By Dr. George Poste, vice chairman of Caris Life Sciences and director of the Complex Adaptive Systems Initiative, Arizona State University.
Cancer care is creating a new era of precision oncology. Profiling the molecular alterations in a patient’s tumor can identify changes that correlate with likely response or resistance to particular therapies. This type of molecular profiling also can lead to more precise treatment, replacing historical “one-size-fits-all” approaches. Advanced analytical technologies are revealing how different genetic mutations in cancer disrupt the molecular signaling (information) pathways that regulate normal cell function and produce specific pathway alterations in different cancers, subtypes of the same malignancy, and individual patients.
There’s no doubt that a spectrum of tumor-profiling technologies is needed to provide oncologists with the most comprehensive information on which to decide treatment options. Comprehensive molecular profiling of this kind is data-intensive, already generating up to a terabyte of data per patient. Refined insights will also come from comparison of individual patient data with the profiling and treatment response data from larger patient populations and from using large-scale analytics to iteratively improve the accuracy of actionable drug to molecular target associations. Thus, adoption of molecular profiling as a routine standard in cancer care will require sophisticated annotation, analysis, and secure curation of petabyte- and potentially exabytescale databases. The opportunities for market expansion are dramatic, but making this a reality will involve a complex interplay of technical, clinical, and economic forces.
The bottleneck no longer resides with profiling technologies; the challenge today is the data processing and corresponding analytics (e.g., identifying new molecular targets for diagnosis, prognostic assessment, and treatment selection). In turn, leveraging the full value of profiling requires integration with a patient’s clinical history and lifestyle data. Doing so helps identify confounding factors that may alter severity of disease and/or therapeutic responses. Currently, however, these data sets are fragmented in disparate systems often with incompatible formats that limit facile interoperability. Additionally, most electronic medical records are not yet designed for seamless extensibility to accommodate large volumes of diverse molecular profiling data.
This rise of Big Data in clinical medicine has created the need for new education and decision-support tools for physicians and payers. It is impossible for these stakeholders to remain aware and interpret the exponential growth in published literature. New services for literature aggregation, analysis, and ranking services will be required to set, and constantly update, evidentiary criteria for treatment and reimbursement decisions, together with automated tools to guide clinical decisions.
More than 1.5 million new cancer cases will be diagnosed and over 800,000 cancer patients will die in the U.S. in 2014. The value proposition for molecular profiling in cancer, as in other data-intensive settings, is the generation of accurate, patient-customized, actionable information that enables physicians and patients to make better–informed, realtime care decisions. Oncology has been in the vanguard of molecular profiling, but the value is not limited to cancer and extends across the entire spectrum of human disease, including profiling neurodevelopmental disorders that arise during fetal development. In addition to providing the intellectual foundation for precision medicine to improve patient care, the economic case for molecular profiling is equally compelling. By enabling high-cost treatment to be directed to only those patients likely to benefit and by eliminating futile interventions, molecular profiling can help determine how to balance infinite demand for care versus finite resources and how to control cost while improving patient care and outcomes.