Cruciferous Vegetables — Body Handbook

Food · §254

Cruciferous Vegetables

Eat broccoli most days and your urine actually carries more of the benzene from city air out of your body, in a clinical trial, on day one. That is not a metaphor — the cabbage family contains a molecule that switches on the liver enzymes that conjugate and excrete carcinogens, and the effect shows up at culinary doses. Stack it daily for years and the cancer and heart-disease numbers move in the cohorts: about 20% lower total mortality at the high end of intake, slower arterial calcification, a couple of points off blood pressure within weeks. The headline trick is to keep the enzyme alive while you cook — chop, wait, then heat lightly.

Do · Daily Evidence Moderate Chapter Food

The case is mostly longevity-shaped — the heart-disease and cancer signals across the big cohorts are some of the most consistent in nutrition. The weeks-scale wins are modest: a small inflammation drop, a couple of mmHg off the systolic. Costs about as much as any other vegetable, takes about as much effort, and the technique that actually matters — light steaming over boiling, with a 40-minute rest after chopping — is a one-time habit shift. The "kale wrecks your thyroid" worry sounds bigger than the human evidence actually says.

What makes broccoli and its relatives different from a carrot is a chemical called sulforaphane — and how the plant hides it. Inside the intact vegetable, sulforaphane doesn't exist yet. Two ingredients sit in separate compartments of the cell: a sulfur-rich storage compound called glucoraphanin and the enzyme that cuts it open, myrosinase. Bite the leaf, chop the floret, slice the cabbage — the walls break and the two finally meet. Sulforaphane forms in your mouth, in seconds.

Once it gets to your liver and your gut wall, sulforaphane does one thing very precisely. It nudges a transcription factor called Nrf2 out of its day-job binding partner and into the cell nucleus, where it turns on a cluster of genes that code for the phase II family of detoxification enzymes — the conjugation crew, the proteins that pin a marker tag onto things like inhaled benzene, charred-meat byproducts, and ordinary metabolic exhaust so the kidneys can flush them. You can think of it as turning the dial up on the body's own carcinogen-excretion machinery. The molecular biology of this has been worked out in unusual detail since the early nineties Zhang et al. 1992.

There is a second arm, which matters mostly for the cancer side. Cruciferous vegetables also carry glucobrassicin, which hydrolyses to indole-3-carbinol and then, in the acid of your stomach, condenses to a molecule called DIM. DIM shifts how your body processes estrogen — toward the less-proliferative metabolite — which is the mechanism most often invoked for the breast and hormone-related cancer signals Higdon et al. 2007.

Both arms require chewing or chopping. A whole, intact, unbroken broccoli floret swallowed by a chimpanzee delivers very little sulforaphane. A floret your knife went through, sat in the air for a moment, then steamed lightly delivers a lot.

Does it actually work?

This is one of the cleaner stories in nutrition. The mechanism is closed, the population data is consistent, and the proximal biomarker — does eating these things actually make the body better at clearing carcinogens? — has been demonstrated in a controlled trial in humans, not just in cells in a dish.

The cohort epidemiology lines up. The Shanghai Women's and Men's Health Studies followed 134,796 Chinese adults for 13 years. The top quintile of cruciferous-vegetable intake had 22% lower total mortality and 31% lower cardiovascular-disease mortality than the bottom quintile, with a clean dose-response on the way up Zhang et al. 2011. A Japanese cohort of 88,184 adults followed for nearly 17 years saw the same pattern, strongest in women, on heart and cerebrovascular endpoints Mori et al. 2019. The signal replicates across geographies that eat completely different cuisines, which is what real effects look like.

For cancer specifically, the colorectal endpoint is the most-studied — Wu and colleagues pulled together 35 studies and found a 18% lower relative risk in the highest-intake group, with the dose-response curve kicking in around 20 g/day (a quarter cup of broccoli, half a cup of cabbage) Wu et al. 2013. A 2024 umbrella review combined 14 meta-analyses across colorectal, lung, breast, bladder, gastric, pancreatic, ovarian, and renal cancer; the relative risks all landed in the 0.75–0.90 range at the high end of intake Aggarwal et al. 2024. None of these trials are randomised. The WCRF's expert panel rates the evidence at probable, not convincing, for exactly that reason WCRF/AICR 2018.

On the heart side, the proximal-mechanism data fills in the cohort signal. Among 684 older Australian women, the ones eating more than 45 g of cruciferous vegetables a day — a quarter cup of steamed broccoli — had 46% lower odds of having extensive build-up of calcium on the abdominal aorta than the women eating almost none Blekkenhorst et al. 2020. In the same cohort, higher intake tracked with thinner carotid arteries — the standard ultrasound marker of generalised atherosclerosis Blekkenhorst et al. 2018. And a small Australian crossover trial in 2024 swapped 18 mildly hypertensive adults between a cruciferous-heavy diet and a root-vegetable control for two weeks each; 24-hour ambulatory systolic blood pressure dropped 2.5 mmHg on the cabbage-family arm Connolly et al. 2024. Small trial, modest number, but the direction and magnitude line up with the cohort and atherosclerosis biomarker data.

