Simple blood tests as noninvasive markers of liver fibrosis in hemodialysis patients with chronic hepatitis C virus infection

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

HCV infection is common among patients with end-stage renal disease (ESRD) on hemodialysis, and it has been considered an independent risk factor for mortality in this setting. Although liver biopsy in ESRD patients with HCV infection is useful before kidney transplantation, it carries a high risk of complications. We sought to assess the diagnostic value of noninvasive markers to stage liver fibrosis in 203 ESRD HCV-infected patients. Univariate and multivariate analysis were used to identify variables associated with significant fibrosis (METAVIR F2, F3, or F4 stages). Significant liver fibrosis was observed in 48 patients (24%). Logistic regression analysis identified AST and platelet count as independent predictors of significant fibrosis (P < 0.001 and P = 0.001, respectively). The area under the receiver operating characteristic curve of the AST to platelet ratio index (APRI) for predicting significant fibrosis was 0.801. An APRI < 0.40 accurately identified patients with fibrosis stage 0 or 1 in 93% of the cases (NPV = 93%), and all misclassified subjects were F2. A cutoff ≥ 0.95 to confirm significant fibrosis had a PPV of 66%. If biopsy indication was restricted to APRI scores in the intermediate range (≥0.40 and < 0.95), 52% of liver biopsies could have been correctly avoided. Conclusion: Stage of liver fibrosis can be reliably predicted in ESRD HCV-infected subjects by simple and widely available blood tests such as AST levels and platelet count. These tests might obviate the requirement for a liver biopsy in a significant proportion of those patients. (HEPATOLOGY 2007.)

The worldwide prevalence of HCV infection among end-stage renal disease (ESRD) patients undergoing hemodialysis ranges from 10% to as high as 59%, according to the geographical area.1, 2 Although transfusion of blood products still plays a significant epidemiological role, the vast majority of HCV chronic infections in the hemodialysis setting are currently attributable to nosocomial transmission.3 HCV infection is an important cause of morbidity and mortality in dialysis patients4 and has been associated with higher rates of graft loss after kidney transplantation.5–8 In addition, given the contraindication of ribavirin in hemodialysis individuals, interferon-alfa represents the standard therapy for those subjects, limiting the efficacy of the treatment.9 Furthermore, dialysis patients often exhibit a higher frequency of adverse events and discontinuation rates than HCV-infected patients with normal renal function.9, 10 Therefore, an accurate characterization of liver disease in those subjects is essential to optimal management and proper selection for kidney transplantation.

Although widely performed and accepted in diagnosing hepatic fibrosis, liver biopsy is an invasive technique with associated morbidity. Minor complications are relatively common, and approximately one fourth of the patients have pain in the right upper quadrant or right shoulder after liver biopsy.11 ESRD individuals frequently have significant hemostatic disorders and hemorrhagic complications, posing additional risks for patients undergoing invasive procedures.12 Transjugular liver biopsy is an alternate route for obtaining liver specimens in patients with coagulation defects13, 14 and has been evaluated in the ESRD population.15, 16 Although safe, this procedure is not widely available and provides smaller samples than the intercostal technique, which might underestimate fibrosis staging. Hence, there is a need to develop accurate noninvasive tests to predict liver fibrosis, especially among dialysis patients with HCV infection in whom a higher risk for liver biopsy complications has been observed.17, 18

In the past few years, several biochemical indices have been proposed for the noninvasive evaluation of liver fibrosis in HCV patients with normal renal function. However, few data are available regarding this issue among the ESRD population, and most of the evaluated tests include nonroutine laboratory assays or use complex and patented models that limit its regular use.19–21 The aim of our study was to assess the diagnostic value of routine laboratory tests as noninvasive markers to stage liver fibrosis in treatment-naive ESRD HCV-infected patients.

Abbreviations

APRI, AST to platelet ratio index; AUROC, area under the receiver operating characteristics curve; ESRD, end-stage renal disease; NPV, negative predictive value; PPV, positive predictive value.

Patients and Methods

Patients.

