The inflammatory phenotype of the fibrous plate is distinct from the liver and correlates with clinical outcome in biliary atresia

Pathology – Research and Practice 211 (2015) 252–260

Contents lists available at ScienceDirect

Pathology – Research and Practice journal homepage: www.elsevier.com/locate/prp

Original Article

The inflammatory phenotype of the fibrous plate is distinct from the liver and correlates with clinical outcome in biliary atresia Nicoleta C. Arva a , Pierre A. Russo a,b , Jessi Erlichman c , Wayne W. Hancock a,b , Barbara A. Haber d , Tricia R. Bhatti a,∗ a

Department of Pathology and Laboratory Medicine, The Children’s Hospital of Philadelphia, Philadelphia, PA, USA Perelman School of Medicine, The University of Pennsylvania, Philadelphia, PA, USA c Department of Pediatrics, Division of Gastroenterology, Hepatology and Nutrition, The Children’s Hospital of Philadelphia, Philadelphia, PA, USA d Merck Sharp and Dohme Corporation, North Wales, PA, USA b

a r t i c l e

i n f o

Article history: Received 15 July 2014 Received in revised form 18 November 2014 Accepted 5 December 2014 Keywords: Cholangiopathy Kasai procedure Transplant-free interval Digital image analysis

a b s t r a c t Biliary atresia is an inflammatory cholangiopathy of still undetermined etiology. Correlations between histologic findings and clinical outcome in this disease have largely been based on evaluation of liver parenchyma. This study aimed to characterize the pattern of inflammation within the biliary remnant and identify associations between the type and degree of inflammation and clinical outcome as reflected by the transplant-free interval. The inflammation within the fibrous plates and livers of 41 patients with biliary atresia was characterized using immunohistochemical markers and the cell populations were digitally quantified. The type and quantity of cells within the infiltrate were then correlated with length of time from Kasai portoenterostomy until transplant. Histologic and immunohistochemical features of the biliary remnant allowed stratification of patients into “inflammatory plate” and “fibrotic plate” groups. Overall there was no significant difference in transplant-free interval between the two cohorts; however, there was a trend towards a longer time to transplant among patients in the “fibrotic plate” group. In addition, the composition of the inflammatory infiltrate in the fibrous plate was distinctly different from that present in the liver and only the characteristics of the inflammation in the fibrous plate, in particular the number of Foxp3+ T regulatory lymphocytes correlated with clinical outcome. The results of this study support the view of the extra-hepatic biliary tree as the primary site of injury in BA with the changes seen in the liver as secondary manifestations of outflow obstruction. The association between specific inflammatory cell subtypes within the fibrous plate and the length of transplant-free interval also supports the role of the immune system in the initial process of bile duct damage in biliary atresia. © 2014 Elsevier GmbH. All rights reserved.

Introduction Biliary atresia (BA) is an inflammatory cholangiopathy of infancy. It is the most common cause of end-stage liver disease in infants and is also the main indication for liver transplantation in children. The incidence of BA is estimated at between 1/5,000 and 1/18,000 live births, being more frequent among non-white

Abbreviations: BA, biliary atresia; Treg, T regulatory lymphocytes. ∗ Correspondence to: The Children’s Hospital of Philadelphia Department of Pathology of Pathology and Laboratory Medicine 324 South 34th Street, Main Building 5NW26 Philadelphia, PA 19104, Pennsylvania. Tel.: +1 215 590 1728; fax: +1 215 590 1736. E-mail address: [email protected] (T.R. Bhatti). http://dx.doi.org/10.1016/j.prp.2014.12.003 0344-0338/© 2014 Elsevier GmbH. All rights reserved.

patients and in girls [1]. Clinically, two types of BA are described: a perinatal form (∼80% of cases) that manifests as jaundice in the first post-natal weeks in initially asymptomatic children and an embryonic form (∼20% of cases). The embryonic form occurs without a jaundice-free interval and is associated with other congenital anomalies including those involving the cardiovascular system (interruption of inferior vena cavae with azygos continuation, pre-duodenal portal vein, hypoplastic left heart syndrome, atrial/ventricular septal defects), digestive system (intestinal malrotation), lungs (bronchial anomalies) or situs anomalies (asplenia or polysplenia) [2,3]. The etiology of BA remains uncertain, although multiple causes have been proposed involving immunologic [4,5], genetic [6,7], vascular [8] and infectious agents [9–11] resulting in a fibro-obliterative process acting upon and destroying a normally

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

developing biliary duct system [12] rather than a primary developmental failure. BA is suspected clinically in neonates with jaundice beyond 14 days of life with elevated conjugated bilirubin, high gammaglutamyl transpeptidase and alkaline phosphatase. Liver biopsy is performed to confirm the diagnosis with histology demonstrating varying degrees of portal tract fibrosis, edema, bile duct proliferation and cholestasis with intracanalicular bile plug formation [13]. Early diagnosis is critical as prompt surgical intervention, typically consisting of Kasai portoenterostomy, directly correlates with longterm prognosis. The procedure involves removal of the fibrotic segment of extra-hepatic bile duct with anastomosis of a jejunal Roux loop to the cut surface of the porta hepatis with the goal of restoring adequate bile flow; however, most patients eventually require liver transplantation [14], particularly in cases with poor early response to Kasai, failure to thrive, late-onset (adolescent) cholestasis, recurrent cholangitis or portal hypertensive bleeding [15]. Only a few studies, concentrating mostly on histopathologic changes in the liver, have attempted to determine relationships between microscopic features and clinical course in patients with BA [16–19]. Further characterization of the changes within the extra-hepatic biliary tree represented in the fibrous plate, the actual target of the fibro-inflammatory process, is needed to increase our understanding of the etiology and clinical course in this disease. This study analyzed the composition of the inflammatory infiltrate of the fibrous plates and livers of BA patients at the time of Kasai portoenterostomy to determine if any associations exist between the pattern of inflammation and clinical outcome as demonstrated by transplant-free interval or bilirubin level. The data obtained revealed differences in the type and degree of inflammation between the biliary remnant and corresponding hepatic parenchyma and uncovered additional factors that correlated with time until liver transplantation following Kasai portoenterostomy in a subset of patients.