Two trials on inflammation. In 81 type-2 diabetics given broccoli-sprout powder for four weeks, hs-CRP — a general inflammation marker — fell about 20% against placebo Bahadoran et al. 2012. In 40 overweight adults given fresh broccoli sprouts for 70 days, CRP and IL-6 both dropped in line with measured urinary sulforaphane Lopez-Chillon et al. 2019. Effect sizes are modest. The direction agrees across populations.

The honest gap: there is no randomised trial with a hard cancer or heart-attack endpoint. There probably never will be — feeding people broccoli for 20 years in a controlled trial is not how the world works. The case rests on mechanism + biomarker + multiple consistent observational cohorts. Among diet-and-disease stories, that's the high end of what evidence ever looks like.

How much, how often, how to cook them

The target dose, where the cohort dose-response curves actually start moving, is somewhere around a small cup a day, three to five days a week. That maps to about 30–45 g — a quarter cup of steamed broccoli florets, half a cup of cooked cabbage, a cup of leafy kale, six or seven Brussels sprouts. The Australian blood-pressure trial used roughly four servings a day for two weeks to land its effect; you don't need that much for the longevity signal, where the cohort benefit climbs through the first ~75 g/day and we don't really know what happens past that.

Pick from the family, rotate. Broccoli, cauliflower, cabbage, kale, Brussels sprouts, bok choy, watercress, arugula, radish, kohlrabi. Different relatives carry different glucosinolate profiles; the variety probably matters at the margin.
Chop first, then wait. Cutting the vegetable lets the enzyme inside meet the sulforaphane precursor. Forty minutes on the cutting board, room temperature, before any heat goes near it. This is the single highest-leverage trick.
Steam light, don't boil long. Three to four minutes of steaming preserves both the enzyme and the conversion. Long boiling destroys both — most of the bioactive ends up in the water you pour down the sink Dosz and Jeffery 2013.
Cheat code if you forgot. A little raw mustard, radish, arugula, or wasabi sprinkled over already-cooked broccoli restores the missing enzyme — the raw plant donates myrosinase to the cooked one and conversion resumes in your gut.
Raw and fermented count too. Slaw, kimchi, sauerkraut, raw cabbage in salads. The glucosinolates are intact; your own chewing activates the enzyme.

One molecular note worth knowing: the heat sensitivity isn't picky. The enzyme inactivates above about 60 °C. Steaming for a few minutes leaves enough enzyme alive and enough vegetable intact; aggressive boiling, frying past golden, or pressure-cooking at full whack does not. If the broccoli is grey-green and limp, the sulforaphane went with it.

What about supplements? Glucoraphanin or broccoli-sprout extract pills exist, and the better ones include myrosinase. The catch is that most over-the-counter versions don't, and without the enzyme the bioavailability collapses — the conversion ends up entirely outsourced to your gut bacteria, which varies tenfold person to person Sikorska-Zimny and Beneduce 2021. Fresh, lightly cooked, real cruciferous vegetables clear the bar; the pills are a coin flip without label-reading.

When to be careful

For everyone else, cruciferous vegetables in normal culinary amounts are about as safe as food gets. The popular fear around kale wrecking your thyroid is addressed in What people get wrong below — short version, the human data does not support it at the doses ordinary people eat. The one case where it becomes a real consideration is in active iodine deficiency, which is uncommon in countries with iodised salt; if you're not eating any iodine source at all and have a thyroid condition, raise it with your clinician before going on a long uncooked-kale-juice streak.

What people get wrong

"Kale is bad for your thyroid." This is the loudest myth in the cruciferous corner of the internet, and the human evidence does not back it up. The mechanism is real in principle: a compound called thiocyanate, made when glucosinolates break down, competes with iodine at the gland. The problem with translating that into a human risk is the dosage maths. Felker, Bunch and Leung modelled blood thiocyanate concentrations from typical cruciferous intake and found them roughly 10 to 100 times below what would meaningfully suppress thyroid function in iodine-replete adults Felker et al. 2016. The one controlled feeding trial in healthy people fed 150 g of cooked Brussels sprouts a day for four weeks and measured no change in thyroid hormones at all Linus Pauling Institute 2017. The case reports of clinical hypothyroidism from cruciferous intake are real but came from extreme, mostly-raw, mostly-juiced kale habits in people who were already iodine-deficient. For a healthy person eating cooked broccoli several times a week, the thyroid claim is noise.

"Raw is better — heat kills the goodness." Partly the opposite, actually. Raw cruciferous gives you intact glucosinolates and a sliver of plant myrosinase from your chewing, but it's hard on the gut and most people can't tolerate much of it. Light cooking — three to four minutes of steaming — keeps enough enzyme alive to do the work and softens the vegetable enough that you'll actually eat a useful amount. Long boiling is what destroys the chemistry; that's the call to avoid, not heat itself.

