Despite decades of scientific progress against cancer, access to treatment remains highly unequal. Some of the reasons — like institutional racism and poverty — are a reflection of our broader society. Other inequities may go unrecognized but are no less damaging. People with underlying health conditions, for example, may be excluded from clinical trials, preventing them from accessing leading-edge treatments and limiting the evidence available to guide their care.
With multiple drivers of inequity, we need multiple solutions. Every sector of the healthcare system — health systems, research sponsors, clinicians, payers, and policymakers alike – must do their part.
Those of us working at the intersection of clinical care and data science are no exception. My organization, for example, is confronting cancer disparities through use of “real-world” data on millions of patients who receive their care outside of clinical trials. Unlike many clinical trial participants, these patients reflect the full diversity of people affected by cancer.
To illustrate how real-world cancer data can inform and improve patient care, consider a fictional patient, “D.” As a 64-year-old woman with advanced lung cancer, she faces two well-documented barriers to high-quality care. First, as a Black woman, she is less likely to receive testing for key molecular biomarkers, which could determine which treatment might be most effective. Second, for the last 17 years, she has been living with lupus.
Lupus, an autoimmune disease, affects as many as 1.5 million Americans. Inevitably, many will also face a cancer diagnosis at some point in their lives. Yet people with autoimmune disease were mostly excluded from clinical trials that led to the approval of immune checkpoint inhibitors, the most important advance in lung cancer treatment in decades. As a result, D’s doctor has no evidence to decide whether she should receive these groundbreaking drugs.
Patients with autoimmune diseases were excluded from trials of checkpoint inhibitors out of concern that they would have more side effects or would not benefit as much as people without these diseases. The exclusion may seem logical, but it was not supported by evidence. In fact, it likely had more to do with ensuring a smoother path to approval by the Food and Drug Administration (FDA). For D and thousands of other patients, the harms of such a narrow approach to drug development are real and long-lasting.
With real-world data pulled from electronic health records (EHRs), however, we can begin to close the knowledge gaps. Recently, we worked with FDA to examine de-identified data on more than 2,400 patients with advanced “non-small cell” lung cancer who were treated with immune checkpoint inhibitors. More than 1 in 5 had evidence of a past or current autoimmune disease. Their doctors had decided to offer the treatments despite their underlying conditions.
Our analysis found that patients with autoimmune disease had similar outcomes when treated with immune checkpoint inhibitors as patients who did not. Although they were more likely to have certain immune-related side effects, they survived just as long and were no more likely to discontinue their treatment.
In other words, the treatment was useful and effective for these patients. Although they were excluded from the original trials, they should not be denied access to checkpoint inhibitors in everyday practice.
For someone like D, our finding opens the door to important treatment options — provided we make the necessary information available to her and her physician. Those of us in the big data field can help ensure that it is.
We can, for example, use electronic prompts to help ensure that D’s doctor tests her tumor for the various molecular characteristics that can inform treatment. (One of these, a protein known as PD-L1, would potentially make her a candidate for immune checkpoint inhibitors.) We can accomplish this, in part, through quality measures that oncology practices can use to ensure their treatment planning and recommendations are consistent with the latest body of evidence.
We are also developing innovative tools to present patient data in a way that is easy for doctors to use without disruption to their daily workflows. This is something that today’s EHRs often fail to do. At best, EHRs contain loosely associated data elements that are not displayed over the trajectory of a patient’s illness. Our task is to assemble these disconnected elements into longitudinal records – coherent “narratives” describing each patient’s diagnosis, treatments, and outcomes over time. Once we do, D’s doctor can gain confidence by comparing her with the aggregated narratives of patients who share her key characteristics and, like her, may not be represented in clinical trials.
These efforts are resource-intensive, and they push the limits of what technology can do to make sense of our disjointed healthcare system. While they cannot undo a long history of inequity in healthcare, they can be used to narrow research gaps and outdated best practices in clinical trials that conspire to limit patients’ access to care.
For all of us working on the frontiers of data and technology in medicine, this should be our guiding light: to help ensure all patients receive the best possible care, no matter who they are or where they live.
About Dr. Miller
Dr. Miller is the Medical Director of CancerLinQ LLC, a non-profit subsidiary of the American Society of Clinical Oncology (ASCO). CancerLinQ is a health technology platform developed and implemented by ASCO to improve the quality of cancer care and advance discovery.