LIVER FAILURE/CIRRHOSIS/PORTAL HYPERTENSION

Computer-Assisted Image Analysis of Liver Collagen: Relationship to Ishak Scoring and Hepatic Venous Pressure Gradient Vincenza Calvaruso,1* Andrew Kenneth Burroughs,1* Richard Standish,2 Pinelopi Manousou,1 Federica Grillo,2 Gioacchino Leandro,3 Sergio Maimone,1 Maria Pleguezuelo,1 Ilias Xirouchakis,1 Gian Piero Guerrini,1 David Patch,1 Dominic Yu,4 James O’Beirne,1 and Amar Paul Dhillon2 Histopathological scoring of disease stage uses descriptive categories without measuring the amount of fibrosis. Collagen, the major component of fibrous tissue, can be quantified by computer-assisted digital image analysis (DIA) using histological sections. We determined relationships between DIA, Ishak stage, and hepatic venous pressure gradient (HVPG) reflecting severity of fibrosis. One hundred fifteen patients with hepatitis C virus (HCV) who had undergone transplantation had 250 consecutive transjugular liver biopsies combined with HVPG (median length, 22 mm; median total portal tracts, 12), evaluated using the Ishak system and stained with Sirus red for DIA. Liver collagen was expressed as collagen proportionate area (CPA). Median CPA was 6% (0.2-45), correlating with Ishak stage (stage 6 range, 13%-45%), and with HVPG (r ⴝ 0.62; P < 0.001). Median CPA was 4.1% when HVPG was less than 6 mm Hg and 13.8% when HVPG was 6 mm Hg or more (P < 0.0001) and 6% when HVPG was less than 10 mm Hg and 17.3% when HVPG was 10 mm Hg or higher (P < 0.0001). Only CPA, not Ishak stage/grade, was independently associated by logistic regression, with HVPG of 6 mm Hg or more [odds ratio, 1.206; 95% confidence interval (CI), 1.094-1.331; P < 0.001], or HVPG of 10 mm Hg or more (odds ratio, 1.105; 95% CI, 1.026-1.191; P ⴝ 0.009). CPA increased by 50% (3.6%) compared with 20% in HVPG (1 mm Hg) in 38 patients with repeated biopsies. Conclusion: CPA assessed by DIA correlated with Ishak stage scores and HVPG measured contemporaneously. CPA was a better histological correlate with HVPG than Ishak stage, had a greater numerical change when HVPG was low, and resulted in further quantitation of fibrosis in cirrhosis. (HEPATOLOGY 2009;49:1236-1244.)

Abbreviations: AUROC, area under the receiver operating curve; CI, confidence interval; CPA, collagen proportionate area; DIA, digital image analysis; HCV, hepatitis C virus; HVPG, hepatic venous pressure gradient; LT, liver transplantation; TJB, transjugular liver biopsy. From 1The Royal Free Sheila Sherlock Liver Centre and Department of Surgery, Royal Free Hospital, London, UK; 2Department of Histopathology, Royal Free Hospital, London, UK; 3UOC di Gastroenterologia 1, Istituto di Ricovero e Cura a Carattere Scientifico “De Bellis” Castellana Grotte, Italy; and the 4Department of Radiology, Royal Free Hospital, London, UK. *These authors contributed equally to this work. Received June 26, 2008; accepted November 11, 2008. Address reprint requests to: Andrew Kenneth Burroughs, Professor of Hepatology, Liver Transplantation and Hepatobiliary Unit, Royal Free Hospital, Pond Street NW3 2QG, Hampstead, London, UK. E-mail: [email protected]; fax: 0044-02074726226. Copyright © 2008 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/hep.22745 Potential conflict of interest: Nothing to report.

