What Can Digital Health Innovators Learn from Oncology? (Sean Khozin)
Therapies for cancer are being developed at light speed and upward of 60 gene and cell therapies are projected to reach regulatory approval in the U.S. by 2030, according to the MIT NEWDIGS collaborative. Due to the nature of cancer, readiness for risks in drug development is much higher here than it might be in other medical fields. Can digital health innovators learn anything from oncology medicines development?
Tune in to the episode in iTunes or Spotify.
Sean Khozin is the CEO of CancerLinQ, a non-profit health technology company focused on improving the quality of care and health outcomes for all patients with cancer. He was the Global Head of Data Strategy and Data Science Innovation at Johnson & Johnson, before that he co-founded Hello Health, a technology company focused on developing integrated telemedicine, point-of-care data visualization, and advanced analytical systems for optimizing patient care and clinical research. He was also the Founding Director of a digital health incubator inside the FDA.
In this episode, he talked about work processes in oncology, innovation in oncology, the promise of decentralized clinical trials, and more.
Specifics of Oncology
Oncology is a rapidly developing medical field. According to the Comparator Report on Cancer in Europe 2020 the absolute number of people diagnosed with cancer rose around 50% in Europe over the past 20 years. However, the number of deaths only increased by 20%. The numbers show we’re making great strides in survival and treatments and early screenings.
Gene therapy is an exciting area of biotechnology, which is seeing continuous progress with many possibilities for the future. According to JAMA; precision oncology has had some major successes. To name a few examples - Imatinib has a 95% response rate in patients with chronic myeloid leukemia and extends quality-adjusted life by about 9 years. Another therapy - venetoclax has an 80% response rate in patients with chronic lymphocytic leukemia (who have a 17p deletion). The CAR-T (new chimeric antigen receptor T cell) therapy tisagenlecleucel has a 62% remission rate at 24 months among patients with acute lymphoblastic leukemia. But the overall effect of precision medicine on care for patients with cancer has been modest. According to this JAMA study, only 8% of patients with cancer are eligible for precision medications approved as of January 2018 and only 5% would actually benefit from them.
The growing number of options and volume of information however has its impacts on the work of oncologists. “Oncologists are quite unique physicians because, unlike other specialists, they end up being primary care physicians for their patients. That means that there are many different aspects of a patient's care that you have to think about as an oncologist. Couple that with the fact that every week there's a new drug that's approved. We have really amazing new tools and technologies like genomic sequencing, but also germline sequencing that has an impact on how we choose a certain drug. All this has reached a point where it’s become challenging for clinicians to figure out how to use all the available information,” says Sean Khozin.
If doctors used to have the option to choose between one or two chemotherapy options, they now have targeted therapies or immunotherapies. In some cases, a number of different therapies exist in the same class. On top of that, most patients nowadays are getting their genome and their tumor sequenced on their own. “All these mutations that we're discovering and variants differ from patient to patient. The totality of the mutations in the tumor at the individual level are quite different. Patients may have the same, what we call driver mutation like GFR, but they have a lot of what we call passenger mutations that may have an impact in terms of how to best design a personalized treatment plan for that patient. Let's add another layer of complexity. Now, patients are also having germline sequencing. Sometimes it is done as part of standard care. But a lot of times patients are just doing germline sequencing through commercial companies. So there's a wealth of information there that is emerging that could have an impact in terms of how you treat a patient. A germline mutation tells you about intrinsic factors, hereditary that is important at the individual level. And then your tumor sequencing information and the mutations that you identify tell you about the characteristics of cancer itself. So now just have to think about, okay, here is the tumor kind of mutation milieu in the background of a germline mutation foundation,” explains Sean Khozin.
Decision Support for Oncology
Clinicians decide on treatment paths based on clinical guidelines. These have been traditionally a way that physicians educate themselves in terms of how to treat patients and to personalize their treatment decisions. But to create a scalable clinical decision support system for oncology can be difficult because guidelines differ from hospital to hospital and country to country. Designing good systems comes down to data that you want to or might want to use. Big data is far from the solution for all the challenges in biomedicine today, Khozin emphasizes. In 2017 he co-authored an article From big data to smart data, published in Nature, about the Information Exchange and Data Transformation (INFORMED) initiative, which aimed to build technical and organizational infrastructure for big data analytics.
“I was trying to point out the fact that just like algorithms, data needs to be engineered to allow training of robust algorithms. And the volume of the data alone is only one dimension of big data. There are other dimensions of big data that are as important. Those dimensions are velocity - how quickly we're measuring the data and analyzing it, variety of types of data we are capturing, and also veracity - computational concepts that define big data beyond just the volume.”
Scaling Big Data and Why Future Guidelines Will be Algorithms
A few years ago the big data hype cycle reached its peak. The prevailing belief was that just having large volumes of data can help improve the signal-to-noise ratio. “In some cases that has actually been quite the case, but we haven't really seen any scalable models that have used noisy data in biomedicine to be able to make predictive algorithms that one could incorporate into existing work workflows with confidence. A lot of these algorithms tend to be quite impressive, but the risks associated with the uncertainties of the prediction in the context of biomedical research and healthcare add way essential benefits. So these algorithms haven't truly scaled,” Sean Khozin elaborates, adding that if the current trend continues, however, the guidelines of the future will be algorithms essentially allowing us to extract data from the real world. The concept of real world data - data gathered after clinical trials - is slowly developing with the use of sensors to measure the quality of life and patient outcomes in their home environments.