Materials and methods Subjects: Information on 41 BA patients enrolled in this study (1/1/1981 – 12/31/2008) was collected under the approval of The Children’s Hospital of Philadelphia Institutional Review Board. In three instances, BA occurred in association with other congenital anomalies: one case of heterotaxy, one case with an unspecified fatty acid oxidation disorder and another with global developmental delay. All 41 patients underwent Kasai portoenterostomy and had slides of the biliary remnant (fibrous plate) for review. In 34 cases, a corresponding liver biopsy obtained at the time of Kasai or shortly before surgery was also available for review. All patients subsequently had liver transplantation after varying lengths of time. The minimal follow-up period after transplantation was 4 years. Histology and immunohistochemistry: Biliary remnants obtained at the time of Kasai were oriented and serially sectioned during macroscopic examination with alternative sections submitted for microscopy. The tissue from the fibrous plates and livers were then routinely processed, embedded in paraffin, sectioned at 4 microns and stained with hematoxylin and eosin (H&E) for light microscopy. All cases were reviewed to confirm the histologic impressions at the time of diagnosis and the most representative section from each case was stained with primary antibodies directed against T lymphocytes (CD4, Leica, 1:100 and CD8, Vector, 1:100); B lymphocytes and plasma cells (CD79A, DAKO 1:100), natural killer (NK) cells (CD56, Cell Marque, 1:400), macrophages (CD163, Vector 1:100) and T-regulatory (Treg) cells (Foxp3, Biolegend, 1:200). Antigen retrieval was performed for

253

20 min with either epitope retrieval 2 reagent (ER2, Leica, for CD4, Foxp3, CD8, CD79A, CD163) or epitope retrieval 1 reagent (ER1, Leica, for CD56). Visualization was obtained with BOND Polymer Refine Detection System and DAB Enhancer. Due to tissue exhaustion, Foxp3 staining was performed on 38 (out of 41) biliary remnant and 29 (out of 34) liver specimens. Image acquisition and digital image analysis: Immunostained slides of the fibrous plates and livers were scanned using the Aperio ScanScope® CS slide scanner (Aperio Technologies, Vista, CA) and the entire digitized sections were analyzed using the Aperio ImageScope software (version 10.0.1346.1807; Aperio Technologies, Vista, CA; http://www.aperio.com/download.asp). For liver sections, all portal tracts available were thus analyzed. Output of the scan analysis consisted of the raw numbers and also the percentage of all cells with membranous (CD4, CD8, CD79A, CD56, CD163) or nuclear (Foxp3) positivity from the total number of cells present on the slide (including fibroblasts, bile duct epithelium, endothelial cells and hepatocytes). The sum of all positive cells was calculated and designated as “percentage of total inflammatory cells”. In some cases, the percentage of total inflammatory cells was greater than 100%, which could be explained by the differences in detection/intensity thresholds used for each antibody as well as non-specific staining of non-inflammatory cells (for example focal CD56 reactivity of bile duct epithelium). However, these cell types were present in all cases, their number was low and it was not felt to contribute to significant differences between the cases. In some cases, the Foxp3+:CD4+ cells ratio was greater than 1, which again can be explained by the differences in the detection threshold between the antibodies. One-third of all immunostained slides for each cell subtype were also reviewed by two pathologists (NA, TRB) blinded to clinical features and digital image analysis results to insure accuracy of the digital system. No significant differences in the quantitation of positive cells using the digitized versus manual method were noted. Statistical analysis: Data generated by digital image analysis had a non-parametric distribution. Spearman correlation coefficient was used to determine the association between the percentage of total inflammatory cells or specific types of inflammatory cells and transplant-free interval (defined as the difference, in days, between the date of transplantation and date of portoenterostomy) or age at Kasai. The Mann – Whitney U test was used to assess the differences in the inflammatory infiltrate composition between the fibrous plate and the liver. The probability of early or late transplant based on the amount of fibrous plate inflammation was determined by Fisher exact probability test. The bilirubin level (<2 or >2 mg/dl at 90 days post Kasai) was also compared between patients with different types of biliary remnant histology. A p value
Results Histologic and immunohistochemical analysis of the biliary remnant allows stratification of patients into “inflammatory plate” and “fibrotic plate” groups. Routine microscopic examination of H&E stained sections of the biliary remnant revealed two primary histologic patterns. One group was characterized by robust inflammation infiltrating around a few residual bile ducts lined by degenerated epithelium (Fig. 1A and B). A second group demonstrated a sparse inflammatory infiltrate and scattered remaining bile ducts of smaller caliber in a more abundant fibrotic background (Fig. 1C and D). The impression was confirmed on review of immunostained slides which also revealed a mixed inflammatory infiltrate composed of mature lymphocytes, macrophages and plasma cells, although the infiltrate was sparse in the latter group with abundant fibrotic stroma (Fig. 1E). Based on quantification