"You can just take a sulforaphane pill." Maybe, but it depends entirely on whether the pill contains active myrosinase, which most don't. Without the enzyme, the glucoraphanin in the capsule has to wait until it reaches your colon and hope your gut bacteria do the conversion — which varies wildly between people. Fresh broccoli sprouts contain about 50 times more glucoraphanin per gram than mature broccoli, which is why the high-end clinical trials use them; for the cohort-level longevity signal, ordinary cooked vegetables are what the evidence is actually about.

"It's a detox food." Closer to the truth than most "detox" claims, but the framing is still loose. There is no toxin reservoir that gets emptied. What sulforaphane does is upregulate the everyday liver enzymes that conjugate carcinogens and oxidative byproducts — the system is running constantly; cruciferous vegetables turn the dial a little higher. The Egner trial showed measurably more benzene leaving the body in the urine; that is not a cleanse, it is a slightly faster processing line.

Where the protocol breaks down in practice

You boil them to oblivion. The single most common failure. Boiled-grey broccoli has lost most of its enzyme and most of its precursor to the water. If your kitchen default is fifteen minutes in a pan of bubbling water, you are effectively eating fibre and very little of what makes the literature work.

You eat them once a week. The cohort signal is dose-responsive and habit-shaped. A virtuous Sunday salad and nothing the rest of the week is not the pattern that moves the numbers — the gut bacteria that handle some of the conversion, and the steady tissue exposure to isothiocyanates, both want regular small doses, not occasional large ones.

You overdo it and pay in gas. Cruciferous vegetables carry a class of fermentable sugars (raffinose-family) that your gut bacteria break down into hydrogen and methane. For people with irritable bowel syndrome or a generally touchy gut, a sudden half-pound of raw kale is a bad afternoon. Smaller portions, cooked longer (which costs some sulforaphane but earns digestibility), and fermented forms — kimchi, sauerkraut, lacto-fermented slaw — are the workaround. The point is to make it sustainable.

You rely on a supplement with no myrosinase. Read the label. If it lists glucoraphanin but not myrosinase or "active enzyme", the conversion is being outsourced to a process that varies tenfold person to person. The brands that do include the enzyme tend to be the more expensive ones, which sometimes makes ordinary broccoli a better deal anyway.

Making it stick

The honest version of "eat cruciferous vegetables daily" is: keep two or three of them in the kitchen at all times, and let convenience pick. A bag of pre-shredded slaw mix in the fridge is half a daily serving handled. Frozen broccoli florets work as well as fresh for the bioactive chemistry — Dosz and Jeffery showed the freeze-thaw cycle, with the right rest time, actually improves sulforaphane formation in some cases Dosz and Jeffery 2013. A bunch of kale lasts a week in the crisper; a head of cabbage lasts a month.

Practical kitchen rhythm that earns the dose: chop the broccoli or shred the cabbage first thing while the pan is heating. Walk away for the duration of the rest of the meal prep — forty minutes if you're cooking properly, twenty if you're not, anything is better than zero. Then a brief steam at the end. The wait isn't dead time; it's the rest of dinner.

For breakfast or lunch where cooking isn't on the table: shredded raw cabbage in a wrap, kale folded into eggs at the very last minute, watercress on a sandwich, kimchi as a topping. The form doesn't matter much; the regularity does.

Cost is low. A head of broccoli is a dollar or two; a head of cabbage feeds a household for a week. Even at premium organic prices the daily portion lands well under a dollar.

What you forfeit by skipping them

The cost of leaving cruciferous vegetables out of the rotation is not something you'll feel next month. There's no syndrome that announces itself. The trade is paid in slow, invisible drift on the long-arc numbers — the ones that matter most when they're already past fixing.

In your fifties, the abdominal CT done for an unrelated complaint shows up with the band of calcium along the aorta that means the vessel has been quietly hardening for two decades. In the women's cohort, the people eating less than 15 g a day of cruciferous vegetables were almost twice as likely to have that result as the women eating more than 45 g Blekkenhorst et al. 2020. The hardening doesn't unhappen — you don't get to walk it back at sixty by starting then.

In your sixties, the version of you that didn't eat the broccoli is in the population whose cancer-screening colonoscopy is more likely to find something. The colorectal cancer risk reduction in the high-intake group of the meta-analysis is about 18% — across a lifetime that's the difference between a polyp and an invasion Wu et al. 2013. The 22% mortality gap in the Shanghai cohort is the gap between the friend who keeps showing up at the reunion and the one who stopped Zhang et al. 2011.

The day-to-day forfeit is smaller and easier to ignore. A blood pressure reading a couple of points higher than it could have been — nothing on one reading, the difference between needing medication and not over time. A C-reactive protein number the doctor circles in the slightly-elevated column. The pollutants from the air in the city you live in, processed a little more slowly than they could have been, conjugating and excreting a little less of what shouldn't have stayed.