This retrospective cross-sectional study included consecutive adult ESRD patients on hemodialysis with HCV infection who underwent percutaneous liver biopsy at the Hospital Sao Paulo, between January 1995 and January 2006, after giving their written informed consent. Liver biopsies were performed using the Tru-Cut 14-gauge biopsy needle. All patients were under regular dialysis for at least 12 months. HCV infection was defined as a positive HCV-RNA by PCR (>50 IU/mL). Patients with the following conditions were excluded: alcohol abuse (ethanol consumption >50 g/day), prior interferon therapy, HBV or human immunodeficiency virus co-infection, insufficient liver tissue for fibrosis staging, and incomplete data on blood counts or liver panel.

The study protocol conformed to the ethical guidelines of the 1975 Helsinki Declaration and was approved by our institutional review board.

Methods.

Information about all ESRD HCV-infected subjects in our institution was reviewed, and demographics, laboratory, and other clinical variables were extracted from medical records.

The duration of HCV infection was estimated based on the assumption that the year of contamination was the first year of dialysis or first blood transfusion (if received before 1992). Serum levels of ALT, AST, gamma-glutamyltransferase, and alkaline phosphatase were expressed as times the upper limit of normal. The other parameters were expressed as absolute values. Only laboratory tests performed within 6 months from the date of the liver biopsy were used for this study.

The AST-to-platelet-ratio index (APRI) was defined as follows: [(AST/upper limit of normal) / platelet count (109/L)] × 100.22

Histological Analysis.

All patients had a liver biopsy irrespective of ALT levels. A single pathologist who was unaware of the clinical data evaluated all slides. Histological features were analyzed using the METAVIR group scoring system. Fibrosis was staged on a scale of F0 to F4, as follows: F0 = no fibrosis, F1 = portal fibrosis without septa, F2 = few septa, F3 = numerous septa without cirrhosis, and F4 = cirrhosis. Significant fibrosis was defined by the presence of F2, F3, or F4 METAVIR stages and the occurrence of F3 or F4 stages characterized advanced fibrosis.

Statistical Analysis.

Continuous variables were compared using the Student's t test, the Mann-Whitney test, or the Kruskal-Wallis test when appropriate. Categorical variables were compared using the chi-square test. A P value of less than 0.05 was considered statistically significant. Univariate and stepwise multivariate analysis were used to identify variables associated with significant fibrosis. The predictive accuracies of these variables were tested by measuring the area under the receiver operating characteristics (AUROC). Based on the receiver operating characteristics curve, the best cutoffs points to predict the absence or presence of significant fibrosis and advanced fibrosis were chosen. Diagnostic accuracy was evaluated by calculating the sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively). Statistical analysis was performed by SPSS software version 11.0 (SPSS Inc., Chicago, IL).

Results

Patient Characteristics.

From January 1993 to July 2006, 2,941 HCV patients were registered on our institution's database. Among these, 455 ESRD patients were considered for enrollment. Two hundred fifty-two patients were excluded because of the following reasons: 60 patients lost to follow-up, 55 had negative serum HCV-RNA, 51 had incomplete data on blood counts or liver panel registered on file, 33 had insufficient liver tissue for fibrosis staging, 23 presented with contraindications to liver biopsy, 23 had active alcohol abuse, 5 had HBV co-infection, 1 had human immunodeficiency virus co-infection, and 1 had prior interferon therapy (Fig. 1). Characteristics of the 252 excluded subjects and of the 203 consecutive patients fulfilling the entry criteria are summarized in Table 1. Overall, demographic and clinical characteristics were similar between included and excluded subjects. Included patients showed higher ALT levels and lower platelet count (P = 0.004 and P = 0.032, respectively). Forty-three excluded patients had liver specimens available for analysis: 1 with HBV co-infection, 1 with human immunodeficiency virus co-infection, and 41 with incomplete laboratory data registered on file within 6 months from the date of the liver biopsy. Of these patients, 20 (46.5%) had no fibrosis (F0), 19 (44.1%) had portal fibrosis (F1), 2 (4.7%) showed septal fibrosis (F2), and 2 (4.7%) had cirrhosis (F4). This frequency distribution was not different from that observed among included subjects (P = 0.144).

Figure 1.

Flow diagram of the potential candidates for participation in the study, reasons for exclusion, and subjects enrolled.