254

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

Fig. 1. Microscopic analysis of sections from fibrous plate obtained at Kasai procedure reveals two histologic patterns. The biliary remnants of both “inflammatory plate” and “fibrotic plate” groups are characterized by mixed inflammatory infiltrates, but the quantity is different. (1A) and (1B) Microscopic field of the biliary remnant from a patient in the “inflammatory plate” group showing fibrotic stroma with robust inflammation and a few residual bile ducts lined by degenerated epithelium ((1A) Hematoxylin–eosin stained section, 200× magnification; (1B) Masson-trichrome stained section, 200× magnification). Whole slide digital image analysis for this particular patient quantified the inflammatory cells as representing 99% from total cells present on the slide; (1C) and (1D) Microscopic field of the biliary remnant from a patient in the “fibrotic plate” group, illustrating sparse inflammatory infiltrate and scattered residual bile ducts in a more abundant fibrotic background. ((1C) Hematoxylin-eosin stained section, 200× magnification; (1D) Masson-trichrome stained section, 200× magnification). Whole slide digital image analysis for this particular patient quantified the inflammatory cells as representing 10% from total cells present on the slide. (1E) Immunostained sections for CD4, Foxp3, CD8 and CD163 epitopes from the extrahepatic biliary duct remnants obtained at Kasai procedure (200× magnification) reveal a mixed inflammatory infiltrate, although the infiltrate was sparse in the cases with abundant fibrotic stroma (slides are from the same cases as Fig. 1A and 1C respectively).

of the percentage of inflammatory cells by digital image analysis, cases were then divided into two groups: an “inflammatory plate” group (n = 20) in which more than 50% of the total cell population were inflammatory cells (ranging from 51.4% to 100%, with an average of 70% and median of 63%) and a “fibrotic plate” group (n = 21) characterized by inflammatory cells contributing to less than half of the total cells present in the section (ranging from 7.5% to 49.8%,

with an average of 33.6% and median of 36.5%). The Mann – Whitney U test revealed that the two groups were statistically different with regards to the percentage of total inflammatory cells, CD4+, CD8+, and CD163+ cells, and had a similar composition of Foxp3+, CD79A+ and CD56+ cells (Supplemental Table 1). The “inflammatory plate” and “fibrotic plate” groups have similar transplant-free intervals. Overall the average transplant-free

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

255

Fig. 2. “Inflammatory plate” and “fibrotic plate” groups do not have different clinical outcome. (2A) The average age at Kasai and transplant-free interval (in days) are similar between the entire cohort of 41 patients, the “inflammatory plate” and “fibrotic plate” groups analyzed separately. (2B) The 34 patients with early need for transplant are almost equally distributed between the “inflammatory plate” and “fibrotic plate” subgroups; from the seven patients with late transplant, the majority (5) shows “fibrotic plate” morphology, whereas only two belong to the “inflammatory plate” group.

interval was 645.8 days (ranging from 126 to 6497 days). When analyzed separately, the “inflammatory plate” group had an average transplant-free interval of 559.8 days (ranging from 126 to 3917 days) versus 727.8 days (ranging from 151 to 6497 days) in the “fibrotic plate” group, but the difference was not statistically significant (Mann – Whitney test: p = 0.16) (Fig. 2A). The entire cohort of patients was also divided into quartiles based on the amount of inflammatory cells in the biliary remnant. Segregated in this manner, five patients had between 0% and 25% inflammatory cells, 16 patients had between 25% and 50% inflammatory cells, 13 patients had between 50% and 75% inflammatory cells and seven patients had between 75% and 100% inflammatory cells. No statistically significant difference in the length of transplant-free interval was identified between the five patients whose fibrous plates contained the least amount of inflammation compared to the seven patients with the most severely inflamed biliary remnants (Mann – Whitney test: p = 0.62). Seven patients underwent liver transplantation after an interval of time greater than the average transplant-free interval for the entire cohort of patients (645.8 days). Five of the seven patients, belonged to the “fibrotic plate” group and two had “inflammatory plate” morphology. For the remaining cases (n = 34) in which the transplant occurred earlier, the distribution of patients was similar between the two groups (16 patients were in the “fibrotic plate” group and 18 were in the “inflammatory plate” group) (Fig. 2B). Fisher exact test did not show a statistically significant difference in predicting the length of interval to transplant between “fibrotic plate” versus “inflammatory plate” morphology (p = 0.40). No significant difference was observed between the two groups in regards to bilirubin level: only three patients demonstrated a bilirubin level below 2 mg/dl at 90 days post Kasai, two from “inflammatory plate” and one from “fibrotic plate” group (Fisher exact test: p = 0.60). The average age at Kasai was 63.8 days (ranging from 12 to 169 days) for the entire cohort of patients. In the “inflammatory plate” group the average age at Kasai was 67.2 days (ranging from 36 to 169 days) and was not statistically different from the average age at Kasai in the “fibrotic plate” group (60.6 days, ranging from 12

