the absence of hepatitis B e antigen (HBeAg), and low or undetectable HBV deoxyribonucleic acid (DNA) levels. However, a histopathologic analysis of such patients has not been undertaken. The aims of this study were 1) to assess severity and histological features of inflammation, 2) to correlate the severity of inflammation with biochemical and virologic parameters, and 3) to define the relationship between inflammation and clinical severity in explanted livers from patients undergoing liver transplantation for HBV cirrhosis. Characteristics of 34 consecutive patients undergoing liver transplantation for HBV cirrhosis were correlated with inflammation and immunohistological findings in the explanted livers. High-grade inflammation (grades 3 and 4) was found in many cases (47.1% interface hepatitis; 14.8% lobular inflammation; and 20.6% portal inflammation). The presence of positive cytoplasmic staining for hepatitis B core antigen (HBcAg) was associated with grade 3 or 4 interface hepatitis (P = .046) and lobular hepatitis (P = .005). There was no correlation between inflammatory activity and age, Asian ethnicity, aminotransferase levels, total bilirubin levels, HBeAg seropositivity, and detectable HBV DNA level. Patients with high-grade inflammation had greater degrees of hepatic decompensation. In conclusion, high-grade inflammation is common in end-stage HBV cirrhosis, but it is not readily detected by biochemical and virologic parameters. High-grade inflammation is associated with a greater degree of hepatic decompensation. (Liver Transpl 2005;11:82–88.)

Biochemical and virologic parameters are useful predictors of the severity of histologic inflammation in patients with chronic hepatitis B virus (HBV). Elevated aminotransferase values reflect increased histologic necroinflammatory disease activity,1–5 and markers of active viral replication, i.e., hepatitis B e antigen (HBeAg) and detectable HBV deoxyribonucleic acid (DNA), are frequently present in patients with active inflammation. Although a pathologic analysis of end-stage HBV cirrhosis has not been reported, necroinflammatory activity is frequently assumed to be mild or absent at this stage due to the presence in many cases of only mildly elevated aminotransferase levels, absence of HBeAg, and low or absent HBV DNA levels.1, 6, 7

A serologic evolution of HBV infection has been proposed in which the presence of HBeAg indicates an early and prolonged phase of active viral replication with high hepatitis B surface antigen (HBsAg) and HBV DNA levels.7–9 With time, many patients spontaneously convert to low viral replication, an event associated with a decrease in HBV DNA levels, loss of HBeAg, development of antibody to hepatitis B e antigen (anti-HBe), and decreased aminotransferase levels.10, 11 Although the presence of HBV DNA and anti-HBe in the absence of HBeAg is usually attributed to viral replication of precore mutant HBV strains, this pattern in Western populations may also be due to the inability to detect low HBeAg levels because of high anti-HBe titers.12 A seroconversion from HBeAg to anti-HBe and improvement in aminotransferase levels are usually associated with a good prognosis.13 However, such is not invariably the case. Inflammatory activity may persist in up to 15% of cases, leading to the development of cirrhosis.7, 14, 15

Because the biochemical and serologic evaluation may not be reliable indicators of the severity of necroinflammatory activity in late-stage infection, we hypothesized that significant inflammation is more frequent in end-stage liver disease due to HBV than is commonly assumed. Because the presence of active necroinflammation in patients with cirrhosis is associated with greater clinical severity in other causes of cirrhosis, i.e., chronic autoimmune hepatitis and alcoholic liver disease, we also speculated that patients with high-grade inflammation would demonstrate a greater degree of hepatic decompensation.16, 17 In this study, we test these hypotheses through a histopathological assessment of liver explants from patients undergoing orthotopic liver transplantation for chronic HBV and correlation of histological changes with the clinical, biochemical and virologic parameters.

Abbreviations

HBV, hepatitis B virus; HBeAg, hepatitis B e antigen; anti-HBe, antibody to hepatitis B e antigen; DNA, deoxyribonucleic acid; HBsAg, hepatitis B surface antigen; HBcAg, hepatitis B core antigen.

