Subscribe
Share
Share
Embed
Truth Seekers: Where Data Meets Reality
Tired of sensational headlines and conflicting health advice? Join Alex Barrett and Bill Morrison as they cut through the noise to uncover what scientific research actually says about the claims flooding your social media feed.
Each week, Alex and Bill tackle a different health, nutrition, or wellness claim that everyone's talking about. From "blue light ruins your sleep" to "seed oils are toxic," they dig into the actual studies, examine the methodologies, and translate the data into plain English.
No agenda. No sponsors to please. No credentials to fake. Just two people committed to finding out what's really true by going straight to the source—the research itself.
Perfect for anyone who's skeptical of influencer health advice but doesn't have time to read every scientific study themselves. New episodes drop regularly, delivering clarity in a world full of clickbait.
Question everything. Verify with data. Find the truth.
Disclaimer: Truth Seekers provides educational content based on published research. Nothing in this podcast should be considered medical, financial, or professional advice. Always consult qualified professionals for decisions affecting your health and wellbeing.
**The 75% Cure That Wasn't: When the FDA Said "Not So Fast"**
Alex: September 2025. Headlines everywhere: "Gene therapy slows Huntington's disease by 75%." Landmark breakthrough. First disease-modifying therapy ever. If you had Huntington's in your family, you saw those headlines and felt something you maybe hadn't felt in years.
Bill: Hope.
Alex: Exactly. And then two months later, the FDA quietly tells the company, "Actually, we don't think your data is good enough to approve this drug."
Bill: That whiplash—from breakthrough to rejection in eight weeks—that's what we're unpacking today.
Alex: Because this isn't a story about a drug that doesn't work. It's about the gap between what a press release says and what the evidence actually shows.
Bill: And why that gap matters so much when real people are making decisions based on these headlines.
Alex: Right, so let's start with what Huntington's disease actually is, because that context is crucial. It's a genetic condition where a mutated protein slowly destroys parts of your brain. You lose motor control, cognitive function, everything. And there's no cure.
Bill: And it's fatal. Average life expectancy after symptoms start is about 15 to 20 years.
Alex: So when September's headlines said a single-dose gene therapy had slowed disease progression by 75%, that's enormous.
Bill: The company, uniQure, put out this press release saying their drug AMT-130 had achieved statistically significant slowing in a trial. They called it a "landmark moment" for Huntington's research.
Alex: And the media ran with it. "Revolutionary gene therapy." "Historic moment." I saw families posting about it on Huntington's forums, asking their neurologists if they could get access.
Bill: Which makes total sense if you're watching someone you love deteriorate and suddenly there's this headline saying 75% slowing.
Alex: But here's where my journalist alarm bells start going off. This wasn't a published study. It was a press release.
Bill: Yeah, no peer review. No full dataset. Just topline results from a company that has a massive financial incentive to—
Alex: Sorry, hang on. When you were in tech, did you lot write press releases about your A/B tests before they were validated?
Bill: Oh God, all the time. We'd have one metric show improvement and marketing would want to blast it out. I'd be like, "Can we at least wait to see if it holds for more than three days?"
Alex: Did they wait?
Bill: Never. Stock price doesn't care about statistical rigor.
Alex: Right. Okay, so press release from a company with every reason to make this sound brilliant. What did the actual trial show?
Bill: Okay, this is where it gets interesting. They treated 29 patients total with this gene therapy—it's a one-time injection directly into the brain.
Alex: Hang on, brain surgery for a one-time treatment?
Bill: Yep. They inject a virus that carries instructions to make a molecule that lowers the huntingtin protein—the toxic one causing the disease.
Alex: Right, that's the theory. Lower the toxic protein, slow the damage.
Bill: So they treated these patients, followed them for 36 months, and measured how fast their disease progressed using something called the composite Unified Huntington's Disease Rating Scale.
Alex: Which measures motor skills, cognitive function, daily living abilities—basically, how well you're functioning overall.
Bill: Exactly. And at 36 months, the high-dose group—12 patients—showed a decline of 0.38 points on that scale.
Alex: Twelve patients.
Bill: Twelve.
Alex: Wait, didn't we—I feel like we've talked about a brain study with 12 people before. Was that the microplastics thing?
