The Clean-Protein Paradox

At the tip of a cacao root there is a protein called NRAMP5. Its job is to pull the metals the plant needs — zinc, manganese, iron — up out of the soil and into the tissue, and it works by shape: it recognises a small, doubly-charged metal ion and draws it inward. Cadmium is a small, doubly-charged metal ion of almost exactly the wrong kind. It does nothing for the plant, has no biological use at all, but it is close enough to zinc that the transporter takes it too, by mistake. From the root the cadmium travels up through the branches and stems, and in the final stage of the bean's development the plant loads it — through its own internal plumbing, the phloem — straight into the seed.¹ By the time the pod is split open, the cadmium is not sitting on the bean. It is inside it.

That is the fact that survives everything that happens next. You can grow the tree without a single synthetic pesticide. You can certify the farm organic, harvest by hand, ferment and roast and mill the beans into the darkest, cleanest-sounding cocoa powder on the shelf — and the cadmium is still there, because the transporter that let it in never asked where the soil's zinc came from. "Organic" describes how the crop was farmed. It has nothing to say about what the crop drew up from the ground.

Hold that molecule, because it is the whole paradox. The words a health-conscious shopper trusts on a protein tub — organic, plant-based, clean, natural — are the words that tell you about the farming. The lead and cadmium in the scoop are the part the farming can't touch. And on the two largest independent tests of the category, those trusted words don't just fail to track the metal. They track it upward.

The word points the wrong way

In October 2025, Consumer Reports put 23 protein powders and shakes through its own lab.² More than two-thirds carried more lead in a single daily serving than the 0.5 micrograms the organisation uses as its level of concern — a figure it borrows from California's warning-label rules, which I'll come back to, because the number matters less than what surrounds it. The pattern in the results is the part that matters here: plant-based powders averaged nine times the lead of dairy-based ones. The two products Consumer Reports told readers to avoid outright were both plant-based — Naked Nutrition's Vegan Mass Gainer, at 7.7 micrograms of lead per serving, and Huel Black Edition, at 6.3.

Huel is a British company, headquartered in Tring, Hertfordshire; its Black Edition is built on pea and rice protein, and in March 2026 the brand was reported to be heading into a roughly one-billion-euro acquisition by Danone, a deal now under review by the UK's competition regulator.³ That pea-and-rice base is not incidental. Rice draws arsenic and cadmium out of flooded paddy soil and concentrates the arsenic in its bran; pea, like most legumes, carries more metal than milk does.⁴ The feedstocks that make a powder plant-based are, by their nature, the higher-metal ones — a matter of which plant, grown in which ground, not of who made the tub. The same Consumer Reports test also found 9.2 micrograms of cadmium in a serving of Huel Black Edition — more than twice the level the organisation uses as its concern threshold for that metal. Both companies answered on the lead result. Huel said its ingredients undergo rigorous testing and that it was confident the product sits "well within the levels set out by NSF" — a statement about lead; the company did not respond to questions about the cadmium finding. Naked Nutrition said it sources from select suppliers with heavy-metal documentation and had requested a fresh third-party test in response. Of the other named brands, Abbott said consumers could be assured its product was safe; Vega disputed the cadmium finding as inconsistent with its own testing, citing a change in its pea sourcing from China to North America; and Momentous said the tested items had been discontinued.²

A separate, larger test points the same way. The Clean Label Project, an independent testing non-profit, screened 160 products in January 2025 and reported that "organic" powders averaged three times the lead of non-organic ones, and that chocolate powders carried up to a hundred and ten times the cadmium of vanilla; across all the metals it screened, roughly 47 percent of the products it tested exceeded at least one federal or state regulatory limit.⁵ I report those figures, but I don't lean the argument on them, and it's worth saying plainly why. The Clean Label Project sells a paid "Certified" mark to manufacturers — it runs the test that concludes you need certified products, and then sells the certification.⁶ Its methodology, its product-selection criteria and its thresholds have never been published or peer-reviewed, and the trade body for the supplement industry has made exactly this point its main line of attack.⁷ The conflict is real and the reader should hold it. What makes the inversion credible is not the Clean Label Project's headline numbers; it is that Consumer Reports' own independent lab, with its own separate sample and no certification mark to sell, found the same direction — plant higher than dairy — in a completely different set of products.

