The American Society of Clinical Oncology is in the midst of developing a rapid-learning system that aims to connect physicians’ practices, measure their performance and quality of care, and advance best practices by providing doctors real-time decision support.
The Learning Intelligence Network for Quality, or CancerLinQ, was formally launched last year and is in the beginning stages of development. The system will be interoperable with the electronic medical records systems across multiple oncology practices.
CancerLinQ will “harness cutting-edge HIT to connect cancer patients, their healthcare providers, and researchers to a central knowledge base; … synthesize information from millions of physician and patient experiences; and … deliver up-to-the-minute, personalized information that allows every patient to receive the highest quality of care,” the society described in a report released in November 2011, titled “Accelerating Progress Against Cancer: ASCO’s Blueprint for Transforming Clinical and Translational Cancer Research.”
Some of the types of information that CancerLinQ will aggregate in an effort to provide real-time decision support include data from patients’ electronic medical records, information on clinical trials, and treatment guidelines. By analysing this information, the system will identify new trends in care and generate hypotheses about best practices. Physicians, in turn, can test these CancerLinQ-generated hypotheses in their practices and figure out whether they improve real-world care. Researchers can then apply those real-world lessons to improving the design of their studies.
ASCO plans to test the feasibility of CancerLinQ through a prototype focused on breast cancer, slated to launch by year end. “ASCO will be calling on you, our members, to help make the assessment phase a success,” the society said in its newsletter, ASCO Daily News, distributed during its annual meeting in Chicago earlier this month. “We will ask you to participate and move the field forward by sharing de-identified information on cases of breast cancer from your own practice.”
ASCO has projected that CancerLinQ will launch in 2020 and cost in excess of $100 million to develop. For the effort, ASCO will use a combination of internal and philanthropic funding. The society has garnered funding from Susan G. Komen for the Cure.
George Sledge, who served as ASCO's president in 2010 and is currently co-chair of the breast cancer program at the Indiana University Cancer Center, discussed with PGx Reporter why oncologists need a CancerLinQ-type system as the disease becomes increasingly characterized by patients’ molecular characteristics. Below is an edited transcript of the interview.
How did CancerLinQ start and why do oncologists need a rapid learning decision support system right now?
The idea of a rapid learning system is not original to us. It’s been floating around for a few years, particularly since there was an Institute of Medicine report [released in June 2010]. We’re moving into an era certainly where the computational ability and the ability that’s offered by the internet gives us new opportunities but also new challenges in terms of being able to bring the [oncology] community together. We felt quite strongly that oncology, like all of medicine, will be wired together, that at some point in time we’re going to have in essence some kind of vast network of physicians, insurers, drug companies all wired together, where you’ll be able to take information basically from every patient and make use of it. The questions are: Who does it? How is it done? Can we do it in such a way that helps us both to effectively maintain our healthcare system and also, hopefully, improve the care of cancer patients?
Molecularly guided personalization of cancer medicines was a big theme at last year’s and this year’s ASCO annual meeting. How will the molecular characterization of a patient’s disease be factored into such a system?
As genomics teaches us more and more about cancer, it's teaching us that there is no such thing as breast cancer or lung cancer or colon cancer, but rather these organ systems harbor a bunch of medically distinct malignancies. Once you start slicing and dicing these cancers into ever smaller subgroups it becomes harder to study these patients at one center or even a few centers like we would have done in the past. Almost out of necessity, from a research standpoint [we] are entering an era where … you’ll need to have multiple centers wired together [to study the association between rare gene markers and cancer]. Say, for example, the genomics of the patient shows that he has a MAP2K4 gene mutation and has colorectal cancer, but there are only one or two patients a year at each center. So, multiple centers will have form a research consortium, have web-based exchange of information, have genomic [analysis capabilities] at multiple sites, have a research enterprise that allows you to identify which trials are studying these subtypes and these diseases and allows you to use novel combination treatments when a patient has multiple … mutations involved. So, the science of this is pushing us in the direction of requiring increasingly the ability to have something like a rapid learning system.
The other thing is, in oncology, we’ve traditionally done these multi-thousand patient trials. Another implication of the genomics era is that we’re not really going to be able to do these types of studies because we’re looking at smaller and smaller and smaller subsets [of patients]. We’re going to, out of necessity, have to find new ways of learning about side effects [of drugs] and other ways of learning about efficacy. Having a CancerLinQ sort of rapid learning approach where every single patient teaches you something about toxicity and something about response could be very useful to the community as a whole.
The community as a whole includes stakeholders that don’t necessarily share patient information. CancerLinQ will try to join physicians, patients, and researchers in sharing data. But other than the research necessities driving them to work together, what other incentives to these groups have to join CancerLinQ?
