Biliary atresia (BA) is a disease of unknown etiology

Serial Circulating Markers of Inflammation in Biliary Atresia Evolution of the Post-Operative Inflammatory Process Bommayya Narayanaswamy, 1 Christopher Gonde, 2 * J. Michael Tredger, 2 Munther Hussain, 2 Diego Vergani, 2 * and Mark Davenport 1 * Biliary atresia (BA) may be characterized as an occlusive cholangiopathy affecting both intraand extra-hepatic parts of the biliary tree, together with a pronounced inflammatory response consisting of hepatic infiltration of (predominantly) CD4 lymphocytes and macrophages. Soluble cellular adhesion molecules are also known to be raised at the time of portoenterostomy, presumably reflecting intrahepatic disease. We investigated this measurable inflammatory component longitudinally by studying a panel of cellular adhesion molecules (soluble intercellular adhesion molecule-1 [sicam-1], soluble vascular cell adhesion molecule-1 [svcam-1]) and soluble proinflammatory mediators (T helper 1 [interleukin {IL}-2 and interferon ] and T helper 2 [IL-4 and IL-10]) cytokines and macrophage markers (tumor necrosis factor [TNF] and IL-18) in 21 consecutive infants with BA post-kasai portoenterostomy (KP). The levels of all adhesion molecules and cytokines (except IL-10) increased progressively by 6 months post-portoenterostomy. The response was non-polarized but with 100-fold increases in IL-2, TNF and IL-18 particularly but only modest elevations in IL-10. When proinflammatory profiles were related to outcome, we found poor discrimination if assessed as clearance of jaundice but markedly higher values for IL-2, interferon, IL-4, IL-10, TNF and sicam-1 for those who would be transplanted by 1 year. Using ROC curve analysis for sicam-1 levels at 1 month post-kp, a cutoff level of 1,779 ng/ml was determined to predict the need for transplantation at 1 year with 92% specificity and 87% sensitivity. Conclusion: The early circulating inflammatory process in BA is persistent, progressive and involves a non-polarized T cell, macrophage and cell adhesion molecule response only partially ameliorated by KP. (HEPATOLOGY 2007;46: 180-187.) See Editorial on Page 6 Abbreviations: BA,biliary atresia; BASM, biliary atresia splenic malformation syndrome; CAM, cell adhesion molecule; ELISA, enzyme-linked immunosorbent assay; IL-2 interleukin-2; IFN, interferon- ; IL-4, interleukin-4; IL-10, interleukin -10; IL-18, interleukin-18; KP, Kasai portoenterostomy; NK cells, natural killer cells; ROC, receiver operating characteristic; sicam-1, soluble intercellular adhesion molecule-1. svcam-1, soluble vascular cell; adhesion molecule-1;tnf, tumor necrosis factor- ; Th1,T helper 1; Th2, T helper 2. From the 1 Departments of Paediatric Surgery and 2 Institute of Liver Studies, King s College Hospital, London SE5 9RS Received November 4, 2006, 2006; accepted February 27, 2007. *These authors contributed equally as senior authors. Address reprint requests to: Mr. M. Davenport, Dept. of Paediatric Surgery, King s College Hospital, Denmark Hill, London SE5 9RS. E-mail: Markdav2@ntlworld.com; fax: 0044 (0) 203 299 4021. Copyright 2007 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/hep.21701 Potential conflict of interest: Nothing to report 180 Biliary atresia (BA) is a disease of unknown etiology with a pre- or peri-natal onset characterized by cholestasis, bile duct proliferation, cellular inflammation and progressive fibrosis. The inflammatory process within the liver is characterized by an initial small cell infiltrate of predominantly CD4 T and natural killer (NK) CD56 lymphocytes 1,2 with a later appearance of cells of the CD68 monocyte/macrophage lineage. 1,3,4 There is also abnormally increased expression of cellular adhesion molecules (CAMs) such as intercellular adhesion molecule (ICAM, CD54), vascular cell adhesion molecule (VCAM, CD106) and E-Selectin (CD62E) on both biliary and vascular endothelium. 1-5 We have also shown abnormally high levels of circulating soluble adhesion molecules (sicam-1 and svcam-1) at the time of Kasai portoenterostomy (KP), with the former having a significant positive correlation with post-natal age and the latter prognostic significance when elevated ( 1380 ng/ml). 2

