It's no surprise that health care is moving toward a future of tailored care for each person diagnosed with cancer. Precision oncology still has much room to grow, but research presented at the 2019 ASCO Annual Meeting shows great promise for realizing the vision of caring for every patient by learning from every patient. But, to drive precision oncology forward, informaticians must be engaged.
Informatician Elmer Bernstam, MD, MSE, MS, FACP, FACMI, Professor and Associate Dean for Research at UTHealth, explains the critical need for informatics to realize precision oncology's potential in personalizing care for every patient diagnosed with cancer. Here's what he told Oncology Times.
1. Why is informatics needed to realize the potential of precision oncology and how will it accelerate the transformation of cancer care?
“Precision oncology requires multiple steps from discovery to practice; each of these steps can benefit from informatics. Precision oncology may leverage a variety of patient-specific information such as behavior, preferences, and exposure history. However, for this discussion, let's focus on genomically targeted therapy.
“First, we must be able to learn what works. For example, which variants are clinically significant and what drugs effectively target the clinically significant variants? To accomplish this requires methods to analyze genomic sequencing data from individual patients. Genomic information must be combined with clinical data to identify useful associations between specific variants and outcomes such as response (or resistance) to a particular drug. Informatics can help standardize the collection of large clinical datasets. Further, routinely generated clinical data from patients' personal medical records have a great deal of potential to inform the care of patients with cancer.
“Second, to realize the promise of precision oncology as ‘standard of care’ for all patients with cancer, rather than a ‘one off’ project for each patient, data must be brought to the point of care and integrated into clinical decision-making. This cannot be done without computerized decision support. There are simply too many variables and the required information change too quickly. Today's variant of unknown significance may be an important target tomorrow. Simply tracking drug-gene associations is a formidable challenge that requires informatics.
“For example, our collaborative team from the UT Health Science Center at Houston and the UT MD Anderson Cancer Center developed an informatics tool called the Automated Identification of Molecular Effects of Drugs (AIMED) to serve as a reference for oncologists, helping to identify actionable drug-gene relationships to inform personalized treatment options for each patient.
“Finally, on a population level, informatics is the key to developing risk prediction models that adapt to changing demographics and treatment strategies for an increasingly diverse population.”
2. What areas of research presented at ASCO 2019 include the work of informatics?
“Researchers have explored various aspects of precision oncology, working towards a future where genomic tests will help identify optimal treatment options for patients, while minimizing ineffective treatments that are not likely to help. But, in order for researchers to develop these treatment options, there must be enough high-quality patient data to provide insight. Though almost 40 percent of Americans will be diagnosed with cancer, only a small minority (3-5%) of adult cancer patients actually participate in clinical trials.
“In an effort to provide clinicians and researchers optimal amounts of high-quality data and to advance cancer data sharing, a core set of data standards—minimal Common Oncology Data Elements (mCODE)—was developed and shared at the 2019 ASCO Annual Meeting. mCODE is essential in capturing and reporting characteristics, treatment, and outcomes of every patient with cancer. Electronic health records (EHRs) contain data from nearly 15 million people living with cancer in the U.S., but because not all EHRs collect data in the same format, they're often incompatible with one another. For that exact reason, cancer researchers are drastically limited in their ability to learn from patient records to later inform treatment options.
“Without compatible EHRs, data sharing is limited and, in some cases, not possible, making it difficult to share patients' medical data with doctors, hospitals, and researchers. mCODE works to combat incompatibility among EHRs by developing common data standards developed by informaticians in the oncology community. With common data standards implemented into EHRs, researchers can determine which treatments work, or don't work, for each person living with cancer.
3. There seems to be a lot of conversation around artificial intelligence's role in precision oncology. Can you speak on the relationship between informatics and AI to advance precision oncology further?
“Today's focused approach to cancer is to develop genetic risk prediction models, prevent or diagnose cancer earlier, and explore targeted treatment options for cancer. This requires genomics, proteomics. and metabolomics data being generated on clinical tumors, which is aided by artificial intelligence (AI), to classify tumors and predict aggressiveness and outcomes, therefore supporting personalized treatments.
“As people become enamored with the idea of AI driving research and providing new treatments, informaticians are working to identify important biomedical problems that can be addressed by AI. For example, many informatics groups are using AI/machine learning to improve early diagnosis of sepsis. At the American Medical Informatics Association 2019 Annual Symposium, informaticians will be gathering to discuss this approach of integrating informatics and AI to help promote precision oncology.
“Informatics helps oncologists use big data to match patients to the appropriate treatments or clinical trials. And in order to successfully implement precision oncology, informatics should be applied every step of the way—collection, understanding, and implementation—to inform the best possible treatment and care for patients.”