to 154 days) (Mann – Whitney test: p = 0.26) (Fig. 2A). Spearman correlation analysis between the age at Kasai and transplant-free interval performed for all patients, and also separately for the “inflammatory plate” and “fibrotic plate” groups, revealed no associations between these two clinical features (r = −0.10, p = 0.50 for all patients; r = −0.33, p = 0.15 for “inflammatory plate” group; r = 0.07, p = 0.73 for “fibrotic plate” group). In addition, the amount of plate inflammation as reflected by the percentage of total inflammatory cells did not correlate with the age at Kasai or transplant-free interval (Spearman correlation analysis, r = 0.2 and p = 0.2 for age at Kasai; r = −0.13 and p = 0.38 for transplant-free interval). African – American patients had a longer transplant-free interval following Kasai procedure, whereas gender does not influence clinical outcome. Of the 41 patients in our study, 13 were identified as African – American and 28 as Caucasian. No significant differences were found in the age at Kasai among African – American patients (average 62.7 days; range: 12 – 154 days) compared to Caucasian patients (average 64.4 days; range: 16 – 169 days) (Mann – Whitney test; p = 0.34) (Supplemental Fig. 1A). However, the transplant-free interval was on average 1025.1 days in African – American patients (range: 203 – 6497 days) which was significantly longer than in Caucasians who had an average of 469.8 days from Kasai until liver transplant (range: 126 – 3917 days) (Mann – Whitney test: p = 0.016) (Supplemental Fig. 1A). This study included 24 female and 17 male patients. The average age at Kasai in boys was 63.7 days (from 12 to 154 days) and was not statistically different from the age of Kasai in girls: average 63.9 days (from 16 to 169 days) (Mann – Whitney test: p = 0.46) (Supplemental Fig. 1B). Boys had an average transplant-free interval of 905.1 days (ranging from 135 to 6497 days) whereas in girls the transplant was performed, on average, 462.2 days after Kasai procedure (ranging from 126 to 2122 days). However, the difference was again not statistically significant (Mann – Whitney test: p = 0.40) (Supplemental Fig. 1B). The quantity of inflammation in the biliary remnant does not correlate with and is not predictive of the inflammation in the hepatic parenchyma. For the majority of the 34 cases with

256

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

Fig. 3. The histology of hepatic parenchyma is similar in patients from “inflammatory plate” and “fibrotic plate” groups. Hematoxylin–eosin stained sections (200× magnification) of liver parenchyma from a patient in the “inflammatory plate” (3A) and “fibrotic plate” (3B) group showing comparable features, with prominent bile duct proliferation, hepatocytic and intracanalicular cholestasis with bile plugs and enlargement of the portal tracts by edema and mixed inflammatory infiltrate (the corresponding biliary remnant for these patients is illustrated in Fig. 1A and B). (3C) Scattered dot-plot showing the percentage of total inflammatory cells in the biliary remnant section vs. corresponding liver parenchyma for each individual patient (n = 34), demonstrating no correlation between the amount on inflammation in the biliary remnant and liver (p value, Spearman correlation analysis).

an accompanying liver biopsy available for review, hepatic tissue was obtained at the time of Kasai; the remaining had biopsies obtained a few days prior to surgery. Histologic examination of the hepatic parenchyma in all 34 cases revealed typical features of extra-hepatic obstruction, including prominent bile duct proliferation, hepatocellular and intracanalicular cholestasis with bile plugs, and enlargement of portal tracts by edema and a mixed inflammatory infiltrate. Marked portal fibrosis was not present in any of the examined liver biopsies. No significant differences in liver histology were seen between patients in the “inflammatory plate” group (n = 18 patients) versus the “fibrotic plate” group (n = 16 patients) (Fig. 3A and B). To confirm this impression, image analysis of the liver sections was performed similar to that done for the biliary remnants. This process included whole slide analysis which allowed determination of an average quantity of portal inflammation despite the fact that the amount may have been variable between the portal areas within any individual case. Dot-plot comparison between the amount of fibrous plate and liver inflammation for each individual patient revealed an overall uniform quantity of hepatic inflammation (ranging from 25% to 65% total inflammatory cells) with no differences between livers from “inflammatory plate” and “fibrotic plate” patients (Fig. 3C). Using the same criteria applied to the fibrous plate (percentage of total inflammatory cells being more or less than 50%)

but irrespective of prior plate classification, the entire cohort of 34 cases was sub-divided into “inflammatory liver” and “noninflammatory liver” categories, based on the amount of portal inflammation. Within the “inflammatory liver” group (n = 13), eight had been previously classified as showing “inflammatory plate” histology and five had demonstrated “fibrotic plate” histology. The “non-inflammatory liver” group consisted of 21 cases, 10 with “inflammatory plate” histology and 11 from the “fibrotic plate” category (Table 1). Fisher exact probability test showed that the amount of inflammation in the biliary remnant was not predictive of the quantity of hepatic inflammation (p = 0.49). The composition of the inflammatory infiltrate in the biliary remnant is distinct from the inflammation in the hepatic parenchyma. Quantification of inflammatory cell subtypes by digital image analysis allowed us to compare the percentages of each inflammatory cell subtype between the biliary remnant and the liver and showed that the fibrous plates had statistically different percentages of Foxp3+, CD8+, CD79A+, CD56+ and CD163+ cells than the corresponding liver parenchyma with comparable percentages of CD4+ lymphocytes (Mann – Whitney test, Table 2). Only the composition of the inflammatory infiltrate in the fibrous plate but not in the liver is associated with clinical outcome. Based on the findings within the fibrous plates, longer transplant-free intervals correlated with a higher Foxp3+:CD4+ cell ratio in all patients (Spearman correlation: r = 0.40; p = 0.01)

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

257

Table 1 The amount of inflammation in the biliary remnant cannot predict the quantity of inflammation in the corresponding liver parenchyma. BILIARY REMNANT “Inflammatory plate” group “Fibrotic plate” group TOTAL

LIVER “Inflammatory liver” group

“Non-inflammatory liver” group

8 5 13

10 11 21

TOTAL 18 16 34

Eight patients in the “inflammatory plate” group had “inflammatory liver” histology, whereas 10 patients had “non-inflammatory liver” morphology. From the 16 patients with “fibrotic plate”, five showed “inflammatory liver” histology, whereas 11 had “non-inflammatory liver” morphology. Fisher exact probability test revealed that the amount of inflammation in the biliary remnant is not predictive of the quantity of liver inflammation.

Table 2 The composition of the inflammatory infiltrates within the biliary remnant and hepatic parenchyma are distinctly different.