HEPATOLOGY, Vol. 46, No. 1, 2007 NARAYANASWAMY ET AL. 181 The aim of this study was to follow the course of the inflammatory process as expressed by levels of soluble cellular adhesion molecules (CAMs) and cytokines into the early post-operative phase of this disease. A range of inflammatory molecules was studied including interleukin (IL)-2 and interferon (IFN) (T helper 1 [Th1] cytokines); IL-4 and IL-10 (T helper 2 [Th2] cytokines); IL-18 and tumor necrosis factor [TNF] (macrophage markers) and soluble intercellular adhesion molecule-1 (sicam-1) and soluble vascular cell adhesion molecule-1 (svcam-1). Materials and Methods Kings College Hospital is a large tertiary center with referrals largely from the South-East of England. The diagnosis of BA was confirmed in all the cases by histological examination of resected biliary remnants. Facilities for both KP and liver transplantation, if required, were available. The KP was performed by one of the authors (M.D.) using previously published techniques. 6 Patients with BA underwent serial blood sampling preoperatively and at 1 week, 1, 3, and 6 months after KP. These were taken at routine clinic visits and patients were otherwise well. Two categories of control subjects of a comparable age group to BA patients were also studied. Cholestatic controls (n 10) included patients with liver disease other than BA ( 1-antitrypsin deficiency, n 4; choledochal malformation, n 3; 1 each of choledocholithiasis, inspissated bile syndrome and parenteral nutrition associated cholestasis). Non-cholestatic (normal) controls (n 10) comprised patients undergoing elective surgery who were not suffering from any infection or inflammatory condition (inguinal hernia, n 6; pyloric stenosis, n 2; 1 each of liver haemangioendothelioma and long-gap oesophageal atresia). All control subjects had a single blood sample taken either at the time of venepuncture for clinical investigation or prior to induction of anesthesia for relevant surgery. Two ml of venous blood were collected into a sterile EDTA (ethylenediamine tetraacetic acid)-containing tube and transported immediately on ice to the laboratory to be processed within approximately 1 h. Each sample was centrifuged twice initially at 1,000g for 15 minutes at 4 C and then again at 10,000g for 10 minutes at 4 C following which the separated plasma was aliquoted and stored at 80 C until analysis. All samples underwent enzyme-linked immunosorbent assay (ELISA) in singlicate for IL-2 and IFN (Th1- type cytokines), IL-4 and IL-10 (Th2-type cytokines), IL-18 and TNF (macrophage markers), as well as si- CAM-1 and svcam-1 (soluble CAMs). Assays for IL-2, IFN, IL-4, IL-10 and TNF were performed using commercially available monoclonal antibody pairs (Duoset, R& D Systems, Abingdon, UK). Microwell plates were coated with capture antibody (monoclonal antibody to specific marker being measured) following which standard, control, and test samples were pipetted into respective wells. Subsequent steps involved addition of biotinylated detection antibody, streptavidinhorseradish peroxidase and substrate, each of which was preceded by incubation and washing to remove unbound reactants. Color development from substrate was terminated by addition of acid and absorbance at 450nm wavelength was measured using a Dynex MRX plate-reader (Prior Laboratory Supplies, West Sussex, UK) with simultaneous derivation of actual levels of marker in samples. Commercially available standard ELISA kits with precoated microwell plates were used in similar fashion to perform assays for IL-18 (Bender MedSystems, GmbH, Vienna, Austria) as well as for sicam-1 and svcam-1 (R & D Systems, Abingdon, UK). Ethical approval was prospectively obtained from the Research and Ethics Committee, King s College Hospital and informed written consent was obtained from parents of all patients and control subjects. Statistical Methods Data are quoted as medians (range) unless otherwise indicated. Most data were confirmed as having a nonnormal distribution by a Shapiro-Wilk test and therefore non-parametric comparisons were used throughout. These included a repeated-measures non-parametric ANOVA (Friedman test) for initial evaluation of the serial median values and Wilcoxon and Mann-Whitney tests for paired and unpaired comparisons respectively. Correlations were studied using the Spearman rank correlation. StatsDirect (version 2.4.5), Wilmslow, UK, and GraphPad Prism (version 3.02), San Diego, CA, software were used. A P value of less than or equal to 0.05 was regarded as significant. Results The study group consisted of 21 (16 female) infants with histologically-proven type 3 (i.e., obliteration of proximal portal plate) BA who were treated between June 2003 June 2004 with a median age at surgery of 59 (range 37-90) days. Biliary Atresia Splenic Malformation syndrome (BASM) 7 was seen in 3 infants. The median follow-up for the group was 18.7 (12-23) months. Eight infants had undergone transplantation at a median age of 9.7 (5.4-13) months. One child died of sepsis at 3 months post KP.

