Original article
Liver steatosis assessment: Correlations among pathology, radiology, clinical data and automated image analysis software

https://doi.org/10.1016/j.prp.2013.04.001Get rights and content

Abstract

Quantitating hepatic steatosis is important in many liver diseases and liver transplantation. Since steatosis estimation by pathologists has inherent intra- and inter-observer variability, we compared and contrasted computerized techniques with magnetic resonance imaging measurements, pathologist visual scoring, and clinical parameters. Computerized methods applied to whole slide images included a commercial positive pixel count algorithm and a custom algorithm programmed at our institution. For all liver samples (n = 59), including pediatric, adult, frozen section, and permanent specimens, statistically significant correlations were observed between pathology, radiology, and each image analysis modality (r = 0.75–0.97, p < 0.0001), with the strongest correlations in the pediatric cohort. Statistically significant relationships were observed between each method and with body mass index (r = 0.37–0.56, p from <0.0001 to <0.05) and with albumin (r = 0.55–0.64, p < 0.05) but not with alanine aminotransferase or aspartate aminotransferase. Although pathologist assessments correlated (r = 0.64–0.86, 0.92–0.97, and 0.78–0.91 for microvesicular, macrovesicular, and overall steatosis, respectively), the absolute values of hepatic steatosis visual assessment were susceptible to intra- and inter-observer variability, particularly for microvesicular steatosis. Image analysis, pathologist assessments, radiology measurements, and several clinical parameters all showed correlations in this study, providing evidence for the utility of each method in different clinical and research settings.

Introduction

Liver steatosis is an abnormal accumulation of lipid in hepatocytes [5], [41]. These lipids include triglycerides and other molecules. This accumulation occurs when there is a defect in the process by which free fatty acids are taken up by the liver and secreted as lipoproteins. The main causes of steatosis include alcohol, obesity, and diabetes mellitus, type II [5], [41]. Liver biopsies and resections are often assessed for steatosis. In many cases, steatosis may eventually lead to a disorder referred to as nonalcoholic fatty liver disease (NAFLD) [41]. In NAFLD, steatosis may lead to inflammation, referred to as steatohepatitis or, more specifically in the absence of alcohol, nonalcoholic steatohepatitis (NASH). NASH may culminate in fibrosis and eventually cirrhosis [41]. With fatty liver disease as the hepatic manifestation of metabolic syndrome, the prevalence of steatosis is only expected to increase in the future, especially in Western countries [15]. In fact, NAFLD is expected to become the number one indication for orthotopic liver transplantation in North America by 2020 [8].

Steatosis is assessed by a number of grading systems in order to determine the severity of involvement of the liver by fat, notably including the Brunt and the NASH Clinical Research Network (NASH CRN)/Kleiner scoring systems [6], [7], [20], [30], [43]. There are two types of steatosis: macro and microvesicular. Macrovesicular steatosis has been defined by a single, large lipid vacuole that displaces the hepatocyte nucleus to the periphery [10], and microvesicular steatosis is typically characterized by small lipid vacuoles in the cytoplasm of the hepatocyte without peripheral displacement of the nucleus [18].

An important role for quantifying the degree of steatosis is also exemplified in liver transplantation [29]. In this setting, steatosis is classified as mild (<30%), moderate (30–60%), or severe (>60%). Macrovesicular steatosis is associated with post-transplant dysfunction, directly proportional to the severity of steatosis. Prior studies have demonstrated that orthotopic transplantation of livers with severe steatosis have lead to primary graft nonfunction rates in approximately 80% of cases [25]. However, the increased demand and relative shortage of donor livers has led to a lower threshold for accepting potential donors with a relatively high degree of fatty changes. These organs may be suboptimal for transplantation but are not altogether disqualified since it is uncertain whether an absolute threshold, if any, exists [9], [21], [27].

During orthotopic liver transplantation, surgeons grossly evaluate the liver for macroscopic signs of steatosis. If the suitability of the donor liver is in doubt, a liver biopsy is taken for intraoperative frozen section evaluation. The pathologist examines hematoxylin and eosin stained slides in order to determine the degree of macro- and microvesicular steatosis. The cutoff for donor disqualification varies between institutions with macrovesicular steatosis ranging from >30% to >60% being cutoffs due to the higher rates of susceptibility to preservation injury, impaired regeneration, and post-transplant dysfunction with increasing levels of steatosis [27], [31], [34]. In a study evaluating liver transplant recipients receiving livers with severe steatosis, it was determined that these patients had a higher rate of primary graft dysfunction and/or renal failure [11], [28]. However, estimating the degree of steatosis may be prone to both intra and inter-observer variability among pathologists [26]. In addition, because frozen section slides evaluated intraoperatively are later converted to paraffin embedded permanent sections, there may be variability in the assessment of steatosis between the two sections [4]. The liver biopsy is an invasive procedure that is prone to sampling error due to the heterogeneity of disease processes and presence of the hepatic capsule. Ultrasound and computed tomography have been used to quantitate steatosis with limited accuracy. Magnetic resonance imaging (MRI) can separate the liver signal into two components: fat and water signals. The portion of the liver MRI signal which is attributable to the fat is known as the signal fat-fraction [35]. Several investigators [13], [32], [33], [39], including authors of this study [33], [39], have developed techniques to measure steatosis and other parameters from MRI examination.