None of this is a single missed dose. It's the version of you that compounded the absence across the decade, on the wrong side of every population curve, watching the divergence at sixty-five from the side that didn't make the small change at thirty-five.

What changes if you start

The first weeks are quiet. Cruciferous vegetables are not caffeine; you don't feel them. If you draw blood at four weeks, your inflammation marker is meaningfully lower — Bahadoran's diabetics dropped CRP about 20% against placebo on broccoli sprouts Bahadoran et al. 2012. If you wear a 24-hour ambulatory blood pressure monitor, your systolic is a couple of mmHg lower than it was — the Connolly trial saw it in two weeks Connolly et al. 2024. Neither of these is something you notice. Both are real.

At a year, the bloodwork your GP runs is in the unremarkable column. The hs-CRP doesn't get circled. The blood pressure cuff doesn't trip the prehypertension threshold this time. None of this gets praise. It's the absence of bad news — the kind of result that doesn't generate a story.

At five to ten years, the divergence starts showing in the imaging. The carotid intima — measured by ultrasound, the standard generalised-atherosclerosis marker — is thinner than the cohort average for your age Blekkenhorst et al. 2018. The aorta isn't laying down calcium at the rate it was going to. The vascular age that diverges from the chronological one — the sixty-year-old whose arteries look fifty — is what this looks like in practice.

The same internal-aging slowdown leaks out into how you look. Less oxidative wear on the collagen, better-perfused skin from arteries that aged kinder, lower baseline inflammation pulling against the skin barrier — none of this is a topical effect and none of it happens in weeks, but at fifty, sixty, seventy, the cumulative version of you who ate plants like these most days looks recognisably less weathered than the version who didn't. It's not the line on the bottle of any skin product; it's the upstream story those products keep gesturing at.

At fifteen to twenty years, the cancer-that-didn't-happen is the payoff. Across colorectal, gastric, bladder, lung, breast, the relative risks in the highest-intake quintile run 10 to 25% below the lowest, replicated across cohorts on three continents Aggarwal et al. 2024. None of this is felt; all of it shows up in the version of your life where the screening colonoscopy is clean, where the smoker-adjacent lung scan finds nothing, where the diagnosis happens to someone else in the room. The Shanghai cohort's 22% lower mortality in the highest cruciferous quintile is what that looks like at population scale Zhang et al. 2011 — the version of you that keeps showing up.

The pace of every payoff here is honest: weeks for the inflammation marker and the blood pressure, years for the vascular calcification and the way you look, decades for the cancer story. The intervention is small and continuous; the result is geological.

Adjacent topics worth knowing about

Vitamin K and vascular calcification. Part of the heart story here is the vitamin K1 in kale and broccoli activating matrix Gla protein — the body's own vascular-calcification brake. Worth its own entry.
Fibre and the gut microbiome. The bacteria that hydrolyse uncooked glucosinolates are a slice of the same population that ferments the rest of your plant intake; the broader gut-microbiome story is what makes this one work.
Broccoli sprouts specifically. Carry ~50× the glucoraphanin of mature broccoli, used in most of the high-dose clinical trials. A more concentrated version of the same effect.
Dietary nitrate (beets, leafy greens). Separate cardiovascular mechanism, complementary to the cruciferous story — different molecule, same direction.
Helicobacter pylori. Sulforaphane has direct antimicrobial activity against the gastric ulcer / cancer bug — small additional reason this all hangs together.

Substance and claimed effects

The cruciferous (Brassica) family eaten regularly: broccoli, cabbage, kale, Brussels sprouts, cauliflower, plus the smaller-tonnage relatives — bok choy, watercress, arugula, kohlrabi, radish, turnip greens, mustard greens, collards. The distinguishing biochemistry is the glucosinolate system: sulfur-and-nitrogen-rich storage compounds (≈120 known, broccoli alone carrying 4-methylsulfinylbutyl glucosinolate / glucoraphanin as the headliner) sequestered in plant cells alongside the enzyme myrosinase, kept in separate compartments. Mechanical damage — chewing, chopping, insect bite — brings them together, and myrosinase hydrolyses glucoraphanin to sulforaphane (and glucobrassicin to indole-3-carbinol, which condenses to DIM in stomach acid) Higdon et al. 2007. Sulforaphane is the most-studied isothiocyanate end-product and the molecule most of the literature traces effects through.