Total % inflammatory cells CD4 Foxp3 CD8 CD79A CD56 CD163

Biliary remnant (average % for all cases from total number of cells on the section) (%)

Liver (average % for all cases from total number of cells on the section) (%)

p value

53.40 11.02 3.16 6.29 3.05 14.69 18.42

45.52 7.34 0.31 0.68 0.63 4.05 32.80

0.22 0.12 <0.001 <0.001 0.002 <0.001 <0.001

Comparison between average percentage (for all cases, relative to the total number of cells present on the section) of different types of inflammatory cells between the biliary remnants and liver parenchyma (n = 34; n = 29 for Foxp3) shows that the composition of the inflammatory infiltrates is different for all types of inflammatory cells, except the percentage of CD4+ T lymphocytes (p value, Mann – Whitney U test).

(Fig. 4A) and also in girls and African – American patients (Spearman correlation: r = 0.45, p = 0.03 and r = 0.69, p = 0.01 respectively; data not shown). When analyzed separately, the “inflammatory plate” group showed a statistically significant association between length of transplant-free interval and percentage of CD79A+ cells (Spearman correlation: r = 0.52; p = 0.002) (Fig. 4B). Interestingly, in African – American patients the transplant-free interval correlated positively with the percentage of Foxp3+ cells (Spearman correlation: r = 0.65, p = 0.01) and negatively with the percentage of CD4+ cells (Spearman correlation: r = −0.62; p = 0.02), CD8+ cells (Spearman correlation: r = −0.59; p = 0.03) and total number of inflammatory cells (Spearman correlation: r = −0.63; p = 0.02) (data not shown). In boys, the length of time from Kasai to transplant showed negative correlation with the percentage of CD163+ cells (Spearman correlation: r = −0.51, p = 0.03, data not shown).

Similar correlation analyses based on the composition of the hepatic inflammatory infiltrate and transplant-free interval showed no relationship between the percentage of any particular inflammatory cell subtype and time to transplant regardless of histologic group (“inflammatory” or “non-inflammatory” liver), patient’s age or gender. Seven patients underwent liver transplantation after an interval of time greater than the average transplant-free interval for the entire cohort of patients (645.8 days). The percentage of CD163+, Foxp3+ cells, as well as Foxp3+:CD4+ cell ratio in the biliary remnant were statistically significant different between patients with early (less than the average of 645 days) versus longer time until transplant (Mann – Whitney test: p = 0.05, p = 0.05 and p = 0.003, respectively): a longer than average time to transplant was associated with fewer CD163+ cells, more Foxp3+ cells and higher

Fig. 4. Only the composition of the inflammatory infiltrate in the fibrous plate correlates with transplant-free interval. (4A) Dot-plot showing the positive correlation between the transplant free interval and Foxp3+:CD4+ cells ratio in the biliary remnant for all patients (Spearman correlation; r = 0.40, p = 0.01). (4B) Dot-plot showing the positive correlation between the transplant free interval and percentage of CD79A+ cells in the biliary remnant for patients in the “inflammatory plate” subgroup (Spearman correlation; r = 0.52, p = 0.002).

258

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

Table 3 The composition of biliary remnant inflammation is different between the patients with early vs. late transplant. Biliary remnant

Early transplant

Late transplant

p value

Average % Foxp3+ cells Average Foxp3+:CD4+ cells ratio Average % CD163+ cells

2.10% 0.44 20.29%

7.99% 1.94 9.48%

0.05 0.003 0.05

Comparison between average percentage (for all cases, relative to the total number of cells present on the section) of Foxp3+ lymphocytes, Foxp3+:CD4+cells ratio and CD163+ macrophages between the patients with early (n = 34) vs. late (n = 7) need for transplant shows that the composition of plate inflammation is different between the two groups of patients (p value, Mann – Whitney U test).

Foxp3+:CD4+ cell ratios (Table 3). No difference in the composition of the hepatic portal inflammatory infiltrate was detected between patients with shorter than average versus longer than average transplant-free interval. Discussion Kasai portoenterostomy is the cornerstone of surgical treatment for BA. However, even with restoration of bile flow at the time of surgery, most patients develop end-stage liver disease with biliary-type cirrhosis and eventually require transplantation. Currently, prognostic indicators post-Kasai are largely based on clinical features, including resolution of jaundice, re-establishment of bile flow, development of cholangitis and serum bilirubin levels [14]. A few prior studies have addressed the relationship between histopathologic changes in the liver and clinical course in patients with biliary atresia. For example, it has been proposed that the size of the residual ducts at the portal plate can predict outcome with a bile duct size greater than 150 ␮m indicating successful postKasai bile drainage [16] and the presence of fewer than five small bile ducts (<100 ␮m) correlating with significantly lower 5-year survival [17]. Other histologic features of the liver that have been associated with clinical outcome include the presence of syncytial giant cells, lobular inflammation, cholangitis, focal and bridging necrosis which were found to be indicative of portoenterostomy failure [18], whereas in another analysis the stage of portal fibrosis was reported to predict transplant-free survival [19]. However, attempts to specifically correlate the inflammatory infiltrate in the liver and/or extrahepatic biliary tree and clinical course have yielded conflicting results, including the finding of an association between low numbers of macrophages and resolution of jaundice in one study [20] contrasted with a possible protective effect of macrophage accumulation in another [21]. Based on gene expression profiling of liver tissue but not histology, Moyer et al. [22] stratified patients into “inflammation” and “fibrosis” groups and found that subjects with a fibrotic pattern had a significantly shorter transplant-free interval. However, a similar characterization of BA patients based on either the molecular or histologic features of the fibrous plate has not previously been performed. The present study found histologic differences in the extrahepatic biliary remnant which allowed separation of patients into two cohorts: “inflammatory plate” and “fibrotic plate”. Both groups showed similar ages at the time of Kasai suggesting that the two groups did not simply represent different time points in the evolution of the disease but perhaps represented different models of disease progression. However, the hypothesis of a continuum in the disease pathogenesis, as suggested in mouse models of rotavirusinduced BA [23] cannot be completed excluded, since the exact timing of disease onset as well as the velocity of the morphologic changes is not known. Previous BA studies have correlated advanced liver fibrosis with a worse clinical course [19,22,24]. However, in our study focusing on the histology of the biliary remnant, no differences were found in the transplant-free interval between the “inflammatory plate’ and “fibrotic plate” groups. In fact, the data in the current study shows