Patients and Methods

Human Subjects

Consecutive patients undergoing orthotopic liver transplantation at the Mount Sinai Medical Center between January 1994 and June 1999 for HBV cirrhosis were evaluated. Pediatric patients and those with recent alcohol use, fulminant or subfulminant disease, hepatocellular carcinoma requiring pretransplant treatment, and concurrent hepatitis delta virus infection were excluded, and all patients tested negative for hepatitis C antibody. The study period ended prior to the frequent treatment of patients with end-stage disease and evidence of active viral replication with lamivudine and / or adefovir. Patients who had received immunosuppressants were excluded, and none had received interferon or other antiviral therapy. This investigation was approved by the Mount Sinai School of Medicine Institutional Review Board.

Clinical Assessment

Patient records were reviewed for the presence and severity of ascites and hepatic encephalopathy. Laboratory parameters assessed included serum alanine aminotransferase, aspartate aminotransferase, albumin, total bilirubin levels, and prothrombin time; all measured by standard methods. To avoid acute changes from complications requiring hospitalization, values at the last outpatient visit prior to liver transplantation were assessed. HBsAg, hepatitis B e antigen (HBeAg), anti-HBe, and antibody to hepatitis delta virus were evaluated using commercially available radioimmunoassays or enzyme linked immunosorbent assays (ELISA; Abbott Laboratories, N. Chicago, IL). HBV DNA level was measured by the liquid hybridization assay that can detect levels >5 pg/mL (median 120 days prior to transplantation) (Abbott Laboratories, N. Chicago, IL). Child-Turcotte-Pugh score was calculated using standard criteria. Percentage of expected liver volume was determined from the explant liver weight and the predicted ideal liver weight based on body height and weight.18

Histological and Immunohistochemical Assessment

Representative sections from total hepatectomy specimens at the time of liver transplantation were fixed in 10% neutral buffered formalin and embedded in paraffin. Sections (6 μm) were stained with hematoxylin and eosin. Histological assessment of inflammatory activity was performed on coded slides (K.I., S.T.) without knowledge of the clinical, biochemical, and virologic data. The extent of inflammatory activity was quantitated by assessing the severity of interface hepatitis (0-4) and lobular inflammation (0-4).19 In addition, the severity of portal inflammation was also assessed (grade 0-4). Sections were carefully analyzed for the presence of ground glass hepatocytes. Immunohistochemical studies for HBsAg and hepatitis B core antigen (HBcAg) were performed on deparaffinized histologic tissue sections using polyclonal antibody to HBsAg and HBcAg (Dako, Carpinteria, CA) as previously described.20 Cytoplasmic staining for HBsAg and nuclear and cytoplasmic staining for HBcAg were interpreted as positive findings.

Statistical Methods

The data were analyzed to test for an association between histologic inflammatory activity and demographics (age, Asian ethnicity), biochemical test results, virologic parameters, immunohistochemical findings, percent predicted liver volume, and positive staining for nuclear and cytoplasmic HBcAg. Student's t-test, Wilcoxon's exact 2-tail test, and the exact Jonckheere-Terpstra test were used for continuous data. Categorical data were tested with exact chi-squared (χ2) or Cochran-Armitage trend tests. Data analysis was implemented with SAS software (SAS Institute, Cary, NC) and Stat-Xact for Windows, version 4 (Cytel Software Corporation, Cambridge, MA).

Results

Clinical Assessment

The study group included 34 patients (29 male, 5 female) with a median age of 54.5 years (range 30-71 years). A total of 9 patients were Asian. Median aspartate aminotransferase and alanine aminotransferase levels were 84.5 IU/mL (23-342 U/mL) and 59.0 IU/mL (23-200 IU/mL), respectively. Median serum albumin level was 2.65 gm/dL (1.4-3.9 gm/dL), and median total bilirubin level was 3.3 mg/dL (1.0-26.1 mg/dL). Median prothrombin time was 16.0 seconds (12.2-26.7 seconds). Median Child-Turcotte-Pugh score was 11 points (range 7-15). Of 29 patients, 10 (34.5%) were HBeAg positive, and 8 of 31 patients (25.8%) had detectable HBV DNA levels (median 8.35, range 5.0-48.5 pg/mL). Median percent predicted liver volume was 63% (range 36-99).

Histologic and Immunohistochemical Assessment

The scores for interface hepatitis, lobular inflammation, and portal inflammation are presented in Table 1. High-grade inflammation (grade 3 or 4) was present in many cases, i.e., 16 of 34 (47.1%) patients with interface hepatitis, 5 of 34 (14.8%) patients with lobular inflammation, and 7 of 34 (20.6%) patients with portal inflammation (Fig. 1). HBsAg immunoreactivity was present in all cases and was associated with ground glass hepatocytes in 32 cases (94.1%). A total of 18 cases stained positively for nuclear HBcAg, and 11 cases stained positively for cytoplasmic HBcAg (Fig. 2).