Claimed effects across the literature, in rough order of evidence strength: induction of phase II conjugation enzymes (glutathione-S-transferases, NQO1, UGTs) via the Keap1/Nrf2 pathway; reduced cancer incidence at multiple sites (colorectal, gastric, bladder, lung, breast, pancreatic, prostate); reduced cardiovascular and total mortality in long prospective cohorts; lower blood pressure and slower vascular calcification; modest reductions in systemic inflammatory markers (hs-CRP, IL-6); modulation of estrogen metabolism via the I3C/DIM arm; gut-microbiome activity (the microbiota itself converts unhydrolysed glucosinolates when plant myrosinase has been heat-killed); and a long-rumoured, weakly-supported risk of thyroid suppression via thiocyanate ions competing for iodine uptake. The entry covers all of these holistically.

Evidence by addressing question

Mechanism

Two parallel mechanistic arms drive most of the catalogue-relevant effects. The first is sulforaphane → Keap1 → Nrf2 → phase II enzymes. Under basal conditions Keap1 tethers the transcription factor Nrf2 to the cytoplasm and targets it for proteasomal degradation. Sulforaphane covalently modifies reactive cysteines on Keap1, releases Nrf2, which translocates to the nucleus and binds antioxidant response elements upstream of phase II detoxification genes: glutathione-S-transferases (GSTs), NAD(P)H:quinone oxidoreductase 1 (NQO1), UDP-glucuronosyltransferases, heme oxygenase-1, gamma-glutamylcysteine ligase. The system upregulates the body's own conjugation-and-excretion machinery for electrophilic carcinogens and oxidative byproducts. Zhang and Talalay's 1992 isolation paper is the foundational mechanism reference — sulforaphane identified, by activity-guided fractionation of broccoli extract, as the major Nrf2-pathway inducer in the plant Zhang et al. 1992.

The second arm is indole-3-carbinol → DIM → aryl hydrocarbon receptor. Glucobrassicin hydrolysis yields indole-3-carbinol, which condenses in stomach acid to 3,3'-diindolylmethane (DIM) and related oligomers. DIM is an AhR ligand and shifts estrogen metabolism toward the 2-hydroxyestrone pathway (relative to 16α-hydroxyestrone) — a less proliferative profile. This is the mechanistic basis for the cruciferous-and-estrogen-driven-cancer signal and for the hormonal-balance claims around DIM supplements.

Glucosinolates themselves are biologically inert; both arms require hydrolysis. Plant myrosinase is destroyed above ~60 °C, so boiled-to-death broccoli yields little sulforaphane from its own enzyme. The gut microbiota can hydrolyse residual glucosinolates — Bacteroides, Lactobacillus, Bifidobacterium, Enterococcus species produce bacterial myrosinase-equivalent activity — but conversion efficiency varies ~tenfold between individuals and a fraction goes toward inactive nitriles rather than the bioactive isothiocyanates Sikorska-Zimny and Beneduce 2021. Bioavailability data: under controlled feeding, ~10–40% of ingested glucoraphanin appears as isothiocyanate mercapturic acids in urine when plant myrosinase is intact, dropping to 1–10% when it has been cooked out Charron et al. 2018.

Evidence — cancer

The strongest, broadest signal in the literature. The 2024 umbrella review pooled 14 meta-analyses covering ~226 case-control and cohort studies and found significant inverse associations for colorectal, lung, breast, bladder, gastric, pancreatic, ovarian, and renal cancers — most relative risks landing in the 0.75–0.90 range at the highest-intake quintile vs the lowest Aggarwal et al. 2024. The colorectal signal is the most replicated single endpoint: Wu et al.'s 2013 meta-analysis of 35 studies put the high-vs-low RR at 0.82 (95% CI 0.75–0.90), with a clear dose-response — each additional 20 g/day buying ~3% lower risk, peak per-gram protection in the 20–40 g/day range Wu et al. 2013.

The lung and bladder signals interact with GSTM1/GSTT1 genotype: carriers of the null deletions (~50% of the population for GSTM1) show larger protection from cruciferous intake, consistent with the Nrf2 mechanism — the people whose alternative detoxification is impaired get the most from a sulforaphane-driven boost Higdon et al. 2007. Egner and Kensler's Qidong intervention trial demonstrates the proximal mechanism in vivo in humans: 291 adults in a heavily air-polluted Chinese region drank a daily broccoli-sprout beverage for 12 weeks; urinary benzene mercapturic acid excretion jumped 61% and acrolein conjugate excretion 23% relative to placebo, both effects rapid (within days) and sustained across the trial Egner et al. 2014. The body really does excrete more inhaled carcinogen when it sees more sulforaphane.

Caveat: the WCRF/AICR 2018 Continuous Update Project ranks the evidence for non-starchy vegetables (a category that includes cruciferous) against cancer at the probable level, not convincing, because the cohort signal is consistent but cannot fully separate cruciferous-specific effects from healthy-diet confounding WCRF/AICR 2018. Sulforaphane RCTs in humans for hard cancer endpoints don't exist — the trials are biomarker studies (Egner, GST/NQO1 induction in nasal cells, urinary detox markers).

