Some combinations of supplements may slightly increase gastro-intestinal cancer risk, whereas selenium may be associated with a risk reduction.

The human diet is a complex mix of oxidants and antioxidants. Excess oxidants can cause cancer by inducing gene mutations. Goran Bjelakovic (University of Niss, Serbia and Montenegro) and colleagues from The Cochrane Hepato-Biliary Group reviewed all randomised trials comparing antioxidant supplements with placebo for prevention of gastrointestinal cancers.

The investigators identified 14 randomised trials (totalling over 170,000 participants). Overall, the results did not show any protective effect of supplementation with b-carotene, vitamins A, C, E, and selenium (alone or in combination) compared with placebo on oesophageal, gastric, colorectal, pancreatic, and liver cancer incidences. In half the trials (categorised as high quality), there was a small but statistically significant increase (6% relative risk value) in mortality among people taking antioxidants compared with placebo; the results also showed that two combinations of supplements were associated with increased mortality risk: b-carotene and vitamin A (30% increase in relative risk), and the combination of b-carotene and vitamin E (10% increase in relative risk). Four of the trials suggested that selenium was associated with a reduction in gastro-intestinal cancer risk, but this may be due to bias.

Dr Bjelakovic comments: We could not find evidence that antioxidant supplements can prevent gastrointestinal cancers; on the contrary, they seem to increase overall mortality. The potential preventive effect of selenium should be studied in adequate randomised trials.

In an accompanying commentary (p 1193), David Forman (University of Leeds, UK) and Douglas Altman (Cancer Research UK) comment: Somewhat chillingly, Bjelakovic and colleagues also estimate that, despite the small size of the relative risk, if their findings are correct, 9000 in every million users of such supplements will die prematurely as a result. The prospect that vitamin pills may not only do no good but also kill their consumers is a scary speculation given the vast quantities that are used in certain communities. However, they conclude that the findings of the study should only be viewed as preliminary: The mortality analysis in Bjelakovic and colleagues ™ review is work in progress, and does not offer convincing proof of hazard. In the event that a hazard is established from a complete review, these researchers will need to identify which specific interventions are associated with any risk. It is unlikely that all supplements will exert a similar effect and it will be vital to establish the safety profile for those with demonstrated benefits.

Note: In the Heart Protection Study (Lancet 2002; 360: 23-33), antioxidant vitamin supplements were assessed among people at high risk of vascular disease. Vitamin supplementation did not produce any significant reductions in the five-year risk of heart attack, stroke, cancer, or other major outcomes.

thelancet