The Dark Side of Antioxidants

The story of the dark side of antioxidant research isn’t well known outside of medical circles. It’s an unseemly story, profoundly unsettling; a story that refuses to be made pretty or happy or uplifting no matter how hard you try to duct-tape a silver lining around it. It doesn’t fit the “antioxidants are good for you” mantra that sells billions of dollars per year of blueberry- and pomegranate-fortified granola bars and tocopherol-enrichened cereals, acai-berry Jell-O mixes, juices and yogurts with added vitamins, organic baby foods, and so forth, not to mention the billions of dollars of nutritional supplements sold each year (to say nothing of the sub-industry of books and magazines devoted to nutrition).

Still, it’s a story that needs to be told. And some of us know where the bodies are buried.

For decades, mainstream medicine pooh-poohed the possibility that vitamins or supplements could “move the needle” on major diseases. Two-time Nobel laureate Linus Pauling was harshly criticized in the 1970s and 80s for suggesting a role for Vitamin C in prevention and treatment of cancer. Even so, laboratory workers had known for years that changes to diet could influence the rate of tumor appearance in lab animals. By the early 1980s, case-control studies and epidemiological evidence from a variety of sources had begun to accumulate, showing that persons who routinely ate large quantities of fresh fruits and vegetables consistently did better with regard to cardiovascular disease (and other diseases) than most people.

In 1981, Sir Richard Peto and colleagues published a paper in Nature that dared asked the simple question: “Can dietary beta-carotene materially reduce human cancer rates?” (Nature, 290:201-208) Shortly thereafter, the National Cancer Institute (whose Chemoprevention branch was headed by Dr. Michael B. Sporn, one of the coauthors of the Nature article) decided to green-light two large intervention-based studies of the cancer-preventing effects of nutritional supplements: a study in Finland involving beta-carotene and alpha-tocopherol (Vitamin E), and a U.S.-based study involving retinol (a form of Vitamin A) and beta-carotene.

The Finland study (conducted by Finland’s National Institute for Health and Welfare) was initially designed to encompass 18,000 male smokers between the ages of 50 and 69. Why just smokers? And why male, and 50+ years old? Lung cancer is ten times more likely to affect smokers; hence a cancer study limited to smokers would need only a tenth as many participants as a study involving the general population. Based on what was known about the age-specific rates of lung cancer among Finnish men, study designers calculated that the desired effect size (a hoped-for 25% decrease in cancer incidence over a period of 6 years) would be measurable with the required level of statistical relevance if 18,000 older male smokers made up the study group. As it turned out, the age distribution of actual volunteers didn’t match the demographics of the eligibility group (volunteers tended to be toward the young end of the eligibility range), and as a result the study’s enrollment target had to be reset to 27,000 in order to get good statistical relevance.

Full-scale recruitment of subjects into the ATBC (Alpha-Tocopherol Beta-Carotene) Lung Cancer Prevention Study began in April 1985 and continued until a final enrollment of 29,246 men occurred in June 1988. Enrollees were randomized into one of four equal-sized groups, receiving either 50 mg/day (about 6 times the RDA) of alpha-tocopherol, or 20 mg/day of beta-carotene (equivalent to around 3 times the RDA of Vitamin A), or AT and BC together, or placebo only.

At the same time, which is to say starting in 1985 (after some very small, very brief pilot studies to validate recruitment mechanics), the Carotene and Retinol Efficacy Trial (CARET) started enrolling volunteers in the U.S. Unlike Finland’s ATBC study, volunteers for CARET were both male and female and were heavy smokers or came from asbestos-exposed workplace environments. They ranged in age from 45 to 69 and were divided initially into four groups (30 mg/day beta carotene only, 25,000 IU retinol-only, carotene plus retinol, or placebo), but in 1988 the treatment groups were consolidated into one group taking both beta-carotene and retinol. The study design called for continuing the vitamin regimen through 1997, with reporting of results to occur in 1998.

Alas, things went horribly awry, and CARET never got that far.

When the Finns reported results from the ATBC study in April 1994, it sent shock waves through the medical world. Not only had alpha-tocopherol and beta-carotene not provided the expected protective effect against lung cancer; the supplement-treated groups actually experienced more cancer than the placebo group—18% more, in fact. 

This was an astonishing result, utterly bewildering, as it contradicted numerous prior animal studies that had shown Vitamin E and beta-carotene to be promising cancer preventatives. Surely an error had occurred. Something had to have gone wrong. One thing it couldn’t be was chance variation: with almost 30,000 participants (three quarters of them in treatment groups), this was not a small study. The results couldn’t be a statistical fluke.

As it turns out, the Finnish investigators had actually done a meticulous job from start to finish. In analyzing their data, they had looked for possible confounding factors. The only thing they found of interest was that heavy drinkers in the treatment group got cancer more often than light drinkers.

Two weeks before the Finnish study hit, the National Cancer Institute was awash in conference calls. Accounts vary as to who knew what, when, but CARET’s lead investigator, who had seen the Finnish group’s data prior to publication, knew that NCI now had a serious problem on its hands. CARET was doing essentially the same experiment the Finns had done, except it was giving even bigger doses of supplements to its U.S. participants, and the study was due to run for another three and a half years. What if CARET’s treatment group was also experiencing elevated cancer rates? Participants might be dying needlessly.

