the same nomenclature and information elements, but many databases use custom nomenclature and ontologies. To address this variability, participants at the ASCO Omics and Precision Oncology Workshop recommended that “standards development organizations should rapidly produce generally accepted, comprehensive standards for transmitting genomic information and should closely collaborate to avoid discrepancies between competing standards. Naming conventions for genes and genomic abnormalities should be harmonized and should be accepted by all except in the most exceptional of circumstances” (Hughes et al., 2017, p. 3154). Without agreed-upon standards, “we will not be able to build large datasets that inform cancer diagnosis, prognosis, and treatment,” Warner said.
Hricak said the complexity of cancer diagnosis and treatment requires a multidisciplinary approach to care that encourages building integrative teams of radiologists, pathologists, oncologists, primary care clinicians, and bioinformaticians. She said that all patients with cancer need access to a highly qualified workforce, but not all members of the care team have formal training in oncology, and the quality of cancer diagnosis and care can vary with factors such as case volume, experience, and time pressures. Hricak stressed that access to high-quality cancer diagnosis and treatment can be affected by a lack of insurance, narrow networks of clinicians within an insurance plan, cost-sharing burdens, preauthorization requirements, and prohibitions on self-referral. She added that these factors add complexity and administrative burdens to the care of patients with cancer, and they disproportionately affect those patients least able to navigate barriers to care.
Hricak summarized a number of strategies that could improve patient access to expertise and technologies in oncologic imaging and pathology, including enhanced education and training, clinical decision support tools, models of care delivery and payment, computational oncology, and data sharing.
In terms of education and training, Hricak said that oncologic imaging needs to be fully integrated into the curriculum of radiology residency programs, and should also be included as a clinical practice area for ongoing longitudinal assessment through MOC. She suggested that ACR create a certificate of special competency in oncologic imaging and recognize radiologists who obtain oncologic expertise, either through fellowship training or CME. Hricak emphasized the importance of reporting standards and com-
munication in pathology education and training, and added that further emphasis on molecular diagnostics and pathology informatics is warranted. For both radiology and pathology, she stressed that peer learning should be encouraged throughout academic and community settings of care.
Clinical decision support can also be leveraged to help non-expert clinicians improve cancer diagnosis and care; reduce unwarranted variations in care; and improve the accuracy, quality, and patient experience of care, Hricak said. To be effective, she said that decision support tools need to be efficient, evidence-based, educational, and either encourage or enforce the adoption of evidence. She acknowledged a number of challenges related to the implementation and scaling of clinical decision support, including clinician acceptance, costs for development and maintenance, and interoperability barriers. Hricak added that ideally, clinical decision support will shorten the amount of time it takes to disseminate new knowledge, but she noted that clinical decision support will only be acceptable to clinicians if it is simple and easy to use.
Hricak highlighted the importance of innovative models of care delivery and payment for improving cancer diagnosis and care. She said Vanderbilt’s DMT approach is an exemplar because it optimizes diagnostic testing, provides a comprehensive, integrated report, and improves the quality of care while also saving time and resources. She reiterated the goal of Project ECHO—to move knowledge, not patients—and encouraged clinician-to-clinician telementoring to build community capacity. She added that value-based payment models are an opportunity to incentivize collaboration among pathology, radiology, and oncology experts.
Hricak reported on essential elements for quality improvement efforts in cancer diagnosis and care, including a constructive, non-punitive culture; clinical and operational leadership; feedback on clinician performance and patient outcomes supported by data systems and measurement infrastructure; and the bandwidth for clinicians to engage in quality improvement efforts by removing burdensome non-patient care activities.
Hricak noted that machine learning and AI have the potential to fundamentally change how cancer care is delivered and will facilitate unified diagnostics and precision oncology: “There’s no question that integrated diagnostics is the future, but we are not there yet.” She noted that interoperability and standardization remain major challenges that need to be addressed in order to facilitate improved data sharing to study the interrelationships among diagnosis, treatment, and patient outcomes at scale.