Prospective Comparison of Fibroscan, King's Score and Liver Biopsy for the Assessment of Cirrhosis in Chronic Hepatitis C Infection
T. J. S. Cross; V. Calvaruso; S. Maimone; I. Carey; T. P. Chang; M. Pleguezuelo; P. Manousou; A. Quaglia; F. Grillo; A. P. Dhillon; G. M. Dusheiko; A. K. Burroughs; P. M. Harrison
Authors and Disclosures
Posted: 02/07/2011; J Viral Hepat.
2010;17(8):546-554. © 2010 Blackwell Publishing
Abstract and Introduction
Abstract
Historically, liver biopsy (LB) was the sole method to evaluate the severity of hepatic fibrosis in patients with chronic hepatitis C infection. However, LB is expensive and associated with a risk of severe complications. Therefore, noninvasive tests have been developed to assess the severity of liver fibrosis. The accuracy of Fibroscan (FS) and King's score (KS) was evaluated individually and in combination using liver histology as the reference standard. One hundred and eighty-seven patients were identified who had undergone a biopsy with a diagnosis of chronic hepatitis C virus (HCV) mono-infection (HCV RNA-positive by RT-PCR), attending King's College Hospital (n = 88) or the Royal Free Hospital (n = 99) (London) between May 2006 and December 2007.
Introduction
Hepatitis C virus (HCV) infection is associated with a wide spectrum of liver disease. Progressive fibrosis is the hallmark of worsening chronic hepatitis C infection (CHC). Cirrhosis leads to the complications of end-stage liver disease including liver failure, ascites, variceal bleeding, porto-systemic hepatic encephalopathy and hepatocellular carcinoma. At presentation, 10–20% of chronic hepatitis C patients are cirrhotic and between 20% and 30% of noncirrhotic patients will develop it within one or more decades.[1–3] Patients with more severe hepatic fibrosis at baseline have a lower rate of sustained virological response in comparison with patients with a lower stage of fibrosis.[4]
Historically, liver biopsy (LB) has been the only accepted method to evaluate the severity of hepatic fibrosis in CHC-infected patients. However, it can be associated with life-threatening complications and pain.[5,6] Furthermore, it is subject to inter-observer variability and sampling error and this can lead to either under- or over-staging of disease.[7–9] Therefore, there is a need for noninvasive tests that accurately and safely measures severity of liver fibrosis. There has been a proliferation in the number of noninvasive tests of liver fibrosis over the last decade some of which have been shown to have clinical utility[10–13]
Fibroscan (FS) or transient elastography [FibroScan (FS); Echosens, Paris, France] is a noninvasive method for measuring liver stiffness. The stiffer the tissue, the faster the shear wave propagates. Studies have confirmed that FS reflects hepatic fibrosis in CHC and has a good correlation with histological staging systems.[13–15] The King's score (KS) is a simple noninvasive test that was developed to predict the likelihood of significant fibrosis or cirrhosis at the bedside or in the clinic. It was derived from a training set of 602 consecutive CHC-infected patients seen at King's College Hospital (KCH) and was validated in a further 102 patients.[16]
A previous study has suggested that a combination of noninvasive tests may be the best strategy to improve the diagnostic accuracy of noninvasive tests.[15] The aim of this study was to determine the diagnostic accuracy for predicting stage of fibrosis of FS and KS individually and in combination in consecutive patients with CHC against LB.
Patients and Methods
One hundred and ninety-nine (199) patients with a diagnosis of CHC infection attending KCH and the Royal Free Hospital (RFH) (London, UK) between May 2006 and December 2007 were identified. Inclusion criteria were as follows: all patients positive for anti-HCV antibodies using a third-generation enzyme immunoassay, positive HCV RNA by reverse-transcription polymerase chain reaction and CHC mono-infection with an available LB that had been performed no more than 6 months before the FS. Exclusion criteria were body mass index (BMI) >30 kg/m2 and patients who were currently on anti-viral therapy or who had completed anti-viral therapy, Patients with co-existent liver disease, e.g. chronic hepatitis B infection, autoimmune liver disease and human immunodeficiency virus (HIV), patients with a history of high alcohol consumption, in men this was defined as patients consuming >80 g alcohol per day, and in women was defined as patients consuming >60 g of alcohol per day. HCV genotypes were measured prior to commencing treatment if patients agreed to have it. The study protocol conformed to the ethical guidelines of the 1975 Helsinki declaration.