YAN's earlier report on the metals in baby-food pouches traced the same class of contaminant through a different aisle and a different testing gap. What follows is about a different object: an adult product, taken by choice, every day, for years — and what "every day, for years" does to a metal once it's in you.

So what? The honest scale, and the thing it can't see

Six micrograms of lead in a scoop — is that a lot? The most careful answer available comes from a 2020 risk assessment that pooled 148 protein powders and did the arithmetic properly.⁸ It built a cumulative hazard index — a chronic model, one that assumes you drink the stuff daily and adds up the exposure — and it found that index sitting below 1, the line where concern begins, even at the 95th percentile and three servings a day. No product in the set pushed a modelled blood-lead level over the reference the US uses for adults. On those numbers, for most products at ordinary servings, the exposure is modest. I'm not going to dress that up as anything else. The industry's rejoinder is, at this level, correct. A metal detected by a modern mass spectrometer is not the same as a metal at a dose that harms you, and the warning thresholds these tests measure against carry safety factors of around a thousandfold.⁷

But look at what that hazard index is, and what it therefore cannot see. It is an average, built on reference doses, describing a general adult. Three things this product's own biology insists on fall straight through it.

The first is the body the label routes toward — and here I'm drawing an inference the data support but do not yet prove, so I'll mark it as one. The groups clinical medicine already flags for lower iron — menstruating and pregnant women, vegetarians — are also, plausibly, much of the plant-based, "clean" protein market. I can't point to a study showing that clean-protein buyers specifically skew lower-iron; that link is unmeasured, and if it turned out to be false, the sharpest claim in this report would soften with it. But the overlap is real enough to take seriously, and it matters, because a low-iron gut absorbs cadmium more readily — iron deficiency turns up the very transporter cadmium hitchhikes on to get across the gut wall.⁹ And those same groups start closer to their ceiling: European vegetarians already take in up to twice the tolerable weekly cadmium the EU sets, from ordinary diet alone, before any powder is added.¹⁰ A hazard index averaged across a whole population smooths exactly this overlap out. It reports the middle of a distribution to someone who may be sitting in its tail.

The second is that lead has no floor. In 2010 the EU's food-safety scientists withdrew the idea of a safe threshold for lead's effect on the developing brain — not because they couldn't find it, but because there isn't one; their reference point is a benchmark dose, an acknowledgement that the honest answer is "less is better, all the way down."¹¹ A hazard index is a ratio against a reference dose. It is not built to represent a substance with no safe level, and it quietly rounds one off.

The third is the simplest. Among those 148 products, one crossed a hazard quotient of 1 — a single cadmium outlier. One in 148 is a reassuring sentence if you are describing a market. It is not a reassuring sentence if you are holding one tub in your kitchen and have no way to find out whether it's the one. And you have no way: nothing on the pack states the per-serving metal, and no test is required to put it there.

That is the sentence the whole thing turns on. The lab measures the tub once and hands back a number. Your body measures the metal every morning for thirty years and hands back nothing. Cadmium banks in the kidney with a biological half-life of six to thirty-eight years — it goes in and it essentially stays, accumulating in the same cells for decades until, if it climbs high enough, they begin to fail.¹² Lead banks in bone, and it does not stay put: during pregnancy and lactation the skeleton remobilises, and the lead comes back out into the blood. In one long-running study of maternal and cord blood, around 79 percent of the lead a mother's skeleton released during pregnancy crossed into the fetus.¹³ The number on the label is a snapshot of the wrong object. The dose that decides anything is the running total in your bones, and no certificate of analysis will ever print it. "Modest on average" and "no way to know if yours is the one" are both true, and together they are the entire case.

