General information, not medical advice. Do not start or stop any supplement on the basis of an article. If you take supplements for a diagnosed condition, that is a conversation for your doctor.
The antioxidant story was one of the most attractive hypotheses in modern nutrition. Free radicals damage cells. Antioxidants neutralise free radicals. People who eat more antioxidant-rich food get less cancer and less heart disease — a finding that replicated across dozens of observational studies.
The inference seemed safe enough to test at scale. Three very large randomised trials were built on it. All three came back badly.
ATBC: Finland, 29,133 smokers
The Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study enrolled 29,133 Finnish male smokers and randomised them, in a two-by-two design, to 50 mg of alpha-tocopherol, 20 mg of beta-carotene, both or neither, for a median of just over six years. Results appeared in the New England Journal of Medicine in 1994.
Beta-carotene increased lung cancer incidence by 18% (95% CI 3% to 36%) and total mortality by 8% (95% CI 1% to 16%), the excess deaths driven by lung cancer and ischaemic heart disease. Vitamin E showed no effect on lung cancer but more deaths from haemorrhagic stroke.
The trial ran to its planned conclusion. Its authors wrote that the results raise the possibility that these supplements may have harmful as well as beneficial effects — the understated phrasing of people who had expected the opposite.
CARET: stopped 21 months early
The Beta-Carotene and Retinol Efficacy Trial enrolled 18,314 people at high risk — 14,254 heavy smokers and 4,060 with asbestos exposure — on 30 mg of beta-carotene plus 25,000 IU of retinyl palmitate daily.
On 11 January 1996, after the second interim analysis, it was halted 21 months ahead of schedule. Lung cancer incidence in the supplement arm carried a relative risk of 1.28 (95% CI 1.04–1.57). All-cause mortality was 1.17 (1.03–1.33). The excess appeared after roughly four years of supplementation.
The published conclusion: the combination had no benefit and may have had an adverse effect on the incidence of lung cancer.
SELECT: 35,533 healthy men
This one matters because the participants were not high-risk. The Selenium and Vitamin E Cancer Prevention Trial randomised 35,533 healthy men to 400 IU of vitamin E, 200 µg of selenium, both or placebo.
Blinded supplementation was stopped in October 2008 for futility. But the 2011 follow-up in JAMA, with data through July 2011, found something worse than nothing: vitamin E alone carried a hazard ratio for prostate cancer of 1.17 (99% CI 1.004–1.36). In absolute terms, 10.9 cases per 1,000 person-years against 9.3 on placebo.
Worth noting that the lower confidence bound is 1.004 — significant by a hair, not by a mile. And oddly, the arm taking vitamin E and selenium showed no significant harm, which nobody has satisfactorily explained.
What the pooled evidence says
Goran Bjelakovic and colleagues have assembled this literature twice. Their JAMA analysis in 2007 covered 68 trials and 232,606 participants; the Cochrane version covers 78 trials and 296,707.
Here the honest reading requires care. The overall mortality estimate is 1.02, with a confidence interval of 0.98 to 1.05 — not significant. Coverage that stops there concludes the supplements are harmless.
Restricted to trials at low risk of bias, the figure becomes 1.04 (1.01–1.07) — significant, and in the wrong direction. By nutrient, beta-carotene came in at 1.05 (1.01–1.09) and vitamin E at 1.03 (1.00–1.05). In the 2007 analysis, vitamin A was worst at 1.16 (1.10–1.24).
Whether to restrict to low-bias trials is a defensible analytic choice, not an obvious one, and the harm conclusion depends on it. Vitamin E’s interval touches 1.00. This is a real signal, but it is a small one, and it should be described that way.
Why the mechanism was wrong
The most interesting explanation is that reactive oxygen species are not purely damage. They are signals.
Michael Ristow’s group gave 39 healthy young men four weeks of supervised training, half of them also taking 1,000 mg of vitamin C and 400 IU of vitamin E daily. Without the supplements, insulin sensitivity improved significantly in both trained and untrained participants. With the supplements, it did not change significantly.
Markers of oxidative stress in muscle more than doubled after exercise in the unsupplemented group and did not rise meaningfully in the supplemented one — and the induction of the body’s own antioxidant defences, along with PGC1α, was blunted alongside it.
The exercise-induced oxidative stress appears to be the signal that triggers the adaptation. Neutralise it and you neutralise the benefit. This is one study of 39 young men on one regimen, and the wider literature on antioxidants and training adaptation is genuinely mixed — but it is a coherent account of why the hypothesis failed.
Where supplements do work
Three settings, all specific.
Documented deficiency. Correcting a measured deficiency is not the same intervention as supplementing a replete person.
Folic acid in pregnancy. The US Preventive Services Task Force gives this a Grade A recommendation: 400 to 800 µg daily for anyone planning or able to become pregnant.
AREDS2 for specific eye disease. The formulation reduces progression from intermediate to advanced age-related macular degeneration by about 25%. It showed no benefit in early or no AMD — and beta-carotene was removed from the formula after former smokers taking it showed higher lung cancer incidence.
For everyone else, the Task Force’s 2022 statement is unusually direct. It recommends against beta-carotene or vitamin E supplements for preventing cardiovascular disease or cancer — a Grade D, meaning there is moderate or high certainty the harms equal or exceed the benefits.
What went wrong, in one sentence
A correlation between eating carotenoid-rich food and getting less lung cancer was converted into a pharmacological dose of one isolated molecule, and the sign of the effect flipped.
That is the general shape of the lesson, and it is not confined to antioxidants. The dose was higher than any diet delivers, the molecule arrived without the thousands of compounds that accompany it in food, and the people most likely to buy it were the people the trials showed were most at risk.
Sources
- Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. The Effect of Vitamin E and Beta Carotene on the Incidence of Lung Cancer and Other Cancers in Male Smokers. New England Journal of Medicine 1994;330(15):1029-1035.
- Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. New England Journal of Medicine 1996;334(18):1150-1155.
- Klein EA, Thompson IM, Tangen CM, et al. Vitamin E and the Risk of Prostate Cancer: the SELECT trial. JAMA 2011;306(14):1549-1556.
- Bjelakovic G, Nikolova D, Gluud LL, Simonetti RG, Gluud C. Antioxidant supplements for prevention of mortality in healthy participants and patients with various diseases. Cochrane Database of Systematic Reviews; and JAMA 2007;297(8):842-857.
- Ristow M, Zarse K, Oberbach A, et al. Antioxidants prevent health-promoting effects of physical exercise in humans. PNAS 2009;106(21):8665-8670.
- US Preventive Services Task Force. Vitamin, Mineral, and Multivitamin Supplementation to Prevent Cardiovascular Disease and Cancer. JAMA 2022;327(23):2326-2333; and Folic Acid Supplementation, 2023.
- National Eye Institute. About AREDS and AREDS2.

