In the past, several morphometric techniques and computerized image analysis programs have been developed to measure the amount of liver steatosis [3], [12], [17], [24], [37], [42], [44], [45]. Several of these methods showed clear advantages over human semi-quantitative scoring [12], [24], [26], [44]. However, none of the image analysis programs developed to measure steatosis are currently in widespread use in clinical practice. The purpose of this study is to compare the different methods of quantifying steatosis and delineate which methodology correlates most accurately with liver function.

Section snippets

Materials and methods

The study cohort included pediatric patients (n = 10), adult patients (n = 13), and transplant biopsies with corresponding frozen section (n = 18) and permanent slides (n = 18) at Emory University Hospital and Children's Hospital of Atlanta. The study was approved by the Emory University Institutional Review Board. All authors had access to the study data and had reviewed and approved the final manuscript. The 18 biopsies from orthotopic liver transplantations were performed at Emory University

Sample cohort

Our cohort consisted of adult patients, pediatric patients, and frozen section slides with matching permanent section slides. In general, the pediatric cohort consisted of a rather homogeneous cohort of patients, most of whom had NAFLD and were on little or no medications. The adult cohort was more heterogeneous since the patients tended to be on more medications and have more comorbid medical conditions. [Demographic data for the patients are shown in supplemental Table 1, and clinical data

Discussion

In this study, we compared multiple methods in assessing liver steatosis: visual inspection, radiology, and automated image analysis software [both commercial and customized algorithms]. We found that visual inspection of macrovesicular and total steatosis had excellent interobserver correlations. When comparing an observer's assessment of macrovesicular steatosis on a frozen section slide vs. a permanent slide, the concordance was strong. However, for total steatosis, assessment on frozen vs.

Disclosure

Authors of this manuscript [P.S. and D.R.M.] have patented some of the radiology methods used in this study [i.e., US Patent # 8,032,335, October 9, 2008]. Otherwise, the authors declare no conflict or duality of interest.

Author contributions

Study concept and design (M.J.L., A.B.F., N.V.A., D.R.M.), acquisition of data (M.J.L., P.B., J.K., M.B.V., P.S., B.K.), analysis and interpretation of data (M.J.L., J.K., A.B.F.), drafting of the manuscript (M.J.L., A.B.F.), critical revision of the manuscript for important intellectual content (M.J.L., P.B., J.K., M.B.V., J.H.S.), statistical analysis (M.J.L., A.B.F.); study supervision (M.J.L., A.B.F.).

Grant support

None.

References (45)

  • A. Nocito et al.

    When is steatosis too much for transplantation?

    J. Hepatol.

    (2006)
  • S. Saadeh et al.

    The utility of radiological imaging in nonalcoholic fatty liver disease

    Gastroenterology

    (2002)
  • Aperio | Support

    (2013)
  • Image Analysis | Aperio |

    (2013)
  • L. Arnould et al.

    Automated morphometric estimation of liver steatosis

    Microsc. Acta Suppl.

    (1979)
  • J.D. Bancroft et al.

    Microtomy: paraffin and frozen

  • E.M. Brunt

    Pathology of nonalcoholic fatty liver disease

    Nat. Rev. Gastroenterol. Hepatol.

    (2010)
  • E.M. Brunt et al.

    Misuse of scoring systems

    Hepatology

    (2011)
  • K.L. Donnelly et al.

    Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease

    J. Clin. Invest.

    (2005)
  • S. Feng

    Steatotic livers for liver transplantation – life-saving but at a cost

    Nat. Clin. Pract. Gastroenterol. Hepatol.

    (2008)
  • R.N. Fiorini et al.

    Development of an unbiased method for the estimation of liver steatosis

    Clin. Transplant.

    (2004)
  • T.M. Fishbein et al.

    Use of livers with microvesicular fat safely expands the donor pool

    Transplantation

    (1997)
  • Cited by (0)

    View full text