You did the careful thing

It's worth being clear about whose fault this isn't. You reached past the loud brand for the quiet one. That is not a mistake — on almost every shelf you shop, it is the single most reliable move you can make. It reads the dyes, the sweeteners, the preservatives, the things a factory added, and it is right about all of them. It has one blind spot, and this hazard lives inside it: nobody added the metal. It was in the soil before the seed. Your instinct is trained to catch what a manufacturer put in; it has no setting for what a plant drew up. The carefulness didn't fail. It just stops at the factory gate, and this one was already upstream, in the ground.

The ceiling that doesn't bite, and the pack that says nothing

Here is where the regulation is supposed to catch what the label doesn't — and where the gap actually sits. It is not that protein powder is unregulated. A legal ceiling for lead in food supplements exists: the EU's contaminants regulation, retained in UK law, sets it at 3.0 milligrams per kilogram.¹⁴ Do the arithmetic on a 40-gram scoop and that ceiling permits 120 micrograms of lead per serving — about nineteen times the 6.3 micrograms Consumer Reports flagged in Huel Black, and two hundred and forty times California's warning trigger. A powder can carry the flagged amount many times over and remain comfortably, lawfully inside the limit.

And that ceiling may not even apply. A "food supplement," in UK law, is a concentrated nutrient source sold in measured dose form — a capsule, a tablet, a sachet. A scooped tub of protein isn't that, and under UK food-law guidance most protein powders are regulated as ordinary food rather than supplements.¹⁵ Ordinary food has no protein-powder-specific metal limit at all. So the honest description is this: a legal ceiling exists, sits about nineteen times above the levels the tests flagged, and may not bind this product in the first place — with nothing on the pack disclosing the per-serving metal and no batch test required to measure it. In the United States the frame is even barer: under the 1994 supplement law, the Food and Drug Administration cannot approve these products before sale and acts only afterward.¹⁶

The one metal signal a shopper ever actually sees is California's Proposition 65 warning, and it is worth understanding exactly what it is, because every frightening percentage in the coverage is measured against it. The Prop 65 level for lead — 0.5 micrograms a day — is a warning-label trigger, built with a large safety margin below the dose at which harm has been observed. Crossing it means a product must carry a warning; it does not mean the product will hurt you.¹⁷ So "47 percent over a regulatory limit" and "1,290 percent of the level of concern" are ratios against a conservative labelling threshold, not measures of poisoning. Set the flagged Huel serving against a different yardstick and the scale looks different again: 6.3 micrograms is roughly thirteen times California's warning trigger, but only about seven-tenths of the interim daily lead figure the FDA uses for women of childbearing age. And taken with food rather than on an empty stomach, most of it never crosses the gut at all.¹⁸ Which is the reassurance, and it is real. It is also, exactly, the problem: the number is small, invisible, unmeasured and permanent, and there is no signal on the pack that separates the small one from the outlier.

The clean sample that still asked for a law

If you want the single study that holds both halves of this honestly, it isn't American. In July 2025, researchers tested 22 protein powders on the Hungarian market — whey, vegan and beef — and found lead, cadmium, arsenic and mercury all within EU limits. A clean result. And in the same paper, having found nothing over the line, the authors called for mandatory heavy-metal testing as routine quality control anyway, because voluntary compliance leaves the question of any individual batch unanswered. They titled it a reassuring snapshot that is not a reassuring quality guarantee.¹⁹ That is the whole stance in one study: the metal is usually modest, and modest-usually is not the same as measured-here. There is no European or UK consumer-group test of protein powders on British shelves — no Which? investigation, nothing equivalent to the American work. That absence is not evidence the shelves are clean. It is the disclosure gap in its purest form: the UK shopper has even less to go on than the American one, and the mechanism that puts cadmium in cacao and arsenic in rice does not stop at the border.

None of this means plant protein is the problem, and the data are clear enough that I won't let the piece imply it. Across every test, whey and egg run lower than plant and cocoa, and plain and vanilla run lower than chocolate. If the metal is what you're minding, those are genuinely the lower-load choices — a preference the evidence supports, not a reason to abandon plants.