Clinical Parameters
The following data were recorded from all patients at the time of FS: age, gender, HCV genotype, alcohol consumption, time since LB, current or previous use of anti-viral therapy, diabetes mellitus and BMI. BMI was calculated as body weight in kilograms divided by the square of height in meters (kg/m2).
Biochemical Parameters
The following data were recorded from patients at the time of LB: full blood count, international normalized ratio (INR) and routine liver function tests including bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), albumin, alkaline phosphatase and gamma-glutamyl transferase. Routine biochemical assessment of renal function was performed. Only AST, INR and platelets were documented for the study. At KCH, platelets were measured using Advia (Siemens, Deerfield, IL, USA, reference range 150–450 × 109/L), and AST was measured using Advia 2400 (Siemens, reference range 10–50 IU/L). INR was measured using Stago Star Evolution (Diagnositica Stago, Asniéres sur seine, France, reference range 0.9–1.2). At RFH, platelets were measured with Advia 2120 (Siemens, reference range 150–450 × 109/L), AST was measured with Roche MODULAR analyzer (Roche diagnostics, Indianapolis, IN, USA, reference range less then 37) and INR was measured using Stago Star Evolution (Diagnositica Stago, reference range 0.9–1.2). The assays were run in separate laboratories. Percutaneous LBs were performed provided patients had platelets more then 80 × 109/L, INR less then 1.3 and haemoglobin more then 10 g/dL. Blood products were given to achieve these targets where necessary. Where coagulation or platelets remained low, transjugular LB was performed: three cores of tissue were taken to provide adequate tissue for histological analysis as previously described.[17]
Paraffin-embedded LB specimens of more than 10 mm or more then 10 portal tracts were analysed by an experienced histopathologist (AQ, FG, APD). The number of LB fragments was not routinely recorded for the purpose of the study. The histopathologists were unaware of patient clinical data, other than the knowledge of chronic HCV infection. LB specimens were stained with hematoxylin and eosin, and fibrosis stage was analysed examining LB specimens stained with reticulin. Histological fibrosis stage was scored using the method described by Ishak.[18] Significant fibrosis was defined as Ishak fibrosis stage F3–F6, and cirrhosis was defined as Ishak fibrosis stage F5–F6. The METAVIR (M) classification has been used in a number of studies, a conversion factor was introduced to standardize results taking into account the relative frequencies of each stage of fibrosis using the method described by Poynard and colleagues:[19] Ishak F0 = METAVIR F0, Ishak F1–2 = METAVIR F1, Ishak F3 = METAVIR F2, Ishak F4–5 = METAVIR F3 and Ishak F6 = METAVIR 4.[20]
Fibroscan (FS) Measurement
This was performed on the right lobe of the liver in the intercostals spaces with the patient in the supine position with the right arm in maximum abduction. The transducer tip was smeared with coupling gel and placed on the skin at the level of the right lobe of the liver. On locating the target area, i.e. an area of liver 6 cm depth with no major vascular structures measurements were started after pressing the probe button. The measurement depth was between 25 and 65 mm. Ten valid measurements were required for a successful acquisition with a greater than 60% success rate being needed to allow inclusion in the final analysis. The results were expressed in kilopascals (kPa). The value recorded was the median measurement of the readings. Measurements were only considered reliable if 10 validated measurements were recorded. Patients, in who FS measurement was not reliable, were not included in the analysis. Patients with BMI>30 kg/m2 were excluded because of the failure of FS to achieve accurate acquisitions in these patients as previously described.[13,22]
King's Score
The results of FS were compared to KS and in combination. The KS is a simple noninvasive test utilizing the contracting elements of age, AST and INR all of which are multiplied together and divided by the platelet count.
.
The accuracy of each test alone and in combination was compared for different stages of liver fibrosis by calculating area under receiver operator characteristics curves (AUROC).