1236

A

ssessment of liver fibrosis is a very important diagnostic and prognostic evaluation in chronic liver disease.1 However, the limitations of liver biopsy scoring systems are becoming more apparent as liver fibrosis is being assessed more frequently in relation to current therapy for chronic viral hepatitis and nonalcoholic steatohepatitis. All histological scoring systems incorporate a categorical description of architectural changes, without reference to quantitative changes in liver collagen (fibrosis) as liver disease stage progresses or regresses.2 Moreover, routine histological evaluation of fibrosis is often performed with trichrome or reticulin stains, which are general connective tissue stains rather than being selective for collagen, so that the amount of trichrome or reticulin staining does not necessarily correspond with the amount of hepatic collagen.2 Computer-assisted digital image analysis (DIA) of histological sections, histochemically stained by the picro-

HEPATOLOGY, Vol. 49, No. 4, 2009

Sirius red technique, is a method for measuring fibrosis morphologically.2,3 PicroSirius red staining identifies tissue collagen primarily.4 The quantity of bound stain correlates well with chemically determined collagen content and morphometrically determined hepatic fibrosis.5,6 DIA uses segmentation of digital images to measure the area of collagen and of tissue, producing a “fibrosis ratio” or collagen proportionate area (CPA). It does not hinder the other necessary evaluations required for routine diagnosis, and collagenous structures irrelevant to the disease process (which contribute to the variability between samples) can be excluded precisely.2 However, this technique has seldom been used clinically, apart from two recent studies. Goodman et al.7 quantified liver fibrosis progression by computer-assisted morphometric image analysis, evaluating 245 patients with treatment-refractory chronic hepatitis C enrolled in a clinical trial of interferon gamma-1b (IFN-␥1b). Each patient had paired biopsies performed before and after treatment. Morphometry was found to be a more sensitive tool than histological staging to demonstrate fibrosis progression, because changes in collagen content between baseline and week 48 were more evident than changes in Ishak score. In the other study, the computerized morphometric assessment was evaluated in 386 patients with chronic hepatitis C with an Ishak fibrosis score of at least 3. It was less well associated with histological stage than a panel of markers of fibrosis.8 In both studies,7,8 all patients had significant fibrosis, either Ishak of 2 or greater7 or Ishak of 3 or greater.8 The image analysis methodology did not express the amount of collagen as a simple proportionate area, but used histological section thickness to calculate the quantity of collagen. Normal human liver is estimated to contain approximately 5.5 mg/g collagen, and cirrhotic liver contains approximately 30 mg/g.4 Correspondingly, CPA increases from approximately 2% in normal liver to approximately 25% in hepatitis C virus (HCV) cirrhosis.2,3 There has been little evaluation of the range of CPA corresponding to each histological stage of hepatic fibrosis. Indeed, although one would expect a close association, this has not been quantified formally. Standish et al.2 from our group used measurement in single cases, only for illustrative purposes. A large number of liver biopsies are needed to have reliable values of CPA to evaluate its relationship with Ishak staging or other histological scores of fibrosis, or indeed noninvasive markers of fibrosis. In addition, the clinical value of CPA measurement also needs to be assessed and established, by its relationship to a known index of clinical progression in chronic liver disease, particularly in cirrhosis. Currently, the most important and well-documented index is hepatic venous

CALVARUSO, BURROUGHS, ET AL.

