Is Oil Red-O Staining and Digital Image Analysis the Gold Standard for Quantifying Steatosis in the Liver?

We read with great interest the randomized placebo-controlled trial by Abdelmalek et al.,1 in which the effects of betaine were studied in nonalcoholic fatty liver disease (NAFLD). The authors used the widely employed NAFLD activity score (NAS)2 as part of the inclusion criteria to their study and as the "gold-standard" endpoint to define treatment response. One of the major outcomes identified in the trial was change in the steatosis grade. The NAS is a well-validated, pragmatic, semiquantitative scoring system for determining the severity of NAFLD.3 The degree of steatosis contributes up to 3 of the 8 points to this score, and hepatocyte ballooning (which may be confused with small-droplet steatosis) contributes a further 2 points. Given the large contribution of steatosis to the overall score, it is important to correctly identify steatosis in a liver biopsy. During the study of both human tissue and tissue from mouse models of NAFLD, we have found that accurately determining the amount of steatosis on hematoxylin and eosin (H&E) staining, especially when the fat droplets are small, is extremely challenging. Consequently, we have explored an alternative approach adopting Oil red-O (ORO) as the "gold standard" histochemical stain for specifically identifying lipids.4 Using steatotic murine liver tissue from C57BL/6 mice, we compared the assessment of steatosis in H&E-stained sections by two expert liver pathologists with digital image analysis (DIA) quantification of ORO-stained sections. Hepatic triglyceride levels were quantified in the same tissues (triglyceride quantification kit, ab65336; Abcam Inc., Cambridge, MA). We found that, although ORO DIA assessment correlates well with the total liver triglyceride concentration and is therefore an accurate reflection of liver steatosis (Pearson correlation R = 0.706, P = 0.001), assessment by the expert pathologists showed poor correlation (R = −0.422). In samples with macrovesicular steatosis, we found that liver pathologists overestimated the amount of steatosis present on H&E stain as compared with ORO DIA (71.8% versus 46.7%, P < 0.01). In 67% of cases, the NAS steatosis score decreased from 3 to 2 when using ORO DIA. We conclude that the ORO DIA technique provides a more accurate quantification of microvesicular and macrovesicular steatosis. The NAS score is a useful and pragmatic tool in clinical practice and for patient selection for trial entry, but when the score is used as an outcome measure in a clinical trial setting, its performance is less robust. When, as Abdelmalek et al.1 report, steatosis changes are the major study outcome, it is therefore difficult to know whether inaccurate scoring of H&E-stained sections by expert pathologists confounds these observations. ORO DIA is the most reliable way to accurately assess and quantify histological liver steatosis. We accept that ORO staining is not practical for everyday use. It is optimally performed on frozen tissue because processing removes the lipids, hence the stain is not routinely performed on liver biopsies. However, its use may be of merit in clinical trials. Indeed, we have been able to successfully perform ORO staining on frozen sections of formalin-fixed human liver biopsies prior to processing, making wider adoption of this technique viable. Adam P. Levene MBChB Hons*, Hiromi Kudo MSc*, Mark R. Thursz M.D, FRCP , Quentin M. Anstee Ph.D., FRCP , Robert D. Goldin M.D., FRCPATH*, * Department of Histopathology, Imperial College Faculty of Medicine at St Mary's Hospital, London, UK, Department of Gastroenterology and Hepatology, Imperial College Faculty of Medicine at St Mary's Hospital, London, UK.