Synergistic effect of different levels of genomic data for cancer clinical outcome prediction.

There have been many attempts in cancer clinical-type classification by using a dataset from a number of molecular layers of biological system. Despite these efforts, however, it still remains difficult to elucidate the cancer phenotypes because the cancer genome is neither simple nor independent but rather complicated and dysregulated by multiple molecular mechanisms. Recently, heterogeneous types of data, generated from all molecular levels of 'omic' dimensions from genome to phenome, for instance, copy number variants at the genome level, DNA methylation at the epigenome level, and gene expression and microRNA at the transcriptome level, have become available. In this paper, we propose an integrated framework that uses multi-level genomic data for prediction of clinical outcomes in brain cancer (glioblastoma multiforme, GBM) and ovarian cancer (serous cystadenocarcinoma, OV). From empirical comparison results on individual genomic data, we provide some preliminary insights about which level of data is more informative to a given clinical-type classification problem and justify these perceptions with the corresponding biological implications for each type of cancer. For GBM, all clinical outcomes had a better the area under the curve (AUC) of receiver operating characteristic when integrating multi-layers of genomic data, 0.876 for survival to 0.832 for recurrence. Moreover, the better AUCs were achieved from the integration approach for all clinical outcomes in OV as well, ranging from 0.787 to 0.893. We found that the opportunity for success in prediction of clinical outcomes in cancer was increased when the prediction was based on the integration of multi-layers of genomic data. This study is expecting to improve comprehension of the molecular pathogenesis and underlying biology of both cancer types.