1237

pressure gradient (HVPG), which reflects prognosis,9 such that HVPG greater than 6 mm Hg indicates portal hypertension, and HVPG above 10 mm Hg represents clinically significant portal hypertension.10 In nontransplant patients with cirrhosis, Ripoll et al.11 found that HVPG greater than 10 mm Hg predicts the likelihood of developing decompensation, and also survival, as has been shown by many studies.9 HVPG has been put forward as a tool to monitor progression of HCV,12 and this has been confirmed in the follow-up of recurrent HCV disease after liver transplantation13,14 and in nontransplant HCV patients in response to interferon.12 In patients with recurrent HCV after transplantation, our group14 and one other13 have confirmed that a HVPG value greater than 6 mm Hg indicates risk of progression, and HVPG greater than 10 mm Hg indicates a risk of decompensation.15 HVPG can be measured at the same time as obtaining a transjugular liver biopsy (TJB). A liver biopsy should be at least 20 to 25 mm long or contain 11 or more complete portal tracts3,16 to reliably assess grade and stage in chronic viral hepatitis. TJB with at least three passes is superior to the average quality of percutaneous liver biopsy reported in the literature17-21 and has the advantage of obtaining several cores, reducing histological variability.19 Recurrent HCV after liver transplantation is the best available patient group to explore the relationship between CPA, histological staging, and HVPG, and changes over time of CPA in individual patients. Numerous studies document rapid histological progression of HCV after liver transplantation,22 resulting in cirrhosis in up to 30% within 5 years.23,24 In this setting, survival is less than for other liver transplantation (LT) indications such as alcohol-related cirrhosis.25 Because the course of HCV reinfection is unpredictable, monitoring with repeated and scheduled liver biopsies26 is frequently undertaken. In all studies, including those mentioned,13,14 that assessed HVPG after LT, categorical histological scoring systems were used to determine liver disease stage, and correlation of HVPG or other indices were made with these categories. However, these categories cannot be evaluated as continuous variables, because, for example, Ishak stage 4 does not represent arithmetically twice as much collagen as stage 2. Correlations of continuous variables such as HVPG and noninvasive indices of fibrosis, including transient elastography, should be made with an arithmetically progressive index, that is, a quantitative measurement of liver fibrosis such as CPA.2 Furthermore, a range of CPA values can be expected within each disease stage category, such that the measured CPA could have additional individual prognostic value within a disease stage

1238

CALVARUSO, BURROUGHS, ET AL.

category such as cirrhosis, or other stage groupings (such as severe fibrosis: Ishak 5 ⫹ 6). Our aim was to evaluate the relationship between the amount of hepatic collagen measured by DIA in liver biopsy samples and the corresponding Ishak grading/staging score and HVPG (measured at the time of TJB) in a cohort of HCV-infected liver transplant recipients followed with serial biopsies. A secondary aim was to evaluate whether DIA of collagen content could ameliorate some of the problems associated with smaller liver biopsy samples.

Patients and Methods Between March 1999 and April 2008, we evaluated prospectively a consecutive cohort of 115 patients who had one or more routine TJBs combined with HVPG measurement (n ⫽ 250); all posttransplantation patients with HCV had solely TJBs. This cohort is similar to the first 131 patients who underwent transplantation for HCV cirrhosis (from October 1988 to February 1999) who did not have HVPG measurement, except that donors were older in the cohort under study. They were followed to their last biopsy for a median period of 48 months (range, 9-168 months). All patients had undergone transplantation for HCV cirrhosis, had received a cadaveric graft, and had been evaluated for at least 6 months posttransplantation. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki. All patients gave written informed consent for both the procedure and the histological evaluation. Repeated HVPG measurements and TJBs were made in 61 patients, usually repeated at 12-month to15-month intervals from transplantation, unless a biopsy at another time was clinically indicated. TJB and Hemodynamic Study. All procedures were performed in the X-ray suite by experienced personnel (D.P., J.O’B., D.Y.) after a 6-hour fast, under local anesthesia (lignocaine 1%, 5-10 mL subcutaneously); 70% of patients received intravenous sedation with 5 to 10 mg diazepam (Diazemulus; Dumex). The procedure has been detailed elsewhere,27 using a 19-gauge Tru-cut type biopsy needle (Quick core; Cook, William Cook Europe, Denmark). We performed three or, more recently, four passes21 through the same hepatic vein wall (right or middle) to optimize sampling of liver tissue.21,27 Hepatic vein pressures were measured using a 5-F balloon catheter (Royal Flush plus Straight Visceral Angiographic Beacon tip catheter-Cook-William Cook Europe) using the technique recently described.28 Three sets of measurements were taken using monitor Datascope Medical 2000A 6189-L7, setting the external 0 point at the mid-axillary line. A difference of 1 mm Hg or less of the free hepatic