Table 1. Demographic, Clinical, and Biochemical Features of Included and Excluded Patients
CharacteristicIncluded Patients (n = 203)Excluded Patients (n = 252)P*
  • NOTE. *Student's t test, Mann-Whitney test, and chi-squared test as appropriate for comparisons of groups.

  • Abbreviation: ULN, upper limit of normal.

  • Availability among excluded patients:

  • a

    201

  • b

    205

  • c

    189

  • d

    187

  • e

    189

  • f

    158

  • g

    156

  • h

    176.

Age (mean ± SD years)45.0 ± 11.046.4 ± 12.50.210
Male gender, n (%)129 (64)142 (56)0.137
Caucasian, n (%)101 (50)130 (52)0.281
Prior renal transplantation, n (%)46 (23)44 (18)0.169
Mode of transmission, n (%)  0.161
 Transfusion42 (21)42 (17) 
 Hemodialysis159 (78)200 (79) 
 Intravenous drug use0 (0)3 (1) 
 Unknown2 (1)7 (3) 
Age at infection (mean ± SD years)35.1 ± 12.637.2 ± 12.90.081
Duration of infection (mean ± SD years)10.0 ± 6.710.4 ± 6.40.686
AST (median × ULN)0.730.68a0.117
ALT (median × ULN)0.900.72b0.004
GGT (median × ULN)1.901.71c0.298
ALP (median × ULN)0.990.94d0.428
Platelet count (median 109/L)176186e0.032
Bilirubin (median mg/dl)0.700.70f0.218
Albumin (median g/dl)4.204.20g0.732
Prothrombin activity (mean ± SD %)92.1 ± 10.293.5 ± 8.7h0.155

The mean age was 45.0 ± 11.0 years, 50% were white, and a male predominance was observed (64%). Twenty-three percent of the patients had been submitted to previous renal transplantation and returned to hemodialysis. Forty-two percent of the patients had received blood transfusions before starting dialytic therapy. For two patients (1%), the route and duration of HCV infection could not be determined. The age at infection and duration of HCV infection were 35.1 ± 12.6 years and 10.0 ± 6.7 years, respectively. As expected, near normal levels of aminotransferases were observed (0.89 ± 0.52 for AST and 1.17 ± 0.94 for ALT). All liver samples were considered adequate based on specimen size (≥10 mm) and number of portal tracts (≥5). The mean biopsy size was 13.7 ± 4.8 mm with a median of 13.0 mm. The incidence of major complications (significant bleeding and/or symptomatic hypotension) was 10%, with no deaths. Significant fibrosis was observed in 24% (48/203) and advanced fibrosis in 9% (19/203) of the patients (Fig. 2). None of the included patients showed any signs or symptoms of hepatic decompensation.

Figure 2.

Frequency distribution of histological stage according to the METAVIR group scoring system in 203 ESRD patients with HCV infection.

Factors Associated With Fibrosis.

When we compared patients with and without significant fibrosis, univariate analysis revealed no differences regarding age, gender, race, prior renal transplantation, mode of HCV acquisition, duration of infection, and AST to ALT ratio. As illustrated on Fig. 3, patients with prior renal transplantation showed similar distribution of fibrosis stage according to the METAVIR group scoring system as compared with those who had not received a renal transplant (P = 0.908). F2-F4 patients had higher serum levels of AST, ALT, and gamma-glutamyltransferase and showed lower platelet count (Table 2). These differences were also observed between patients with and without advanced fibrosis (F3-F4). Stepwise multiple logistic regression analysis identified AST (OR = 4.607, 95% CI 2.274-9.335, P < 0.001) and platelet count (OR = 0.988, 95% CI 0.891–0.995, P = 0.001) as independent predictors of significant fibrosis.

Figure 3.

Distribution of METAVIR fibrosis scores according to history of previous kidney transplantation. Fibrosis stage was similar in patients with (gray bars) and without (white bars) prior renal transplantation (P = 0.908).

Table 2. Univariate Analysis of Variables Associated with the Presence of Significant Fibrosis and Advanced Fibrosis
VariableNo Significant Fibrosis METAVIR F0-F1 (n = 155)Significant Fibrosis METAVIR F2-F4 (n = 48)P*No Advanced Fibrosis METAVIR F0-F2 (n = 184)Advanced Fibrosis METAVIR F3-F4 (n = 19)P*
  • *

    Student's t test, Mann-Whitney test, and chi-squared test as appropriate for comparisons of groups.

