Principal Investigator: Dr Rajarshi Banerjee
Department: Oxford Centre for Innovation
Perspectum Diagnostics Ltd, Oxford Centre for Innovation, New Road, Oxford, OX1 1BY, United Kingdom
Collaborating leads –
Dr Pradeep Natarajan – Broad Institute, Boston, Massachusetts, USA
Dr Nicola Guess, King’s College London
Tags: 9914, disease, Fat, Fibrosis, inflammation, Iron, Liver
Professor Stefan Neubauer – University of Oxford, Oxford, UK
Dr Constantinos Parisinos – University College London, London, UK
Dr Jake Mann – University of Cambridge, Cambridge, UK
Professor Jimmy Bell – University of Westminster, London, UK
Dr Theresa Tuthill – Pfizer Inc, USA, New York, New York, USA
Dr Stephen Harrison – San Antonio Military Medical Center, Fort Sam Houston, Texas, USA
Funding body: Perspectum Diagnostics Ltd
1a: Perspectum Diagnostics has developed a method of analysing magnetic resonance imaging (MRI) data that gives an accurate estimate of the amount of liver fat, the amount of liver iron, and the extent of inflammation and scarring in the liver. These three characteristics of the liver are also the most important in the diagnosis of liver disease. By analysing the abdominal MR images from all UK Biobank participants, we can determine approximately how many have abnormal liver composition, and the distribution of each of these measures in the population. Finally, and most importantly, we can examine the outcomes of the participants with liver disease, and determine which biomarkers are predictive of these outcomes.
1b: New, clinically meaningful data will be generated from the existing DICOM images, and fed back in to the UK Biobank data repository. These data will be directly relevant to future health outcomes and of use to other researchers. Excess liver fat is associated with coronary artery atheroma and metabolic syndrome, and is strongly associated with obesity-related disease. Liver fibrosis and inflammation are both associated with adverse outcomes, which is especially relevant in those with fatty liver disease. We will be able to show which patients have liver disease, and future researchers can link these findings to specific outcomes.
1c: The MRI scans from the imaging enhancement study will be analysed by LiverMultiScan to determine liver fat, iron, inflammation and fibrosis (LIF score). These measures have separately been validated against liver biopsies from patients.
These data will then be compared to measures of body composition, serum markers (lipid profile, iron stores, CRP and others) and habits associated with liver disease (eg alcohol intake, exercise and diet).
We will follow up all patients and identify those with a liver-related clinical outcome (eg liver failure, hepatic encephalopathy), and determine which prognostic factors best predict these outcomes in this population.
1d: All 100,000 participants from the UK Biobank imaging enhancement study (ie the full cohort from the imaging enhancement study) will be analysed to determine the baseline liver health profiles of the population.
Clinical outcomes data will be collected, with the aim of capturing
– every liver-related death
– every episode of oesophageal variceal bleeding
– every new diagnosis of cirrhosis
– every new diagnosis of liver failure or gross ascites due to liver disease (excluding malignant ascites)
– every new primary hepatocellular carcinoma and cholangiocarcinoma
– every new pancreatic carcinoma
Application Number / Title: 9914 – Determining the Outcomes of People with Liver Disease
Applicant PI: Dr Rajarshi Banerjee
Applicant Institution: Perspectum Diagnostics Ltd
PROJECT EXTENSION – APPROVED BY UK BIOBANK 01.12.2016:
“We would like to further request the fields within categories 10060 (Mental Health) and 125 (Bone size, mineral and density by DXA). These would be in relation to work to establish the effect of iron overload on various outcomes at a population level. On the latter category, 125, EASL 2010 guidelines state a link – referencing papers stating a link, though none with a potential sample size of that now available in biobank (i.e Guggenbuhl et al 2005).
In relation to category 10060 – interest is two-fold. Firstly, reports from patient surveys from major patient groups like the Haemochromatosis Society indicate a link between poor mental health and iron overload (Haemochromatosis Society 2016). Other reports indicate a link with conditions such as bipolar disorder (Serata et al 2012, Feifel et al 1997) – but state further study is required. On another level, links between mental and physical health, and separate mental health burdens, have often been ignored by societal norms in the past – and warrant further investigation as a matter of course.
Furthermore, haemochromatosis, and hepatic iron overload in general, have been shown to predispose people to the development of hepatocellular carcinoma (HCC). Given this, early symptom detection would be useful to aid the identification of at risk populations.
PROJECT EXTENSION – APPROVED BY UK BIOBANK 08.10.2016:
Liver disease has been cited as a major cause of mortality and morbidity in the UK and across the globe. The Chief Medical Officer has named liver disease as one of her prime targets (CMO reports 2012 and 2013). Imaging methodologies have the unique capability of providing information about internal structures without being invasive, therefore such an approach enables the potential of advanced image phenotyping to be explored.
There is an increasing awareness that excessive liver fat (termed steatosis), is a precursor for insulin resistance and type II diabetes (Roden, 2006; Stefan, Kantartzis, & Haring, 2008), as well as to liver inflammation (non-alcoholic steatohepatitis – NASH) and cirrhosis. The recent rise in obesity and insulin resistance (Anstee et al. 2013), has contributed to an increase in the occurrence and changes in the progression of excess liver fat into liver inflammation and cirrhosis (Hart et al. 2010, Liu et al. 2010). Additionally, liver inflammation and cirrhosis are associated with cardiovascular disease (Hu, et al., 2014; Targher, Day, & Bonora, 2010; Adams, et al., 2005). Moreover, NASH and fibrosis predispose individuals to hepatocellular carcinoma (Lade, Noon, & Friedman, 2014; Oda, Uto, Mwatari, & Ido, 2015).
While there are some known genetic factors for liver disease (e.g. PNPLA3 and TM6SF2) the mechanism of action is not fully understood, and there are many more factors that merit further investigation (Anstee et al., 2016). Many of these genetic factors have a role in cardiovascular disease, inflammation and lipid metabolism.
Several blood biomarkers, including platelets and white blood cell counts, are correlated with liver disease (Vallet-Pichard et al., 2007, Miyake 2013). So too are they explanatory variables for cardiovascular disease.
Using genotypes, phenotypes (including our MR derived phenotypes), and lifestyle choices we would like to develop accurate risk profiling for fatty liver disease, fibrosis, and HCC. Initially, this will be with principal components analyses, genetic association studies, and multivariable analyses. Further work will involve building more complex models to handle the complex interactions between liver disease, metabolism and genetics.
Perspectum Diagnostics has developed a method of analysing magnetic resonance imaging (MRI) data that gives an accurate estimate of the amount of liver fat, the amount of liver iron, and the extent of inflammation and scarring in the liver. These three characteristics of the liver are also the most important in the diagnosis of liver disease. By analysing the abdominal MR images from all UK Biobank participants, we can determine approximately how many have abnormal liver composition, and the distribution of each of these measures in the population. Finally, and most importantly, we can examine the outcomes of the participants with liver disease, and determine which biomarkers are predictive of these outcomes.
In addition, as part of this Research Project the Applicant will use the Data to develop and validate semi-automated and automated methods for calculating fat, iron, and corrected T1 in both the liver and the pancreas. This will include the development of semi-automated and automated methods to assess data quality, identify image artefacts, and identify biological features within the liver and pancreas.
Last updated Jan 15, 2020