HEPATOLOGY, April 2009

pressure compared with the inferior vena cava pressure was considered acceptable. Wedged hepatic vein pressure was measured for at least 1 minute each time. The trace was examined to ensure “occlusion.” If there was patient movement or breathing artifacts, measurements were repeated. HVPG (the difference between wedged hepatic vein pressure and free hepatic pressure) was calculated as the mean of the three measurements. Biopsy Specimen Study. Liver biopsy samples were formalin fixed, paraffin-embedded, and stained with hematoxylin-eosin, Gordon and Sweet staining for reticulin, and chromotrope aniline blue. Another tissue section was stained with picroSirius red for collagen quantification and determination of CPA by DIA. Each biopsy sample was evaluated histologically according to Ishak et al.29 for disease stage and grade of necroinflammatory activity. For each biopsy, the total length (lengths of each fragment summed) and total number of portal tracts were recorded.27 We excluded liver biopsies less than 12 mm long. Portal tracts were defined according to Crawford et al.,30 that is, “focus of connective tissue containing at least two luminal structures (either bile duct, portal vein, or hepatic artery).” Portal tracts were not counted in biopsies with severe distortion of the liver architecture, such as in cirrhosis or marked nodularity, because recognition and enumeration of individual portal tracts in these circumstances is impossible. The sections of each biopsy stained with picroSirius red were used for DIA, which was performed by one author (V.C.). The equipment setup used consisted of a digital camera (Canon Powershot A640 attached to a close-up copystand with backlighting) connected to a compatible personal computer. Calibration of camera setup was: 154 ⫻ 154 pixels ⫽ 23,716 ⫽ 1 mm2. After whole section digital image capture, CPA was measured with Zeiss KS300 image analysis software. The CPA measurement included editing steps to eliminate image artifacts and structural collagen in large portal tracts and blood vessel walls (which do not represent disease-related liver fibrosis). Unfilled natural spaces such as vascular cavities were not included in the measurements and, because noncollagenous cellular areas such as lymphoid aggregates in portal tracts are not stained (red) with picroSirius red, these also were not included. The software allows slider adjustment to view the output overlay at different detection thresholds for binary (grayscale); or color (R, G, B). This permits optimization of the detection overlay by direct visual comparison with the original picroSirius red image (Figs. 1 and 2). We assessed intraobserver variability by repeating the CPA assessment using 20 biopsies (10% of cohort) selected at random by A.P.D. for a second evaluation by

HEPATOLOGY, Vol. 49, No. 4, 2009

CALVARUSO, BURROUGHS, ET AL.

1239

Logistic regression analysis was used to determine independent associations with the HVPG cutoffs of at least 6 mm Hg and at least 10 mm Hg representing the presence of portal hypertension28 and clinically significant portal hypertension, respectively.10 We also evaluated area under the receiver operating curve (AUROC) for the CPA in relation to HVPG cutoffs. The best cutoff of the curve (Youden index) was determined by the software program.

Results

Fig. 1. PicroSirius red stained transjugular liver biopsy photographed with a Canon “Powershot” A640 digital camera (macro lens setting). Collagen is stained red, and the rest of the tissue is yellow. Part of the wall of a large blood vessel (black arrow), and a septal portal tract (blue arrow) is seen. The lumens of blood vessels in the septal portal tract are not stained. Cellular, noncollagenous components of portal tracts are yellow (for example, lymphoid aggregates, green arrows).

V.C., and interobserver error by using a different observer (F.G.) unaware of V.C.’s assessments, using a separate randomly selected group of 20 biopsies. Statistical Analysis. All data were analyzed using the statistical package SPSS (version 15.0; SPSS Inc, Chicago, IL) and MedCalc statistical software(version 9.3.3). Correlation between variables was evaluated by Spearman correlation. Significance testing was two-sided, and type 1 error rates were set to 0.05. Comparisons between unpaired samples for HVPG and CPA were made using the Mann-Whitney test, whereas comparisons for HVPG and CPA were made between first and last (paired) biopsies in the same patient using the Wilcoxon test. Intraobserver and interobserver variability was assessed by calculation of the concordance co-efficient.

