“…sulforaphane is the most potent naturally occurring dietary activator of the genetic pathway Nrf2, which regulates over 200 different genes.” FoundMyFitness, 2016 [1]
“…resulted in the slowing of the doubling rate of a cancer biomarker known as prostate specific antigen or PSA by 86% compared to placebo… That’s pretty gigantic.” FoundMyFitness, 2016 [1]
“…there’s just so much data on animals and lots of mechanisms at play. But ultimately, I think the bottom line here is that we should be eating either our cruciferous vegetables or supplementing with sulforaphane or eating the broccoli sprouts.” FoundMyFitness, 2025 [2]
That’s Dr Rhonda Patrick, across nearly ten years of talks. She argues that a molecule from broccoli sprouts, sulforaphane, flips a switch inside your cells called NRF2, and your body turns up its own defences and detox enzymes. Astaxanthin, the red pigment in salmon, is also marketed as an NRF2 switch. And at the end of 2023, it made male mice live longer [29].
So should we supplement with these molecules?
Table of Contents
- Rats and a Switch
- Getting It Into People
- Switching NRF2 On Hard
- Broccoli Sprout Drinks
- The Case at Its Peak
- A Blood-Sugar Treatment on the Way
- Testing the Switch
- The Story Changes
- Astaxanthin and Skin
- Pooling the Skin Trials
- Astaxanthin in Mice
- The Bottom Line
- Other Studies
- References
Rats and a Switch
Excitement grew in 1994 when researchers at Johns Hopkins gave sulforaphane to rats around the time they were exposed to a cancer-causing chemical. The rats grew fewer breast tumours, and smaller ones [3].
In 2002, the team tied it to a switch. Sulforaphane protected mice from a chemical that causes stomach tumours. But in mice bred without the NRF2 gene, the protection vanished [4].
Would any of it work in people?
Getting It Into People
The first problem was getting it in. Sulforaphane is unstable, so broccoli doesn’t store it. The plant stores a stable raw material called glucoraphanin, and young broccoli sprouts are especially rich in it [5]. In the plant, sulforaphane forms when that raw material meets an enzyme called myrosinase. That happens when you crush the plant. Or chew it [6][7].
Heat destroys myrosinase, so cooking works the other way. When men ate crushed broccoli raw, about 37% of the sulforaphane reached their blood and urine. Cooked, it was 3.4% [8].
But if it does get in, does it help?
Switching NRF2 On Hard
One way to test the idea is with a drug. In 2013, a drug designed to switch NRF2 on strongly, bardoxolone methyl, was tested in 2,185 people with type 2 diabetes and advanced kidney disease. But it didn’t protect their kidneys, and the trial stopped early after more heart failure on the drug, much of it linked to fluid build-up in people already at risk [9][10]. It’s a result you rarely hear about from people promoting NRF2 supplements.
Broccoli Sprout Drinks
But maybe there was something wrong with that NRF2 drug? What if we look specifically at broccoli sprout drinks?
The most convincing human result I’ve found is from 2014.
In Qidong, China, a region with heavy air pollution, 291 adults drank a broccoli sprout drink or a placebo every day for 12 weeks. The sprout group passed 61% more of a benzene breakdown product in their urine. Benzene is a toxic chemical in polluted air. The detox chemistry that NRF2 controls was at work in people [11]. So is Dr Rhonda Patrick right? Let’s keep going to find out.
That second quote is about a 2015 trial, in men whose PSA, a blood marker for prostate cancer, had started rising again after prostate surgery. They took stabilised sulforaphane tablets or a placebo.
The 86% comes from the trial’s main measure, and it fell short of statistical significance. Several secondary and re-calculated PSA analyses did reach it. It was a six-month proof-of-concept trial in 78 men, funded by the company that developed the tablets. But a trial that short can’t tell us whether anyone lives longer [12].
