Curing Cancer with Big Data

How big data analytics is speeding cancer tumor profiling.

Cancer is a formidable foe.

Oncologists have long known that cancers arising in different body organs, or in the same organ in different patients, progress and respond differently to treatment.

Pullquote share icon. Share

“One size fits all” approaches are being replaced by strategies reflecting the unique molecular profile in a patient’s tumor.

Profiling the alterations in molecular signaling (information) pathways responsible for these differences is revealing additional complexity with major implications for how cancer is diagnosed and treated.

Cancers can no longer be described based simply on site of origin.  Rather, cancers occurring in the same organ are now known to comprise multiple cancer subtypes distinguished by distinct patterns of altered molecular signaling pathways.

Additionally, we also know that cancer is not static. The mutations and other genetic alterations that drive cancer formation continue to change with tumor progression and metastatic spread to other organs.

This leads to the relentless emergence of tumor cells with widely differing properties and drug responses within both the primary tumor and different metastases in the same patient.

As more is learned about different cancer sub-types and the extravagant patterns of tumor cell heterogeneity, the cancer research and clinical oncology communities have recognized that profiling perturbations in molecular signaling networks is necessary to better predict tumor behavior and — of particular importance — understand how these alterations are associated with response (or resistance) to specific anti-cancer drugs.

These insights are forging a new era of precision oncology in which the historical “one size fits all” treatment approaches are being progressively eclipsed by tailored strategies reflecting the unique molecular profile identified in a patient’s tumor.

Although genetic changes underlie cancer’s initiation and progression, genomic information alone is insufficient to fully understand (or predict) the altered patterns in molecular signaling networks that determine tumor behavior (the phenotype).

Connecting genotype and phenotype requires profiling cancer cell functions at multiple levels: genome; transcriptome; proteome; metabolome (“panOmics”).

Caris Molecular Intelligence™, which uses panOmics profiling to provide oncologists clinically useful treatment information, identifies approximately 70 molecular targets in different cancers that are linked to 55 FDA-approved drugs.

However, only 19 of these drugs are identified by genome sequencing, with the remaining majority profiled by protein analysis.

Caris pioneered the use of panOmics profiling, with more than 65,000 cancer patients profiled to date. Caris Molecular Intelligence™ uses advanced profiling technologies, including genome sequencing, immunohistochemistry, fluorescent and chromogenic in situ hybridization and flow cytometry, to provide a comprehensive evaluation of the molecular alterations found in the cancer.

Pullquote share icon. Share

Patient care will  require integration with data on patient history, treatment outcomes and environmental or lifestyle factors.

This profile is then interpreted in the context of published clinical guidelines, drug compendia and the published clinical literature to provide oncologists with information on which of the 55 FDA-approved cancer therapies are potentially most relevant (or potentially ineffective) for the patient and/or assess opportunities for enrolling patients in clinical trials for new drugs being developed by the pharmaceutical industry.

In common with many aspects of contemporary biomedicine, the future landscape for molecular profiling and clinical oncology will be dominated by big data. Caris already generates terabytes of data per day on patients profiled.

As market penetration for profiling grows, and new high-dimensionality panOmics biomarkers are added to the profiling menu, generation of ever increasing amounts of data per patient and the need for management of multi-petabyte scale profiling databases will become the reality.

Proficient use of this information in patient care will also require integration with data on patient history, treatment outcomes and information on environmental or lifestyle factors that may affect disease risk and treatment responses, along with seamless migration of these integrated databases into electronic medical records.

New tools will also be needed for visualization of increasingly complex profiling data to build facile decision support tools for already over-burdened physicians.

These data-intensive trends demand establishment of better infrastructure and capabilities in large scale data annotation, analysis and curation.

Caris Life Sciences partnership with IBM and IBM Premier Business Partner Re-Store to implement a secure scalable and data-aware infrastructure provides Caris the significant scale and performance gains that will be critical to enabling more precise cancer diagnosis and care decisions. goldbrown2

This article first appeared on September 17, 2014 on IBM Smarter Planet Blog and was republished with permission.

Click here for more on Caris Life Sciences.

pain in the chainUPS recently launched it’s Seventh Annual Pain in the (Supply) Chain Survey, which takes a deep dive into the untapped opportunities in the healthcare industry. Click the image to access the full report.

Dr. George Poste is the vice-chairman of Caris Life Sciences. He also serves as Chair of the Caris Life Sciences Scientific Advisory Board and as Director of the Complex Adaptive Systems Institute at Arizona State University.

Click the RSS icon to subscribe to future articles by this author. RSS Feed