Development of an unbiased method for the estimation of liver steatosis

Clin Transplant. 2004 Dec;18(6):700-6. doi: 10.1111/j.1399-0012.2004.00282.x.


Background: Steatosis significantly contributes to an organ's transplantability. Livers with >30% fat content have a 25% chance of developing primary non-function (PNF). The current practice of evaluating a hematoxylin and eosin (H&E) stained donor biopsy by visual interpretation is subjective. We hypothesized that H&E staining of frozen sections fails to accurately estimate the degree of steatosis present within a given liver biopsy. To address this problem of evaluating steatosis in prospective donor organs, we developed a fast, user friendly computer methodology to objectively assess fat content based on the differential quantification of color pixels in Oil Red O (ORO) stained liver biopsies.

Methods: The accuracy of human visual estimation of fat content by H&E and ORO stains was compared with computer-based measurements of the same slides from 25 frozen sections of donor biopsies.

Results: Samples with a fat content >20% showed marked variation between human interpretation and computer analysis. There was also a significant difference in the human interpretation of fat based on the method of staining. This difference ranged from 3 to 37% with H&E.

Discussion: Use of ORO resulted in a more consistent estimation of liver steatosis compared with H&E, but human interpretations failed to correlate with computer measurements. Such differences in fat content estimations might result in the rejection of a potentially transplantable organ or the acceptance of a marginal one. Ideally, our protocol can rapidly be applied to clinical practice for accurate and consistent measurement of fat in liver sections for the ultimate purpose of increasing the number of successful transplantable organs.

MeSH terms

  • Algorithms*
  • Fatty Liver / pathology*
  • Humans
  • Liver Transplantation
  • Tissue Donors