182 NARAYANASWAMY ET AL. HEPATOLOGY, July 2007 Table 1. Baseline Values for Th1, Th2 and Macrophage Cytokines and Cellular Adhesion Molecules in Biliary Atresia (n 21) Versus Age-Matched Normal (n 10) and Cholestatic Controls (n 10).* Median (Range) Controls (Normal) P Value (vs. BA) Controls (cholestatic) P Value (vs. BA) Th1 cytokines IL-2 (pg/ml) 8 (0 960) 8 (0 45) 0.5 0 (0 32) 0.04 IFN (pg/ml) 1187 (0 5967) 1134 (0 1688) 0.68 665 (0 2705) 0.79 Th2 cytokines IL-4 (pg/ml) 18 (0 362) 0 (0 39) 0.14 32 (0 235) 0.32 IL-10 (pg/ml) 396 (0 3130) 1073 (0 2544) 0.8 123 (0 670) 0.05 Macrophage cytokines IL-18 (pg/ml) 48 (0 6035) 48 (6 96) 0.64 102 (3 541) 0.2 TNF (pg/ml) 8 (0 1594) 0 (0 55) 0.16 1 (0 74) 0.18 Cellular adhesion molecules sicam-1 (ng/ml) 1099 (348 2325) 164 (20 343) 0.001 805 (503 1868) 0.04 svcam-1 (ng/ml) 1054 (737 2947) 991 (167 1287) 0.05 1062 (830 1693) 0.6 *All comparisons are one-tailed M-W. Baseline Values Table 1 illustrates the baseline values for Th1, Th2, and macrophage cytokines together with CAMs for subjects and the 2 control groups. There were no significant differences in baseline plasma cytokine levels between subjects and normal controls although plasma IL-2 (P 0.04) and IL-10 (P 0.05) levels were raised compared to cholestatic controls. Median sicam-1 and svcam-1 levels were significantly elevated compared to normal (P 0.0001 and P 0.05 respectively) and sicam-1 was elevated compared to cholestatic controls (P 0.04). There was a significant positive correlation of age at KP with TNF levels only ( 0.56, P 0.009). Serial Values Post-Kasai Portoenterostomy Table 2 illustrates the serial median values for Th1, Th2, and macrophage cytokines together with CAMs pre- and post-kp (1 week, 1, 3, and 6 months). There were highly significant progressive increases in median levels of IL-2 (P 0.0002) and IFN (P 0.04), as well as IL-4 (P 0.0001) but not IL-10 (P 0.43) levels. Although there was a 100-fold increase in median TNF levels (P 0.0001) over the period, serial values for IL-18 showed significant variation (P 0.004), without an overall trend. Median values of both sicam-1 (P 0.007) and svcam-1 (P 0.0001) increased significantly up to 6 months. At 4, 12 and 24 weeks post-kp, total bilirubin had significant positive correlation with sicam-1 ( 0.50, P 0.02; z 0.61, P 0.005 and 0.86, P 0.0001, respectively). There was no other significant correlation between bilirubin and any other marker. Relationship to Outcome The relationship of these inflammatory markers to the effect of the KP was explored in two ways. Firstly the study group was divided on the basis of serum bilirubin at 6 months ( 50 mol/l ( 3mg/dl), n 11 (good outcome) versus 50 mol/l, n 10 (poor outcome)) to establish the relationship to the key deter- Table 2. Longitudinal Values (Median (Interquartile Range)) for Th1, Th2 and Macrophage Cytokines and Cellular Adhesion Molecules in Biliary Atresia (n 21)* Pre 1 Week 