Beyond the Maximum Tolerated Dose: What Project Optimus Changed in Cancer Drug Trials

Project Optimus is an initiative of the FDA's Oncology Center of Excellence, launched in 2021, that ended oncology's decades-long habit of picking a dose at the highest level patients can tolerate. Its central instruction, formalized in a final guidance issued in August 2024, is that sponsors should study more than one dose, compare them directly (ideally in a randomized trial), and defend the choice with pharmacokinetic, pharmacodynamic, and biomarker data rather than toxicity alone. The goal is an optimized dose that balances benefit against harm, not simply the largest dose a body can withstand.

This is a genuine change in how cancer drugs are developed, and its logic reaches well beyond oncology.

Why the maximum tolerated dose stopped making sense

For most of the modern era, early cancer trials were built around cytotoxic chemotherapy, where more drug generally meant more tumor kill. Dose-finding studies climbed the dose ladder until side effects became unacceptable, then declared the step just below that the maximum tolerated dose, or MTD. The assumption baked into that design was simple: efficacy rises with dose, so the highest safe dose is the best dose.

Targeted therapies and immunotherapies broke that assumption. Many of these drugs saturate their target or their intended biological effect well below the dose that causes serious toxicity. Pushing to the MTD then buys no extra benefit while adding avoidable harm, dose interruptions, and discontinuations. The FDA's guidance and the surrounding literature describe exactly this pattern: doses carried forward from small MTD-driven studies were sometimes higher than necessary, only for post-marketing studies to later identify a lower, equally effective dose. That is an expensive and patient-unfriendly way to learn.

What the 2024 guidance actually asks for

The final guidance, titled "Optimizing the Dosage of Human Prescription Drugs and Biological Products for the Treatment of Oncologic Diseases," was announced in the Federal Register in August 2024 and codifies the Project Optimus expectations. A few themes stand out.

First, dose selection should be evidence-based and prospective. Sponsors are expected to characterize the relationship between dose or exposure and both efficacy and safety before committing to a dose, rather than treating dose optimization as an afterthought.

Second, more than one dosage should usually be evaluated in the clinic. A common structure is to carry two dosages into a randomized comparison so that efficacy, safety, tolerability, and patient-reported outcomes can be weighed side by side. The guidance does not mandate a single trial template; it asks for a design capable of distinguishing between candidate doses.

Third, the rationale should integrate multiple data streams. Pharmacokinetics (what the body does to the drug), pharmacodynamics (what the drug does to the body and the target), biomarkers of target engagement, and tolerability over time all feed the decision. The concept often described as the optimal biological dose reflects this shift: the aim is the dose that best engages the biology at acceptable cost, which may sit below the MTD.

Is the field really changing, or just the paperwork?

The more interesting question is whether trial designs on the ground have moved. Published analysis suggests they have. A 2025 study in JCO Oncology Advances reviewed 367 industry-sponsored phase I protocols activated at one research network between 2021 and 2024 and documented measurable shifts. The authors reported a move away from the traditional rule-based 3-plus-3 escalation toward Bayesian dose-finding designs, whose use rose from 48 percent of protocols in 2021 to 75 percent in 2024, alongside a rise in protocols that included an explicit dose-optimization plan. The same analysis found that patient-centered elements, such as patient-reported outcomes, changed little over the period. In other words, the incentive structure has begun to reshape early-phase protocols, though unevenly.

That matters because design choices upstream determine what regulators, clinicians, and patients can learn. A trial that never compares doses cannot tell you whether a lower dose would have worked as well with fewer harms. Building that comparison in early is cheaper and faster than discovering it years later in a post-approval commitment.

The trade-offs worth naming

None of this is free. Evaluating multiple doses can require more patients, more time, and more complex statistics in early development, which sponsors and investigators have to plan and fund. There is also a scientific tension: for a drug with a wide gap between the biologically active dose and the toxic dose, a formal randomized comparison may add cost without changing the answer, while for a drug with a narrow window it may be indispensable. The guidance leaves room for judgment, which places weight on the quality of the dose-response modeling and the biomarker strategy that supports it.

There is a broader lesson here that applies to drug development generally, including in metabolic disease. Dose is not a detail to be settled late; it is a hypothesis to be tested with the same rigor as the molecule itself. Exposure-response thinking, pharmacodynamic anchoring, and randomized dose comparison are transferable tools. The reason they surfaced first and most forcefully in oncology is that the older MTD logic failed most visibly there, but the underlying discipline, choosing a dose because the data support it rather than because the body tolerated it, is good practice across therapeutic areas.

For readers, the practical takeaway is narrow and worth stating plainly: a higher dose is not automatically a better dose, and modern regulation increasingly expects developers to prove which dose is right. This article is educational and not medical advice.