The Case at Its Peak
Doesn’t look convincing so far. But the case for sulforaphane didn’t stop there. It also leaned on studies of people who eat a lot of vegetables. Her 2016 video opens with a study of about 135,000 adults in Shanghai, where the fifth who ate the most cabbage-family vegetables had a 22% lower death rate than the fifth who ate the least [15]. Then she connects that to sulforaphane: “These are some pretty powerful associations. But before we can begin to establish causality, we must establish a plausible mechanism. Our most likely candidate? Isothiocyanates.” [14] That’s sulforaphane’s chemical family. In a dish, sulforaphane really is a potent switch [13]. That paper’s first author runs a company that makes a broccoli sprout ingredient.
But people who eat more broccoli differ in lots of other ways, so this can’t tell us sulforaphane was the reason.
A Blood-Sugar Treatment on the Way
So in 2017, a team at Lund University in Sweden took it further. They screened the gene fingerprints of 3,800 drugs by computer, looking for one that might reverse the liver’s gene pattern in type 2 diabetes. Sulforaphane came out on top and worked about as well as metformin in diabetic rodents. Rather than another NRF2 drug like the one that failed in 2013, this Swedish team tested sulforaphane itself, in a 12-week trial in people with type 2 diabetes. The paper’s summary reported lower fasting blood sugar and HbA1c, a longer-term blood-sugar test, in obese patients whose diabetes was poorly controlled [16].
The university announced that a food product was planned within two years. Things are sounding good for Dr Patrick’s theory [17].
Testing the Switch
The human trials kept coming. By 2019, the field counted more than 50 [6].
But unfortunately there were problems with many of these studies. As I mentioned earlier, sulforaphane is very unstable, and many of the studies didn’t deliver sulforaphane to their participants in a way that got around that instability [6].
And even when it did reach the blood, the NRF2 changes were small or absent. As such, a review of the human studies in 2021 concluded there was “little evidence that sulforaphane activates Nrf2 in humans.” The Qidong pollutant result was real, but direct measures of NRF2 in people mostly came up empty [18][19][20].
There were also small trials in autism, depression and schizophrenia, with mixed results. They’re covered in the Other Studies section below.
The Story Changes
So did that Swedish blood-sugar result hold up? In February 2025, the same Lund team tested its broccoli sprout extract for 12 weeks in adults whose fasting blood sugar was mildly raised, 6.1 to 6.9 millimoles per litre (about 110 to 124 mg/dL). Before starting, they set a target: it should fall 0.3 more than on placebo, (about 5.4 mg/dL) based on what they’d seen in diabetes. It fell by 0.2 (about 3.6 mg/dL). That was statistically significant, but short of their own target, and HbA1c, the longer-term blood-sugar test, didn’t change [21].
The team’s own explanation involves gut bacteria, which do some of the converting when the plant’s enzyme isn’t there. They think it works in some people and not others. People with a particular profile, mildly obese but with low insulin resistance, saw a bigger drop, and they carried more of a gut bacterial gene needed to turn the raw material into sulforaphane [21]. That may turn out to be true. But those responders were picked out after the results were in, the 2017 group was defined differently, and as far as I can find, nobody has yet run a trial in those people from the start.
In August 2026, the newest trial reported on 99 adults, average age 74, with type 2 diabetes and advanced kidney disease, much like the patients in the bardoxolone trial. For 12 weeks they took broccoli sprout extract, rising from one sachet a day to three, or a placebo. The sulforaphane got into their blood, but there was no difference in fasting blood sugar, insulin or HbA1c [20].
So the blood-sugar promise from 2017 has shrunk to something small or nothing. So what about a more stable molecule?
Astaxanthin and Skin
Which brings us to astaxanthin. Astaxanthin is much more stable as a supplement than sulforaphane, and it can actually reach people intact: it’s sold as an oil in softgels, and studies have measured it in people’s blood, especially when it’s taken with some fat [22]. Its NRF2 evidence in people is thinner [23]. And a lot of people now take it for their skin. So what does the evidence show?
In 2017, a trial in Japan gave 65 women 6 milligrams of astaxanthin, 12 milligrams, or a placebo for 16 weeks. Fuji Chemical paid for it, and every author worked for AstaReal, whose astaxanthin they tested. The summary says wrinkles and moisture got worse in the placebo group but not on astaxanthin. That compares each group with its own starting point. When the groups were compared with each other, none of the measurements differed. So not exactly a glowing endorsement [24].