that a fibrotic morphology of the biliary remnant is more likely to be associated with a longer transplant-free interval as reflected by the higher proportion of the “fibrotic plate” patients undergoing transplantation later than the study average of 645 days post-Kasai. However, the number of patients in our cohort was too small to reach statistical significance. Similar to the work of Moyer, it would be of future interest to perform molecular analysis specifically of the fibrous plate to determine if a molecular signature exists that is predictive of clinical course. It would also be interesting to compare the histologic features of the fibrous plate between patients that progress to end-stage disease requiring liver transplant after Kasai and those who survive without transplantation. Moyer et al. [22] also demonstrated that the need for transplantation was influenced by molecular groupings as a function of age at portoenterostomy. In the present analysis, the age at Kasai did not predict the need for transplant. This finding may also be explained by differences in methodology between the studies (histologic versus molecular) used to segregate the patients. The older age at diagnosis of patients in the present study may also reflect the population of patients referred to our institution including those whose initial biopsies were inconclusive or whose surgical management was felt likely to be more complicated. The finding that the “inflammatory plate” and “fibrotic plate” groups had similar hepatic histology and similar quantities of portal inflammation would support the view that the changes observed in the liver are secondary to bile duct obstruction and do not necessarily reflect the changes in the fibrous plate. It would also be of interest to compare the inflammatory infiltrates in the initial liver biopsies to that present in the liver explant to characterize changes that may be responsible for disease progression in some patients. In our study, hepatic inflammation did not predict outcome, as previously reported [20,21], which may be a reflection of different measurements used to define clinical course: transplantfree interval in our work versus clearance of jaundice and good bile flow in prior studies. Although the etiology of BA is still undetermined, an immune process has been frequently proposed as a causative factor [25–31]. It has also been hypothesized that the initiating injury may be a primary perinatal hepatobiliary viral infection with generation of secondary immune-mediated bile duct injury [32]. This process may involve the interplay of macrophages that can act as antigenpresenting cells, activated T lymphocytes which effect the cellular damage and Foxp3+ Tregs (CD4+ T-cells that are indispensable for maintenance of peripheral tolerance and prevention of autoimmune disease) that regulate the activity of the T-effector cells. In support of an immune-mediated process, our study confirmed the presence of a mixed inflammatory infiltrate in the fibrous plate although sparse in some cases. Similar percentages of CD4+ T-cells in the fibrous plate and liver were also detected suggesting that these cells might play a role in initiating and/or maintaining the immune response in BA. The number of Foxp3+ Tregs as a proportion of CD4+ cells was associated with a longer transplant-free interval in all patients as well as among female and African – American subgroups suggesting a protective effect of Tregs in these cases, findings in accordance with prior reports [33]. Post-translational modifications of Foxp3 through acetylation and

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260

methylation have also been shown to have significant impact on the functional activity of Tregs [34,35]. These epigenetic changes would not have been discernible within the current study which detected only the presence of total Foxp3 protein expression by immunostaining, but would be of interest in future studies to determine if these alterations by themselves or in association with other inflammatory markers correlate with outcome. We also found a positive relationship between the number of CD79A+ cells and time from portoenterostomy until transplant, albeit only in the “inflammatory plate” group. The role of Blymphocytes and plasma cells in BA has not yet been clearly defined, but the results of this study support a possible protective effect and raise the possibility of an antibody-mediated component to the inflammatory process. Also, the percentage of CD163+ cells inversely correlated with the transplant-free interval in boys; in addition, patients with longer than average transplant-free intervals had lower percentages of CD163+ cells. This data is consistent, in part, with prior studies showing an inverse relationship between the number of macrophages and outcome [20,28]. The fact that some of the correlations were influenced by patient ethnicity or gender may be a consequence of immune system differences between members of different ethnic and gender groups. In support of this finding, allelic dissimilarities and copy number variations of the immunoglobulin heavy-chain genes [36] as well as T-cell receptor repertoire differences based on genetic polymorphisms [37] have been found between African – Americans and Caucasians, suggesting ethnic differences in both humoral and cellular components of the immune system. Hormones have also been shown to influence immunity resulting in gender dissimilarities: under the influence of estrogen and progesterone females tend to mount higher innate and adaptive immune responses, which may be advantageous upon pathogen encounter, but also come at the expense of a higher risk for autoimmune diseases. In males, testosterone hormone leads to inhibition of T-cell activation, dampening of Th1 differentiation, increased T-cell apoptosis and also to inhibition of B-cell development and antibody production [38,39]. Conclusions This analysis found differences in the composition of the inflammatory infiltrate between the fibrous plate and the liver in BA. Only the composition of the inflammation in the biliary remnant correlated with transplant-free interval, with specific inflammatory cell subtypes (Tregs, B cells and macrophages) showing prognostic significance in the clinical course of this disease. Disclosures The authors have no conflict of interests to disclose. Acknowledgments We would like to thank Dr. Eduardo Ruchelli for insightful comments and suggestions. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.prp.2014.12.003. References [1] P.W. Yoon, J.S. Bresee, R.S. Olney, L.M. James, M.J. Khoury, Epidemiology of biliary atresia: a population-based study, Pediatrics 99 (3 (Mar)) (1997) 376–382.

