Translating Computational Precision Oncology Into Clinical Practice Altitude Blog Post Aptitude Health 2020

Translating Computational Precision Oncology Into Clinical Practice

Precision oncology computational methods for diagnosis and treatment selection are generating increased attention in parallel with the increased use of targeted cancer therapies. To outline best practices for ensuring the clinical utility of predictive computational methods in the oncology clinic, the National Cancer Policy Forum and the Board on Mathematical Sciences and Analytics at the National Academies of Sciences, Engineering, and Medicine hosted a workshop in October 2018. A diverse group of stakeholders and experts were in attendance, including computer scientists, clinicians, industry leaders, health systems, federal agencies, and others.

Workshop attendees identified several factors critical to the effective translation of computational precision oncology into clinical practice: enhanced quality of data collection and management, validation and reproducibility of computational algorithms, regulatory oversight, and development of patient-centered and clinician-friendly tools. The group discussed the need for national standards for streamlining data collection to help ensure reliability and completeness of clinical data. Workshop attendees also proposed that computational methods be classified as medical device software, to be evaluated by the US Food and Drug Administration for potential risks they may pose to individuals. Validation of computational methods, including comparison against standards of care and clinical practice guidelines, as well as periodic recertifications should be ongoing to ensure sustained quality and accuracy.

High Altitude: As precision oncology becomes a mainstay of cancer therapy, oncologists now have access to a vast number of data regarding gene mutations and molecular structure of a patient’s cancer. The reliability of the data used to develop and apply computational algorithms is fully dependent on their completeness, quality, diversity, relevancy, timeliness, and accuracy. Concerted efforts to collect and align data from diverse groups can help reduce the risk of creating invalid and biased algorithms for eventual clinical application. Cancer clinics and hospitals can ensure sufficient support of computational oncology by hiring staff and acquiring technology capable of compiling and structuring data, generating reports to support oncologists, and communicating securely with other computational systems. Additionally, research training in informatics may serve as a foundation for accelerated training of clinical oncology fellows.

Ground Level: Interpreting biomarker test results and other computational methods, and using that information to select a particular treatment, can be daunting in the oncology clinic, especially when time is limited. To facilitate appropriate use of computational methods, treating physicians need to access multidisciplinary teams of pathologists and bioinformatics engineers with broad expertise and deep training in clinical informatics. This can be facilitated through the use of tumor boards or precision oncology boards that meet regularly, focus on the patient, and are designed to elicit communal feedback.