On sunburn, the claim I see most in my comments, there’s one small trial from 2018, run by employees of the company that funded it, with 22 people analysed. By the end, both groups needed about the same dose of UV light before their skin reddened. But compared with placebo, the astaxanthin group’s threshold rose significantly more from where it started. So on that measure, astaxanthin beat placebo [25].
Maybe you’ve taken astaxanthin and your skin seemed better. I don’t think you imagined it. But skin changes with the seasons on its own. In that 2017 trial, the placebo group’s skin got worse from August to December, just from the weather. And people often start a supplement when their skin is at its worst, so it tends to improve anyway. That’s why the fair test is against placebo, and against placebo the differences were small [24].
Pooling the Skin Trials
When all the small skin trials were pooled in 2021, there was a signal: about 300 people, with moisture and elasticity improving when measured with skin instruments. Wrinkles didn’t change significantly. The authors called the findings preliminary, and noted that several trials were run or funded by companies [26].
And a 2025 review of supplements for sun-damaged skin found too little evidence to recommend astaxanthin [27].
For skin, the evidence is a small change in moisture and one small sunburn result, from short trials mostly paid for by the people selling it. I don’t think that’s worth paying for every month.
Astaxanthin in Mice
But then astaxanthin got a big result in mice. The Interventions Testing Program, run by the US National Institute on Aging, tests compounds for lifespan in parallel at three separate labs, so researchers can check whether a result holds up at each one. It publishes everything [28].
At the end of 2023, the program reported that astaxanthin, added to the diet from middle age, made male mice live 12% longer at the median, and the gain was significant at all three labs. Females didn’t benefit. The astaxanthin came from Cardax, a pharmaceutical company that supplies it in a special synthetic form [29].
Then, in March 2026, the program reported a second astaxanthin test. This was not an exact repeat. The food held about half as much astaxanthin, and the main group started at 11 months instead of 12, with a second group of males starting at 16 months. Neither group of males lived meaningfully longer [30].
The authors give two explanations. Maybe the benefit needs a particular dose and starting age. Or maybe the original 12% was chance, because it sat near the smallest effect the study could reliably detect [30].
The Bottom Line
There might be something to the NRF2 literature, but to me that’s more reasons to eat vegetables and food sources of astaxanthin, rather than a reason to buy these supplements. I don’t take either, after a careful review of the literature.
On vegetables, Dr Patrick and I agree: eat plenty, and a wide variety [31][32]. Eat sprouts raw and chew them well, and on cooked broccoli, ground mustard seed may help some conversion [33]. If you’re pregnant, older or have a weakened immune system, cooked broccoli with ground mustard seed is the safer route [36]. Astaxanthin is the pigment in salmon, trout, shrimp and crab [22].
We also want to be careful about taking too many antioxidants, because they can interfere with the response to exercise. Equally, for a select group, antioxidants used correctly may actually improve exercise performance [34][35].
Other Studies
These studies didn’t make it into the video, but they belong to the same story.
In 2012, a team in Tehran gave 81 people with type 2 diabetes broccoli sprout powder or a placebo for four weeks. At the higher dose, insulin and a measure of insulin resistance fell, just reaching statistical significance [37].
Autism and mental health followed a similar path. A 2014 trial in 44 young men with autism reported big improvements in behaviour scores [38], but most later trials, including two larger ones, missed their main goals [39][40][41], and a 2025 pooled analysis found only a small improvement [42]. In a 2021 trial in depression after heart procedures, 30% responded on sulforaphane against under 7% on placebo [43]. But a 172-person trial in early schizophrenia found no benefit on its main thinking-skills score, though a couple of sub-scores improved [44].
Two trials of a drug-grade, stabilised sulforaphane measured whether people got better. In Dundee, during COVID, 133 people in hospital with pneumonia got the drug or a placebo. The 2024 report: stopped early for lack of benefit, with no difference in recovery, time in hospital or deaths at 28 days. A third of the people on sulforaphane stopped taking it, mostly because of gut side effects, against fewer than one in ten on placebo. LifeArc funded it, and the drug’s maker supplied the drug, the placebo and staff support. One author advises that company, which the paper declares as outside this work [45]. A second, smaller trial of the same drug, in 105 people after bleeding around the brain, didn’t reduce spasm in the large arteries or improve recovery [46].
