Assessment of a transient elastography in predicting the graft fibrosis after liver transplantation and validation of the Venturi score in an adult population.

Introduction: Progressive liver graft fibrosis is a cause of graft failure and late death in the follow-up of transplant patients. Early detection of liver graft fibrosis is crucial to improve survival of transplant patients. Liver biopsy is still the gold standard to detect fibrosis of the graft but is an invasive, costly and not complication-free procedure. Transient elastography (TE) might measure graft fibrosis but data are lacking for this population. Regarding the assessment of fibrosis, the Metavir score is still used, but not designed for a transplant cohort. The liver allograft fibrosis semiquantitative scoring system (LAFSc or Venturi Score) was designed and validated to assess the fibrosis in a pediatric transplant population better than with the Metavir score. Aim: Our aims are to determinate if TE could be correlated to histological scores and to validate LAFSc in an adult population. Methods: We included 52 liver transplant recipients followed-up at our institution who underwent per-protocol liver biopsies between 02/2021 and 10/2021. The biopsy was a 6-months, 1-year, 5-, 10-, 15- or 20-years follow-up procedure, depending on the date of the transplantation. A TE (FibroScan®) was performed before each protocol liver biopsy. With TE, we measured liver stiffness and liver steatosis severity assessed by the controlled attenuation parameter (CAP). Each biopsy sample was examined in the department of pathology of our institution, by one experienced pathologist. Graft fibrosis was classified according to Metavir score and LAFSc. The association between Metavir and LAFSc, and between histological scores and TE was tested through the Spearman’s rank correlation coefficient (rho), which was reported alongside the 95% confidence interval (CI). The relationship between the different scores was modelled through linear regressions and r2 were reported. All tests were two-tailed. Significance was retained at p < 0.05. Analyses were run with GraphPad Prism 8.0 (Grap