Bill: Oh yeah, the postmortem study. Same sample size.
Alex: Which we absolutely tore apart for being too small to draw conclusions from.
Bill: Well, but hang on. That was postmortem tissue with no controls. This is a prospective trial with—
Alex: With twelve people, Bill.
Bill: Fair. Fair. But let me finish the methodology before we get into whether it's enough, because it gets more complicated.
Alex: Okay.
Bill: So they compared these 12 patients to 940 patients from a historical database called Enroll-HD.
Alex: Wait. Historical database?
Bill: Yeah, this is where the methodology gets really important. They didn't compare the treated patients to a placebo group in the same trial. They compared them to data from other patients collected previously.
Alex: So not people who enrolled in this trial, got a sham surgery, and then got measured the same way at the same time?
Bill: Correct. These are external controls. People from a separate database who were matched on things like age, disease stage, genetic markers.
Alex: But they're not in the trial.
Bill: Right. And the external controls showed a decline of 1.52 points over 36 months. So the company says 75% slowing—you take 0.38, divide by 1.52, subtract from one, and that's your effect size.
Alex: And that gave them a p-value of 0.003, which is statistically significant.
Bill: On paper, yes.
Alex: But?
Bill: But this trial had internal controls. About ten patients got sham surgery—all the brain surgery without the actual drug.
Alex: So they could compare apples to apples.
Bill: Exactly. Same trial, same surgical experience, same measurement protocols, same time period. And the company never presented that comparison.
Alex: Sorry, what?
Bill: The 36-month data for those sham surgery patients? Not in the press release. Not in the slide deck they showed at conferences.
Alex: That's... why would you not show the most direct comparison you have?
Bill: That's what I want to know. If your drug works, you'd want to show it works compared to the placebo group in your own trial, right?
Alex: Unless that comparison doesn't look as impressive.
Bill: I'm not saying that's what happened, but it's a massive red flag when the data that should be most convincing is the data that's missing.
Alex: Okay, but hold on. Let me play devil's advocate for a second. External controls—they're not ideal, but they're not automatically rubbish, are they? I mean, if you match carefully on all the important factors...
Bill: You're right that they're not automatically invalid. In rare diseases, sometimes you can't ethically withhold treatment once you have a signal. So external controls are sometimes your only option.
Alex: Right, because you can't ask someone with a fatal disease to volunteer for sham brain surgery if there's a chance the real thing works.
Bill: Exactly. And I do think the researchers here tried to match carefully. Age, disease stage, CAG repeat length—that's the genetic marker that predicts severity—they tried to control for confounding.
Alex: But?
Bill: But external controls can't account for unmeasured differences. You can match on age, disease stage, genetic profile—but you can't match on things like patient motivation, subtle differences in how raters score symptoms, changes in supportive care over time.
Alex: And people who volunteer for experimental brain surgery are self-selected.
Bill: Right. They might be more health-literate, more engaged, more hopeful. That psychological difference could affect how they report symptoms or engage with rehab.
Alex: Okay, but then why did they have internal controls and not show them? That's what I don't get. If the external controls are necessary because of ethics, fine. But they did sham surgeries on ten people.
Bill: That's the question I keep coming back to.
Alex: Because if you're going to ask ten people to have sham brain surgery—which is not nothing—
Bill: It's definitely not nothing.
Alex: Then you bloody well better use that data. Otherwise, what was the point?
Bill: Unless the comparison to internal controls doesn't support your press release narrative.
Alex: Which brings us back to: this is starting to sound like the trial design has some serious limitations.
Bill: And the sample size makes it worse. We're talking about 12 patients in the high-dose group.
Alex: Which we've established is not a lot.
Bill: With that small a number, random variation can look like a treatment effect. You need replication in a larger sample to be confident.
Alex: So 75% slowing based on 12 patients compared to a historical database, with the internal control data mysteriously absent.
Bill: And that's just the primary endpoint. They measured five different clinical outcomes.
Alex: Oh no.
Bill: What?
Alex: Multiple comparisons. This is going to be a multiple comparisons thing, isn't it?
Bill: It is absolutely a multiple comparisons thing.
Alex: Go on then. How many reached significance?
Bill: Two out of five.