  • Abbreviation: ULN, upper limit of normal.

Age (mean ± SD years)44.6 ± 11.546.2 ± 8.90.30744.7 ± 11.347.5 ± 6.90.123
Male sex, n (%)99 (64)30 (63)0.863117 (64)12 (63)0.970
Caucasian, n (%)75 (48)26 (54)0.48491 (50)10 (53)0.792
Prior renal transplantation, n (%)36 (23)10 (21)0.72942 (23)4 (21)0.562
Mode of transmission, n (%)      
 Transfusion32 (21)10 (21)0.73138 (21)4 (21)0.901
 Hemodialysis121 (78)38 (79) 144 (78)15 (79) 
 Unknown2 (1)0 (0) 2 (1)0 (0) 
Age at infection (mean ± SD years)34.7 ± 12.836.4 ± 11.90.40134.8 ± 12.737.5 ± 11.10.370
Duration of infection (median years)8.008.500.7048.007.000.522
AST (median × ULN)0.711.14<0.0010.711.47<0.001
ALT (median × ULN)0.841.160.0020.841.400.002
GGT (median × ULN)1.862.500.0061.883.790.004
ALP (median × ULN)0.971.040.3540.981.050.979
Platelet count (median 109/L)184.00135.00<0.001179.00151.000.003
Bilirubin (median mg/dl)0.700.700.2720.700.700.888
Albumin (mean ± SD g/dl)4,2 ± 0.54.3 ± 0.60.2614.2 ± 0.54.2 ± 0.70.915
Prothrombin activity (mean ± SD %)92.3 ± 9.891.4± 11.60.58992.5 ± 9.887.8 ± 13.40.175
AST/ALT (median)0.810.890.1790.810.980.154

Figure 4 summarizes the distribution of APRI in relation to the METAVIR fibrosis score. APRI values significantly increased with higher fibrosis stages (P < 0.001, Kruskal-Wallis test).

Figure 4.

Box plot of AST to platelet ratio index in relation to METAVIR fibrosis score. The line across the box indicates the median value; the box contains the 25% to 75% interquartile range; and the whiskers represent the highest and the lower values. APRI values significantly increased with higher fibrosis stages (P < 0.001, Kruskal-Wallis test).

Predictive Value of APRI.

The median APRI was significantly higher in the F2–F4 group than in F0–F1 patients (0.89 versus 0.37, respectively, P < 0.001). To discriminate subjects with significant fibrosis, the AUROC of APRI was 0.801 ± 0.038. Based on the receiver operating characteristics curve, two cutoffs were chosen to predict the absence or presence of significant fibrosis: APRI < 0.40 and ≥ 0.95, respectively. The diagnostic accuracy of the APRI model in discriminating significant liver fibrosis and advanced fibrosis by applying the proposed cutoffs are shown in Table 3. Among patients with APRI lower than 0.40, 84 of 90 (93%) exhibited no or mild fibrosis (F0 or F1). Only 6 of 48 patients (12%) who had significant fibrosis would have APRI lower than 0.40, and all of them showed a METAVIR score of F2. Twenty-one of 48 patients (44%) with significant fibrosis exhibited APRI greater than 0.95, and only 11 of 155 (7%) without significant fibrosis would be classified inaccurately. By comparison with liver biopsy, an APRI ≥ 0.95 showed a PPV of 66% for the presence of significant fibrosis, and an index < 0.40 excluded significant fibrosis with a NPV of 93%. Among patients with scores < 0.40 or ≥ 0.95, concordant results were found in 86% (105/122). If biopsy indication was based only on APRI and restricted to scores in the intermediate range (≥0.40 and <0.95), 52% of liver biopsies could have been correctly avoided.

Table 3. Diagnostic Accuracy of the APRI Model in Predicting Significant Fibrosis (METAVIR F2–F4) and Advanced Fibrosis (METAVIR F3–F4)
  All Patients (n = 203) n (%)Actual FibrosisAccuracy (%)Sensitivity (%)Specificity (%)PPV (%)NPV (%)
F0–F1 (n = 155)F2–F4 (n = 48)
  1. Abbreviations: PPV, positive predictive value; NPV, negative predictive value.