We excluded six biopsy specimens because they were 12 mm long or shorter. Thus, we evaluated 244 procedures in 115 patients (85 men, 30 women; median age, 52 years; range, 33-66). Sixty-one patients had repeated biopsies with a median of 25.5 months between the first and last (range, 8-76); baseline clinical characteristics were similar to those of the cohort who only had a single biopsy. The median length of the liver biopsies was 22 mm (mean, 22.03 mm; range, 12-46); 199 (80%) of the liver biopsies were 15 mm length or longer, and 142(57%) were 20 mm in length. The median number of portal tracts was 12 (range, 4-27). In 64 biopsies (22%) from 38 patients, the portal tracts were not counted because of cirrhosis or severe nodularity. At the first biopsy, 81 patients had Ishak stage score of 3 or less, and 34 patients had Ishak stage 4 or above. In the entire cohort, the median HVPG was 4 mm Hg (range, 1-24). The median free hepatic venous pressure was 9 mm Hg (range, 1-21), and the occluded hepatic venous pressure was 14 mm Hg (range, 3-41). HVPG was at least 6 mm Hg in 51patients (36%; 89 procedures), with 36 (14%) patients at first measurement. In 21 patients (8%; 29 procedures), HVPG was 10 mm Hg or higher; with 11 (4%) patients at first measurement. Nineteen biopsy specimens were excluded from DIA evaluation because insufficient tissue was left in the tissue block to perform the picroSirius red stain. The remaining 225 liver biopsies were evaluated by DIA. The median CPA was 6% (range: 0.2%-45%). The median CPA analyzed either in biopsies of 15 mm in length or longer, or those of 20-mm length, was 5.7% in both groups. The concordance coefficients between intraobserver and interobserver evaluation were 0.98 and 0.97, respectively. Relationship Between CPA and Ishak Scoring. The CPA values increased with worsening Ishak stage as expected (Fig. 3). For significant fibrosis (Ishak stage ⬎ 2), the AUROC for CPA was 0.84 [95% confidence interval (CI), 0.780.88], and the best cutoff value was 6% CPA, with a 78% sensitivity and 80% specificity.

1240

CALVARUSO, BURROUGHS, ET AL.

HEPATOLOGY, April 2009

Fig. 2. (A) The liver biopsy tissue area has been defined by using KS300 image analysis software, using image grayscale detection (and operator-determined thresholding) by matching the tissue area overlay (green) against the original image of the picroSirius red preparation. Spaces (such as the lumens of blood vessels noted in Fig. 1) are not included in the tissue area to be measured, whereas lymphoid aggregates within portal tracts are included in the overall biopsy tissue area. (B) The collagenous regions of the biopsy specimen (green) have been defined with the image analysis software, using image RGB color detection (with operator-determined thresholding) by matching the collagen area overlay (green) against the original image of the picroSirius red preparation. Noncollagenous portal tract lymphoid aggregates are not included in the measurement of the liver biopsy collagen. The structural collagen (large blood vessel wall, and septal portal tract, which are unrelated to the disease-related liver fibrosis; Standish et al.2), and other tissue section preparation and histochemical staining technical artefacts are manually outlined and edited subsequently. Without editing, the collagen proportionate area (CPA) of this biopsy is 12.5%; after removal of the areas of the biopsy containing the large blood vessel wall and the septal portal tract, the CPA is 5.7%. With regard to these “irrelevant” regions, the picroSirius red–stained area is removed from the numerator, and the corresponding biopsy tissue area is removed from the denominator for the calculation of the edited CPA.

