What was so revolutionary about the Oncotype DX test, and what impact has it had on cost and treatment?
Before Oncotype DX, the vast majority of women with tumors over a centimeter received cytotoxic chemotherapy because there was no way to systematically identify those who would vs wouldn’t benefit from chemotherapy treatment using clinical or pathologic features. Large studies at that time suggested that, depending on age, the number needed to treat (NNT) with chemotherapy was between 15 to 33 women to save just one life. Because it can be physically draining, has multiple side effects, and is costly, there was significant interest in identifying women who could safely forego the treatment.
Along comes the Oncotype DX test, a 21 gene assay that provides a low-, medium-, or high-risk score that is predictive and prognostic for women who are more or less likely to benefit from chemotherapy. It also revolutionized the upfront management of women with invasive, hormone receptor-positive, breast cancer.
One goal of the test was to reduce the use of chemotherapy at the population level, but because studies were largely limited to clinical trials or single academic institutions, they didn’t reflect the general population. So we embarked on the first nationally representative investigation of patients who received an Oncotype DX test using Medicare claims data in women aged 65. It was originally thought we couldn’t use claims data to reliably detect test receipt because the diagnostic is billed with a generic CPT code. Yet, in looking closely at the claims, I determined the test always had the same cost, zip code, and provider code, so I developed a simple algorithm to detect its receipt. Researchers have since used this approach and variations, thereby confirming its validity.
The three questions I was interested in were 1) who received Oncotype DX, 2) what was its impact on chemotherapy utilization, and 3) what was its impact on cost? Our initial findings were nuanced and context-dependent. For example, if a woman had high-risk disease based on clinical factors (i.e., involved lymph nodes), receipt of the test was associated with less frequent chemotherapy utilization. However, if she had low-risk disease based on clinical factors (i.e., the tumor was less than a centimeter in size), then receipt of the test was associated with more frequent chemotherapy utilization. Following the initial adoption of Oncotype DX testing, there wasn’t a discernable net impact on overall chemotherapy use, but after the test had been around for a while, we started seeing evidence of a population-level decrease in chemotherapy utilization and inpatient costs. We suspect it took a while to see the impact materialize due to more selective use of Oncotype DX testing during initial adoption, where younger and healthier women were more likely to be tested and receive chemotherapy. However, after the initial uptake bias lessened, we saw a beneficial impact of Oncotype DX on chemotherapy use at the population level. It should be noted that my studies were limited to older women (over age 65), and we think the benefit of Oncotype DX in reducing chemotherapy might have been strongest in younger patient populations.
New treatments may have the potential to lessen heath and treatment disparities, from your research can you say whether this is happening?
As a society, if we are to reduce disparities in care, we have to know all the drivers, be able to intervene at all levels of care, and understand all relevant issues. We know common barriers to cutting-edge care—whether your doctor is up-to-date on the latest technologies, if your hospital is involved in clinical trials and has the latest equipment, or if you’re being treated at a top-level cancer center. We also know patient adherence affects disparities—if medicine is prescribed, and whether or not the patient can afford their co-pay or withstand the side effects. We also know that medical disparities reflect a broader systemic and institutionalized racism that permeates American culture.
But identifying what causes care inequality is a complex question that I ask in my currently funded R01. I’m looking at disparities in oral cancer agent utilization and adherence in renal cell carcinoma from the perspective of the patient, provider, and system-level characteristics and pinpointing the drivers to intervene at all care levels. Additionally, in another grant proposal under review, l hope to study patients with renal cell carcinoma within a prospectively created registry that ties social, clinical, demographic, biologic, and genomic factors into a single, diverse dataset.
I think people are looking for a single answer when it comes to health disparities, but once again the data are more nuanced. For instance, if you look at PET scan utilization by race and ethnicity, we continually see persistent differential usage by race because the test requires substantial infrastructure, investment, and maintenance by health care facilities. Telling a different story, we don’t see racial disparities with Oncotype DX usage data, most likely because patients don’t encounter treatment barriers, such as facility-level infrastructure—anyone can ship the diagnostic to Redwood, CA. But when a student of mine was researching a different Oncotype DX data set and concordant care—she saw disparities emerge and found that compared to minorities, white women were more frequently tested outside of what was considered guideline-concordant care, which amounted to better-quality care. We know this because later, guidelines were expanded to include the care white women received before it was recognized as standard.
In your estimation, what has had the greatest impact on cancer care?
Personalized medicine and immune-oncology treatments have been real game-changers in cancer care. Unfortunately, immune-oncology agents, such as checkpoint inhibitors, don’t work on all cancers and are only effective in a small subset, maybe 20% of the population, for most cancer types. But for whom it does work, it can be akin to a cure. Unlike the typical, blunt-tool cancer treatments—chemotherapy, surgery, and radiation—these new immune-oncology drugs use one’s immune system to recognize and attack cancer as a foreign invader. Expanding on this new family of drugs, researchers are now developing more immuno-oncology agents, doing clinical trials to come up with different cocktails, creating diagnostics to predict who will benefit, and determining how immune-oncology agents fit in with chemotherapy and radiation. We now have treatments, like CAR T therapy for example, that can genetically engineer a patient’s immune cells to recognize cancer cells. Although exciting, these types of therapies haven’t been developed and/or shown to work in many other cancer types just yet.
I have a pending R01 application that will study the survival, cost, and sequencing of treatment in patients 65 years and older who receive immune-oncology agents. I’m interested in this age group for several reasons—number one, they aren’t represented in clinical trials, yet most people receive a cancer diagnosis when they are 65 and older. Number two, we’ll soon be hit with a “silver tsunami” (baby-boomers turning 65) causing a huge jump in cancer diagnoses that will tax the oncology system and force us to determine ways cancer survivors are brought back into the normal healthcare system.