Efforts should be focused on Clinical Decision Support for medical oncologists, surgeons and pathologists that is informed by cancer informatics and directed towards one or more of the following goals shared by patients and physicians:

The program will support a variety of approaches including the following:

  • Improved diagnosis, including via novel clinical assays, so that patients and physicians have confidence that the selected treatment strategy is appropriate for them and their disease.
  • Improved methods for assessing the likely success or failure of lines of therapy, especially via reliable biomarkers, so that decisions regarding choice of therapy can be made with confidence.
  • Reliable methods for defining high-risk populations for effective screening, so that cancer can be detected and treated as early as possible.
A genitourinary clinician’s perspective illuminates the value of these approaches  for patients:

With current diagnostic biomarkers, patients and physicians have limited tools to help them know whether the patient’s disease is aggressive and whether it will respond to specific therapies.  Early-stage patients treated with active surveillance have improved quality of life and in many cases equal or greater long-term survival than those treated with surgery, radiation, and/or androgen deprivation therapy. Understanding which prostate cancer patients have a cancer that may be so slow-growing that they will die of other causes long before their prostate cancer becomes fatal could help patients avoid the life-altering side effects of surgery and/or radiation. Current decision support techniques do not provide sufficient information to allow patients to make that decision confidently.  Prostate biopsies also pose a significant risk to patients; reliable analytics based on less invasive tools such as blood-based biomarkers or imaging may be helpful in stratifying patients to inform treatment decisions.

Further, metastatic patients may be treated with second-line androgen deprivation and chemotherapy that have debilitating and potentially fatal side effects. Patients with particular mutations may benefit from more targeted approaches such as immunotherapy or PARP inhibitors.

Potentially promising clinical questions include

  • How can methods for assessing “exceptional responders” in clinical trials be improved? Can cancer informatics enable genomic or multi-omic characterizations for comparisons of clinical trial patient sub-populations? In many “negative” phase II and III trials, a few patients may do very well on the experimental treatment (and are called “exceptional responders”), but the number of responders is too small to consider the drug worthy of further testing or of FDA approval. In this era of precision oncology, however, a new question is being asked: Do those “exceptional responders” have an “omics” profile that could be used as an inclusion criterion in a targeted clinical trial that would give patients with that profile or a closely related profile a promising new treatment option, through the continuing precision clinical trial and lead to a targeted therapy approved by the FDA? (reduce this down)
  • How can we leverage existing high-content (-omic, imaging, etc) data to identify profiles of patients likely to respond to, or develop resistance to, currently approved treatments?
  • Can the number of patients accruing to omics-informed clinical trials be substantially increased, thereby improving the chances of early approval of targeted treatments?
  • How can the presentation of actionable -omics data to oncologists be improved sufficiently to substantially increase their use in treatment selection and monitoring?
  • Can improved risk assessment for specific cancer types from clinical records lead to substantial improvement in screening and prevention programs and early detection?
  • How can liquid biopsy data be used to track response to treatments and monitoring and integrated with the other data used by oncologists?
  • Can databases be shared, merged or made more broadly available to accelerate research or support better decision making? Federated searches may also be an option.

Illustrative research approaches that could be considered to answer these questions:

  • Application of machine learning (ML/AI).
  • Technical and semantic linking of different data types, such as genetic and molecular profiles, images, disease states.
  • Multi-dimensional characterization of transitions between disease states and discovery of clinically actionable therapeutic biomarkers.
  • Improved analytics to support clinical use of non-invasive assays of potential biomarkers (e.g. circulating tumor DNA, microbiome-based, microRNAs etc) for disease diagnosis, disease monitoring, and response to treatment.
  • Develop methods to use liquid biopsy results and integrate with other omics data for decision making.
  • Develop methods to leverage heterogeneous omics, drug response and clinical data for deciphering regulatory pathways to study: (i) inter-patient population subtypes and intra-tumor heterogeneity, and (ii) mechanisms of drug resistance and response.
  • Develop methods for single-cell data analysis (including imaging data, e.g., multiplexed IHC) that have a direct impact on the selection of combination therapies.
  • Efficient data standardization within the scope of specific questions with clinical impact.
  • Augment available patient clinical data by expanding direct patient participation and empower patients to share their genomic and clinical data through novel informatics techniques.
  • Develop methods of predicting cancer susceptibility, progression, and therapeutic response based on environmental exposure and socio-economic determinants (e.g. living in a food desert, a high pollution area, distance to pharmacy) based on existing data sets.

Collaborative opportunities:

In addition to inter-institution collaborations, other types of partnerships can be considered:

  • Partnerships with specific disease foundations that can increase the funding available to support proposed projects
  • Collaboration with groups that have large databases that include clinical (EHR) data as well as omic data
  • Community engagement for the adoption of standardized data formats