Beta-Carotene: The ATBC and CARET Lung-Cancer Trials, Explained

By Erin Rose · Updated July 19, 2026 · Methodology

Educational overview — not medical advice. This page describes two of the most consequential negative-result RCTs in supplement history. This statement has not been evaluated by the FDA. This product is not intended to diagnose, treat, cure, or prevent any disease.

Why anyone expected beta-carotene to prevent cancer in the first place

By the late 1980s, a substantial body of observational (cohort and case-control) research had linked diets higher in fruits and vegetables — and higher in beta-carotene specifically — to lower rates of lung and other cancers. Beta-carotene is also an antioxidant in laboratory settings, capable of neutralizing reactive oxygen species that can damage DNA, which gave the hypothesis a plausible biological mechanism on top of the dietary correlation. The logic seemed straightforward: if diets rich in beta-carotene associate with less cancer, and beta-carotene is an antioxidant in vitro, then a concentrated beta-carotene supplement should protect high-risk populations even more directly. Two large, well-funded, methodologically rigorous randomized controlled trials were built specifically to test that hypothesis in the people who stood to benefit most: smokers.

ATBC (1994): the first shock result

The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study (ATBC Study Group, 1994, New England Journal of Medicine) randomized 29,133 Finnish male smokers aged 50-69 into four groups: alpha-tocopherol (vitamin E, 50mg/day) alone, beta-carotene (20mg/day) alone, both together, or placebo, followed for 5-8 years. The primary endpoint was lung cancer incidence. The beta-carotene groups had an 18% higher lung cancer incidence and 8% higher total mortality than the groups not receiving beta-carotene, with the excess mortality driven by lung cancer and ischemic heart disease. Alpha-tocopherol showed no effect on lung cancer either way. This was not a small or marginal result in a small trial — it was a large, well-powered RCT finding real harm from a supplement everyone expected to help.

CARET (1996): stopped 21 months early

The Beta Carotene and Retinol Efficacy Trial (CARET; Omenn, Goodman, Thornquist, et al., 1996, New England Journal of Medicine) enrolled 18,314 people at even higher lung-cancer risk: current and former heavy smokers plus workers with occupational asbestos exposure. Participants received 30mg beta-carotene plus 25,000 IU retinyl palmitate (preformed vitamin A) daily, or placebo. CARET's independent data and safety monitoring board halted the trial 21 months ahead of its planned end date when an interim analysis showed the treatment arm had lung cancer incidence up 28% (relative risk 1.28), lung cancer mortality up 46%, total mortality up 17%, and cardiovascular mortality up 26% compared to placebo. Continuing to expose participants to a treatment associated with that much excess death was no longer considered ethical. The trial authors' own conclusion was blunt: the combination "had no benefit and may have had an adverse effect."

Why it backfired: the leading mechanistic explanations

Neither trial answers the "why" question definitively, but two mechanistic hypotheses have the most support. First, high-dose isolated beta-carotene may act as a pro-oxidant rather than an antioxidant in the specific environment of a smoker's lung — an oxygen-rich tissue already under chronic oxidative and carcinogen damage from tobacco smoke, where a molecule that neutralizes free radicals in a test tube may instead generate harmful oxidation byproducts at high concentrations in that particular tissue context. Second, beta-carotene's oxidative breakdown products may interfere with retinoic acid signaling, a cellular pathway that regulates cell growth and differentiation and is already implicated in cancer biology. Both explanations point toward the same broader lesson: a compound behaving one way in an epidemiological association across a whole dietary pattern does not guarantee the same compound, isolated and concentrated into a pill, will behave identically inside a specific, already-damaged tissue. Mechanism from mouse or cell studies should never be read as proof of a human outcome on its own — here, the human RCT data is what actually established the harm; the mechanism is the leading explanation for it, not independent proof.

Confirmed twice more: Bjelakovic and AREDS2

ATBC and CARET could, in principle, have been two unlucky trials. They weren't. A 2012 Cochrane meta-analysis of 78 randomized controlled trials pooling 296,707 participants (Bjelakovic et al.) examined antioxidant supplements — beta-carotene, vitamin A, vitamin C, vitamin E, and selenium — against mortality, and found beta-carotene supplementation (alone or combined) was associated with significantly increased all-cause mortality across the pooled trials. Separately, in 2013 the AREDS2 eye-health trial — testing lutein/zeaxanthin as a substitute for beta-carotene in the original AREDS formula — found more lung cancers in its own beta-carotene arm than its no-beta-carotene arm (23 vs 11 cases), concentrated in former smokers. That's a third, independent RCT signal, from a trial not even designed to study lung cancer as its primary question, pointing the exact same direction as ATBC and CARET. See the full AREDS2 breakdown.

Frequently asked questions

Related

• Beta-carotene: what the lung-cancer trials showed
• Dosage guide — how close the retail dose sits to the harm-trial doses
• Beta-carotene & eye health — the AREDS2 postscript, in full
• Best beta-carotene — the honest buying framework, if you need one at all

Sources