Estimated reading time: 4 minutes
Oncology research continues to push scientific boundaries, from biomarker-driven therapies to increasingly personalized treatment strategies. Operationally, however, many oncology trials are straining under the weight of that progress. That tension between scientific ambition and real-world feasibility was the focus of Advarra and Fierce Life Sciences’ latest webinar, “Building More Feasible Oncology Trials: Data-Informed Approaches to Study Design.”
Panelists included Paige Gonzalez, patient recruitment lead, oncology, at Bristol Myers Squibb; Shantheri Pai, associate director of clinical operations, oncology, at Gilead Sciences; and Laura Russell, senior vice president of AI and data products at Advarra. Mike Eckrote, senior vice president of strategic solutions and technology at Advarra, moderated the discussion, which explored what the data shows, how those patterns play out in day-to-day trial operations, and where AI can help teams make more informed design decisions.
Oncology trials are different, and the data shows why
A central theme of the discussion was what large-scale trial data reveals about the operational realities of oncology research and how that evidence can guide more feasible study design. Russell addressed this directly, explaining that oncology trials consistently carry greater complexity than studies in other therapeutic areas. Her observations align with patterns seen across Advarra’s dataset of digitized protocol-related documents and operational data from more than 30,000 historical studies conducted by approximately 3,500 sponsors.
Russell noted that oncology protocols average nearly three amendments per study, compared with roughly one and a half to two in non-oncology trials—an increase of about 40%. She also pointed out that participants in oncology trials undergo an average of 138 assessments, versus roughly 100 elsewhere, a level of intensity that directly increases both patient and site burden.
Enrollment expectations are higher as well. Oncology trials plan for average enrollment of just over 400 participants compared with about 380 in non-oncology research. At the same time, recruitment is often more difficult because eligibility criteria tend to be narrower and include complex biomarker-driven requirements.
Russell explained that the operational impact of complexity is not “concentrated in one phase of the conduct life cycle,” describing how early design decisions continue to create strain throughout feasibility, startup, amendments, and study conduct.
Many amendments are predictable
From the sponsor perspective, Gonzalez said the data mirrors day-to-day experience in oncology operations. “Most … oncology amendments are not surprises; they’re missed conversations,” she said. This can be especially true when eligibility criteria and assessment schedules are not pressure-tested early with operational and recruitment teams.
She pointed out that “overly narrow eligibility criteria, especially when we’re talking about biomarker-driven requirements and also the high-frequency and low-value assessments” can reduce the number of viable sites and eligible patients. What may appear scientifically sound can become operationally fragile when applied in real-world settings.
Precision medicine raises feasibility stakes
Pai highlighted how oncology’s precision focus further complicates feasibility, especially in global studies. Precision medicine limits the pool of eligible patients, which in turn complicates recruitment strategy, enrollment projections, and timelines. She also addressed the operational impact of emerging safety data in early oncology development, noting that frequent updates can require repeated informed consent revisions. Consolidating updates, when possible, can help reduce burden for sites and patients.
AI’s role is foresight, not replacement
AI featured prominently in the conversation, but panelists emphasized that its role is to inform human decision-making, not replace it.
Russell described AI as a way to unlock historical data and help teams anticipate challenges earlier in the design process. Gonzalez reinforced that perspective, saying, “AI is most powerful when it augments, not replaces, human judgment.” She noted that AI can help benchmark eligibility complexity, stress-test visit schedules, and flag design elements historically linked to delays.
Pai added an important human perspective: “AI is not going to build the element of empathy.” Her comment underscores that patient experience, site realities, and practical trial execution still depend on experienced professionals.
Feasibility must be a design principle
Across the discussion, panelists described a shift toward earlier and deeper cross-functional collaboration in protocol development. Pai summarized this evolution clearly: “Gone are the days when clinical operations [was] given the protocol and [told], ‘Now run this trial.’”
Instead, successful oncology study design increasingly depends on involving operations, recruitment, regulatory, biomarker, biostatistics, and site perspectives early in the process, before protocols are finalized.
The bottom line
Oncology trials will always be complex, and some amendments are inevitable. As the panel discussed, however, many operational challenges are predictable. With better use of historical data, earlier cross-functional collaboration, and thoughtful application of AI, sponsors can design oncology trials that balance scientific rigor with operational feasibility and improve the experience for both sites and patients.