When statisticians presented interim results to CARET’s Safety Endpoint Monitoring Committee in August 1994, four months after the Finnish study appeared in print, it became clear that CARET participants were, if anything, faring worse than the patients in the ATBC study. Even so, the safety committee found itself deadlocked on whether to call a premature halt to CARET. The study’s formal stopping criteria (as given by something called the O’Brien–Fleming early-stopping boundary) had not been met. Ultimately a decision was made to continue to accumulate more data.

A second interim statistical analysis was presented to CARET’s safety committee in September 1995, one year after the first analysis. According to the committee:

At that time it was clear that the excess of lung cancer had continued to accumulate in the intervention regimen at about the same rate during the time since the first interim analysis. Further, the cardiovascular disease excess persisted. The conditional power calculations showed that it was extremely unlikely that the trial could show a beneficial effect of the intervention, even if the adverse effect ceased to occur and a delayed protective effect began to appear.Therefore the SEMC voted unanimously to recommend to NCI that the trial regimen should be stopped but the follow-up should continue.

The study was halted—but not until January 1996, nearly two years after final publication of the Finnish results. (Even then, CARET participants were contacted by snail mail to let them know of the study’s early termination and the reasons for it. See this writeup for details.)

CARET’s results were published in The New England Journal of Medicine in May 1996. Once again, shock waves reverberated throughout the medical world. Participants who took beta-carotene and Vitamin A supplements had shown a 28% higher rate of lung cancer. They also fared 26% worse for cardiovascular-related mortality, and 17% worse for all-cause mortality.

There was great reluctance in the medical community to believe the results. Perhaps the even-worse results of the CARET study (relative to the Finnish experiment) had to do with the decision to include 2,044 asbestos-exposed individuals in the treatment group of 9,241 persons? Not so, it turns out. Segment analysis of the asbestos group’s data relative to the heavy-smoker group showed that “There was no statistical evidence of heterogeneity of the relative risk among these subgroups.”

What the CARET study had, in fact, done was not just replicate the ATBC results but provide the beginnings of a dose-response curve. The Finns had used 20 mg/day of beta-carotene; CARET employed a 50% higher dose. The result had been 50% more cancer.

It was hard to understand the results of the ATBC and CARET studies in light of the fact that another large trial involving beta-carotene, the Physicians’ Health Study, had reported neither harm nor benefit from 50 mg of beta carotene taken every other day for 12 years. However, the Physicians’ Health Study population was younger and healthier than ATBC or CARET study groups and was predominantly (89%) made up of non-smokers. This turned out to be quite important. (Read on.)

It’s been almost 20 years since the ATBC and CARET results were reported. What have we learned in that time?

In 2007, Bjelakovic et al. undertook a systematic review of existing literature on antioxidant studies covering the time frame 1977 to 2006. The systematic review procedure was conducted using the well-regarded methodology of the Cochrane Collaboration, a group that specializes in (and is known for) high-quality meta-analyses. In analyzing the 47 most rigorously designed studies of supplement effectiveness, Bjelakovic et al. found that 15,366 study subjects (out of a total treatment population of 99,095 persons) died while taking antioxidants, whereas 9,131 placebo-takers, in control groups totalling 81,843 persons, died in those same studies. (This is not including ATBC or CARET results.) The studies in question used beta-carotene, Vitamin E, Vitamin A, Vitamin C, and/or selenium.

In a separate meta-analysis, Miller et al. found a dose-dependent relationship of Vitamin E with all-cause mortality for 135,967 participants in 19 clinical trials. At daily doses below about 150 International Units, Vitamin E appears to be helpful; above that, harmful. Miller et al. concluded:

In view of the increased mortality associated with high dosages of beta-carotene and now vitamin E, use of any high-dosage vitamin supplements should be discouraged until evidence of efficacy is documented from appropriately designed clinical trials.

How are we to make sense of these results? Why have so many studies shown a harmful effect for antioxidants when so many other studies (particularly those carried out in animals, but also those carried out in predominantly healthy human populations) have shown a clear benefit?

The answer may have to do with something called apoptosis, otherwise known as programmed cell death. The body has ways of determining when cells have become dysfunctional to the point of needing to be told to shut down. Most cancer therapies exert their effect by inducing apoptosis, and it’s fairly well accepted that in normal, healthy individuals, precancerous cells are constantly being formed, then destroyed through apoptosis. Antioxidants are known to interfere with apoptosis. In essence, they promote the survival of normal cells as well as cells that shouldn’t be allowed to live.

If you’re a young non-smoker in good health, the level of cell turnover (from apoptosis) in your body is nowhere near as high as the level of turnover in an older person, or someone at high risk of cancer. Therefore, antioxidants are apt to do more good than harm in a young, healthy person. But if your body is harboring cancer cells, you don’t want antioxidants to encourage their growth by interfering with their apoptosis. That’s the real lesson of antioxidant research.

The food industry and the people who make nutritional supplements have no interest in telling you any of the things you’ve read here. But now that you know the story of the dark side of antioxidants (a story made possible by thousands of ordinary people who died in the name of science), you owe it to yourself to take the story to heart. If you’re a smoker or at high risk for heart disease or cancer, consider scaling back your use of antioxidant supplements (Vitamins A and E in particular); it could save your life. And please, if you found any of this information helpful, share it with family, friends, Facebook and Twitter followers, and others. The story needs to get out.