259

[2] R. Carmi, C.A. Magee, C.A. Neill, F.M. Karrer, Extrahepatic biliary atresia and associated anomalies: etiologic heterogeneity suggested by distinctive patterns of associations, Am. J. Med. Genet. 45 (1993) 683–693. [3] T.R. Silveira, F.M. Salzano, E.R. Howard, A.P. Mowat, Congenital structural abnormalities in biliary atresia: evidence for etiopathogenic heterogeneity and therapeutic implications, Acta Paediatr. Scand. 80 (1991) 1192–1199. [4] B.R. Lu, C.L. Mack, Inflammation and biliary tract injury, Curr. Opin. Gastroenterol. 25 (3 (May)) (2009) 260–264. [5] K. Harada, Y. Nakanuma, Biliary innate immunity and cholangiopathy, Hepatol. Res. 37 (Suppl 3 (Oct)) (2007) S430–S437. [6] T. Kohsaka, Z.R. Yuan, S.X. Guo, M. Tagawa, A. Nakamura, M. Nakano, H. Kawasasaki, Y. Inomata, K. Tanaka, J. Miyauchi, The significance of human jagged 1 mutations detected in severe cases of extrahepatic biliary atresia, Hepatology 36 (4 Pt 1 (Oct)) (2002) 904–912. [7] R.N. Bamford, E. Roessler, R.D. Burdine, U. Saplako˘glu, J. dela Cruz, M. Splitt, J.A. Goodship, J. Towbin, P. Bowers, G.B. Ferrero, B. Marino, A.F. Schier, M.M. Shen, M. Muenke, B. Casey, Loss-of-function mutations in the EGF-CFC gene CFC1 are associated with human left-right laterality defects, Nat. Genet. 26 (3 (Nov)) (2000) 365–369. [8] L.K. Pickett, H.C. Briggs, Biliary obstruction secondary to hepatic vascular ligation in fetal sheep, J. Pediatr. Surg. 4 (1 (Feb)) (1969) 95–101, 17. [9] B. Bangaru, R. Morecki, J.H. Glaser, L.M. Gartner, M.S. Horwitz, Comparative studies of biliary atresia in the human newborn and reovirus-induced cholangitis in weanling mice, Lab. Invest. 43 (5 (Nov)) (1980) 456–462. [10] M. Riepenhoff-Talty, V. Gouvea, M.J. Evans, L. Svensson, E. Hoffenberg, R.J. Sokol, I. Uhnoo, S.J. Greenberg, K. Schäkel, G. Zhaori, J. Fitzgerald, S. Chong, M. elYousef, A. Nemeth, M. Brown, D. Piccoli, J. Hyams, D. Ruffin, T. Rossi, Detection of group C rotavirus in infants with extrahepatic biliary atresia, J. Infect. Dis. 174 (1 (Jul)) (1996) 8–15. [11] G.P. Jevon, J.E. Dimmick, Biliary atresia and cytomegalovirus infection: a DNA study, Pediatr. Dev. Pathol. 2 (1 (Jan–Feb)) (1999) 11–14. [12] M. Gautier, N. Eliot, Extrahepatic biliary atresia. Morphological study of 98 biliary remnants, Arch. Pathol. Lab. Med. 105 (8 (Aug)) (1981) 397–402. [13] J.L. Hartley, M. Davenport, D.A. Kelly, Biliary atresia, Lancet 374 (9702 (Nov 14)) (2009) 1704–1713 (Review). [14] B.E. Wildhaber, A.G. Coran, R.A. Drongowski, R.B. Hirschl, J.D. Geiger, J.L. Lelli, D.H. Teitelbaum, The kasai portoenterostomy for biliary atresia: a review of a 27-year experience with 81 patients, J. Pediatr. Surg. 38 (10 (Oct)) (2003) 1480–1485. [15] B.L. Shneider, G.V. Mazariegos, Biliary atresia: a transplant perspective, Liver Transpl. 13 (11 (Nov)) (2007) 1482–1495. [16] R.S. Chandra, R.P. Altman, Ductal remnants in extrahepatic biliary atresia: a histopathologic study with clinical correlation, J. Pediatr. 93 (2 (Aug)) (1978) 196–200. [17] C.E. Tan, M. Davenport, M. Driver, E.R. Howard, Does the morphology of the extrahepatic biliary remnants in biliary atresia influence survival? A review of 205 cases, J. Pediatr. Surg. 29 (11 (Nov)) (1994) 1459–1464. [18] K.S. Azarow, M.J. Phillips, A.D. Sandler, I. Hagerstrand, R.A. Superina, Biliary atresia: should all patients undergo a portoenterostomy? J. Pediatr. Surg. 32 (2 (Feb)) (1997) 168–172. [19] L. Pape, K. Olsson, C. Petersen, R. von Wasilewski, M. Melter, Prognostic value of computerized quantification of liver fibrosis in children with biliary atresia, Liver Transpl. 15 (8 (Aug)) (2009) 876–882 (18). [20] M. Davenport, C. Gonde, R. Redkar, G. Koukoulis, M. Tredger, G. Mieli-Vergani, B. Portmann, E.R. Howard, Immunohistochemistry of the liver and biliary tree in extrahepatic biliary atresia, J. Pediatr. Surg. 36 (7 (Jul)) (2001) 1017–1025. [21] M.A. Kotb, A. El Henawy, S. Talaat, M. Aziz, G.H. El Tagy, M.M. El Barbary, W. Mostafa, Immune-mediated liver injury: prognostic value of CD4+, CD8+, and CD68+ in infants with extrahepatic biliary atresia, J. Pediatr. Surg. 40 (8 (Aug)) (2005) 1252–1257. [22] K. Moyer, V. Kaimal, C. Pacheco, R. Mourya, H. Xu, P. Shivakumar, R. Chakraborty, M. Rao, J.C. Magee, K. Bove, B.J. Aronow, A.G. Jegga, J.A. Bezerra, Staging of biliary atresia at diagnosis by molecular profiling of the liver, Genome Med. 2 (5 (May 13)) (2010) 33. [23] P. Shivakumar, K.M. Campbell, G.E. Sabla, A. Miethke, G. Tiao, M.M. McNeal, R.L. Ward, J.A. Bezerra, Obstruction of extrahepatic bile ducts by lymphocytes is regulated by IFN-␥ in experimental biliary atresia, J. Clin. Invest. 114 (3 (August 1)) (2004) 322–329. [24] V.S. Weerasooriya, F.V. White, R.W. Shepherd, Hepatic fibrosis and survival in biliary atresia, J. Pediatr. 144 (1 (Jan)) (2004) 123–125. [25] C.L. Mack, R.M. Tucker, R.J. Sokol, F.M. Karrer, B.L. Kotzin, P.F. Whitington, S.D. Miller, Biliary atresia is associated with CD4+ Th1 cell-mediated portal tract inflammation, Pediatr. Res. 56 (1 (Jul)) (2004) 79–87. [26] M. Shinkai, T. Shinkai, P. Puri, M.D. Stringer, Elevated expression of IL2 is associated with increased infiltration of CD8+ T cells in biliary atresia, J. Pediatr. Surg. 41 (2 (Feb)) (2006) 300–305. [27] B. Narayanaswamy, C. Gonde, J.M. Tredger, M. Hussain, D. Vergani, M. Davenport, Serial circulating markers of inflammation in biliary atresia—evolution of the post-operative inflammatory process, Hepatology 46 (1 (Jul)) (2007) 180–187. [28] H. Kobayashi, P. Puri, D.S. O’Briain, R. Surana, T. Miyano, Hepatic overexpression of MHC class II antigens and macrophage-associated antigens (CD68) in patients with biliary atresia of poor prognosis, J. Pediatr. Surg. 32 (4 (Apr)) (1997) 590–593. [29] P. Shivakumar, G.E. Sabla, P. Whitington, C.A. Chougnet, J.A. Bezerra, Neonatal NK cells target the mouse duct epithelium via Nkg2d and drive tissue-specific

