Integrin β-8, But Not β-5 or-6, Protein Expression Is Increased in Livers of Children With Biliary Atresia

Objectives: Our previous work demonstrated altered messenger RNA expression of integrin beta-5 and -8, using an in silico analysis of publically available data from patients with biliary atresia (BA); however, we were unable to demonstrate statistically significant differences in protein expression because of sample size. In the present study, we repeated the analysis of liver fibrosis and protein expression of the integrins in a larger cohort of patients with BA and compared them with patients undergoing liver biopsy for other diagnoses, with the hypothesis that >= 1 of the integrins would be differentially expressed. Methods: Liver specimens were obtained at 2 collaborating institutions. Samples from infants with BA (n = 23) were compared with samples from those who underwent liver biopsy for neonatal hepatitis (n = 9). All of the specimens were analyzed by 2 pathologists (C.R. and R.A.), who were blinded to the diagnoses. Standard Ishak scoring was performed to evaluate fibrosis and inflammation, and immunohistochemical (IHC) positivity was graded from 0 to 4. Comparisons between the IHC positivity and Ishak scoring for the BA and control groups were performed using the Student t test with P < 0.01 considered significant because of the multiple comparisons. Interobserver variability was assessed by intraclass correlation (ICC). Results: Pooled analysis from specimens from patients with BA showed significantly more fibrosis than controls based on Ishak scores (3.21 +/- 1.82 vs 1.17 +/- 1.00, P < 0.005). IHC evaluation showed increased integrin alpha nu beta 8 protein expression when compared with controls (2.67 +/- 0.81 vs 1.72 +/- 0.62, P < 0.005); however, there were no significant differences in integrin alpha nu beta 5 (1.93 +/- 0.84 vs 1.50 +/- 0.90, P = 0.23) or integrin alpha nu beta 6 (0.85 +/- 1.20 vs 0.94 +/- 0.85, P = 0.82) expression. These data were confirmed on individual analysis. Interobserver agreement was fair for integrin alpha nu beta 5 (ICC 0.52), good for integrin alpha nu beta 6 (ICC 0.72), and excellent for integrin alpha nu beta 8 (ICC 0.79) and fibrosis (ICC 0.89). Conclusions: Our data show that integrin alpha nu beta 8, but not integrin alpha nu beta 5 or integrin alpha nu beta 6, protein expression is increased in liver specimens of patients with BA. These data support the mounting evidence that transforming growth factor-beta (TGF-beta) activation is responsible for the fibrosis found in BA. Anti-integrin alpha nu beta 8 or more global integrin blocking strategies may be therapeutic options in BA, but further work is clearly needed.