Table 1. Grades of Interface Hepatitis, Lobular, and Portal Inflammation in 34 Consecutive Patients Undergoing Orthotopic Liver Transplantation for HBV Cirrhosis
Grade of histologic activity1234
Interface hepatitis315151
Lobular inflammation20941
Portal inflammation91861
Figure 1.

Photomicrograph of liver explant from 57-year-old Asian woman with end-stage liver disease secondary to chronic HBV and cirrhosis, demonstrating severe inflammatory activity (alanine aminotransferase 55 U/L; aspartate aminotransferase 72 U/L; HBV DNA level and HBeAg not detected) (hematoxylin and eosin stain, ×100).

Figure 2.

Immunohistochemical stains demonstrating positive staining for nuclear and cytoplasmic HBcAg (A) and for HBsAg (B) (counterstained with hematoxylin, ×200).

Association of Histologic Parameters of Inflammation With Demographics, Clinical Features, and Immunohistologic Findings

Patients with higher grades of inflammation had greater degrees of hepatic decompensation. Increasing grades of interface hepatitis correlated with increasing prothrombin times (P = .035) and Child-Turcotte-Pugh scores (P = .051) (Table 2A). Higher grades of lobular activity were correlated with lower albumin levels (P = .079) and increasing Child-Turcotte-Pugh scores (P = .084) (Table 2B). Grade of portal inflammation inversely correlated with albumin level (P = .025) (Table 2C). No parameter of inflammatory activity was correlated with age, Asian ethnicity, aminotransferase levels, total bilirubin levels, HBeAg seropositivity, detectable HBV DNA level, or percent expected liver volume (Tables 2A-C). Cytoplasmic staining for HBcAg was associated with grade 3 or 4 interface hepatitis (P = .046) and lobular inflammation (P = .005) (Tables 2A and B).

Table 2. Clinical, Biochemical, Virologic, and Immunohistochemical Characteristics by Grade of Inflammatory Activity in 34 Consecutive Patients Undergoing Orthotopic Liver Transplantation of Interface Hepatitis (A), Lobular Inflammation (B), and Portal Inflammation (C)*
 Grade 1Grade 2Grade 3 or 4Significance
  • *

    Values presented are number (percent) or median (minimum, maximum).

  • Abbreviations: NSS, not statistically significant; ALT, alanine aminotransferase; AST, aspartate aminotransferase.