260

[30]

[31]

[32] [33]

[34]

N.C. Arva et al. / Pathology – Research and Practice 211 (2015) 252–260 injury in experimental biliary atresia, J. Clin. Invest. 119 (8 (Aug)) (2009) 2281–2290. C.L. Mack, R.M. Tucker, R.J. Sokol, B.L. Kotzin, Armed CD4+ Th1 effector cells and activated macrophages participate in bile duct injury in murine biliary atresia, Clin. Immunol. 115 (2 (May)) (2005) 200–209. J.A. Bezerra, G. Tiao, F.C. Ryckman, M. Alonso, G.E. Sabla, B. Shneider, R.J. Sokol, B.J. Aronow, Genetic induction of proinflammatory immunity in children with biliary atresia, Lancet 360 (9346 (Nov 23)) (2002) 1653–1659. C.L. Mack, The pathogenesis of biliary atresia: evidence for a virus-induced autoimmune disease, Semin. Liver Dis. 27 (3 (Aug)) (2007) 233–242. A.G. Miethke, V. Saxena, P. Shivakumar, G.E. Sabla, J. Simmons, C.A. Chougnet, Post-natal paucity of regulatory T cells and control of NK cell activation in experimental biliary atresia, J. Hepatol. 52 (5 (May)) (2010) 718–726. G. Lal, J.S. Bromberg, Epigenetic mechanisms of regulation of Foxp3 expression, Blood 114 (18 (Oct)) (2009) 3727–3735 (20).

[35] Y. Xiao, B. Li, Z. Zhou, W.W. Hancock, M.I. Greene, Histone acetyltransferase mediated regulation of Foxp3 acetylation and Treg function, Curr. Opin. Immunol. 22 (5 (Oct)) (2010) 583–591. [36] C.T. Watson, K.M. Steinberg, J. Huddleston, R.L. Warren, M. Malig, J. Schein, A.J. Willsey, J.B. Joy, J.K. Scott, T.A. Graves, R.K. Wilson, R.A. Holt, E.E. Eichler, F. Breden, Complete haplotype sequence of the human immunoglobulin heavychain variable, diversity, and joining genes and characterization of allelic and copy-number variation, Am. J. Hum. Genet. 92 (4 (Apr 4)) (2013) 530–546. [37] J.D. Inocencio, E. Choi, D.N. Glass, R. Hirsch, T cell receptor repertoire differences between African Americans and Caucasians associated with polymorphism of the TCRBV3S1 (V beta 3.1) gene, J. Immunol. 154 (9 (May 1)) (1995) 4836–4841. [38] G. Gabriel, P.C. Arck, Sex, immunity and influenza, J. Infect. Dis. 209 (Suppl 3 (Jul 15)) (2014) S93–S99. [39] G. Zandman-Goddard, E. Peeva, Y. Shoenfeld, Gender and autoimmunity, Autoimmun. Rev. 6 (6 (Jun)) (2007) 366–372.