Surgical assessment of significant steatosis in donor livers: The beginning of the end for frozen-section analysis?


  • See Article on Page 437

Address reprint requests to Elizabeth M. Brunt, M.D., Department of Pathology and Immunology, Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8118, St. Louis, MO 63110. Telephone: 314-362-0101; FAX: 314-362-8950; E-mail:

Communication between surgeons and pathologists has heightened significance in the frozen-section suite when critical decisions are being made. In liver transplantation, the assessment of a donor liver frequently includes the need for the assessment of steatosis because it has long been recognized that short-term allograft function may be adversely affected by >30% accumulation of steatosis, and the liver is a known target of lipid dysmetabolism, fat accumulation, and lipotoxicity in our population seemingly ubiquitously affected by obesity and the complexities of metabolic syndrome. The current study from the University of California Los Angeles[1] is the product of a prospective study of 201 procured donor livers. Features grossly observed by the retrieving surgeons, who were blinded to any results of possible contemporaneous liver biopsy, were catalogued. These features, used to determine the presence of ≤/≥30% macrovesicular steatosis and overall organ quality, were then compared with carefully determined histological findings from liver biopsy samples obtained from right and left livers and read by 2 pathologists blinded to the clinical results at a later time. Among the interesting findings were the following: experienced surgeons' observations of liver yellowness, rounded edges, and an absence of capsular scratch marks were all predictive of the presence of macrovesicular steatosis (large-droplet type). The corollary was that sharp edges and the presence of scratch marks correlated with nonsteatotic livers, regardless of observed yellowness. A decision tree analysis confirmed that all liver texture characteristics are more important as correlates with moderate or severe steatosis than any single correlate such as yellowness. As for predictions of the amount of steatosis, the study showed excellent prediction by the surgeons of <30% steatosis, but there was a lack of a significant correlation when macrosteatosis was >30%; in those cases, an underestimation of the amount of steatosis was found. The final recommendation of the study is for liver biopsy to be performed if there are concerns based on the aforementioned gross findings, but in the clinical setting in which pathology is not readily available, sharp liver edges and the presence of capsular scratch marks could be an assurance of a lack of significant steatosis (ie, >30%).

Before there is a hopeful sigh of relief from the pathologists in the community who deal with these challenging donor biopsy samples as frozen sections (and often during nontraditional hours), a few comments in perspective are warranted. First, the histological findings of this study by Yersiz et al.[1] came entirely from formalin-fixed, paraffin-embedded, routinely stained liver biopsy samples that were presumably read during normal working (daylight) hours. Any frozen sections that had been taken during organ procurement were, by study design, not reviewed or otherwise included in the study. Although it is not specifically stated, the inferred rationale for this must have been that because liver biopsy was considered the gold standard for the determination of the amount of steatosis by the study's investigators, all involved understood and agreed that formalin-fixed, paraffin-embedded, routinely stained sections would provide the best results for the analysis. Somewhat ironically, however, a conclusion of the study is that if the procuring surgeon is adequately concerned about the possibility of >30% steatosis on the basis of the gross criteria discussed, immediate liver biopsy (ie, frozen-section analysis) is recommended. Although the authors acknowledge well-known shortcomings of frozen-section liver biopsy samples, they do not suggest how to overcome them. Thus, practicing pathologists will remain challenged by the distinction between fat vacuoles and artifactual spaces in frozen-section liver biopsy samples. This writer is not a proponent of the use of stains for fat such as Oil Red O, however, because they only delay interpretation, significantly increase the amount of steatosis identified, and cannot differentiate large- and small-droplet steatosis.

The second and equally significant point of this article is its insistence on the proper use of terminology for fat droplets in the liver. The 3 types of steatosis histologically observed, large- and small-droplet macrovesicular steatosis and true microvesicular steatosis, were defined and separately studied. The authors nicely illustrate the 2 types of macrovesicular steatosis in Fig. 1 of their article: large-droplet macrovesicular steatosis is characterized by 1 large vacuole or a few large vacuoles that replace the cytoplasm and cause the nucleus to be in an eccentric location, whereas the nucleus (not nucleolus) retains its central location and a significant amount of cytoplasm is likewise retained with the few and discrete vacuoles of small-droplet macrovesicular steatosis. True microvesicular steatosis, found in only 2 of their cases, is not illustrated. Microvesicular steatosis results in a foamy appearance for affected hepatocytes because of the intracytoplasmic accumulation of tiny vesicles, which may be too numerous to individually discern. The process is commonly related to hepatic encephalopathy and liver failure, and the underlying disease processes are those of severe mitochondrial dysfunction but not insulin resistance and lipotoxicity, as in macrovesicular steatosis.[2] Although it is true that patches of true microvesicular steatosis may occur in both nonalcoholic fatty liver disease and alcoholic fatty liver disease,[3, 4] it is highly unlikely that a liver from an affected patient (eg, one with acute fatty liver of pregnancy or Reye's syndrome) would even be considered as a donor liver; thus, liver pathologists may remain confident in the appropriate use of terminology. The authors have further confirmed that the assessment of large-droplet macrovesicular steatosis correlates with various surgical observations and thus with initial graft function. It is hoped that the findings of this study will continue the clarification of the types of steatosis because this feature will remain significant in the future of liver transplantation.

Finally, this study highlights the fact that experienced pathologists can and do agree on an important topic: the amounts of macrovesicular steatosis (large-droplet type) in liver biopsy samples. This fact has been published for many nonallograft scenarios as well.[5] The true challenge, as mentioned previously and unfortunately not answered by this otherwise interesting, helpful, and hopeful study, is the interpretation of the ever-present vacuoles in the frozen-section liver biopsy sample.