Evidence — cardiovascular and longevity

The Shanghai Women's and Men's Health Studies (n=134,796; 13 years of follow-up) found the highest quintile of cruciferous intake associated with a 22% lower total mortality risk and 31% lower cardiovascular-disease mortality vs the lowest, with a clear dose-response. The inverse association was specifically driven by cardiovascular mortality, not cancer mortality, in this cohort Zhang et al. 2011. A Japanese cohort (JPHC Study, n=88,184, 16.9-year follow-up) replicated the mortality finding, with the strongest effects on cardiovascular and cerebrovascular endpoints in women Mori et al. 2019. Aune's 2017 dose-response meta-analysis of all-cause mortality showed cruciferous vegetables outperforming most vegetable categories on a per-100-g/day basis Aune et al. 2017.

The proximal-mechanism data backs this up. Blekkenhorst et al.'s 2018 cross-sectional analysis (n=684 older Australian women) found higher cruciferous intake inversely associated with carotid intima-media thickness — a measure of generalised atherosclerosis Blekkenhorst et al. 2018. The 2020 follow-up in the same cohort found women eating >45 g/day of cruciferous vegetables had 46% lower odds of extensive abdominal aortic calcification vs the <15 g/day group, independent of total vegetable intake Blekkenhorst et al. 2020. The plausible mediators are vitamin K (cruciferous vegetables are major dietary sources of vitamin K1, which activates matrix Gla protein — a vascular-calcification inhibitor), nitrate (modest in cruciferous, larger in beets/leafy greens), and the sulforaphane-driven anti-inflammatory effect on the endothelium.

For blood pressure specifically, the 2024 VESSEL crossover trial (Connolly et al., n=18 Australian adults with mildly elevated BP) compared 2 weeks of a cruciferous-rich diet (~4 servings/day of broccoli, cabbage, cauliflower, kale) against 2 weeks of root-vegetable controls. Twenty-four-hour ambulatory systolic BP fell 2.5 mmHg on cruciferous relative to control — a magnitude that translates, at population scale, to roughly 5% lower risk of major cardiovascular events Connolly et al. 2024. The trial is small but cleanly designed and the effect direction matches the cohort and atherosclerosis literature.

Evidence — inflammation

Bahadoran et al.'s 2012 RCT in 81 type-2 diabetics gave broccoli-sprout powder (the high-sulforaphane formulation) for 4 weeks vs placebo and observed hs-CRP fall 20.5% and 16.4% in the two active-treatment arms, with non-significant trends down in IL-6 and TNF-α Bahadoran et al. 2012. Lopez-Chillon et al.'s 2019 trial in 40 overweight subjects fed 30 g/day of fresh broccoli sprouts for 70 days saw plasma IL-6 and CRP fall in parallel with increased urinary sulforaphane metabolites Lopez-Chillon et al. 2019. The effect size is modest; the consistent direction across two well-controlled trials in different populations is what matters.

Evidence — thyroid

The goitrogen story is mostly theoretical in humans at culinary doses. Progoitrin (a glucosinolate found in higher concentration in older Brassica cultivars and seeds) hydrolyses to goitrin, which can suppress thyroid hormone synthesis. Thiocyanate (another hydrolysis product) competes with iodide at the sodium-iodide symporter. Felker, Bunch, and Leung's 2016 quantitative analysis modelled exposure: typical cruciferous-vegetable intakes in iodine-replete adults produce thiocyanate plasma concentrations roughly 10–100× below those known to affect thyroid function, and the goitrin content of common varieties (broccoli, kale, cabbage) is negligible to begin with Felker et al. 2016. The Linus Pauling Institute summary: no controlled human study has demonstrated clinical hypothyroidism from dietary cruciferous intake at any realistic level, and the one bounded human feeding trial — 150 g/day cooked Brussels sprouts for 4 weeks — produced no measurable change in thyroid function Linus Pauling Institute 2017. The risk is real in two narrow contexts: severe iodine deficiency (thyrocyte iodine uptake becomes the rate-limiting step) and extreme uncooked-juiced-kale-style intakes documented in case reports.

Evidence — gut

Two distinct effects. First, the microbiome does hydrolyse glucosinolates the plant didn't — bacterial myrosinase-equivalent activity in Bacteroides thetaiotaomicron, lactobacilli, and bifidobacteria recovers some of the bioavailable isothiocyanate yield from cooked cruciferous vegetables, with the conversion efficiency varying ~tenfold between individuals based on microbiome composition Sikorska-Zimny and Beneduce 2021. Second, sulforaphane has direct antimicrobial activity against Helicobacter pylori — Fahey et al.'s 2002 in vitro and rodent work demonstrated bactericidal effect at micromolar concentrations and reduced gastric-tumour incidence in benzo[a]pyrene-exposed mice Fahey et al. 2002. The downside: the same fermentable carbohydrate (raffinose-family oligosaccharides) that feeds gut bacteria produces hydrogen and methane along the way — hence the well-known gas-and-bloat from large servings, particularly in IBS populations with visceral hypersensitivity.