1 Month 3 Months 6 Months P Value Th1 cytokines IL-2 (pg/ml) 8 (0 28) 15 (7 30) 48 (3 200) 83 (0 530) 400 (35 11068) 0.0002 IFN (pg/ml) 1187 (0 1810) 1200 (658 2805) 1188 (413 3442) 875 (0 5397) 2320 (1086 6280) 0.04 Th2 cytokines IL-4 (pg/ml) 18(0 54) 32 (25 98) 124 (35 354) 223 (98 2112) 563 (160 2335) 0.0001 IL-10 (pg/ml) 396 (0 1680) 553 (90 1860) 470 (135 1231) 336 (0 1138) 1660 (0 5514) 0.43 Macrophage cytokines IL-18 (pg/ml) 48 (28 101) 132 (71 162) 58 (41 106) 88 (53 224) 177 (113 286) 0.0004 TNF (pg/ml) 8 (0 20) 64 (36 100) 108 (42 652) 900 (183 2818) 2370 (266 7293) 0.0001 Cellular adhesion molecules sicam-1 (ng/ml) 1099 (780 1534) 1392 (1130 1687) 1447 (1334 2543) 1793 (1395 2003) 1848 (1066 2152) 0.007 svcam-1 (ng/ml) 1054 (946 1246) 1402 (1109 1659) 1498 (1262 2611) 2083 (1551 2608) 2406 (1751 2844) 0.0001 *Overall significance assessed using non-parametric two-way Friedman test.

HEPATOLOGY, Vol. 46, No. 1, 2007 NARAYANASWAMY ET AL. 183 Fig. 1. Serial Th1 cytokine response to Kasai portoenterostomy. (A) Plasma IL-2 (median interquartile range): good (n 11) versus poor (n 10) response (n.b. logarithmic y axis). (B) Plasma interferon-. Fig. 2. Serial Th2 cytokine response to Kasai portoenterostomy. (A) Plasma IL-4 (median interquartile): good (n 11) versus poor outcome (n 10) (n.b. logarithmic y axis). (B) Plasma IL-10 (n.b. logarithmic y axis). minant of operative success (i.e., alleviation of cholestasis). Only median pre-operative IL-18 levels were significantly higher in the good outcome group (101 versus 28 ng/ml, P 0.05 (2-sided MW)). Median preoperative s-vcam-1 levels were lower in the same group, but this did not reach significance (1001 versus 1230 ng/ml, P 0.09). Over the 6-month post-operative course, median values of sicam-1 were significantly higher in the poor outcome group (at 3 months, P 0.02 and 6 months, P 0.001) (Fig. 4A). Otherwise, there were no significant differences in IL-2, IFN, IL-4, IL-10, IL-18, TNF, or svcam-1 levels between the good and poor response groups (Figs. 1 to 4 illustrate this analysis). Secondly, the study group was divided on the basis of outcome at 1 year into those who were alive with their native livers (n 12) and those who had been transplanted (n 8). One child who had died of sepsis at 3 months was excluded. Serial biochemistry is illustrated in Fig. 5. As anticipated, there were significant differences in bilirubin levels at 1, 3, and 6 months but unexpectedly no significant difference in the plasma AST levels at any time point. In contrast to the analysis presented above, there was much more difference between the groups. In the transplant group, levels of s-icam-1 were significantly raised at 1 month; IL-2, IL-4 and TNF at 3 months; Fig. 3. Serial macrophage cytokine response to Kasai portoenterostomy. (A) Plasma IL-18 (median interquartile range): good (n 11) versus poor (n 10) response. (B) Plasma TNF- (n.b. logarithmic y axis).

