Mining molecular data for cancer prognosis

How will any one patient fare during and after cancer treatment? It’s a question with no easy answer.

But a mega-database known as The Cancer Genome Atlas may provide insight into why outcomes differ among patients.

The atlas, a project of the National Cancer Institute and National Human Genome Research Institute, is cataloging genomic changes in various types of cancer. Begun in 2005, the project's goal is to complete genomic sequencing and analysis of 20 to 25 cancers by the end of 2014. Such an analysis will provide a better understanding of the basis of the disease and will help scientists diagnose, treat and prevent cancer.

Han Liang, Ph.D., assistant professor in Bioinformatics and Computational Biology at MD Anderson Cancer Center, believes that such comprehensive “molecular profiling” could one day help identify patients who may be resistant to certain therapies or may guide clinicians in tailoring treatment strategies to each patient’s unique genetic makeup.

Liang conducted a multi-institutional study of data from the Cancer Genome Atlas, including 953 samples from four cancer types — lung, ovarian, brain and kidney.

The study’s goal, he said, was to address how and to what extent molecular data from the Cancer Genome Atlas could impact cancer treatment. It revealed that molecular profiling, while still in its infancy, may eventually hold promise for cancer patient prognosis.

An article detailing his findings, “Assessing the Clinical Utility of Cancer Genomic and Proteomic Data across Tumor Types,” is published in the June online version of the journal Nature Biotechnology.

Focusing on the four cancers examined in the study, Liang looked at the potential for predicting patient survival by evaluating molecular data from multiple tumors, both alone or in combination with clinical variables such as patients’ age and tumor stage. He then developed an open-access database, allowing researchers to build and evaluate cancer survival prediction models based on the data.

“By analyzing data from multiple cancer types, we were able to evaluate prognostic models and identify gene alterations that led to tumor formation,” he said. “This would not have been obtained by looking at tumor data from just one cancer type.”

By combining molecular data and clinical variables, Liang observed a better prediction of cancer prognosis in three of the four cancers: kidney, ovarian and lung. He cautions, however, that further analysis is needed.

Other MD Anderson collaborators included Gordon Mills, M.D., Ph.D., chair of Systems Biology; Lauren Byers, M.D., assistant professor in Thoracic Head & Neck Medical Oncology, and John Weinstein, M.D., Ph.D., chair of Bioinformatics and Computational Biology. Other participating institutions included Baylor College of Medicine, Dana-Farber Cancer Institute in Boston, the University of California, Santa Cruz, Oregon Health & Science University in Portland, Massachusetts General Hospital in Boston, and The University of Texas at Austin.

The study was funded by grants from the National Institutes of Health (NIH) (CA143883), (CA175486), (CA016672); the National Cancer Institute-MD Anderson Cancer Center Uterine SPORE career development award; and from the Estate of George S. Hogan, and the Lorraine Dell Program in Bioinformatics for Personalization of Cancer Medicine funded by the Michael & Susan Dell Foundation.