Lila Soto: Hugo, good to see you — I ran an errand this morning and ended up in the vitamin aisle for way too long, just staring at dosages, and it sent me into a spiral I cannot get out of.
Hugo Vance: A spiral about what, exactly?
Lila Soto: About the vitamin E meta-analysis — the one that found high-dose supplementation increased all-cause mortality. I keep turning it over because the bottles are still on the shelf, the dosages are still high, and the cultural story is still 'antioxidants protect you.' But the data said the protective intervention became harmful past a certain point.
Hugo Vance: The benefit curve plateaued before anyone noticed.
Lila Soto: Right — and people kept dosing past it. Which makes me want to ask a much more basic question: what is the dose-response curve actually telling us, and why are we so bad at reading it?
Hugo Vance: Well, the curve has a shape — threshold, acceleration, plateau. That's not contested. What's contested is whether we build clinical and regulatory thinking around that shape, or whether we default to the linear assumption because it's simpler to communicate. Hippocrates was already gesturing at dose-dependent effects; A.J. Clark made it rigorous in the nineteen-twenties. The concept is ancient.
Lila Soto: And we still manage to miss it with vitamin E in, what, the early two-thousands.
Hugo Vance: Which is precisely why the therapeutic window — the narrow band between minimum effective dose and toxicity — matters more than the question of how much you can take. That tension is what we're really sitting with here.
Lila Soto: But that window framing — it makes it sound tidy, and it isn't, right? Because the body doesn't experience dose like a ruler. It's more like... okay, think of a thermostat. Nothing happens until the temperature crosses a threshold. Then the system kicks in fast. Then it maxes out and adding more heat does nothing. That's the actual shape.
Hugo Vance: That is exactly the shape. Three phases — a floor where nothing registers, a steep climb where tiny increments produce large effects, and a ceiling where more dose buys you nothing. The Minimum Effective Concentration is the floor. Below it, the drug is biologically invisible.
Lila Soto: Biologically invisible. Hm.
Hugo Vance: And yet we've known this — I mean, not as metaphor but as measured data — for a century. A.J. Clark, 1885 to 1941, took acetylcholine and applied it to frog heart tissue and showed the response curve wasn't linear. It was sigmoidal. S-shaped. And the explanation was receptor occupancy — as you fill more receptors, response climbs steeply, then the receptors saturate and the curve flattens. That's not abstract. That's a curve drawn from an actual frog heart in the nineteen-twenties.
Lila Soto: Wait — Clark got there in the twenties?
Hugo Vance: Yes. And Hippocrates was gesturing at dose-dependent effects centuries before that — the idea that the same substance heals at one dose and harms at another. So this isn't modern pharmacology discovering something subtle. It's ancient intuition confirmed empirically and then, you see, somehow underweighted anyway.
Lila Soto: Which is the part that I can't get past — if Clark drew this in the nineteen-twenties, and Hippocrates had the instinct millennia before that, why does the vitamin E story happen at all? Like, who looked at that curve and thought the plateau wasn't real?
Hugo Vance: That's the harder question, and I think it's not about ignorance of the biology. It's that we had a cultural story — antioxidants protect, more protection is better — that ran faster than the receptor data could slow it down. The science was there. The institutional weight wasn't. And that gap is, I'd argue, where the real problem lives.
Lila Soto: And that gap is exactly where I want to pull at something — because the cultural story treats dose like a single track, one curve, more gets you more. But benefits and toxicity aren't even riding the same biological road, are they?
Hugo Vance: No. They're not. That's the part that gets underplayed. The efficacy curve and the toxicity curve are driven by different mechanisms — different receptors, different tissue targets — so they rise and fall independently. When the toxicity curve starts steepening right around the efficacy plateau, every milligram you add past that point is almost no additional benefit against a sharply rising risk.
Lila Soto: So the therapeutic window is literally the gap between those two curves.
Hugo Vance: The favorable gap, yes. Minimum Effective Concentration on one side, toxic threshold on the other. That's the target. Not the maximum tolerable dose — the window.