For astaxanthin, the one randomised trial I could find that measured NRF2 in people was in 49 women with polycystic ovary syndrome going through IVF (in vitro fertilisation). NRF2-related genes rose in the cells around their eggs, and egg and embryo quality scores improved. Pregnancy rates didn’t change [23].
The same mouse-lifespan program had earlier tested Protandim, a plant mix sold to switch NRF2 on. Male mice lived 7% longer at the median, but that was significant at only one of the three labs, and females got nothing [28].
References
1. https://www.youtube.com/watch?v=zz4YVJ4aRfg
2. https://www.youtube.com/watch?v=NoL7kQtmrOw
3. https://www.pnas.org/doi/10.1073/pnas.91.8.3147
4. https://www.pnas.org/doi/10.1073/pnas.112203099
5. https://www.pnas.org/doi/10.1073/pnas.94.19.10367
6. https://pmc.ncbi.nlm.nih.gov/articles/PMC6804255/
7. https://pubmed.ncbi.nlm.nih.gov/11352861/
8. https://pubmed.ncbi.nlm.nih.gov/18950181/
9. https://www.nejm.org/doi/full/10.1056/NEJMoa1306033
10. https://pubmed.ncbi.nlm.nih.gov/25307295/
11. https://pmc.ncbi.nlm.nih.gov/articles/PMC4125483/
12. https://aacrjournals.org/cancerpreventionresearch/article/8/8/712/
13. https://pmc.ncbi.nlm.nih.gov/articles/PMC4736808/
14. https://www.foundmyfitness.com/episodes/sulforaphane
15. https://pubmed.ncbi.nlm.nih.gov/21593509/
16. https://www.science.org/doi/10.1126/scitranslmed.aah4477
18. https://pmc.ncbi.nlm.nih.gov/articles/PMC7925463/
19. https://doi.org/10.1371/journal.pone.0163716
20. https://pubmed.ncbi.nlm.nih.gov/42641715/
21. https://pmc.ncbi.nlm.nih.gov/articles/PMC11879859/
22. https://pmc.ncbi.nlm.nih.gov/articles/PMC3917265/
23. https://pmc.ncbi.nlm.nih.gov/articles/PMC9050983/
24. https://pmc.ncbi.nlm.nih.gov/articles/PMC5525019/
25. https://pmc.ncbi.nlm.nih.gov/articles/PMC6073124/
26. https://pmc.ncbi.nlm.nih.gov/articles/PMC8472736/
27. https://pmc.ncbi.nlm.nih.gov/articles/PMC12318760/
28. https://pmc.ncbi.nlm.nih.gov/articles/PMC5013015/
29. https://pmc.ncbi.nlm.nih.gov/articles/PMC10828146/
30. https://pmc.ncbi.nlm.nih.gov/articles/PMC13356140/
31. https://www.who.int/news-room/fact-sheets/detail/healthy-diet
32. https://cdn.realfood.gov/DGA.pdf
33. https://pubmed.ncbi.nlm.nih.gov/29806738/
34. https://pmc.ncbi.nlm.nih.gov/articles/PMC4001759/
35. https://pmc.ncbi.nlm.nih.gov/articles/PMC12491480/
36. https://www.cdc.gov/food-safety/foods/safer-food-choices.html
37. https://pubmed.ncbi.nlm.nih.gov/22537070/
38. https://www.pnas.org/doi/10.1073/pnas.1416940111
39. https://pubmed.ncbi.nlm.nih.gov/34034808/
40. https://pubmed.ncbi.nlm.nih.gov/36427174/
41. https://pubmed.ncbi.nlm.nih.gov/36771424/
42. https://pubmed.ncbi.nlm.nih.gov/40458076/
43. https://pubmed.ncbi.nlm.nih.gov/34033171/
44. https://pubmed.ncbi.nlm.nih.gov/39144775/