Decentralized Clinical Trials
Clinical trials are usually designed and executed in a very controlled clinical environment, with a very rigorously chosen group of patients, for the pharma company to be able to prove a therapeutic effect. “The way we conduct clinical trials has remained more or less the same since Second World War. We have made tremendous improvements in the ethical conduct of clinical research. We have good clinical practice guidelines, which have emerged out of the tenants of enormous trials. But for the most part, the way that we conduct clinical trials has remained the same. So it’s time for an upgrade and I think decentralized clinical trials can accommodate that,” says Sean Khozin.
Incorporating digital health in oncology clinical trials can help extrapolate from that experience and even change what should be measured as improving patients’ quality of life. “A lot of cancer patients in clinical trials believe that maybe the endpoints we're using are not important to them. If their tumor is five centimeters versus three centimeters, what does that really mean? So can we capture their experience with digital health solutions? Patients and advocacy groups are starting to talk more and more about the patient experience and capturing the patient experience. This has had a tremendous amount of impact on on thinking about how to incorporate digital health solutions into oncology clinical development programs, but also at the point of which you care.”
So what are decentralized clinical trials? In short, this refers to data gathering and therapy administration that is not done in a centralized location. “Especially since the beginning of the pandemic, a lot of what is done in clinical trials has been decentralized. What I mean by decentralization is where you collect the data and how you collect the data. What method for collecting do you use? Historically a lot of the data along the method of data collection access has been collected in a decentralized way. A lot of times you make a phone call to administer a questionnaire. That's an example of decentralized data collection that excludes asking the patient to come to a centralized location. Something people tend to forget is that telemedicine is a virtual synchronous remote connection to the patient. This can be a telephone call or a text message, or video. The next step in decentralized clinical trials is to standardize their design and what are the logistics that are involved.”
How Can Regulators Support Innovation?
Among his past roles, Sean Khozin was also the Founding Director of a digital health incubator inside the FDA. He built a multidisciplinary team of physicians, biostatisticians, and entrepreneurs and gave them a sandbox to collaborate and to ideate. “We also started to develop a portfolio of public-private collaborations and we wanted to de-risk innovations that were consistent with the mission of the FDA, which was to which is to advance public health and to advance things that can expedite the development and delivery of safe and effective therapies for patients.”
Among the programs aiming to help with the faster introduction of new solutions in clinical practice, is the FDAs Breakthrough Device Designation Program or Breakthrough Therapy Designation program. “They are intended to mobilize the expertise of the FDA to advance the development of breakthrough innovations. And so the FDA puts additional resources behind working with companies that have a breakthrough designated device to help them figure out how to design their studies and clinical trials to bring these solutions to the market. It is helpful and the FDA tries to be much more proactive and flexible in working with companies that have these breakthrough designated devices or drugs. When a technology enterprise fails or when a clinical trial fails, nobody wins. Clearly, the investors and the founders don't win, but most importantly, patients don't win. So if we don't help de-risk these innovations then we prolonged the period of experimentation with patients and if they're not deriving any benefit from these solutions, it's better to know that fast.”
This is only an except. Tune in to the episode in iTunes or Spotify.
Questions addressed:
You are a trained oncology specialist who is also a programmer, in essence, a dream employee for any health tech company because of the clinical and tech knowledge. For starters, how do you see that your perspective on digital health differs from the perspective of doctors without a tech background and tech people entering healthcare?
You were the Founding Executive Director of the FDA’s first multidisciplinary science and technology incubator INFORMED (Information Exchange and Data Transformation). What did an incubator inside the FDA look like and what’s happening with it now?
We want technology to enter clinical practice faster. The FDAs Breakthrough Devices Program is trying to achieve that by helping device manufacturers efficiently address topics as they arise during the premarket review phase. In essence, these is a highway to an FDA approval. replaces the Expedited Access Pathway and Priority Review for medical devices. Follow-up comment on the breakthrough device designation and what does it actually mean/not mean, apart from it being a great PR for any company that gets approval.
We met at The Economist Impact Cancer Series conference last year. You mentioned that the risk averseness in oncology is lower than in other medical fields, because of the nature of cancer. Many patients will have poor prognoses, have nothing to lose, only gain by entering clinical trials and testing new potentially breakthrough therapies. Can you talk a bit about the drug development process in oncology and reflect on the potential of transferring processes/thinking in oncology innovation to digital health innovation? To rephrase: In oncology, drug development is happening at warp speed. There’s a lot of interest from the industry to develop new therapies because cancer is on the rise and targeted therapies are profitable. What can we in digital health learn from oncology about the development of solutions for medicine?
Can you describe oncology treatment from the physician’s perspective? What kind of decision support systems are they using today?
How does all this impact the development of technologies for oncology?
In one of your past roles, you were the cofounder of Hello Health. A technology company optimising patient care and clinical research with smart use of data and analytics. focused on developing integrated telemedicine, point-of-care data visualization, and advanced analytical systems for optimizing patient care and clinical research. What specific needs must be addressed by healthcare IT solutions for oncology?
To step back a bit: new oncology treatments are entering the market very fast. Upward of 60 gene and cell therapies are projected to reach regulatory approval in the U.S. by 2030, according to the MIT NEWDIGS collaborative. But how would you describe the actual changes in cancer treatments in the last decades? To which extent are survival rates improving and to which extent are new therapies improving the quality of life of patients but not longevity?
Among the popular topics in digital health are also decentralized clinical trials. How would you describe their potential?