From Data to Insights: Using the Totality of Evidence for Better Oncology Dose Selection

Under Project Optimus, the FDA now expects oncology sponsors to evaluate multiple clinically relevant doses using the totality of the evidence and to identify the dosage that “maximizes not only the efficacy of a drug but the safety and tolerability as well.”1 This regulatory evolution was driven by repeated cases in which suboptimal dose explorations undermined drug development.

For decades, oncology dose selection has relied on escalating treatment intensity until reaching the maximum tolerated dose (MTD), a paradigm created for cytotoxic chemotherapy.2 The assumption was simple: higher exposure yields greater antitumor activity. Yet, in a recent paper, the authors note that this “more is better” logic is misconceived for most modern agents, including targeted therapies, antibody–drug conjugates, and immunotherapies, many of which have flat or shallow dose–response curves and wide therapeutic indices.

This historical reliance on the MTD is increasingly incompatible with contemporary science and regulatory expectations.2 MTD-based designs focus almost exclusively on severe toxicities, overlooking the chronic, cumulative, and lower-grade toxicities that determine whether a patient can remain on therapy long enough to benefit. These toxicities often drive dose reductions, interruptions, and discontinuations in clinical practice and were likely a motivation behind Project Optimus.

Reflecting this reality, the goal of Project Optimus is to educate, innovate, and collaborate with drug developers and patients to move forward with a dose-finding and dose optimization paradigm across oncology that emphasizes selection of a dose or doses that maximizes not only the efficacy of a drug but the safety and tolerability as well.1

How Does Evidence Inform Dose Selection?

After Phase I studies have identified one or more doses that appear safe and pharmacologically active, the central question becomes which of these should advance into Phase III. The choice is ideally evaluated in randomized Phase II trials, where two candidate doses can be compared under conditions that approximate their intended clinical use. Yet these studies often assess efficacy and toxicity on separate tracks. Activity is summarized through endpoints such as response rate or progression-based measures, while toxicity is reported descriptively or through adverse-event tables.

Evaluating these components independently makes it difficult to determine which dose offers the most appropriate overall profile. Two doses may deliver similar antitumor activity but differ substantially in chronic or cumulative toxicity. Conversely, a higher dose may show a modest numerical advantage in efficacy while imposing side effects that shorten treatment duration, undermine long-term tolerability, and ultimately reduce patients’ ability to remain on therapy long enough to benefit. These are precisely the trade-offs that Project Optimus now asks sponsors to characterize before selecting the dose that proceeds into pivotal testing.

The limitations of evaluating benefit and toxicity separately have been highlighted repeatedly in academic literature. In a recent Lancet Oncology review, Tannock and colleagues argued that many approved anticancer regimens are unnecessarily intensive, and that lower-intensity alternatives often deliver “near-equivalent” overall outcomes once toxicity is taken into account.3 A conclusion that is only visible when efficacy and harm are assessed jointly rather than in isolation. They explicitly point to innovative statistical methods, such as Net Treatment Benefit (NTB), as ways to quantify these multidimensional trade-offs. Similar arguments were made in another recent article of the same journal, where the authors emphasize that optimal drug use requires integrating efficacy, toxicity, and treatment burden into a single assessment, not treating each domain as a separate exercise.4 Together, these insights reinforce the need for analytical approaches capable of capturing the full balance of benefit and harm when choosing the dose that should advance to Phase III.

The Net Treatment Benefit (NTB) allows one such approach. NTB summarizes the probability that one dose provides a better overall clinical outcome than another, based on a prespecified hierarchy of outcomes that can reflect clinical priorities.5 Because it can consider efficacy and safety within a unified metric, NTB complements traditional endpoint analyses and provides a transparent way to inform and document dose-selection decisions in randomized Phase II trials. Importantly, NTB can help synthesize the totality of evidence needed to justify which dose should advance into Phase III.

What Does This Mean for Oncology Development?

Dose selection has become a consequential decision in modern oncology development. The shift toward more deliberate, evidence-based dosing reflects a broad recognition that efficacy alone or safety alone cannot dictate this choice. FDA’s Project Optimus has made clear that dose selection must reflect the full spectrum of clinical evidence, integrating benefit and harm rather than relying on isolated endpoint summaries.

Innovative approaches such as the Net Treatment Benefit can provide a practical way to unify multiple outcomes into a single, interpretable metric, aligning with the clinical trade-offs that matter for dose optimization. As regulatory expectations continue to evolve, integrating these multidimensional assessments will be essential for selecting doses that are not only active, but also tolerable and clinically sustainable.

References

1. US Food and Drug Administration. Project Optimus. Silver Spring (MD): US Food and Drug Administration. Available from: https://www.fda.gov/about-fda/oncology-center-excellence/project-optimus
2. Roberts SA, McKee AE, Pazdur R. Totality of the evidence: optimizing dosage selection strategies in oncology drug development. J Clin Oncol. 2025;43. doi:10.1200/JCO-25-00488. Available from: https://ascopubs.org/doi/pdf/10.1200/JCO-25-00488
3. Gyawali B, Kesselheim AS. Dose optimisation to improve access to effective cancer medicines. Lancet Oncol. 2024;25. Available from: https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(24)00648-X/abstract
4. Banerjee S, et al. Determining the optimal use of approved drugs in oncology. Lancet Oncol. 2025;26. Available from: https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(25)00037-3/abstract
5. Buyse M, Verbeeck J, Saad ED, Backer M, Deltuvaite-Thomas A, Molenberghs G, editors. Handbook of generalized pairwise comparisons: methods for patient-centric analysis. 1st ed. Boca Raton (FL): CRC Press; 2025. Available from: https://www.routledge.com/Handbook-of-Generalized-Pairwise-Comparisons-Methods-for-Patient-Centric-Analysis/Buyse-Verbeeck-Saad-Backer-Deltuvaite-Thomas-Molenberghs/p/book/9781032488059