InForm software: a semi-automated research tool to identify presumptive human hepatic progenitor cells, and other histological features of pathological significance

Anne S. Kramer,Luke A. Diepeveen,Bruce Latham,Caryn Elsegood,Lingjun Mou,Geoffrey J. Laurent,George C. Yeoh,Laura Ochoa-Callejero

doi:10.1038/s41598-018-21757-4

Abstract

Hepatic progenitor cells (HPCs) play an important regenerative role in acute and chronic liver pathologies. Liver disease research often necessitates the grading of disease severity, and pathologists’ reports are the current gold-standard for assessment. However, it is often impractical to recruit pathologists in large cohort studies. In this study we utilise PerkinElmer’s “InForm” software package to semi-automate the scoring of patient liver biopsies, and compare outputs to a pathologist’s assessment. We examined a cohort of eleven acute hepatitis samples and three non-alcoholic fatty liver disease (NAFLD) samples, stained with HPC markers (GCTM-5 and Pan Cytokeratin), an inflammatory marker (CD45), Sirius Red to detect collagen and haematoxylin/eosin for general histology. InForm was configured to identify presumptive HPCs, CD45+ve inflammatory cells, areas of necrosis, fat and collagen deposition (p < 0.0001). Hepatitis samples were then evaluated both by a pathologist using the Ishak-Knodell scoring system, and by InForm through customised algorithms. Necroinflammation as evaluated by a pathologist, correlated with InForm outputs (r2 = 0.8192, p < 0.05). This study demonstrates that the InForm software package provides a useful tool for liver disease research, allowing rapid, and objective quantification of the presumptive HPCs and identifies histological features that assist with assessing liver disease severity, and potentially can facilitate diagnosis.

Highlights

Hepatic progenitor cells (HPCs) are a heterogeneous population, expressing immature and intermediate phenotypes of biliary and hepatic lineages[1]
We use custom designed algorithms to determine whether InForm can (i) identify and quantitate presumptive HPCs comparably to trained investigators (ii) identify histological features including inflammation, fibrosis and fat which are important in grading liver disease, and known to influence HPCs, and (iii) score the necroinflammatory activity in acute hepatitis patients consistent with a pathologist’s assessment using the Ishak-Knodell scale
Three normal livers and four livers from the hepatitis cohort were stained for Pan Cytokeratin (PCK); a general epithelial stain that cross reacts with a wide range of cytokeratins, which have been used as a marker for cholangiocytes and HPCs35–37

Summary

Introduction

HPCs are a heterogeneous population, expressing immature and intermediate phenotypes of biliary and hepatic lineages[1] They are small ovoid cells with a high nuclear-to-cytoplasmic ratio. Due to the intricate interactions of HPCs with inflammation, fibrosis and fat, HPC research often necessitates the assessment of these parameters. The Ishak-Knodell system grades necroinflammatory activity using five categories; piecemeal necrosis, confluent necrosis, lobular necrosis and portal inflammation. We use custom designed algorithms to determine whether InForm can (i) identify and quantitate presumptive HPCs comparably to trained investigators (ii) identify histological features including inflammation, fibrosis and fat which are important in grading liver disease, and known to influence HPCs, and (iii) score the necroinflammatory activity in acute hepatitis patients consistent with a pathologist’s assessment using the Ishak-Knodell scale

Methods

Results

Conclusion