What the tub can and can't tell you next time

The correction that works here is small and specific: for this hazard, stop asking whether the tub is natural and start asking whether anyone measured it. Concretely —

• Turn the tub over and read what the words certify. "Organic," "plant-based," "clean," "vegan," "natural" describe the farming and the additives. None of them is a statement about soil metal, and on the testing record they can run opposite to it. They are not the signal you want for this.
• Look instead for a batch-testing mark, or a published number. NSF Certified for Sport and Informed Sport both test individual production lots for heavy metals, not just the recipe once — the presence of the mark means the brand paid to measure what's actually in the batch.²⁰ Better still is a brand that publishes a per-serving heavy-metals figure on a certificate of analysis you can find. That is the only signal that tracks the hazard, because it is the only one that measures it.
• If you use a powder, lean the feedstock lower and vary it. Whey or egg over plant, plain over chocolate, if the metal is your concern; and don't drink the same single blend every day for years — the banking is a function of the same source repeated, so rotating sources spreads the load.
• Take it with a meal, not fasted. Lead absorption drops roughly threefold to sixfold when the shake goes down with food rather than on an empty stomach.¹⁸ It is the one free lever in the whole story.
• And the option that costs nothing and needs no label decode: for most people who already eat enough protein, the powder is optional. Whole food — dairy, eggs, fish, legumes, meat — carries no concentration step and comes with no per-serving metal question to answer. If you don't need the scoop, the cleanest tub is no tub.

The fix that would make all of this unnecessary is not a per-tub tax that would fall hardest on small honest brands. It is a required per-serving heavy-metals disclosure and a mandatory batch test, its cost spread across the industry the way any floor cost is — the thing the Hungarian researchers asked for, and the thing Consumer Reports petitioned the FDA to impose. It would put the one signal that tracks the hazard onto the one surface the shopper actually reads.

What would turn this back into a footnote

The strongest version of the case against this report is worth stating in full, because a reasonable person could hold it. Heavy metals are in virtually all soil-grown food; a modern spectrometer detects them far below any harm threshold, so "detected in protein powder" is a statement about instruments, not danger. Every alarming figure here is a ratio against a deliberately conservative warning trigger. The one peer-reviewed cumulative assessment put the hazard index below 1 even at the 95th percentile; the nearest European test came back clean; and the headline inversion leans heavily on a non-peer-reviewed body that profits from the certification it recommends. The practical advice — take it with food, prefer plain, choose a tested brand, or eat whole food — is cheap and already available. A fair reader could conclude: a genuine information gap, worth a testing rule, but not a reason to change what most people do tomorrow morning.

I think that reading is half right, and I've tried to give the half that's right its full weight. What it can't answer is the individual question. A population-average hazard index cannot tell any single shopper whether she holds the median tub or the one outlier, and it averages away the lower-iron, higher-absorbing groups who sit closest to the cadmium ceiling and who may overlap most with the plant-protein market. Lead's developmental endpoint has no safe floor for a hazard index to sit on. And the single signal that would resolve any of it — a per-serving number, measured per batch — is precisely the signal a brand can lawfully leave off for free.

So what would change this analysis? A published mass-balance following a named crop from field to finished isolate, showing the powder concentrates metal only trivially, would shrink the concern. A randomised sample — not the top-sellers these testing bodies chose, but a representative draw — that failed to reproduce the plant-over-dairy pattern would weaken the inversion. The Clean Label Project publishing its full methodology for independent scrutiny would let its numbers carry weight they currently can't. A UK-market test finding British tubs clustered low would answer the border question the American data can't. And a demographic study showing that plant-protein buyers do not, in fact, skew toward the lower-iron, higher-absorbing groups would collapse the sharpest claim in this report back to a coincidence. Each of those is obtainable. None of them exists yet — and until one does, the honest position is the mature one: modest per serving, permanent in aggregate, no safe floor. Reduce the avoidable, calmly. Don't panic, and don't shrug.