Cancer histopathologic subtypes prediction from miRNA expression data using pattern recognition

• Main Authors: Razak, Eliza, Yusof, Faridah, Ahmad Raus, Raha
• Format: Article
• Language: English; English
• Published: RGN Publications 2018
• Subjects: TA164 Bioengineering
• Online Access: http://irep.iium.edu.my/64231/1/subtypes.pdf; http://irep.iium.edu.my/64231/7/acceptance.pdf; http://irep.iium.edu.my/64231/; http://www.rgnpublications.com/journals/index.php/jims/index

Despite numerous breakthroughs in personalized medicine, cancer still remains as a major leading cause of death globally. Cancer occurrence rate is increasing at an alarming rate in the world and it accounts for around 13% of all deaths worldwide. Having anaplastic, dysplastic, and metastatic properties in normal cells, are the initiative signs of carcinogenesis and metastasis. Conventional diagnosis of cancer subtypes are still based on pathologic guided histopathological diagnosis. However, certain clinical signs and symptoms usually appear in late stages of cancer and hence the vast majority of patients are diagnosed with cancer very late. Therefore, it is of paramount importance to predict the type of cancer and its histopathologic subtypes in early stages so that specific treatments can be sought. Furthermore, it has been demonstrated in the medical literature that specific treatments based on histopathologic types could dramatically increase the survival rate of the patients. Nevertheless, clinicians cannot fully rely on blood-based protein biomarkers in actual clinical practice because of low sensitivity and specificity, time consuming and intensive laboratory sampling procedures, and fail to provide early cancer detection. Circulating miRNAs have gained great interest in medical field because of their higher sensitivity, specificity and potential for minimally invasive sampling procedures. Furthermore, miRNA expression profiling from body fluid samples using high-throughput approaches is a promising technology that could predict cancer histopathologic subtypes. This paper describes a pattern recognition based approach called one-dependent estimator to predict cancer histopathologic subtypes from miRNA expression data. The proposed framework will predict particular cancer subtypes. To select relevant cancer subtypes associated miRNAs, we employ an entropy-based marker selection approach. This proposed system has achieved an average accuracy of 96.21% in predicting cancer subtypes over 4 datasets, namely pancreatic cancer, ovarian cancer, lymphoma, and renal cell carcinoma (RCC).The experimental results strongly suggest the efficacy of the entire proposed framework.