Statistical Methods
Between May 2006 and December 2007, 199 consecutive patients were identified who fulfilled inclusion criteria. Twelve patients (6%) were excluded for the following reasons; patient declining LB (n = 2), unable to find LB result (n = 1), unable to acquire transient elastography reading (n = 1), unreliable liver stiffness reading (n = 8). All patients in whom FS measurement was unsuccessful were overweight (BMI more then 25 kg/m2). The median age was 49 years (43–54), and there were 111 men (59%). The median length of LB samples was 15 mm (13–17 mm).
The distribution of Ishak fibrosis stage was as follows: F0, n = 6; F1, n = 51; F2, n = 41; F3, n = 28; F4, n = 11; F5, n = 11, F6, n = 39. Fifty patients (27%) had cirrhosis based on Ishak fibrosis stage. Adjusting for METAVIR units, the frequencies were F0 = 6 (3%), F1 = 92 (49%), F2 = 28 (15%), F3 = 22 (12%) and F4 = 39 (21%). The mean fibrosis stage for advanced fibrosis using the METAVIR score was (0.15 × 2) + (0.12 × 3) + (0.21 × 4)/0.48 = 3.13. The mean fibrosis stage for nonadvanced fibrosis was 0.49/0.52 = 0.94.
Thus, for this cohort the DANA = 3.13 – 0.94 = 2.19). The genotype was available for 148 patients: genotype 1 (n = 81, 55%), genotype 2 (n = 10, 7%), genotype 3 (n = 43, 29%), genotype 4 (n = 12, 8%) and genotype 5 (n = 2, 1%). Twenty-one per cent of patients were cirrhotic.
The baseline characteristics for the study population are summarized in Table 1, and the baseline characteristics according to liver centre are shown in Table 2. The correlations between Ishak fibrosis stages and noninvasive tests, alone and in combination were FS, (r = 0.72, P less then 0.0001), KS (r = 0.66, P less then 0.0001) and FS + KS (r = 0.71, P less then 0.0001). The box-plots are shown in Figs 1–3.
Figure 1.
Box-plot for Fibroscan for each Ishak fibrosis stage.
Figure 2.
Box-plot for King's score for each Ishak fibrosis stage.
Figure 3.
Box-plot for combined Fibroscan + King's score.
Liver stiffness measurements ranged from 3.0 to 73.5 kPa (median 6.8). The readings for KS ranged from 1.1 to 441.7 (median 12.5), and for the combined score FS + KS, the range of scores was 3.2–476.5 (median 19.8). The diagnostic accuracy of the tests, alone and in combination to predict liver fibrosis is shown in Table 3. The AUROC for more then F2 was for FS, KS and FS + KS: 0.86, 0.79 and 0.82, respectively. The AUROC for more then F3 was for FS, KS and FS + KS: 0.89, 0.83 and 0.86, respectively. The AUROC for cirrhosis (more then F5) for FS, KS and FS + KS was 0.97, 0.88 and 0.92, respectively. The AUROC curves of this analysis are shown in Figs 4–6. The best discriminator cut-offs to predict the worsening stages of liver fibrosis are shown in Table 4 together with sensitivity, specificity, PPV and NPV.
Figure 4.
AUROC curve of Fibroscan (FS), King's score (KS) and combined FS + KS for the prediction of hepatic fibrosis stage ≥F2.
Figure 5.
AUROC curve of Fibroscan (FS), King's score (KS) and combined FS + KS for the prediction of hepatic fibrosis stage ≥F3.
Figure 6.
AUROC curve of Fibroscan (FS), King's score (KS) and combined FS + KS for the prediction of hepatic fibrosis stage F5–6 (cirrhosis).
.
Using METAVIR fibrosis (MF) stage significant fibrosis (more then MF2), the AUROC for FS, KS, and FS + KS were 0.89 (0.83–0.94, SE 0.03, P less then 0.0001, standardized ROC = 0.89), 0.83 (0.77–0.89, SE = 0.03, P less then 0.0001, Standardized ROC = 0.89) and 0.86 (0.80–0.91, SE 0.03, P less then 0.0001, standardized ROC = 0.84), respectively. There was no statistical difference between the accuracy of FS vs KS (P = 0.08) alone.
Effect of Aspartate Aminotransferase on Accuracy of KS and Combined FS + KS
.
