Reply: Is oil red-O staining and digital image analysis the gold standard for quantifying steatosis in the liver?^†

The suggestion by Levene et al.1 to further investigate new approaches to the quantification of the histological features of nonalcoholic fatty liver disease (NAFLD) for the purpose of inclusion in clinical studies and the assessment of the response to treatment interventions is interesting and appropriate. The authors have correctly pointed out the limitations of the NAFLD activity score (NAS) scoring system. In our study, a treatment response to betaine (if one existed) may have been missed with standard techniques used to interpret liver histology.2 The NAS scoring system, although well validated, is a semiquantitative score system that may not accurately quantify the amount of steatosis on a hematoxylin and eosin stain, especially when fat droplets are small and/or hepatocyte ballooning is present. More importantly, it remains unknown whether a change in the NAS score reported in such treatment trials influences the natural history of NAFLD and/or alters the risk of NAFLD-related morbidity or mortality.

In the validation study by Kleiner et al.,3 agreement on scoring and diagnostic categorization [nonalcoholic steatohepatitis (NASH), borderline, or not NASH] was evaluated with weighted kappa statistics. Inter-rater agreement on adult cases was 0.84 for fibrosis, 0.79 for steatosis, 0.56 for ballooning, and 0.45 for lobular inflammation. The validation of the NAS scoring system demonstrated reasonable inter-rater agreement among experienced hepatopathologists that was similar to the agreement found in other studies of variability in fatty liver disease.4, 5 The NAS scoring system is simple, can readily be used by practicing pathologists, and requires only routine histochemical stains (hematoxylin and eosin and Masson trichrome stains). As proposed by Levene et al.,1 other special stains, such as Oil Red O and digital image analysis, can increase the diagnostic accuracy for identifying lipids and provide a more accurate quantification of microvesicular and macrovesicular steatosis. Studies in animal models of NASH have often used Oil Red O staining to quantify hepatic steatosis and have demonstrated good control with triglyceride stores, and this suggests that this approach has merit. However, the extension of this application to routine clinical practice has several handicaps.6 Of greatest importance are the poor tissue detail of the frozen section and the added burden of storing and processing frozen tissue. Although this is feasible, it is not routinely done now. Nonetheless, the addition of special staining to better define and quantitate the histological features of NASH (e.g., Oil Red O and cytokeratin 8/18) is worthy of consideration for defining treatment endpoints in future clinical studies of NASH.