David Sterling: Megan, I want to start with an arithmetic problem, because I think it's the most unsettling way into this.
Megan Skiendel: I'm awake, go.
David Sterling: 133 versus 509. Lartruvo, approved 2016 for soft-tissue sarcoma — the approval rested on a 133-patient study that suggested a survival benefit. The confirmatory trial had 509 patients. It found no benefit. The drug was withdrawn. Now — what's the difference between those two numbers?
Megan Skiendel: 376 patients worth of evidence that the first answer was wrong.
David Sterling: And years of prescriptions in between, with no proof the drug helped anyone. That's not a rounding error — that's the system operating as designed. Accelerated Approval lets the FDA approve on a surrogate endpoint, something that predicts benefit without being the actual outcome measure, and requires the sponsor to confirm later. The uncertainty doesn't disappear at approval. It relocates.
Megan Skiendel: It relocates onto patients who are already taking the drug.
David Sterling: That's the mechanism. And CDER ran that mechanism 290 times through December 31, 2022. So Lartruvo is one case — but 290 is the scale at which this experiment is actually running.
Megan Skiendel: And that's the thing we're really trying to work out — not whether Accelerated Approval is good or bad, but what it actually is. Because I don't think 'faster access' is the whole description. The more honest description is: the FDA is making a deliberate choice to trade certainty for speed, and the question is whether anyone is honestly accounting for what that costs.
David Sterling: But that framing — trade certainty for speed — actually flattens something important. Because most people hear 'expedited FDA approval' and assume the bar came down across the board. It didn't. Three of the four pathways don't touch what counts as proof.
Megan Skiendel: Wait — which three?
David Sterling: Fast Track, Breakthrough Therapy Designation, Priority Review. Fast Track is the lowest threshold — you just have to address an unmet medical need for a serious condition, no clinical evidence of a treatment effect required at all. Priority Review just compresses the review clock. Six months instead of ten to twelve. Breakthrough Therapy adds intensive FDA involvement, senior management in the room. None of them change what you have to prove to get approved.
Megan Skiendel: So the bar is — identical. You still need what the standard pathway needs. Two well-designed trials, substantial evidence, the full benefit-risk calculus.
David Sterling: Right — the FDA is just talking to you more, or moving faster. Accelerated Approval is the only one that actually changes the evidentiary standard. And the change is specific: instead of a confirmed clinical outcome, you can use a surrogate endpoint. A biomarker. Tumor shrinkage instead of survival.
Megan Skiendel: It's like — honestly, imagine a contractor who doesn't wait ten years to see if the house stands. They pass the foundation stress-test, call that sufficient, and you move in. The ten-year data comes later. Maybe.
David Sterling: And the 'maybe' is where bevacizumab and atezolizumab sit. Both got Accelerated Approval for breast cancer indications on surrogate endpoints. Both got withdrawn after post-market confirmatory trials found no verified clinical benefit. The surrogate didn't predict the outcome it was supposed to predict.
Megan Skiendel: Which is — I mean, that's the complication no one wants to say out loud. A single drug can stack all four designations simultaneously. Fast Track, Breakthrough, Priority Review, Accelerated Approval, all at once. So you have a drug moving faster, with more FDA handholding, that also replaced the clinical outcome with a biomarker — and from the outside it all looks like one thing called 'expedited.'
David Sterling: So the question that actually bites: if Fast Track and Breakthrough Therapy don't lower the bar, what exactly is getting accelerated for those drugs? The relationship with the FDA? That's not nothing, frankly — but it's also not what patients think they're getting.
Megan Skiendel: What they're getting is the moral signal. The FDA's stamp says: this condition is serious enough that we moved. And that logic has a specific birthplace — the AIDS crisis. Patient advocates in the 1980s didn't ask nicely. They showed up. They made dying visible. And the FDA created the Accelerated Approval pathway in 1992 because the counterfactual was genuinely — you die in eighteen months with nothing.
David Sterling: That's the load-bearing ethical premise. When the alternative is death, higher uncertainty is defensible. That's not a regulatory convenience — that's a moral claim.