For severe fibrosis (Ishak stage 5 ⫹ 6), the AUROC for CPA was 0.90 (95% CI, 0.86-0.94), and the best cutoff value was 9% CPA with a 78% sensitivity and 88% specificity. To compare evaluations made in previous publications regarding the relationship between CPA and the Ishak stage score, we analyzed the correlation coefficient (despite this being incorrect methodologically because Ishak stage scores are categorical data and CPA is a continuous measurement): for first biopsy (n ⫽ 115) r ⫽ 0.67, P ⬍

0.001, and for total cohort (n ⫽ 244) r ⫽ 0.72, P ⬍ 0.001. The median of CPA in Ishak stage 6 was 17%, with a range of 13% to 45%. Relationship Between HVPG, CPA, and Ishak Scoring. We subdivided Ishak fibrosis stages (0, 1, 2, 3 versus 4, 5, 6) as previously published.14 The differences between the medians of HVPG and CPA were found to be statistically significant: HVPG: 4 mm Hg versus 8 mm Hg, P ⬍ 0.001 and CPA: 5% versus 16.2%, P ⬍ 0.001.

HEPATOLOGY, Vol. 49, No. 4, 2009

CALVARUSO, BURROUGHS, ET AL.

1241

Fig. 4. Correlation between HVPG and Ishak stage score in the entire biopsy/patient cohort (n ⫽ 244).

Fig. 3. CPA and HVPG values according to the Ishak stage score in 225 liver biopsies from 115 patients who underwent transplantation for HCV-related cirrhosis.

When subdivided into three categories (0, 1, 2 versus 3, 4 versus 5, 6), the differences between the medians of HVPG and CPA were also statistically significant: HVPG mm Hg: 3 versus 5 versus 8; P ⬍ 0.001, CPA (%) 3.8 versus 6.5 versus 16; P ⬍ 0.001); Overall, we found a similar relationship between HVPG and Ishak stage score (r ⫽ 0.605; P ⬍ 0.001) and Ishak grade score (r ⫽ 0.47; P ⬍ 0.001) as previously documented14 (Fig. 4). Relationship Between CPA and HVPG. We assessed the correlation between CPA and HVPG in the first biopsy for each patient: r ⫽ 0.55, P ⬍ 0.001 (115 patients: 64 ⫽ stages 0, 1, 2; 28 ⫽ stages 3, 4; 23 ⫽ stages 5, 6). Similarly, considering all 225 procedures, we found that CPA correlated significantly with HVPG: r ⫽ 0.61; P ⬍ 0.001 (110 ⫽ stages 0, 1, 2; 51 ⫽ stages 3, 4; 64 ⫽ stages 5, 6) (Fig. 5). Median CPA was significantly different between patients with HVPG of at lease 6 mm Hg or less than 6 mm Hg (4.1% versus 13.8%; P ⬍ 0.0001) and between patients with HVPG of at least 10 mm Hg or less than 10 mm Hg (6% versus 17.3%; P ⬍ 0.0001) (Fig. 6). Considering only biopsies of at least 15 mm length or at least 20 mm length, the correlation coefficients of CPA with HVPG were r ⫽ 0.58 (P ⬍ 0.001) and r ⫽ 0.62 (P ⬍ 0.001), respectively (in other words, similar to the total cohort).

Using the best cutoff values of CPA from the AUROC described previously, for Ishak stage score greater than 2 (CPA 6%) and for Ishak stage 5 ⫹ 6 (CPA 9%), the median HVPG was significantly different between patients with CPA of at least 6% or less than 6% (3 mm Hg versus 7 mm Hg; P ⬍ 0.001), and similarly the median HVPG was significantly different between patients with CPA of at least 9% or less than 9% (4 mm Hg versus 8 mm Hg; P ⬍ 0.001). For portal hypertension (HVPG ⱖ 6 mm Hg), the AUROC for CPA was 0.85 (95% CI: 0.79-0.89) and the best cutoff value for CPA was 7.2%, with 80% sensitivity and 79% specificity.

