Major liver diseases such as hepatocellular carcinoma, cirrhosis, and hepatitis affect huge numbers of people. For example, each year, death from liver cancer alone is 662,000 worldwide. Early diagnosis of liver disease is crucial. If identified in time, hepatitis C patients can be treated with interferon to lower risk of developing cirrhosis; and cirrhosis patients can be closely monitored and treated to stop progression to liver cancer. Previous gene expression analysis of liver diseases mainly focuses on identifying transcriptomic signatures that differentiate a single liver disease (such as hepatocellular carcinoma) from normal liver. Disease signatures identified in this way may not be specific to a particular liver disease, but may be a common signature for multiple liver pathological states because other liver diseases are not considered in a single unifying context. In our study, we collected microarray data of 365 liver biopsy samples from Gene Expression Omnibus, representing major liver pathophysiological states such as hepatocellular carcinoma , cirrhosis with or without concomitant hepatocellular carcinoma, hepatitis C and normal liver. Then we used a Top Scoring Pair based decision tree algorithm to classify all these liver diseases simultaneously and found a unique transcriptomic signature for each liver disease. The algorithm finds gene pairs that show relative expression reversal among different liver diseases and builds the decision tree in a bottom up way, with highest accuracy at the top, and lowest accuracy at the bottom. Classifying all these major liver pathophysiological states simultaneously ensures that the identified disease signatures truly represent the unique properties of each liver disease, because each signature is identified against all major liver pathological states, not just normal liver. The identified unique transcriptomic signatures may provide valuable information for targeted biomarker discovery for liver diseases, which would enable the early diagnosis and stratification of patients for the most appropriate treatment.
You will be able to download and print a certificate for these PDH credits once the content has been viewed.
If you have already viewed this content,
please click here