Alex: There it is.
Bill: One of them just barely—p-value of 0.033.
Alex: And the other three?
Bill: One missed significance with p equals 0.057—so close it screams multiple testing problem. Another was marked as "nominal significance," which is statistician-speak for "we're not confident this is real." And the motor score? P equals 0.174. Not even close.
Alex: Wait, so the motor symptoms—which are one of the most visible parts of Huntington's—didn't slow significantly?
Bill: Nope. But the composite score that includes motor, cognitive, and functional measures did reach significance.
Alex: That feels like... you're mixing together things that worked and things that didn't, and calling the average a win.
Bill: That's the risk with composite endpoints. They can smooth over failures in individual measures.
Alex: Hold on though. Didn't they pre-register which outcome they were measuring? I thought that was the whole point of pre-registration—you commit to what you're testing before you see the data.
Bill: They did pre-register this analysis plan with the FDA. They weren't making it up after the fact.
Alex: So they followed their protocol.
Bill: They did. And some really respected Huntington's researchers believe this finding is real. Sarah Tabrizi at University College London, Ed Wild—these are not company shills. They're serious scientists who think the drug works.
Alex: Okay, so we have serious researchers who think it's real, a pre-registered analysis plan, and a statistically significant finding on the primary outcome. But you're still skeptical.
Bill: I'm skeptical that 12 patients with external controls is enough to be confident. Yeah.
Alex: Even with the statistical significance?
Bill: Statistical significance tells you the finding is unlikely to be due to chance if your methodology is sound. But with external controls and a tiny sample, I don't know if the methodology is sound enough.
Alex: Huh.
Bill: What?
Alex: I think I might actually disagree with you here.
Bill: Okay.
Alex: Look, I'm usually the one saying we need more skepticism. But in this case, if the experts in the field think the signal is real, if the analysis plan was pre-registered, if the matching was done carefully—maybe the problem isn't that the evidence is weak, it's that the press release oversold how strong it was.
Bill: That's... actually a fair distinction.
Alex: Like, there's a difference between "this evidence suggests the drug might work and we should do a bigger trial" and "landmark breakthrough, 75% cure."
Bill: Right. The evidence might be good enough to justify further research but not good enough to justify approval.
Alex: Which is exactly what the FDA ended up saying.
Bill: Wait, we haven't gotten to the FDA part yet.
Alex: Right, sorry. I'm jumping ahead. What happened with the FDA?
Bill: Okay, so in April 2025, the FDA gave AMT-130 something called Breakthrough Therapy designation.
Alex: Which means the FDA thought the approach was promising enough to fast-track it.
Bill: Right. And as part of that designation, FDA says, "Yes, we'll accept external controls for your approval package."
Alex: So in April, the FDA is on board with this methodology.
Bill: Correct. Then in September, the company announces these landmark results.
Alex: With the 75% figure.
Bill: Yep. And in November—just two months later—the FDA tells the company, "Actually, we've changed our mind. Your Phase 1/2 data with external controls is not sufficient to support approval."
Alex: Hang on, the FDA reversed course?
Bill: Completely. They essentially said, "Small sample size plus external controls plus no internal placebo comparison equals you can't prove causation."
Alex: That is a massive credibility hit. I mean, if your own regulator says the evidence isn't good enough...
Bill: It tells you that despite the careful matching, despite the pre-registered analysis, despite the respected investigators—experts who review this stuff for a living don't think the methodology meets the standard.
Alex: What changed between April and November?
Bill: The data didn't change. The FDA's confidence in external controls as proof changed.
Alex: Because they looked at the actual numbers and realized 12 patients compared to a historical database can't rule out confounding.
Bill: That's my read, yeah.
Alex: See, this is what I was trying to say earlier. The evidence might be suggestive, but it's not proof. And the FDA is drawing the line at proof.
Bill: Yeah. And here's something else that's missing that really bothers me as someone who used to look at data pipelines—target engagement.
Alex: What's that?
Bill: The whole point of this drug is to lower the huntingtin protein in the brain. That's the mechanism. You inject genetic instructions, those instructions make a molecule, the molecule lowers huntingtin, less toxic protein equals slower disease.
Alex: Right, that's the theory.