Fig. 5. Correlation between CPA and HVPG in 225 procedures (TJB and HVPG measurement) according to the number of biopsies in each patient (115) with recurrent HCV hepatitis after transplantation.

1242

CALVARUSO, BURROUGHS, ET AL.

HEPATOLOGY, April 2009

Fig. 6. Collagen proportionate area values in 225 liver biopsy specimens from 115 patients with recurrent HCV hepatitis after liver transplantation related to the presence of portal hypertension (HVPG ⱖ 6 mm Hg) or its absence (HVPG ⬍ 6 mm Hg) and to the presence of clinically significant portal hypertension (HVPG ⱖ 10 mm Hg) or its absence (HVPG ⬍ 10 mm Hg).

For clinically significant portal hypertension (HVPG ⱖ 10 mm Hg), the AUROC for CPA was 0.87 (95% CI: 0.81-0.91), and the best cutoff value for CPA was 12.5%, with a 78% sensitivity and 86% specificity. Logistic Regression Analysis. We evaluated whether the histological scores, Ishak grade, Ishak stage, and CPA were independently associated with the diagnostically important cutoff values of HVPG, that is, portal hypertension of at least 6 mm Hg and clinically significant portal hypertension of at least 10 mm Hg. In the univariate analysis, both Ishak grading and staging score, as well CPA, were statistically associated with HVPG of at least 6 or at least 10 mm Hg, but in the multivariate analysis only CPA was significantly associated with HVPG of at least 6 mm Hg (odds ratio, 1.206; 95% CI: 1.094-1.331; P ⬍ 0.001) or HVPG of at least 10 mm Hg (odds ratio, 1.105; 95% CI, 1.026-1.191; P ⫽ 0.009). Data with odds ratios, confidence intervals and P values are shown in Table 1. Patients with Repeated Biopsies. As a cohort, the patients acquired more collagen with time. Among the 61 patients who had repeated biopsies, 22 had two and 39 had three or more. The mean HVPG and CPA were sig-

nificantly different between the first and the last biopsy: 4.7 mm Hg versus 6.2 mm Hg; P ⫽ 0.005, and CPA 6.5% versus 10.6%, P ⫽ 0.003. As expected, the CPA value remained the same or increased between the first and the last determination in almost all patients. However, in nine patients, we found that the amount of collagen decreased over time. Only 2 of these 10 patients also had a reduction of HVPG (⫺3 mm Hg for both), five patients had the same HVPG, and in two the HVPG increased by 3 and 4 mm Hg. Among these patients with repeated biopsies, only five had antiviral treatment between the first and the last biopsy, but only one achieved sustained virological response, and in this patient there was no difference in the amount of collagen between the two specimens. In the other four patients, the CPA remained the same or increased. Thirty-eight patients had three biopsies at similar intervals of 15, 30, and 45 months after LT. The mean HVPG only changed by 1 mm Hg over 45 months, from 5 to 6 mm Hg, that is, 20%, the CPA changed by 3.6 units (from 7.3 to 10.9) that is, 50%, suggesting that at low HVPG values CPA may detect progression of liver disease more effectively.

Table 1. Histological Associations with HVPG > 6 mm Hg and HVPG > 10 mm Hg (CSPH) in 225 Procedures Performed in 115 Patients with Recurrent HCV Infection After Liver Transplantation Univariate Analysis OR (95% CI)

Predictor of HVPG > 6 mm Hg Ishak grading score Ishak staging score Collagen proportionate area (%) Predictor HVPG > 10 mm Hg Ishak grading score Ishak staging score Collagen proportionate area (%)

Multivariate Analysis P Value

OR (95% CI)

P Value

⬍0.0001 ⬍0.0001 ⬍0.0001

1.214 (0.940-1.567) 1.372 (0.979-1.923) 1.206 (1.094-1.331)

0.138 0.067