Megan Skiendel: And it's a legitimate one, in that context. But then — think about what actually happens in a Tuesday morning clinic. An oncologist is sitting across from a 54-year-old soft-tissue sarcoma patient. There's an approved drug. The oncologist says there's an approved drug. Doesn't mention the evidence rests on 133 patients. Because the alternative is hospice. The counterfactual is doing real moral work in that room whether anyone names it.
David Sterling: Right — but does the counterfactual still hold when existing treatments are already on the table?
Megan Skiendel: That's exactly where it gets hollowed out. Critics — the National Center for Health Research has documented this — argue the pathway now covers drugs offering incremental improvement over treatments that already exist. The original calculus assumed genuine absence of alternatives.
David Sterling: And Marty Makary's April 18 announcement pushes it further still. A proposed pathway for rare diseases based on 'plausible mechanism' — not a surrogate endpoint, a mechanism — with post-approval adverse event monitoring as the primary safeguard. PhRMA calls that a patient-access lifeline. I'd call it the 1992 logic running on fumes.
Megan Skiendel: On fumes — or on faith that the monitoring actually catches the failure fast enough.
David Sterling: Which it didn't, with bevacizumab and atezolizumab. Both withdrawn from breast cancer indications after confirmatory trials. The mechanism worked — eventually. But who bore the cost of 'eventually'?
Megan Skiendel: That's actually what we need to dig into next — because there's a framing I've heard, I think it's from Mahnum Shahzad at the Institute for Population Medicine, that those withdrawals are simultaneously proof the system self-corrects and proof that patients were the error-detection instrument while it did.
David Sterling: That framing is sharper than it first sounds. Because if patients are the error-detection instrument, then the confirmatory trial isn't really the safeguard — it's the receipt. The harm already happened.
Megan Skiendel: That's Shahzad's point exactly. It's not a calibration problem — she calls it a societal choice about who bears the cost of uncertainty. And we've made that choice implicitly, without ever voting on it.
David Sterling: Hold on. 'Societal choice' is doing a lot of work there.
Megan Skiendel: It is, but — look, bevacizumab was on the market for breast cancer for years. Prescribed. Patients taking it. Then the post-market confirmatory trials came back, no verified clinical benefit, and the FDA withdrew the indication. The system self-corrected. That's the clean version. But the messier version is: who were the people taking bevacizumab during that correction window? They were the data.
David Sterling: And atezolizumab — same story, same breast cancer context, withdrawn after post-market trials failed to confirm the surrogate-endpoint evidence. Two drugs, same mechanism, same gap between approval and withdrawal.
Megan Skiendel: Right — but the enforcement was inconsistent even then. Some drugs stayed on market longer than the mechanism was ever intended to allow. So it's not just slow. It's unevenly slow.
David Sterling: Which brings me to the 2024 enforcement tightening. If you compress those timelines — force confirmatory trials to resolve faster — you also force withdrawal decisions earlier. And here's the problem I can't resolve: are you withdrawing because the evidence says no, or because the clock ran out before the evidence said anything?
Megan Skiendel: That's — honestly, that might be the harder question than anything about surrogate endpoints.
David Sterling: The stated standard is that the safety threshold doesn't move under Accelerated Approval. But we know post-approval safety label changes run higher for these drugs than for standard approvals. The stated standard and the observed outcome diverge. That gap is the actual cost Shahzad is naming — and nobody in the benefit-risk calculus is formally pricing it.
Megan Skiendel: The thing I can't put down — and I've been sitting with it this whole conversation — is those people between 133 and 509. They're not in the regulatory record as people. They're in it as prescriptions, maybe as adverse event reports. But the Lartruvo withdrawal notice doesn't carry their names.
David Sterling: And the 2024 enforcement tightening doesn't fix that. It compresses the timeline. Maybe it means the gap between a failed confirmatory trial and a withdrawal shrinks. But the logic is identical — patients are still the medium through which the uncertainty resolves. Faster doesn't change that.
Megan Skiendel: No. It doesn't.
David Sterling: I think that's — frankly, that's where I land. Not with an answer. Just with the shape of what we haven't figured out.
Megan Skiendel: 133 to approve it. 509 to undo it. And somewhere in there, a Tuesday morning clinic. That's enough to sit with.
