

AstraZeneca is pushing two drugs toward approval despite Phase 3 failures, betting that subgroup data and supplemental analyses can overcome the biggest hurdle in drug development. The outcome could redefine what "enough evidence" means for the entire industry.
Most companies get a "no" from their biggest clinical trial and quietly move on. AstraZeneca is doing the opposite.
The pharma giant is pushing ahead with not one but two regulatory approval bids built on Phase 3 trials that missed their primary endpoints. One involves a breast cancer drug where FDA advisors already voted against it. The other involves a rare disease drug that flopped in its overall patient population but looked promising in a tiny slice of patients.
It's a strategy that's equal parts audacious and risky. And it could reshape how the entire industry thinks about what counts as "enough" evidence to get a drug approved.
Let's start with camizestrant, an oral drug designed for patients with ER-positive, HER2-negative advanced breast cancer who develop a specific genetic mutation called ESR1.
The Phase 3 SERENA-6 trial actually showed a 56% reduction in the risk of disease progression or death. That sounds great, right? The problem wasn't the numbers; it was everything around them. FDA advisors had deep concerns about the trial's design: how patients were tested, when they were switched to the drug, and whether the results truly reflected a meaningful benefit in real life.
The advisory committee voted 6-3 against concluding that camizestrant provides clinically meaningful benefit. That's the kind of score that usually sends a drug back to the drawing board.
But AstraZeneca didn't flinch. Instead, the company submitted additional analyses, including data on ctDNA clearance (essentially tracking whether tumor-related DNA fragments disappear from the bloodstream, which can signal the drug is working at a molecular level). On May 27, 2026, the FDA extended its review timeline to evaluate this new evidence. No new decision date was announced.
The twist? Europe already said yes. The EMA's drug evaluation committee issued a positive opinion on camizestrant (to be marketed as Etcamah) for essentially the same patient population, using the same data package that the U.S. panel rejected. Same drug, same evidence, opposite conclusions. It's like two teachers grading the same paper and one giving it a B+ while the other hands it back with "See me after class."

Agios' next-gen pyruvate kinase activator tebapivat just flunked its phase 2b trial in a blood cancer called lower-risk MDS, and the company is scrapping the entire program. The failure exposes a classic biotech trap: assuming a mechanism that works in genetic anemias will translate to cancer.


Join thousands of biotech professionals who start their day with our free, daily briefing.
If camizestrant is the bold play, anselamimab is the Hail Mary.
AstraZeneca's Phase 3 trial of anselamimab in AL amyloidosis (a rare disease where misfolded proteins damage the heart and other organs) missed its primary endpoint in July 2025. In the overall trial population, the drug didn't outperform placebo on the metrics that mattered most.
But buried in the data was a prespecified subgroup: patients with a "kappa-predominant" light chain type. Think of it like this: AL amyloidosis comes in two biological flavors (kappa and lambda), and the drug appeared to work dramatically better in one of them.
In those 72 kappa patients, 31.3% on anselamimab died compared to 58.3% on placebo. Annual cardiovascular hospitalization rates were 0.41 versus 1.4. Those are striking numbers. In the much larger lambda group of 328 patients? No difference whatsoever between the drug and placebo.
AstraZeneca is now building its entire regulatory case around this small subgroup, presenting the data at ASCO 2026 and signaling it could seek approval this year.
The question every regulator will ask: is this a genuine biological insight, or is it what statisticians politely call "data fishing"?
Surprisingly, AstraZeneca isn't as far out on a limb as it seems. A 2023 study in JAMA Internal Medicine found that 10% of new drugs approved between 2018 and 2021 had at least one pivotal trial with a null primary endpoint. That's 21 out of 210 drugs.
Four of those were approved based on a single pivotal study where every primary endpoint failed. The poster child is aducanumab (Biogen's Alzheimer's drug), which won accelerated approval in 2021 after both its Phase 3 trials were stopped for futility. FDA pivoted to a surrogate biomarker (amyloid plaque reduction) to justify the green light.
Tofersen for SOD1-ALS followed a similar script: the Phase 3 missed its clinical endpoint, but FDA approved it based on biomarker reductions that were "reasonably likely to predict clinical benefit."
The common thread in these cases is high unmet need plus a plausible biological story. When patients are dying and there's nothing else available, regulators have shown they'll accept a lower bar of evidence, provided the benefit-risk math still works.
Regulators know that every exception they make becomes a template. If the FDA approves anselamimab based on 72 patients from a failed trial, every pharma company with a disappointing Phase 3 will start slicing their data into subgroups looking for a rescue narrative.
That awareness tends to make agencies more conservative, not less. And the timing isn't great for AstraZeneca: the camizestrant ODAC pushback shows the FDA is in a skeptical mood when it comes to creative interpretations of trial data.
Analysts are reportedly treating the anselamimab filing as a binary, high-risk event. The base case for many models assigns a 25-50% probability of approval, with the most likely positive outcome being a narrow label restricted to kappa patients plus mandatory post-approval confirmatory studies.
The bear case? FDA issues a Complete Response Letter and asks AstraZeneca to run a new trial specifically in kappa patients. That would push any launch out by years.
AstraZeneca's double-barreled strategy is essentially a stress test for regulatory standards. The company is asking a fundamental question: in diseases where patients are desperate, how much evidence is enough?
If both programs eventually win approval, it sends a signal that the "totality of evidence" framework is expanding. Drug developers will feel empowered to file even when their pivotal trials stumble, as long as the supporting data tell a coherent biological story.
If both get rejected, the message is equally clear: a Phase 3 primary endpoint still means what it says, and no amount of creative analysis can substitute for a clean win.
Either way, the industry is watching. With more than 20 Phase 3 readouts expected from AstraZeneca alone by the end of 2026, the company has essentially made itself the test case for how flexible modern drug approval can be. The next few months will tell us whether that flexibility has limits, or whether the rules of the game are quietly being rewritten.
Bristol Myers Squibb is betting up to $15.2 billion on 13 preclinical drugs from China's Hengrui Medicine, none of which have been tested in humans. It's one of the largest China-originated pharma deals ever, and it's landing right in the middle of a geopolitical minefield.