Protocol and practicalities

Effective dose floor from the cohort dose-response curves: ~30–40 g/day, roughly a small cup of broccoli florets, half a cup of cabbage, or a cup of leafy kale, three to five days a week. The Connolly trial used ~300 g/day in the active arm (well above usual intake) over two weeks to see a measurable BP effect; the Shanghai cohort signal kicks in at ~75 g/day at the high end. Preparation matters more than is widely appreciated: plant myrosinase is heat-labile, so vigorously boiled broccoli yields little sulforaphane. Steaming for 3–4 minutes preserves both myrosinase and the active conversion. Chopping broccoli and letting it sit ~40 minutes before cooking lets the enzyme work first, then mild heat afterward doesn't undo it Dosz and Jeffery 2013. Adding a small amount of raw cruciferous (mustard, radish, or arugula) as a topping to cooked cruciferous restores myrosinase via a simple kitchen workaround. Raw, fermented (sauerkraut, kimchi), and lightly steamed all clear the bar; long-boiled and overcooked do not Linus Pauling Institute 2017.

Contraindications and failure modes

The clinically real interaction is with warfarin. Cruciferous vegetables — kale in particular — are high in vitamin K1, which warfarin works against. The clinically managed solution is consistent intake, not avoidance: warfarin patients can eat them, but day-to-day variability swings INR, so the practical rule is "the same amount most days" Linus Pauling Institute 2017. Direct oral anticoagulants (apixaban, rivaroxaban, dabigatran) don't have this interaction.

The two failure modes for the substance itself: (1) cooked into oblivion, the glucosinolate-to-sulforaphane conversion collapses, and the microbiome rescue is partial and individual-variable; (2) the gas-and-bloat from raffinose fermentation can be limiting for IBS or visceral-hypersensitivity readers, where cooking longer (which trades sulforaphane for digestibility), smaller portions, or kimchi/sauerkraut forms can solve the discomfort but at some bioactive cost.

Credibility range

Optimist case

The mechanism is unusually well-characterised: a known molecule (sulforaphane) hits a known pathway (Keap1/Nrf2) and upregulates a known set of effector enzymes (the phase II detox cluster) that are central to how the body neutralises and excretes both ingested and inhaled carcinogens. The biomarker chain is closed in humans — broccoli sprout intake demonstrably raises urinary excretion of benzene and acrolein conjugates in a controlled trial (Egner 2014). The epidemiology is consistent and dose-responsive across cancer sites and across geographically distinct cohorts (Shanghai, EPIC, Japanese, US Nurses', Australian). The cardiovascular signal is independent of the cancer signal and is mediated through both vascular-calcification (vitamin K) and blood-pressure (Connolly 2024) routes. Modest anti-inflammatory effects appear in RCTs at sub-pharmacological doses. The package is the cleanest "real food does this" story in nutrition outside the fish-and-omega-3 literature.

Skeptic case

Almost the entire cancer and mortality signal is observational. People who eat cruciferous vegetables exercise more, smoke less, weigh less, and eat more of every healthy food group; residual confounding can drive hazard-ratio reductions of 0.8–0.9 in any cohort. The RCT evidence for hard endpoints does not exist — every trial is biomarker, surrogate, or short-term. The Connolly BP trial is n=18, two weeks, against a deliberately mediocre comparator (potatoes and squash). The phase II induction in human nasal cells is real but the leap from "GST mRNA goes up" to "you get less cancer over decades" is large. Sulforaphane bioavailability from cooked vegetables — what most people actually eat — is single-digit-percent, and individual microbiome variability swings that further. The thyroid risk is dismissed too easily: the Felker analysis assumes iodine sufficiency, which is not universal even in developed populations. And the entire literature has industry-adjacent ties — supplement manufacturers fund a chunk of the sulforaphane work.

The author's call

Mid-strength. The mechanism is real and the molecular biology is unambiguous; the cancer epidemiology is the most consistent diet-and-cancer signal there is, and the cardiovascular cohort + atherosclerosis-biomarker + small BP trial triangulate cleanly. But the effect size for any individual eater is modest — single-digit-percent absolute risk reductions on a multi-decade horizon, not a transformative single-intervention story. The honest framing is "real food, real mechanism, real population-scale effect, modest individual-scale effect, near-zero downside in normal preparations". Score the evidence high (4), the longevity meaningful (3), the inflammation/short-term health modest (2), and the cardiovascular-mediated cumulative aesthetic effect modest (2). Don't oversell as a magic bullet; don't underplay as just-another-vegetable.

