Abstract
Introduction
Karen E. Knudsen, MBA, PhD, is a world-renowned leader and expert in cancer research. She has recently taken on several new roles, most notably the Chief Executive Officer (CEO) of the American Cancer Society (ACS). She was previously the Executive Vice President of Oncology Services for Jefferson Health and Director of the Sidney Kimmel Cancer Center.
She also serves on the board of advisors for the National Cancer Institute (NCI) as well as numerous advisory boards for NCI-designated cancer centers. She is an active member of several committees with the American Society for Clinical Oncology, in addition to serving on other academic and for-profit advisory boards.
In this interview, AI in Precision Oncology, Editor-in-Chief, Doug Flora, asks Dr. Knudsen about her career journey, her challenging new role at ACS, and her thoughts on the impact AI will have on oncology research and patient care.
This interview has been lightly edited for length and clarity.
One of the things that drove me, not only with my team at Jefferson but now also at the American Cancer Society (ACS), is innovating so that we are advancing progress against cancer at every possible level including prevention, detection, cure, and survivorship. And then, increasing access to that care. I view my role now as CEO of ACS as taking that goal, which for me was a Philadelphia-focused endeavor, to a national and even a global endeavor. At ACS, we strive every day to improve the lives of cancer patients and their families through research, advocacy, and direct patient support in 5,000 communities across the country. ACS had been my long-term partner and now it’s the best job I never thought I’d have!
We also recognize that we can do more, that we can incorporate new innovations in technology in our grant mechanisms, but also in the science that we conduct assessing cancer risk.
Incorporating AI will allow greater integration among complex datasets—that’s one of the ways we can make formidable gains in enhancing understanding of risk. But it’s also a way for us to enhance discovery at the level of precision medicine, especially if we’re looking at large, complex clinical trial data of responders and nonresponders. The ability to take not just subcomponents and analyze them but the entirety of the data collected on those patients, to try to determine the best hypothesis about who was responding, who didn't, and what that might look like in terms of a hypothesis for the next clinical trial.
But AI is also being used in precision medicine to escalate drug discovery in a targeted way—what is the protein or antigen that we’re after? We’re now able to incorporate AI to more quickly develop potential strategies to target or to deliver a target to that particular antigen.
We’re seeing this across the board at the level of discovery, but only a handful, less than 5% of adults, are enrolled in a clinical trial, which we know to be the most advanced form of care for cancer patients.
So how we handle that part of the problem now is that someone, to get on a clinical trial, needs to be a unicorn, given the very significant investment that it takes to assess someone for clinical trial eligibility before they walk into the clinic door. I remember trying to staff that up for every part of my organization, and it was very difficult. AI could potentially help us with that but also help determine much more quickly whether someone would actually meet the inclusion criteria for a clinical study. Patients want it. It’s just not simple for them to navigate on their own, and it is not simple for health systems to devote the kind of resources needed to do that manually.
But these are the guidelines that are for individuals of average risk. What you’re describing, which is so correct, is that we have more information now so that we can determine whether someone truly is of average risk versus our ability to make that assessment 10 years ago.
Why? Because we know that risk is not just your age and your cancer history, your family history, and your genetics—if you know it—but your exposures. Many different factors go into whether someone is of average risk. So, making adaptable guidelines in the future is where we are going as well at ACS. AI will enable that, and that includes polygenic risk scores. We are in a much more sophisticated way going to be able to tailor cancer screening strategies to the individual instead of it just being pinned on what was your last birthday.
We are keen to ensure that that same fate does not befall what portends to be a potentially game-changing way to identify cancer risk in individuals. Multi-cancer-early detection tests (MCEDs) have a potential to truly identify cancers at an earlier stage when individuals are more likely to have a successful therapeutic intervention. It's also the case that MCEDs, by nature of the ease of testing, have an ability to extend cancer screening into populations that currently don’t have access to screening at the level they should. This includes rural populations and individuals who are on hourly wage and can’t take time off work to go to cancer screening. These are real barriers that could be overcome by MCEDs.