An analysis was performed to assess the impact of normal and elevated AST on the diagnostic accuracy of each noninvasive modality. Sixty-seven patients had normal AST level. The AUROC for more then F3 fibrosis for FS, KS and FS + KS was 0.80 (0.65–0.95, P = 0.002), 0.83 (0.68–0.99, P = 0.001) and 0.88, respectively (0.75–1.0, P ≤ 0.0001). The AUROC for patients with elevated AST (n = 120) for more then F3 for FS, KS and FS + KS was 0.84 (0.76–0.93, P less then 0.0001), 0.79 (0.69–0.89, P less then 0.0001) and 0.84 (0.75–0.92, P less then 0.0001), respectively.
.
Sixty-seven patients had normal AST level with cirrhosis. The AUROC for ≥F5 fibrosis for FS, KS and FS + KS was 0.68 (0.54–0.79, P = 0.57), 0.96 (0.91–1.0, P = 0.12) and 0.94, respectively (0.88–1.0, P = 0.13). The AUROC for patients with elevated AST (n = 120) for more then F5 for FS, KS and FS + KS was 0.89 (0.83–0.95, P less then 0.0001), 0.78 (0.67–0.88, P less then 0.0001) and 0.84 (0.76–0.92, P less then 0.0001), respectively. The results are shown in Table five.
Effect of Liver Biopsy on Accuracy of KS and Combined FS + KS
Finally, the authors assessed the effect of LB length on diagnostic accuracy. For a LB length less then 15 mm, the accuracy in detecting significant fibrosis (F3–F6) for FS, KS and FS + KS was 0.84 (0.70–0.98, SE 0.07, P = 0.001), 0.84 (0.70–0.98, SE 0.07, P = 0.001) and 0.87 (0.74–0.99, SE 0.07, P less then 0.001), respectively vs FS, 0.81 (0.70–0.91, SE 0.05, P less then 0.0001), KS, 0.83 (0.72–0.93, SE 0.05, P less then 0.0001) and FS + KS0.86 (0.76–0.95, SE 0.05, P less then 0.0001) for LBs more then15 mm. For a LB length less then 15 mm, the accuracy in detecting cirrhosis (F3–F6) for FS, KS and FS + KS was 0.98 (0.93–1.0, SE 0.02, P = 0.001), 0.94 (0.87–1.0, SE 0.04, P less then 0.0001) and 0.97 (0.92–1.0, SE = 0.03, P less then 0.0001), respectively vs FS, 0.94 (0.88–1.0, SE = 0.03, P less then 0.0001), KS, 0.82 (0.71–0.9, SE 0.06, P less then 0.0001) and FS + KS 0.88 (0.79–0.97, SE 0.05, P less then 0.0001) for LBs more the 15 mm.
Discussion
The United Kingdom is moving away from the decision to treat chronic hepatitis C based on the presence of moderate or severe fibrosis. In the United Kingdom, in August 2006, the National Institute of Clinical Excellence (NICE) issued updated guidance on treatment for patients with chronic hepatitis C infection permitting treatment of all chronic HCV-infected patients to be offered anti-viral therapy irrespective of fibrosis stage dramatically reducing the need for LB.[23]
This study evaluated the performance of FS, KS and FS + KS in 187 consecutive patients with CHC. The diagnostic performance of each test was good with clinical utility (AUROC more then 0.8) for all three tests alone and in combination across the range of fibrosis studied. The combined use of FS and KS offered the most accurate test for less then F2, but this failed to reach statistical significance in comparison with FS or KS alone. For all stages of fibrosis more then F3 fibrosis, FS alone was the best test. Twenty-seven per cent of patients had cirrhosis based on Ishak fibrosis, in comparison with 27%,[14] 25.1%[15] and 19.5%[13] reported in previous FS studies.
The AUROC curves and cut-offs from the present study compare favourably to those from previous studies. The AUROC for significant fibrosis and cirrhosis defined in the studies by Ziol, Castera and Foucher were 0.79 and 0.97, 0.83 and 0.95, and 0.80 and 0.96, respectively. The optimal cut-off value for more then F3 (8.9 kPa) was similar to those described (Foucher, 8.1 kPa, Ziol, 8.7 kPa; and Castera, 7.1 kPa), but the cut-off for cirrhosis (10.1 kPa) was lower than that described in similar studies (Foucher, 17.6 kPa; Ziol, 14.5 kPa; and Castera, 12.5 kPa).