Stakeholder and incentive map

Sulforaphane supplement industry — broccoli-sprout extracts, glucoraphanin pills, and DIM capsules form a real-money supplement category. Manufacturers fund some clinical trials and amplify the Nrf2 story. The honest read on supplement evidence: glucoraphanin-only capsules without co-administered myrosinase (which most commercial pills lack) have markedly worse bioavailability than fresh sprouts.
Public-health and dietary-guidelines bodies — WCRF/AICR, AHA, NICE all recommend cruciferous vegetables as part of overall vegetable intake, generally without singling them out. The institutional caution is appropriate to the observational evidence.
Functional-medicine and integrative-health practitioners — strong adoption of the "detox" framing, sometimes overstating the case (e.g., implied liver-cleanse claims that exceed what the phase II literature supports).
Carnivore / paleo communities — counter-camp, occasionally amplifying the goitrogen story to argue plants are harmful. The clinical evidence in humans does not support this framing at culinary doses.
Thyroid-condition online communities — long-running anxiety about cruciferous and thyroid that the human data does not actually support; needs gentle correction in the article.

Population variability

GSTM1 / GSTT1 genotype — the null deletions affect roughly half the population for GSTM1; carriers may get more cancer-protective benefit per gram of cruciferous vegetable, consistent with the Nrf2 / phase II mechanism doing more work where alternative detoxification is impaired.
Gut microbiome composition — bacterial myrosinase activity varies ~tenfold between people; matters most for those who eat cruciferous mostly well-cooked.
Iodine status — only relevant to thyroid risk in actually iodine-deficient populations.
Existing IBS / visceral hypersensitivity — gas, bloating, distention is rate-limiting at higher intakes; cooking, smaller portions, and fermented forms are workarounds.
Anticoagulation status — the consistency-of-intake rule is specific to vitamin-K-antagonist anticoagulation (warfarin); DOACs are unaffected.
Sex — observational signal on cardiovascular mortality is somewhat stronger in women in several cohorts (Mori 2019; Blekkenhorst's atherosclerosis work is women-only). Likely sample-size and follow-up-length artefacts rather than biological dimorphism, but worth noting.

Knowledge gaps

No RCT exists with hard cancer-incidence or cardiovascular-event endpoints. The cancer-prevention case is built entirely on cohort epidemiology plus biomarker trials.
The dose-response curve above ~75 g/day is poorly characterised — most cohorts cap their top quintile there, and whether more is better, plateau, or hormetic past that point is unknown.
Microbiome-mediated conversion is the dominant pathway for cooked cruciferous but the determinants of who's a high vs low converter are not well mapped; no validated way to predict an individual's bioavailability.
The relative contribution of sulforaphane vs indole-3-carbinol/DIM vs vitamin K vs fibre vs nitrate to the observed cardiovascular and mortality signals isn't decomposed — the cohorts measure "cruciferous intake", not the specific bioactive pathway.
Long-term safety of concentrated sulforaphane supplementation (multi-year, supraphysiological doses) is not established; the literature is short and trial-grade.

Scope. Brief named broccoli, cabbage, kale, Brussels sprouts, cauliflower, plus the six named consequences: phase II detox, cancer, cardiovascular, thyroid, gut, inflammation. All six are covered — phase II in mechanism and evidence, cancer + cardiovascular as the spine of evidence / stakes / payoff, thyroid in contraindications + misconceptions, gut in mechanism (microbiome-mediated conversion) and failure-modes (FODMAP gas), inflammation in evidence. Nothing from the brief was silently dropped.

The thyroid call. Tempting to add thyroid-condition to contraindications because of the loud lay-discourse anxiety. Left it off — the Felker 2016 quantitative analysis and the LPI summary are clear that culinary intake does not produce clinical hypothyroidism in iodine-replete adults, and listing it would mislead by formalising a non-risk. Instead handled in misconceptions with the actual evidence framing and the iodine-deficiency narrow exception flagged.

Longevity score. Considered 4 (large-effect). Settled on 3 (meaningful) because the cohort signal is real but cannot be cleanly isolated from healthy-diet confounding; WCRF rates the cancer association at probable, not convincing, and that floor caps the honest longevity call here. Anyone making the case for 4 would point at the breadth (cancer + CVD signals in the same cohorts) — not unreasonable, but 3 is the conservative honest call.

Dream narrative. Score landed at 38 (below the 40 obligatory threshold). Wrote one anyway — the substance has a genuinely aspirational vascular-age and cancer-that-didn't-happen story that wouldn't be served by a flat dek. The dek and opening carry it lightly; tagline is straight per ./headline.md §2 at this tier.

Sulforaphane supplements vs whole food. Deliberately did not break sulforaphane / broccoli-sprout extracts out as a separate entry candidate — the supplement case differs enough (myrosinase-status, bioavailability) to warrant its own entry eventually. Flagged in the in-text protocol section briefly; backlog item rather than full coverage.

Future-link candidates: vitamin K1 and vascular calcification, broccoli sprouts as a concentrated form, dietary nitrate, H. pylori, fibre and gut microbiome (mentioned in out-of-scope).
Separate-entry candidates: sulforaphane / broccoli-sprout extract supplementation as a distinct supplement entry; I3C/DIM as a hormone-modulation supplement.