The Best Drug Isn't Always the Most Powerful One
I’ve sat through a lot of pitches where the slide that gets the room’s attention is the one showing the steepest dose-response curve. Bigger effect, more tumor shrinkage, a more dramatic line on the chart. It’s an easy thing to get excited about, and for a long time it was treated as a reasonable proxy for “this is going to be a good drug.” I’ve come to think it’s one of the more persistent ways biotech fools itself.
The tell is usually in how a founder answers a simple question: why this dose, and not a lower one? Too often the honest answer is that the highest dose tested happened to be the one patients could still tolerate, and nobody went back to check whether a smaller amount would have done nearly as much good with meaningfully less harm. That’s a very different thing from evidence that more drug produces a better outcome for the person taking it.
Where the Habit Came From
The instinct has a real origin. Traditional chemotherapy kills cells indiscriminately, cancerous and healthy alike, so for decades the working assumption in oncology was straightforward: push the dose as close to the maximum a patient can physically tolerate, because more drug generally means more cancer cells die. That logic produced an entire regulatory and clinical culture built around finding the maximum tolerated dose, or MTD, in early trials and then carrying that number forward, largely unquestioned, into the late-stage studies that decide whether a drug gets approved.
The problem is that most of the interesting drugs coming out of biotech today aren’t traditional chemotherapy. Targeted therapies and immuno-oncology agents often have a completely different relationship between dose and effect. Past a certain point, giving more of the drug doesn’t kill more cancer cells. It just adds more toxicity for a patient to absorb, more dose interruptions, more reasons to quietly stop taking a treatment that might otherwise be working. Carrying over the MTD instinct from chemotherapy to these newer drug classes was never really justified. It just took the industry a while to admit it.
Regulators Caught Up to This Before Most Companies Did
The FDA’s Oncology Center of Excellence made this official in 2021 with an initiative it calls Project Optimus, built specifically to reform how companies select doses for cancer drugs. The agency’s own language on this is blunt: the old paradigm “leads to doses and schedules of molecularly targeted therapies that are inadequately characterized before initiating registration trials,” and poorly characterized dosing can produce “toxicity without additional efficacy,” premature discontinuation, and lost opportunities for patients to benefit from a drug that might have worked at a gentler dose. The agency turned that philosophy into final guidance in August 2024, formally asking sponsors to identify an optimized dosage before ever entering the trials meant to win approval, not after.
I don’t think most biotech founders have caught up to what that actually requires yet. It means running real dose-ranging studies earlier, when a program has the least amount of proof that it works and the most pressure to move fast. It means resisting the version of the story where the highest dose produces the best-looking early data, because best-looking and best are not the same claim.
The specific mechanics of Project Optimus are worth sitting with for a second, because they describe a genuine shift in what regulators expect a company to have already done by the time it walks into a meeting about registration trials. FDA reviewers now want sponsors showing up early, well before the studies meant to support approval, to talk through dose-finding as its own workstream. They want randomized comparisons across a real range of doses rather than a single arm carried forward from an early-phase study because it was the one that didn’t cause too much trouble. And they want that dose-finding work leaning on the pharmacology data a company already has, not treated as a box to check after the biology has been settled.
A Drug That Got Better by Getting Smaller
Cabazitaxel makes a useful example precisely because it’s such an old-fashioned drug. This is a conventional chemotherapy, approved in 2010 at 25 mg/m² for men with metastatic castration-resistant prostate cancer, with none of the novel biology that makes targeted agents behave so differently from traditional cytotoxics. Seven years later, the FDA approved a second, lower dose after a trial called PROSELICA compared 25 mg/m² against 20 mg/m² head-to-head in 1,200 patients. The FDA’s own summary of that trial shows overall survival that was statistically indistinguishable between the two doses, 13.4 months at the lower dose versus 14.5 months at the higher one. What wasn’t indistinguishable was the toxicity. Severe infections occurred in 20% of patients on the higher dose against 10% on the lower one. Febrile neutropenia hit 9% of patients at 25 mg/m² and just 2% at 20 mg/m². Deaths within 30 days of the last dose were also more common on the higher dose. Same cancer, same drug, notably fewer dead and hospitalized patients, just by giving less of it.
Sometimes the Answer Is That the Higher Dose Was Right
I don’t want to oversell the lesson into “always go lower,” because the more useful example might be the one where that instinct got tested and didn’t hold up. Sotorasib, a targeted lung cancer therapy, won accelerated approval in 2021 at 960 mg once daily, but the FDA required its maker to run a randomized trial comparing that dose against 240 mg, because the pharmacology data at the time didn’t clearly justify why the higher dose was necessary. When the head-to-head results came in, the higher dose actually produced a better response rate and longer survival, alongside more gastrointestinal toxicity. A 2024 analysis in the Journal of Clinical Oncology, cited among the FDA’s own Project Optimus publications, described it as a dosing conundrum precisely because the answer wasn’t obvious in either direction. The point of the exercise was never to prove low doses are always better. It was to stop assuming an answer nobody had actually gone and checked.
This Isn’t Only an Oncology Problem
Project Optimus is written specifically for cancer drugs, but the underlying idea doesn’t stop at oncology’s border. Any chronic-disease therapy has the same tension built into it: a dose aggressive enough to produce a dramatic result in a short trial isn’t automatically the dose a patient can keep taking for years without the side effects eventually winning the argument. A therapy for a metabolic condition, an autoimmune disease, or a neurological disorder faces the same question cabazitaxel and sotorasib faced. The dose that looks best on a chart covering 12 weeks and the dose that actually keeps someone healthy and compliant over a decade are not guaranteed to be the same number, and a company that never separates those two questions is gambling that they happen to line up.
What I Look For Now
That distinction, between assuming and checking, is what I actually watch for when I’m evaluating a company. I want to see a management team that treats dose optimization as a real scientific question worth its own dedicated study, with its own budget line and its own timeline, rather than an afterthought bolted on after regulators ask for it. I want founders who can tell me why their chosen dose is the right one for a patient’s whole treatment experience, not just the one that produced the most impressive number in an early cohort of a dozen people.
The habit of chasing potency is seductive because it photographs well. A steep dose-response curve looks like conviction. A therapeutic index built around what a patient can actually sustain, refill after refill, looks like caution, or worse, like a company that lacks confidence in its own drug. I think that reading has it backward. The founders who understand what a body can tolerate over months and years, not just what a tumor does in the first six weeks of a trial, are the ones building drugs that actually reach the finish line their investors and their patients are both hoping for.
None of this is an argument for timidity. Sotorasib’s story is proof enough that the highest dose sometimes really is the right one, and a company shouldn’t talk itself out of real efficacy because caution has become fashionable. The argument is narrower than that, and I think it’s a better one: know the difference between a dose you chose and a dose you merely survived long enough to call an answer. Every drug that has ever made it to a pharmacy shelf had to clear that bar eventually, whether at approval or, as cabazitaxel shows, years afterward. The companies I want to back are the ones asking the question on their own timeline, before a regulator has to ask it for them.
Leen Kawas, Ph.D., is the co-founder and managing general partner of Propel Bio Partners, a Los Angeles–based life sciences venture firm.