184 NARAYANASWAMY ET AL. HEPATOLOGY, July 2007 Fig. 4. Serial cellular adhesion molecule response to Kasai portoenterostomy. (A) Plasma sicam-1 (median interquartile range): good (n 11) versus poor (n 10) response. (B) Plasma svcam-1. Fig. 5. Liver biochemistry by outcome at 1 year. (A) Plasma bilirubin (median interquartile range): alive with native liver (n 12) versus OLT at 1 year (n 8). (B) Aspartate aminotrasferase. and IFN, IL-10 and s-vcam-1 at 6 months. There was no difference, however, in serial IL-18 values. Figs. 6 to 9 illustrate this analysis. Although there was an overlap in the infants in the two analyses; 7/20 (35%) infants changed their group depending on the definition of outcome. As sicam-1 was the earliest marker (at 1 month post- KP) showing a significant difference between outcome groups, we explored its utility in predicting the need for transplantation. Using ROC curve analysis (area under curve 0.88, 95% CI 0.42-0.98), we determined that a cutoff level of sicam-1 of 1779 ng/ml would predict the need for transplantation with 87% sensitivity and 92% specificity. Discussion Whatever its origin, biliary atresia is not a simple mechanical obstruction of the extrahepatic biliary tree. There is a distinct inflammatory component to the disease, which is obvious histologically within both the liver and the biliary tree. 1 The process appears to be driven, and possibly initiated, by abnormal expression of adhesion molecules (e.g., ICAM -1 and VCAM-1) and MHC-class II molecules (e.g., HLA-DR) on biliary and vascular epithelium with a (presumed) influx of CD4, 1,8 CD8 9 Fig. 6. Th1 cytokines by outcome at 1 year. (A) Plasma IL-2 (median interquartile range): alive with native liver (n 12) versus OLT at 1 year (n 8). (n.b. logarithmic y axis). (B) Plasma Interferon-.

HEPATOLOGY, Vol. 46, No. 1, 2007 NARAYANASWAMY ET AL. 185 Fig. 7. Th2 cytokines by outcome at 1 year. (A) Plasma IL-4 (median interquartile range): alive with native liver (n 12) versus OLT at 1 year (n 8). (n.b. logarithmic y axis). (B) Plasma IL-10 (n.b. logarithmic y axis) Fig. 9. Soluble cellular adhesion molecule levels by outcome at 1 year. (A) Plasma sicam-1 (median interquartile range): alive with native liver (n 12) versus OLT at 1 year (n 8). (B) Plasma svcam-1. Fig. 8. Macrophage cytokines by outcome at 1 year. (A) Plasma IL-18 (median interquartile range): alive with native liver (n 12) versus OLT at 1 year (n 8). (B) Plasma TNF (n.b. logarithmic y axis). and NK cells. 1 Such naïve CD4 T cells can differentiate into either Th1 effector cells (driven by IL-12 and producing pro-inflammatory cytokines IFN, IL-2, TNF and TNF ) or Th2 effector cells (driven by IL-4 and producing IL-4, IL-5, and IL-10). One step removed from this initial phase is activation, infiltration and proliferation of macrophage/monocytes (producing IL-18 and TNF ), which are able to initiate fibrosis, and produce or control changes in the extracellular matrix. 10 We postulated that established immune / inflammatory activity within the liver involves the release of various soluble mediators and is therefore likely to be paralleled by a circulating component. There have been few reports where soluble cytokines have been measured in BA. 11-14 These have been limited either to a single cytokine 11,12 or have looked only at a cross-section of an older BA population. 11,13,14,15 We therefore measured a representative panel of circulating mediators as markers of activity of the respective parent immune-competent cells (ICCs) and examined the evolution of this circulating immune / inflammatory response from diagnosis to, in some cases, endstage disease and transplantation. In our study, the baseline values for the range of cytokines studied were little different from age-matched and cholestatic controls at the time of KP although the intragroup variation was wide (particularly for cytokines), and

186 NARAYANASWAMY ET AL. HEPATOLOGY, July 2007 certainly extreme values were only seen in the BA group. Baseline levels of both sicam-1 and svcam-1 were more constrained and elevated compared to controls and in the case of sicam-1, other cholestatic infants. Our previous cross-sectional study 2 had shown a positive correlation of sicam-1 with age at KP (seen only with TNF in this study), suggesting some relationship with the degree of inate liver disease. It is not really clear whether the inflammatory process seen in BA is specific or related in some way to neonatal cholestasis per se. 4,16 The lack of difference in the initial profile of soluble cytokines between BA patients and controls certainly argues against such specificity. In the first 6 months following the KP rising levels of Th1 cytokines and CAMs were clearly seen. The