A. Interface hepatitis
 Patients (n)3 (8.8%)15 (44.1%)16 (47.1%) 
 Asian05 (26.7%)4 (25%)NSS
 Age (years)58 (57–62)48 (30–70)56.5 (30–71)NSS
 ALT (U/L)53 (37–200)65 (23–145)59 (26–84)NSS
 AST (U/L)85 (48–342)87 (23–206)84 (41–153)NSS
 Albumin (gm/dL)2.8 (2.6–3.1)2.6 (1.4–3.9)2.6 (2–3.6)NSS
 Total bilirubin (mg/dL)2.2 (1–13.6)2.4 (1–26.1)4.2 (.8–10.6)NSS
 Prothrombin time (seconds)13.2 (13–14.4)16 (11.8–26.7)17.3 (12.2–20.5)p = .035
 Child-Turcotte-Pugh score9 (8–9)10 (7–13)11 (8–15)p = .051
 HBeAg+0/ 2 (0%)4/ 13 (30.8%)6/ 14 (42.9%)NSS
 HBV DNA+0 / 2 (0%)4 / 14 (28.6%)4 / 15 (26%)NSS
 % Predicted liver volume82 (63–97)58 (41–93)70 (36–99)NSS
 HBcAg nuclear staining positivity33.3%46.7%62.5%NSS
 HBcAg core staining positivity0%20%50%p = .046
B. Lobular inflammation
 Patients (n)20 (58.9%)9 (26.5%)5 (14.8%) 
 Asian5 / 20 (25%)2 / 9 (22.2%)2 / 5 (40%)NSS
 Age (years)52.5 (30–71)57 (30–68)55 (36–66)NSS
 ALT (U/L)53 (23–200)67 (26–107)55 (36–66)NSS
 AST (U/L)84 (23–342)87.5 (41–156)91 (50–129)NSS
 Albumin (gm/dL)2.8 (2.2–3.6)2.4 (1.4–3.9)2.2 (2–3.2)p = .079
 Total bilirubin (mg/dL)2.6 (1.0–26.1)4.2 (1.0–12.6)4.9 (.8–10.6)NSS
 Prothrombin time (seconds)16 (11.8–19.2)17.7 (13.8–26.7)16.6 (12.2–20.5)NSS
 Child-Turcotte-Pugh Score10 (7–13)11 (8–15)11 (8–13)p = .084
 HBeAg+6 / 18 (33%)3 / 8 (37.5%)1 / 3 (33.3%)NSS
 HBV DNA +3 / 18 (16.7%)3 / 9 (33.3%)2 / 4 (50%)p = .16
 % Predicted liver volume62 (43–97)50 (36–92)92 (58–99)NSS
 HBcAg nuclear staining positivity45%55.67%80%NSS
 HBcAg core staining positivity15%44.4%80%p = .005
C. Portal inflammation
 Patients (n)9 (26.5%)18 (52.9%)7 (20.6%) 
 Asian2 / 9 (22.2%)6 (33.3%)1 / 7 (14.3%)NSS
 Age (years)52 (30–66)54 (30–71)56 (40–84)NSS
 ALT (U/L)65 (28–145)53 (23–200)64 (45–84)NSS
 AST (U/L)135 (23–206)71 (41–342)83 (47–153)NSS
 Albumin (gm/dL)3.0 (2.3–3.9)2.6 (1.4–3.4)2.4 (2–3.6)p = .025
 Total bilirubin (mg/dL)3.2 (1–26.1)2.8 (.8–13.6)4 (1.5–6.5)NSS
 Prothrombin time (seconds)16.0 (11.8–26.7)16.0 (12.2–19.5)16.9 (14.7–20.5)NSS
 Child-Turcotte-Pugh Score10 (7–13)10.5 (8–13)16.9 (14.7–20.5)NSS
 HBeAg+2 / 7 (28.6%)7 / 16 (43.8%)1 / 6 (16.7%)NSS
 HBV DNA +2 / 8 (25%)3 / 16 (18.8%)3 / 7 (42.9%)NSS
 % Predicted liver volume63 (43–93)67 (36–99)58 (40–96)NSS
 HBcAg nuclear staining positivity44.4%61.1%42.9%NSS
 HBcAg core staining positivity22%38.9%28.6%NSS

Discussion

In this study, we report the unexpected finding that significant hepatic inflammation is common in patients with end-stage HBV cirrhosis undergoing orthotopic liver transplantation, but it is not reliably detected by biochemical and virologic measures. Despite an undetectable serum HBV DNA level in many of our cases, the associated hepatic inflammation can be attributed to HBV infection. Other etiologies for inflammation, e.g., recent alcohol consumption, HCV and hepatitis delta virus infection, were excluded. There was expression of HBsAg in all cases, and there was a significant correlation between expression of cytoplasmic HBcAg staining with interface hepatitis and lobular inflammation.21–23 Although detectable HBV DNA levels were not present in many cases, it is important to emphasize that the less sensitive liquid hybridization assay for HBV DNA previously employed for clinical indications was used rather than the more sensitive polymerase chain reaction assay that is currently available.

Biochemical and virological parameters are important predictors of inflammatory activity in patients with chronic HBV. Inflammation is frequently mild or absent in those with normal aminotransferase levels.24–27 Abnormal aminotransferase levels reflect severity of lobular activity and interface hepatitis, and patients with high aminotransferase activities more commonly develop cirrhosis.1–5 HBeAg and HBV DNA levels greater than 5 pg/mL are also useful indicators of inflammation in chronic hepatitis.23 The presence of HBeAg is associated with elevated aminotransferase levels and active inflammation, and HBeAg to anti-HBe seroconversion is usually accompanied by normalization of aminotransferase levels and decreased inflammation.6, 13, 14, 22 Likewise, detectable serum HBV DNA is associated with lobular inflammation, and its persistence predicts histologic progression.2, 3, 10, 28, 29