 < 0.4090 (44)84 (54)6 (12)6288543793
Significant fibrosis≥ 0.40113 (56)71 (46)42 (88)     
 < 0.95171 (84)144 (93)27 (56)     
 ≥ 0.9532 (16)11 (7)21 (44)8144936684
   F0–F2 (n = 184)F3–F4 (n = 19)     
 ≤ 0.55127 (63)126 (69)1 (5)7195692499
Avanced fibrosis> 0.5576 (37)58 (31)18 (95)     
 < 1.00175 (86)164 (89)11 (58)     
 ≥ 1.0028 (14)20 (11)8 (42)8542892994

Comparable results were obtained for the detection of advanced fibrosis (F3–F4). The AUROC of APRI was 0.844 ± 0.034. An APRI value ≤ 0.55 for excluding advanced fibrosis achieved an accuracy of 71%, a sensitivity of 95%, a specificity of 69%, a PPV of 24%, and an NPV of 99%. Scores ≥ 1.00 could detect liver F3–F4 stages with an accuracy of 85%, a sensitivity of 42%, a specificity of 89%, a PPV of 29%, and an NPV of 94%.

As shown in Table 4, the ability of an APRI < 0.40 to rule out METAVIR stage ≥ F2 was not influenced by sex (NPV = 93% in males and 96% in females).

Table 4. Diagnostic Accuracy of the APRI Model in Excluding Significant Fibrosis (METAVIR F2–F4) According to Sex
  All Patients (n = 203) n (%)Actual FibrosisAccuracy (%)Sensitivity (%)Specificity (%)PPV (%)NPV (%)
F0–F1 (n = 155)F2–F4 (n = 48)
  1. Abbreviations: PPV, positive predictive value; NPV, negative predictive value.

Female (n = 74)< 0.4023 (31)22 (39)1 (6)5394393396
 ≥ 0.4051 (69)34 (61)17 (94)     
Male (n = 129)< 0.4067 (52)62 (63)5 (17)6783634093
 ≥ 0.4062 (48)37 (37)25 (83)     

Discussion

In ESRD patients with liver disease, the assessment of fibrosis is currently considered of great relevance for prognosis, indication of therapy, and because cirrhosis is usually a contraindication to kidney transplantation alone.23 Recently, the accuracy of liver biopsy in staging liver disease has been a matter of discussion. Because biopsy represents 1/50,000 of the liver, the heterogenicity of liver fibrosis in HCV infection and the inadequacy of sample size can cause considerable bias in the assessment of hepatic histology.24–26 A recent study on virtual liver biopsy has indicated that a nonfragmented specimen of at least 25 mm in length would be necessary to correctly evaluate fibrosis.27 However, trying to achieve this goal may increase the risk of complications, especially among hemodialysis patients who have increased risk of bleeding complications after percutaneous liver biopsy.17, 18 Hence, serum markers can theoretically offer a more accurate view of fibrogenic events occurring in the entire liver with the advantage of providing frequent fibrosis evaluation without additional risk. Ideally, noninvasive markers of liver fibrosis should not only have high sensitivity and specificity, but also should be safe, inexpensive, reproducible, readily available, and able to follow disease progression.

Currently, a variety of noninvasive tests are proposed to estimate liver fibrosis in HCV patients with normal renal function. Nevertheless, few data are available regarding the utility of those tests in ESRD patients with HCV chronic infection. In an attempt to determine possible predictors of liver fibrosis, Boyacioglu et al.21 found no correlations between fibrosis stage and the parameters investigated in 95 hemodialysis patients with HCV infection. However, the analysis included only clinical data, HCV-RNA titer, and serum ferritin and ALT levels, not evaluating the most accepted markers of fibrosis. The AST/ALT ratio was studied as a noninvasive marker of liver fibrosis in 49 ESRD patients with HCV infection.28 Despite the significant differences in AST/ALT ratio found between different fibrosis stages, the usefulness of this index may be limited by the absence of adjusted cutoffs in ESRD patients in whom lower aminotransferase activity is expected.29 In addition, we found no association between AST/ALT ratio and the degree of liver fibrosis in our cohort. The FibroTest (BioPredictive, France), a marker of fibrosis composed by α2 macroglobulin, haptoglobin, γ-glutamyl transpeptidase, total bilirubin, and apolipoprotein A1 levels, has been evaluated in both hemodialysis patients and kidney transplant recipients with HCV infection.19 For the 50 ESRD patients, the AUROC of this index to detect METAVIR F2-F4 fibrosis was 0.470, with a PPV of 75% for scores greater than 0.60 and an NPV of 71% for scores lower than 0.20. Discordant results were found in 28% of the cases. Besides its low accuracy, FibroTest applicability in clinical practice may be limited by the fact that this test is only commercially available and uses non-routine laboratory assays.