pattern of Th2 and macrophage cytokines was more mixed with obvious progressive increases in IL-4 but not IL-10, and TNF but not IL-18 levels. In general, there were few differences between those who cleared their jaundice by 6 months and those who did not. It is remarkable that while bilirubin and AST levels decreased post-kp, this did not parallel the time-course of any of the observed cytokines or CAMs, which tended to increase (vide infra). In contrast, when analysed on the basis of the need for early transplantation, more striking differences emerged. Earliest discrimination occurred with sicam-1 (but not sv- CAM-1) by 1 month, followed by IL-2, IL-4, and TNF levels. This strongly suggests that factors other than the degree of cholestasis or its clearance, are responsible for driving or maintaining the observed inflammatory process. The concept of balance between the T-helper subsets Th1 and Th2 originated from research into murine T- helper cells 17,18 and has been applied to the human immune response. However, the validity of rationalizing a given immune response exclusively to one or other T- helper subtype has been questioned. 19 There has been some clinical and experimental evidence that BA exhibits a polarized, specifically Th1 immune response, at least in the early stages of the disease (i.e., up to the time of KP). 16,20,21,22 For instance, Mack et al. 16 using immunocytochemistry and RT-PCR showed a CD4 and CD8 T cell infiltrate within portal tracts with a localized increased expression of the cytokines IL-2, IFN, TNF and IL-12 (but not IL-4 or IL-5). Experimentally (using a rotavirus-induced murine BA model) the same group 20 has shown significant increases in CD4 T cells producing IFN (i.e., Th1 response) with a later influx of activated macrophages producing TNF. It is noteworthy that this cytokine response persisted despite viral clearance. This extended phase of elevated soluble cytokines also appears to be a feature of our soluble inflammatory profile, at least to 6 months. Evidence from other series where cytokines have been measured in much older children suggests a return toward normal values. 13 Our data suggest that following KP there is a mixed circulating cytokine response with clear progressive elevation in both Th1 effectors (IL-2 and IFN ), some Th2 effectors (IL-4) as well as a macrophage marker (TNF ) but no clear trend in others (e.g., IL-10, IL-18). Nonetheless, it is likely that BA is a disease phenotype exhibiting etiological heterogeneity from a variety of insults to the immature or developing liver and biliary tract and given the complexity of the immune response involved, it may therefore be an oversimplification to attribute its pathogenesis solely to a polarized Th1 immune response. Increased expression of soluble CAMs appears to be a consistent feature in BA with progressive increase in both sicam-1 and svcam-1 levels. This probably reflects ongoing recruitment of circulating inflammatory/immune-competent cells into target tissues. Although clearly established pre-operatively, this process persists post-kp and notably enhances (at least for sicam-1) in those with a poor outcome. It is noteworthy that levels of sicam-1 (but not svcam-1) were alone in paralleling the time course of bilirubin values, with a return to normal in those who cleared their jaundice. From our data, sicam-1 appears to be a candidate marker with the clinical potential for prediction of later outcome. However, this must be interpreted with a degree of caution due to the constraints of the small numbers involved. It is evident that studies on cytokines and CAMs both clinically and in the experimental model may help to identify more pieces of the inflammatory jigsaw in BA, but, the challenge remains to piece them together as a consistent unifying hypothesis on their actual role. Our study clearly shows a persisting inflammatory response in BA, which is largely unaffected by portoenterostomy. This phase persists for at least 6 months after surgery and may (if regarded as detrimental) explain why some cases have an initially good response (in terms of falling bilirubin) but then go on to develop end-stage liver disease. Acknowledgment: The authors thank the Childrens Liver Disease Foundation who generously supported this study as well as Paul Cheeseman for his help in the laboratory. References 1. Davenport M, Gonde C, Redkar R, Koukoulis G, Tredger JM, Portmann B, et al. Immunohistochemistry of the liver and biliary tree in extrahepatic biliary atresia. J Pediatr Surg 2001;36:1017-1025. 2. Davenport M, Gonde C, Narayanaswamy B, Mieli-Vergani G, Tredger JM. Soluble adhesion molecule profiling in preoperative infants with biliary atresia. J Pediatr Surg 2005;40:1464-1469.