Lila Soto: Okay, and — I mean, for most drugs that window is wide enough that we don't panic. But warfarin. Lithium. Digoxin. The window is so narrow that... think about what that actually looks like on a Tuesday morning. Someone on warfarin for atrial fibrillation wakes up, notices a small bruise on her forearm. That's it. That's the signal. Her INR has drifted just past the therapeutic range and the difference between clot prevention and internal bleeding is measured in milligrams.
Hugo Vance: And the therapeutic index — the ratio of the adverse-effect dose to the therapeutic dose — puts a number on exactly how thin that margin is. A narrow TI means those two doses are close. Uncomfortably close.
Lila Soto: How close are we talking in practice, though? Like — is there a case where we actually measured the cliff?
Hugo Vance: Yes. The floxuridine Phase I trial — the Roswell Park regimen. Three patients at 1,750 milligrams per square meter. All three developed grade IV diarrhea. All three. So the maximum tolerated dose was set at 1,650. One hundred milligrams per square meter separated tolerable from catastrophic.
Lila Soto: A hundred milligrams. That's — yeah. That's the cliff being measured in real time, on real people.
Hugo Vance: Which is why the FDA embeds this logic — NOAEL, the therapeutic index — directly into how doses get approved. Goodman and Gilman codifies all of it as canonical standard. The regulatory goal was never maximum dose. It was always the window. Though — and we'll get to this — hormesis complicates whether the window is even a stable concept when the same substance helps at low doses and harms at high doses through mechanisms we still don't fully understand.
Lila Soto: But the window being 'not stable' — that's the part that doesn't quite fit yet. Because the vitamin E data wasn't showing diminishing returns. It was showing a direction flip. Mortality went up. That's not just a plateau, that's the curve turning back on itself.
Hugo Vance: That is exactly what hormesis names. And it's the wrinkle that breaks the sigmoidal model altogether — not just flattens it. The sigmoid assumes monotonic movement. One direction. Threshold, rise, plateau. Hormesis says: low dose, stimulatory benefit. High dose, inhibition or harm. U-shaped. The curve reverses.
Lila Soto: So A.J. Clark's whole receptor-occupancy framework — the thing that gave us the sigmoid — doesn't actually account for this.
Hugo Vance: It doesn't. Clark's model describes saturation. It has no mechanism for reversal. Which is why you need something like the DRDM — a model built specifically to fit non-monotonic, U-shaped data. And it does fit it. Better than the sigmoid. But, you see, fitting a curve and explaining why the biology reverses — those are entirely different problems.
Lila Soto: That gap kind of terrifies me, actually. Because if we're pattern-matching without knowing the mechanism — I mean, is it that low doses activate repair pathways and high doses just overwhelm them? Or are we just seeing a shape we want to see?
Hugo Vance: That's where I'd be cautious. The U-shaped pattern appears across exercise physiology, antioxidant research, neuroprotection — it's not one dataset. But a prospective mechanistic study — not retrospective curve-fitting — showing precisely why a low-dose carcinogen, say, stimulates repair while a high dose overwhelms it? We don't have that cleanly yet.
Lila Soto: Hold on. So the regulatory consequence is — what? The FDA's NOAEL framework assumes the curve only moves one direction.
Hugo Vance: Yes. Built entirely on monotonic assumptions. Which means if a meaningful subset of dose-response curves actually reverses direction, the safety standards derived from that framework could be systematically wrong for that subset. Not slightly off. Wrong in the wrong direction.
Lila Soto: And we genuinely don't know how large that subset is.
Hugo Vance: We don't. And that uncertainty scales up — because individual genetic variability means the window itself shifts per person. The same dose sits safely inside the therapeutic range for one patient and above the toxicity threshold for another. The Human Genome Project was supposed to address exactly that — personalized, gene-based dosing rather than population averages.
Lila Soto: And that's where I keep getting stuck. Because the FDA's population-level standards — they're hard-won, they're defensible, they protect people. But if the therapeutic window is genuinely different for every patient, then the recommended dose is... what, exactly? A useful fiction?
Hugo Vance: Well — a necessary one, I think. You can't regulate a curve per person. Not yet. So the population standard holds because the alternative is no floor at all.
Lila Soto: A necessary fiction. Yeah. I think that's the thought I'm going to be carrying out of this.
Hugo Vance: It's an honest place to land. Uncomfortable, but honest. Thank you for the vitamin aisle spiral — it went somewhere worth going.