Bill: The company never proved it actually lowers huntingtin in humans.
Alex: Sorry, what?
Bill: They never measured whether the drug does what it's supposed to do at the molecular level.
Alex: How is that possible? Isn't that the first thing you'd check?
Bill: They said the tools to measure huntingtin levels in human brains are "too noisy." The measurements are unreliable.
Alex: But they're confident enough to claim 75% slowing of disease without confirming the drug does what it's designed to do?
Bill: That's the gap. They have clinical measures showing slowing—or at least, measures that correlate with slowing. But no proof the mechanism is what they think it is.
Alex: So the benefit could be from something else. Immune response to the virus, inflammation reduction from the surgery itself—
Bill: Placebo effect, even. When you have brain surgery and get told you're receiving a cutting-edge gene therapy, that psychological impact is real.
Alex: And that's exactly why you need the internal placebo control.
Bill: Which they had and didn't show us.
Alex: Okay, I'm back to being fully skeptical now. That's actually quite damning.
Bill: Yeah.
Alex: Because if you're a family dealing with Huntington's, you saw those September headlines and felt hope for the first time in maybe years.
Bill: And then two months later, you see the FDA said no, and that hope gets crushed without understanding why.
Alex: So let's be clear about what this actually means. Is the drug useless?
Bill: Probably not. The signal is real enough that serious scientists believe it. The problem is the evidence isn't strong enough to be confident.
Alex: Meaning the drug might work, but we can't tell the difference between a real effect and methodological noise with this trial design.
Bill: Right. And the FDA is saying, "Come back with a Phase 3 trial that has a proper randomized placebo control and a bigger sample."
Alex: Which the company is now planning.
Bill: Yeah, but that means patients wait another two, three years for approval instead of getting it in late 2026.
Alex: Which is heartbreaking if you're declining fast. But it's also protective, because approving a drug that doesn't actually work would be worse.
Bill: Exactly. You'd have people undergoing brain surgery for a treatment that might be no better than placebo.
Alex: What does this tell us about how to read medical headlines?
Bill: Press releases are marketing. They're not peer-reviewed science.
Alex: And "statistically significant" doesn't mean "definitely true," especially with small samples.
Bill: When you see a small trial with external controls claiming a huge effect, the appropriate response is "interesting, let's see if it replicates," not "landmark breakthrough."
Alex: And if a company measures five different outcomes and only two reach significance, that's a red flag.
Bill: Because when you test multiple things hoping for an effect, you'll find something just by chance. That's why replication matters so much.
Alex: The FDA reversal is the key piece here. That tells you that experts who review this stuff professionally don't think the evidence is sufficient.
Bill: Even though they initially thought it would be.
Alex: Which shows how hard it is to evaluate evidence in real time, even for regulators.
Bill: And that should make us all a bit more humble about claiming breakthroughs before the data is fully vetted.
Alex: So for families waiting—AMT-130 might still get approved. It's not dead.
Bill: Right. If the Phase 3 trial replicates the finding with better methodology, this could absolutely become a real treatment.
Alex: But September's "landmark breakthrough" was overselling evidence that wasn't ready for that framing.
Bill: And the cost of that overselling is families who got their hopes up and then crashed when reality hit in November.
Alex: That's the gap we're trying to translate here. Between the hype and the actual state of the evidence.
Bill: So the takeaway isn't "don't trust medical research." It's "understand what strength of evidence you're looking at."
Alex: Press release with 12 patients and external controls? Promising signal. Not proof.
Bill: Published peer-reviewed study with hundreds of patients and randomized placebo control? Much stronger.
Alex: And regulatory rejection two months after the hype? That's the market correction. That's reality reasserting itself.
Bill: Which is frustrating, but it's also the system working the way it should.
Alex: Protecting patients from treatments that haven't been proven yet, even when we desperately want them to work.
Bill: Especially when we desperately want them to work.
Alex: That's the translation. Hope is not the same as evidence. And knowing the difference protects you from the whiplash of hype and disappointment.
Bill: Next time you see a headline claiming a breakthrough for a devastating disease, check the sample size, check the control group, and check if the FDA agrees.
Alex: Because if it sounds too good to be true, it might just be too early to know if it's true.