HEPATOLOGY, Vol. 46, No. 1, 2007 NARAYANASWAMY ET AL. 187 3. Tracy TF, Dillon P, Fox ES, Minnick K, Vogler C. The inflammatory response in pediatric biliary disease: macrophage phenotype and distribution. J Pediatr Surg 1996;31:121-126. 4. Dillon P, Belchis D, Tracy T, Cilley R, Hafer L, Krummel T. Increased expression of intercellular adhesion molecules in biliary atresia. Am J Pathol 1994;145:263-267. 5. Kobayashi H, Puri P, O Brian S, Surana R, Miyano T. Hepatic overexpression of MHC Class II antigens and macrophage-associated antigens (CD68) in patients with biliary atresia of poor prognosis. J Pediatr Surg 1997;32;596-593. 6. Davenport M. Biliary atresia. In: Spitz L, Coran AG, editors. Operative Pediatric Surgery. 6th ed. London: Hodder Arnold, 2006:661-672. 7. Davenport M, Savage M, Mowat AP, Howard ER. Biliary Atresia Splenic Malformation syndrome: an etiologic and prognostic subgroup. Surgery 1993;113:662-668. 8. Schweizer P, Petersen M, Jeszberger N, Ruck P, Dietz K. Immunohistochemical and molecular biological investigations regarding the pathogenesis of extrahepatic biliary atresia (Part 1: immunohistochemical studies). Eur J Pediatr Surg 2003;13:7-15. 9. Ahmed AF, Ohtani H, Nio M, Funaki N, Shimaoka S, Nagura H, et al. CD8 T cells infiltrating into bile ducts in biliary atresia do not appear to function as cytotoxic T cells: a clinicopathological analysis. J Pathol 2001; 193:383-389. 10. Tracy TF, Dillon P, Fox ES, Minnick K, Vogler C. The inflammatory response in pediatric biliary disease: macrophage phenotype and distribution. J Pediatr Surg 1996;31:121-125. 11. Honsawek S, Chongsrisawat V, Vejchapipat P, Thawornsuk N, Tangkijvanich P, Poovorawan Y. Serum interleukin-8 in children with biliary atresia: relationship with disease stage and biochemical parameters. Pediatr Surg Int 2005;21:73-77. 12. Urushihara N, Iwagaki H, Yagi T, Kohka H, Kobashi K, Morimoto Y, et al. Elevation of serum interleukin-18 levels and activation of Kupffer cells in biliary atresia. J Pediatr Surg 2000;35:446-449. 13. Nobili V, Marcellini M, Giovannelli L, Girolami E, Muratori F, Giannone G, et al. Association of serum interleukin-8 levels with the degree of fibrosis in infants with chronic liver disease. J Pediatr Gastroenterol Nutr 2004; 39:540-44. 14. Kobayashi H, Horikoshi K, Long L, Yamataka A, Lane GJ, Miyano T. Serum concentration of adhesion molecules in postoperative biliary atresia patients: relationship to disease activity and cirrhosis. J Pediatr Surg 2001; 36:1297-1301. 15. Kobayashi H, Yamataka A, Lane GJ, Miyano T. Levels of circulating antiinflammatory cytokine interleukin-1 receptor antagonist and proinflammatory cytokines at different stages of biliary atresia. J Pediatr Surg 2002;37:1038-1041. 16. Mack CL, Tucker RM, Sokol RJ, Karrer FM, Kotzin BL, Whitington PF, et al. Biliary atresia is associated with CD4 Th1 cell-mediated portal tract inflammation. Pediatric Res 2004;56:79-87. 17. Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Coffman RL. Two types of murine helper T cell clone. I. Definition according to profiles of lymphokine activities and secreted proteins. J Immunol 1986;136: 2348-2357. 18. Mosmann TR, Coffman RL. TH1 and TH2 cells: different patterns of lymphokine secretion lead to different functional properties. Annu Rev Immunol 1989;7:145-173. 19. Kidd P. Th1/Th2 balance: The hypothesis, its limitations, and implications for health and disease. Altern Med Rev 2003;8:223-246. 20. Mack CL, Tucker RM, Sokol RJ, Kotzin BL. Armed CD4 Th1 effector cells and activated macrophages participate in bile duct injury in murine biliary atresia. Clin Immunol 2005;115:200-209. 21. Bezerra JA, Tiao G, Ryckman FC, Alonso M, Sabla GE, Shneider B, et al. Genetic induction of proinflammatory immunity in children with biliary atresia. Lancet 2002;360:1653-1659. 22. Shinkai M, Shinkai T, Puri P, Stringer MD. Elevated expression of IL2 is associated with increased infiltration of CD8 T cells in biliary atresia. J Pediatr Surg 2006;41:300-305.