At the less severe grades of fibrosis, the PPV was high, whereas for cirrhosis the noninvasive tests were stronger with a high NPV. The main strength of FS is its high PPV and specificity at lower stages of fibrosis and its high NPV together with excellent sensitivity and specificity in detecting cirrhosis. The KS has the advantage that no additional equipment is required, and it is quick and easy to calculate. This may be particularly advantageous in parts of the world where FS is not routinely available. But, the test is not without its problems. There may not be standardization between different assays for AST, INR and platelets used between hospitals. Thus, the cut-off derived from one institution may differ from that of a neighbouring institution. In addition, age is taken as a surrogate marker of disease duration, and this assumption may be prone to error.
Furthermore, medication or conditions that lead to prolongation of the INR, e.g. warfarin treatment, or low platelets or other factors that may lead to a rise in AST (e.g. steatohepatitis) may lead to an inaccurate assessment of disease severity. The problems with standardization of routine biochemical, haematological and coagulation tests have been described previously.[19]
The use of LB in chronic HCV is declining. Not only is LB associated with complications including bleeding, pain, pneumothorax, haemothorax, bile peritonitis, haemobilia, kidney and intestinal puncture and death,[5,24] but also is associated with problems of sampling error because a LB sample is only 1/50 000th of total liver volume. Regev et al. [7] showed a difference of 33% in fibrosis stage in left and right lobe LBs, while Siddique et al. [25] showed that even from biopsies taken from the same puncture site, 45% of patients could display at least one fibrosis stage of difference. Further problems occur because of intra-and inter-observer variability assessing fibrosis stage.[26] The length of LB can also be crucial in determining diagnostic accuracy. Regev showed only 65% of 15 -mm-staged liver fibrosis stage correctly, this increasing to 75% for 25mm sample. The length of LB may also be significant because fibrosis is likely to be under-staged as LB length diminishes.[9] More recently, work has been performed showing that length of LB and the number of fragments should be taken into account when evaluating the diagnostic accuracy of a test using AUROC curves.[19] This will obviously have an effect on the accuracy of LB as the gold standard and may explain in part why no test is likely to be perfect. For this reason, a sensitivity analysis on the effect on LB length above or below 15 mm was performed. This failed to demonstrate any statistical difference between the two groups in this study population. A potential advantage of FS is that it effectively samples an area of liver stiffness that is approximately of 1 cm diameter and 2 cm long, leading to a volume 100 times larger than a standard LB specimen.
As has been confirmed from previous studies, AST appears to have an impact on FS readings with elevation of transaminases being mirrored by increase in liver stiffness. It has been postulated that high transaminases are a reflection of hepatitis activity, and thus points to oedema and inflammatory infiltrate contributing to the transient elastographic measure.[27] For patients with normal AST, the KS had greater diagnostic accuracy than FS alone. However, the majority of patients had abnormal AST and thus, overall FS performed better, particularly for cirrhosis. Sebastiani et al. assessed the performance of AST-to-platelet ratio index (APRI), Forn's index, AST-to-ALT ratio (AAR), Fibrotest and Fibroindex in CHC patients with normal (n = 80) and elevated ALT (n = 164) in discriminating patients with significant fibrosis (METAVIR more then F2).[28] They found that performance across all the tests was better in patients with elevated ALT. Interestingly, our study showed that KS was less accurate in patients with elevated AST. A possible explanation is that patient age and prolonged INR may counterbalance the effect of a lower AST, thereby preserving accuracy.
In summary, FS and KS have clinical utility in identifying patients with significant fibrosis and cirrhosis. For the lower stages of fibrosis (less then F2), the combination of FS + KS had the best AUROC curve, but it was not statistically different to FS or KS alone. FS alone was the best test for predicting more then F3 fibrosis. FS measures liver stiffness and not fibrosis, and thus factors that cause inflammation or add to oedema may elevate FS readings. All tests perform better at the more severe scale of fibrosis, and overall FS is a superior test, but in circumstances where FS is not available, KS is a suitable alternative.
Abstract and Introduction
Patients and Methods
Results
Discussion References
No comments:
Post a Comment