Where we have taken a position on MCEDs is multifold. First, we’re investing in research and conducting our own research to determine whether we could have predicted meaningful cancers earlier and as well predict whether we could have identified cancers among nonscreenables like pancreatic or ovarian cancer. We are also looking from an advocacy perspective to lay the groundwork for a pathway for reimbursement. We're not endorsing any particular MCED but rather have sufficient confidence that this is the future to lay down the potential groundwork for reimbursement once two criteria have been met—FDA approval and evidence of clinical benefit.
We’re keen to ensure that MCEDs are utilized in a way that add to our current body of knowledge and do not shift the balance toward overtreatment.
We have had three major wins in the past 2 years. The first had to do with how it is that you get access to an innovation. There is a way to do remote cancer risk testing for colorectal cancer—the at-home stool-based tests. But if someone screened positive, now their colonoscopy—the thing that they needed next—was no longer completely reimbursed. That was a barrier for people going to get life-saving colonoscopy. So, we went to the Biden Administration and asked for the removal of this financial penalty that truly was a barrier to care. As of January 1, 2023, that was realized. We called that our first ground shot! We love a good moon shot, but ground shots are things we can do right now.
Second was January 1, 2024—patient navigation as a reimbursed component of complex cancer care. We know that patient navigators change lives and enhance understanding of care. It improves adherence to a care plan, reduces the overall cost of care, and is associated with enhanced survival. Again, through data and through research, we were able to see that navigation is something that is mission-critical for cancer patients. This was incorporated into a reimbursable component for cancer care. ACS is helping to train those navigators through best practices so that they can maximize their ability to help cancer patients.
The third win was at the state level. We’re 14 states down regarding reimbursement for biomarkers. We know that 50% of FDA-approved oncology agents in the past few years require or recommend biomarker testing to better match patients to therapy the first time. But it wasn’t happening because it wasn’t being reimbursed at the state level. We went to work—first in Illinois and now 14 states have had this passed into law – to require reimbursement for biomarker testing.
This is one of those scenarios where we think this is a win for everyone. Getting someone to the right care the first time can offer a phenomenal opportunity for survivorship and generally reduces the cost of care. We’ll continue to work on the other states that have not yet passed biomarker legislation.
[My patient] is coming in: She’s got breast cancer; she’s been on hormone therapy; she’s starting to recover. We need to think about what the next stage is for her. Here’s everything we need to know about the quality of life for this patient. Here are the clinical trials she’s eligible for. These are the things that we need to prioritize.
Practitioner-assisted decision making is rapidly what we’re seeing. There is no better example of that than the inability to go through an electronic health record and galvanize information in a way that is time efficient to allow the oncologists and oncology team to spend more time doing what they want to do, talking to the patient about options.
It will also change the way that we train medical students. It’s actually one of the things we were proud of at Jefferson—we already started to shift away from just the didactic how did you do to how is your emotional intelligence/emotional quotient? Back to allowing practitioners to do the things that they want to do—spend more time in front of patients. Return medicine to being an art so that you evaluate that person sitting in front of you, complemented with the data that you learned about them from their EHR. And then have a conversation with them about their goals for care becomes an informed decision about the path forward.
We also know that AI has promise for helping us understand disparities that we currently cannot understand. For example, why are black women disproportionately burdened by triple-negative breast cancer? What is the underpinning cause? At ACS, we are about to start a 100,000-woman cohort called VOICES, specific for black women, to assess all the information when they enroll in this study to understand triple-negative breast cancer risk. We will not be able to make that assessment without AI-enabled decision making and data analysis.
As a former health care executive, the ability of AI to allow me to go back and ask complex questions about the care that was happening in my 16 hospitals will be critical. For everybody who came in with a diagnosis of triple-negative breast cancer, did they all get offered the same plan? Did everybody get offered a standard of care and if not, why not? Did everyone get considered for a clinical trial? If not, why not?
I think one thing that is probably not thought about as much is the ability of a health system to determine whether the health system works and whether they are delivering on their own commitment to equitable cancer care in every nook and cranny. I think that kind of insight will be really important as well.