First of all, we’re going to have to have affordable technology. It’s no mystery that a lot of groups have resisted getting into the electronic health record age simply because it’s very, very expensive. One of the necessities going forward is going to have to be things that incentivize physicians to be part of such a group. Those incentives may well be financial incentives but they won’t necessarily be shoving money at people. [For example,] instead of having two people devoted solely to arguing with insurance companies over payments, if you’re linked together and there is something that says, “You are within appropriate guidelines,” and then your payments will come through automatically, that sort of thing has the potential to make doctors’ lives a whole lot easier. So, that’s going to be a real incentive going forward that we should be able to cut out a lot of the very complex payment system that has developed in cancer [care] over the last 30 years.
So how will payors be involved in CancerLinQ? Insurance companies also collect patient data and usage data, and they have their own criteria for making coverage decisions. How do you plan to align payors’ coverage decisions with physicians’ treatment decisions?
Payors have large armies of people who do nothing but try and decide who gets what. So, there is absolutely a role for including payors in such a system [as Cancer LinQ]. We’ve certainly had multiple discussions with many payors and what we hear over and over again is that they don’t like the current system any more than we do … But if you have a national system, where you had agreed upon quality standards and the researchers, physicians, payors were all linked together, it would certainly make their job a lot easier, as well as ours.
So, are you planning to involve payors in CancerLinQ in some way in the near future?
ASCO has an annual summit with payors, called the Payor-Provider Initiative, where we bring payors together [with oncologists] to discuss areas of common interest. This is something we’ve done in the past and we’ll continue to do in the future. And certainly we’ll have extensive discussions with them over CancerLinQ.
In terms of health IT, what kinds of investments would physicians need to make in their practice? You mentioned that the expense of adding EMR capabilities to their practices has been a disincentive for physicians.
A lot of the things we currently do on a paper-based nature we need to replace with something that is seamless and electronic. If you’re a physician, you’re required to do board certification every 10 years. That is presently a process that involves countless hours tallying up stuff that you have to submit to the appropriate boards. If you could have a CancerLinQ-type system that automatically gave you practice patterns that a lot of boards ask for these days … that actually would make their lives simpler. If we could have built into the system — and it is certainly our intent to do so — decision support, so that when a patient came in and you put some data into the system and it said that the standard-of-care therapy is X, the doses of the drugs were X, and [this information] would automatically be populated by the system, so you wouldn’t spend half-an-hour writing up orders — that would make physicians’ lives simpler. Then imagine if the system could also inform doctors that a particular patient was a candidate for a national cooperative group trial. Having all this information available seamless and in the background ready for use, then it saves the most valuable commodity that every physician has, and that’s the physicians’ time. So, that’s going to be the major incentive going forward. But because there are multiple electronic health records systems, this is going to have to something that is interoperable with other systems.
Are you reaching out to practices that have some sort of EMR already implemented?
Yes, we certainly are.
Where are you seeing the most interest in joining CancerLinQ? Among academics or community oncologists?
We have discussed CancerLinQ with both academic and private practice centers. We already have a system called Quality Oncology Practice Initiative, which ... involves about 20 percent of the oncologists in the US. In this system, we access patients’ medical records and do quality analysis. We’re in the process of trying to move that from a paper-based system to an electronic system, which hopefully will be a whole lot easier for the practitioners. So, we already have a significant network of academic and practice physicians who we interact with and do quality measures on.
Then, is the idea to just migrate the QOPI network to the CancerLinQ system?
QOPI would become e-QOPI.
If we could go through a hypothetical example it would help illustrate how CancerLinQ could incorporate patients’ molecular data into care. There was a lot of excitement at the ASCO annual meeting over a small extension study that found that non-small cell lung cancer patients with rearrangements in the ROS1 gene responded to the Pfizer drug Xalkori similarly well as patients with ALK rearrangements. Although Xalkori is not approved for ROS1-positive patients, some oncologists said that the data presented at the meeting were sufficient for them to give Xalkori to patients with ROS1 rearrangements. What type of information would a system like CancerLinQ give doctors in such a situation where there is early but compelling data supporting a certain treatment strategy for a patient population with limited treatment options?
Any system like this will have to have several layers of decision support. One of those layers is for [treatment strategies] in guidelines with a high level of evidence, such as a Phase III trial. There is a lower level of evidence, for example, if there has been a presentation of a Phase II trial at ASCO that’s not yet peer reviewed, it’s not yet FDA approved, there aren’t yet any guidelines. It’s a reasonably straightforward matter to populate your system with that information so that it’s available. Of course, there is no guarantee that UnitedHealthcare is going to pay for it.