A transformative application of artificial intelligence (AI) in drug development lies in its potential to match the right patient population with the right drugs, and this should be a primary focus of innovation efforts, said experts at a recent oncology-focused conference.
AI’s greatest impact could come from improving trial success rates by ensuring therapies are tested in the populations most likely to benefit, said Steve Warner, Senior Vice President and Head of Translational Research & Medicine and Oncology at Sumitomo Pharma America. Warner was speaking on a panel that delved into the impact of AI and global events on the oncology clinical trial industry at the 15th Annual Clinical Trial in Oncology (CTO) West Coast conference, which took place 14–15 May in Burlingame, California.
Warner outlined the transformative impact and practical applications of AI in his company, sharing that AI is allowing them to address very large and growing datasets to identify the right patient population and design trials that address their needs. He mentioned that Sumitomo has “dipped [its] toes in the water” with digital twins, taking preclinical and clinical study data to simulate trials. He stressed the significance of AI use in early clinical development, highlighting the adaptability that it brings to trial designs.
Telehealth as a patient-centric strategy in oncology trials
Another central theme of the panel was patient centricity and how patient burden can be reduced to improve trial adherence in oncology trials.
Some of the solutions that Jennifer Willert, Global Clinical Development Medical Director, Novartis Cell and Gene/Hematology, has integrated at institutions like University of California San Francisco (UCSF) and Stanford University, involve telehealth. Even prior to the pandemic, due to the large geographic spread of patients across California, Willert had experience with projects that incorporated telehealth into clinical trials. These projects embedded virtual consent conferences to make trial participation more accessible to patients and reduced the burden of travelling to sites multiple times, she said. With this approach, laboratory tests also do not have to be carried out at the clinical study site, she noted.
In clinical trials, there is often an unrealistic expectation that patients will regularly travel for tests and visits, Willert mentioned. Clinical trials are also frequently designed without sufficient consideration for the patient’s experience, leading to poor patient retention and incomplete data. Moreover, when a commercial treatment is already available nearby, it is increasingly difficult to justify asking patients to travel long distances for a trial, she highlighted.
Recognising this, some institutions now allow participants to transition into a “survival follow-up” phase, where fewer in-person requirements are needed and virtual options are provided, Willert added.
Applications of wearables in clinical trials
Building on earlier points about the transformative impact of AI, Warner’s team has begun incorporating wearables into studies, for indications such as myelofibrosis, where tracking symptoms, such as fatigue and night sweats, is critical and is traditionally done based on patient-reported outcomes (PROs). Wearables offer a more data-driven approach to capturing endpoints through step counts or heart rate variation, which may improve both insight and patient compliance, Warner said.
Whether wearable data can provide a push for novel regulatory endpoints is still a matter under discussion. However, Warner said the case of myelofibrosis research is unique because symptom score and spleen volume reduction endpoints were originally established for the first JAK inhibitor drug approval for the indication, which may not be appropriate for new agents with other mechanisms of action. According to him, sponsors now have the opportunity to contribute to the evolving framework that can advance novel endpoints through wearable use, but that remains an evolving discussion with regulatory agencies. While wearable data is not typically used as a primary endpoint, it is increasingly being considered as a valuable data.
Ethical and logistical considerations with wearables
While wearables have elicited a lot of interest, a member of the audience shared that in a breast cancer study from UCSF, only 16% of participants used wearables, emphasising how better adherence strategies are needed. Willert also brought up the ethical considerations of using wearables, like concerns about privacy and data collection.
Acknowledging the growing complexity around these issues, Warner stressed that sponsors often do not need or want access to all collected data that deals with things like location and other personal information. In indications where trials are focused on symptom tracking, wearable devices can relieve the burden of frequent questionnaires and may improve compliance, but concerns around data volume and privacy remain valid.
Willert advocated for the importance of transparency and boundaries in data handling. Encouragingly, there are new regulations that now require patient letters to be sent at the end of clinical trials, explaining the trial results in accessible language, she noted. This marks a meaningful step towards improving transparency, she said, adding that she is also an advocate for sharing insights with patients even earlier, in the primary and interim analysis stages.
On the topic of the impact of global geopolitical changes, Anthony Fermin, Clinical Program Director at Genentech, a part of Roche, said that some sites have stated they are starting to lose funding or coordinators for clinical trials, particularly in the US. He added that the industry is still figuring out how to navigate these challenges and questioned whether site augmentation services should be considered amidst the uncertainty. However, Fermin noted that the manufacturing sector is drawing more investment, especially from larger pharmaceutical companies, who have the resources to bring manufacturing back to the US or to invest in it.