In this study, the APRI predicted the degree of liver fibrosis in a considerable number of ESRD patients with HCV infection. This index was originally described for HCV patients with normal renal function.22 In this setting, the AUROC of APRI for predicting Ishak fibrosis stage ≥ 3 in the training set was 0.800, with an NPV of 86% for scores of 0.5 or less and a PPV of 88% for scores greater than 1.5. The APRI is based on the rationale that worsening of fibrosis and increasing of portal pressure is associated with reduced production of thrombopoietin by the hepatocytes, increased platelet sequestration within the spleen, and reduced clearance of AST.30–32 However, many factors may affect the diagnostic performance of APRI in hemodialysis subjects, including heparin-induced thrombocytopenia, lower AST activity, and the coexistence of drug-induced liver disease.29, 33 For these reasons, adjustment of the cutoff values for the APRI in ESRD patients with HCV infection is probably required for optimal performance. In this study, we sought to validate the APRI in evaluating the extent of liver fibrosis and to propose new cutoffs for ESRD HCV-infected patients. The AUROC of APRI for identifying clinically significant liver fibrosis was 0.801. Values < 0.40 accurately identified patients with fibrosis stage 0 or 1 in 93% of the cases (NPV = 93%) and, importantly, the six misclassified subjects were all F2. Less impressive results were observed with the PPV of scores ≥ 0.95 for the prediction of significant fibrosis, because 11 of 32 patients (34%) were erroneously classified. Overall, if biopsy indication was restricted to APRI scores in the intermediate range (≥0.40 and <0.95), 52% of liver biopsies could have been correctly avoided. In addition, from a more practical point of view, a strategy in which the indication of liver biopsy is restricted to patients with APRI ≥ 0.40 would save 44% of biopsies, with an acceptable margin of error (7%).

We acknowledge some limitations to our analysis. First, the use of retrospectively collected data might have led to selection bias. However, this was unlikely to have occurred in this study because included patients showed no significant differences from excluded ones regarding demographic, clinical, and histological characteristics. The higher ALT levels and lower platelet count observed in included patients were expected, due to the presence of a considerable number of non-viremic patients in the exclusion group. Second, the relatively small number of patients with liver cirrhosis also implies possible selection bias, which could restrict the ability to generalize these findings to other populations. Nevertheless, our figures are in keeping with previous case-control studies that evaluated renal transplant candidates.15, 34–38 With very few exceptions, these studies have demonstrated that HCV subjects with ESRD show less progressive liver disease and lower prevalences of advanced liver fibrosis than those with normal renal function. Taking into account that predictive values of tests are affected by disease prevalence, we consider that the 9.4% prevalence of F3/F4 derived from the large cohort presented reflects the actual profile of HCV chronic infection within ESRD population and therefore provides adequate generalizability of the findings to clinical practice. Finally, our study is based on the premise that liver biopsy is the gold standard for assessing hepatic fibrosis, and the correlations between histology and noninvasive markers can be complicated by sampling error as well as intraobserver and interobserver variability.24, 39 Nevertheless, this issue is common to all studies aimed at finding or developing noninvasive methodologies for the evaluation of hepatic fibrosis.

In conclusion, we showed that the stage of liver fibrosis can be reliably predicted in treatment-naive ESRD HCV-infected subjects by simple and widely available blood tests such as AST levels and platelet count. By using APRI with adjusted cutoffs, the need for a liver biopsy might be obviated in a significant proportion of these individuals, particularly among those with APRI < 0.40. If additional studies in patients with ESRD support our findings, the APRI could be used to identify those who might benefit from HCV therapy and those who could be referred for renal transplantation.

Ancillary