Our results indicate that the conventional assumption that histologic inflammatory activity is mild in patients with end-stage cirrhosis awaiting liver transplantation is incorrect. We report that grade 3 and 4 interface hepatitis is present in 47.0% of cases, lobular inflammation in 14.8% of cases and portal inflammation in 20.6% of cases undergoing liver transplantation for end-stage disease. In addition, our findings concur with previous studies that indicate that aminotransferase levels may not reflect inflammatory severity in end-stage liver disease.7 Similarly, the HBeAg status is not a useful parameter in predicting the severity of necroinflammatory activity in cirrhotic patients, and it has even been reported that cirrhotic patients with decompensated disease are more commonly HBeAg negative than those with compensated cirrhosis.5, 7, 8, 12, 30

Treatment of patients with cirrhosis and active inflammation may lead to improved liver function and decreased portal venous hypertension.16, 17 For example, survival and liver function in patients with autoimmune hepatitis is improved with immunosuppression, and patients with alcoholic hepatitis benefit from abstinence and corticosteroids. In patients with decompensated HBV cirrhosis and active viral replication, treatment with adefovir or lamivudine may lead to potentially marked clinical and biochemical improvement.31–35 Antiviral therapy with adefovir and lamivudine markedly improve necroinflammatory activity and can reverse fibrosis in patients, even in those with cirrhosis.36, 37 Patients with decompensated cirrhosis and evidence of viral replication have improved to such an extent that they have been removed from liver transplant waiting lists.34 Although response is associated with suppression of viral replication, the mechanism by which antiviral therapy leads to improved liver function has not been determined, but possibilities include a decrease in the severity of necroinflammatory activity, a direct effect on viral replication, and reversal of fibrosis.

Based on our findings that significant inflammation is frequent and associated with greater degrees of hepatic decompensation, we propose that the response to antiviral therapy is due to a decrease in inflammatory severity. Inhibition of viral replication is probably not the immediate cause. Injury from HBV is due to immune-mediated damage to infected hepatocytes.38 Severity of liver disease does not correlate with viral load, and many patients have low HBV DNA levels prior to therapy.34 Finally, improved liver function may persist despite the development of resistant mutants.32, 34 Reversal of fibrosis is also unlikely to be the reason for early improvement due to the time-course. Years of effective antiviral therapy are required for cirrhosis to regress.39 However, liver function begins to slowly but progressively improve within only 3 months to achieve a potentially well compensated state by 6 to 9 months of therapy.31, 34, 35 In contrast, severe deterioration in the clinical status may rapidly develop with the emergence of resistant mutants.33

It has recently been proposed that all decompensated patients should be considered for antiviral therapy.40 Unfortunately, only approximately two-thirds of decompensated patients have a clinical response. Most deaths occur within the first 6 months of therapy, and no predictive factors are currently available that identify patients who may ultimately respond at the time treatment is initiated.32, 34, 35 As a result, the decision about whether to proceed directly to liver transplantation for listed patients or wait for the outcome of a trial of medical therapy for patients with decompensated cirrhosis can be quite difficult.

At this time, simultaneous listing for liver transplantation and antiviral treatment until evidence of clinical improvement or the availability of an allograft for transplantation is the most prudent course. To translate our hypothesis into clinical practice, it will be important to confirm the proposal that the clinical response in patients with end-stage HBV cirrhosis is greatest in those patients with significant histologic inflammatory activity. Unfortunately, liver biopsy is frequently problematic in patients with end-stage liver disease due to the presence of ascites and coagulopathy. As a result, the development of a noninvasive marker for inflammatory activity would also be required. Although we did not find the HBV DNA level helpful in predicting the severity of inflammatory activity, a significant limitation of this study is the absence of HBV DNA levels based on polymerase chain reaction technology, to further evaluate those with negative HBV DNA levels by liquid hybridization assay. To avoid a selection bias and the effect of antiviral therapy on the histopathologic and clinical findings, this study was limited to consecutive cases of patients undergoing orthotopic liver transplantation from a time period prior to the use of antiviral therapy for patients with end-stage disease and prior to the availability of the ultrasensitive polymerase chain reaction assay, and stored serum was not available for reanalysis. Whether there is a threshold HBV DNA level for the presence of significant inflammatory activity is unknown.40 Because many patients with advanced HBV cirrhosis are currently treated with antiviral therapy while awaiting transplantation, however, it is no longer possible to acquire such “natural history” data in an untreated cohort of patients. Whether other markers, such as those related to tumor necrosis factor α that directly assess inflammatory activity, are useful in predicting clinical response could be evaluated in a prospective study.41

Ancillary