Assessment of donor liver steatosis: pathologist or automated software?

Steatosis in donor liver biopsy specimens has been shown to correlate with graft dysfunction after orthotopic liver transplantation. This 2-part (laboratory pilot, clinical retrospective) study compared the traditional interpretation of steatosis by a pathologist with an automated measurement determined by an image analysis system. In our pilot study, Sprague-Dawley rats were studied prospectively by feeding them a choline-deficient diet for up to 7 days. In our clinical group, data from 49 consecutive recipients of cadaveric liver transplantation were reviewed retrospectively. In both studies, the percentages of microvesicular fat, macrovesicular fat, and total fat content within liver biopsy specimens were determined by an automated image analysis software program and a pathologist using the same set of slides. The association between fat content of the donor liver and patient survival and graft survival, along with levels of aspartate aminotransferase, alanine aminotransferase, prothrombin time, and total bilirubin after transplantation, were also examined in the clinical study. A direct correlation was observed between levels of macrovesicular fat determined by a pathologist and the automated software using livers from rats fed a choline-deficient diet and livers from deceased donors. A significant association was observed between macrovesicular fat content in the donor liver biopsy and graft survival by both techniques. We conclude that an image analysis system can be used to automate the determination of fat content in liver biopsy specimens, and that its findings correlate with both the visual interpretation by a pathologist and graft survival. Further study is needed to determine the role of an automated technique in the evaluation of donor livers for transplantation.