Headlines all over the press today were “5 a day doesn’t prevent cancer”. So after all these years set in stone it’s all been a waste of time? Hard to tell really, the paper in question is all about cancer but as the Walter Willett editorial points out the same study group provided evidence that 5 a day reduces stroke and heart disease by 30%. But maybe that’s not right either, maybe Dr Aragon in Woody Allen’s Sleeper was right (this is from an interesting article about nutritional genomics by the way):
The study: Fruit and Vegetable Intake and Overall Cancer Risk in the European Prospective Investigation Into Cancer and Nutrition (EPIC) by Boffetta et al, published online in the Journal of the National Cancer Institute. It involved almost 500,000 people in a prospective study looking at nutrition and cancer between 1992-2000. Over 30,000 developed cancer but the detailed analysis revealed only a minor F&V protective effect of a few percent, if any at all.
Disappointing to say the least, an EPIC failure on the Cecil B. DeMille scale as far as cancer is concerned. So what about cancer, we assume it is preventable but is it (apart from smoking and sunlight)? Is there just so much endogenously generated DNA damage that we are really at the mercy of our genetics with no environmental components to modify for the majority of cancers? That’s hard to swallow, surely nutrition should modify cancer risk shouldn’t it? The most likely, or rather the more acceptable alternative is that this study is not really helpful either way. “Cancer” and “5 a day” – in between those inverted commas there are a whole host of biochemical processes and an even greater number of nutritional components.
- What fruit, what veg?
- Where from, how processed, cooked etc?
- The subjects were aged between 25-70
- Self-reported consumption data (which was just a single assessment of past 12 months using a food frequency questionnaire)
- 10 different countries from Europe, north to south
- Etc.
Doesn’t make for precise results. An ambitious, extremely expensive study but it’s really a rather blunt tool.
This is such a big problem with nutrition, it’s so hard to design studies with precise hypotheses to be able to pin down what exactly is going on. The most important aspect of any experiment is to control for all the variables – even a simple lab experiment needs about 10 control results per single test result.
Nowadays it seems that then most you can say about nutrition research is that it’s “mostly harmless”. It had its heyday in the first half of the last century – hunting down deficiencies in vitamins and minerals. Pure biochemistry with powerful results. It was important research, one of my biochem lecturers would recount his stories about the research he did on trying to create vitamin E deficiencies in WW2 conscientious objectors – the things they were able to get away with (but they failed to induce vit E deficiency, never found out what it’s for).
Today it’s not so powerful, it’s about optimising nutrition rather than curing malnutrition. A child with rickets is a tragedy, a 65 yr old with colon cancer is a pity. In fact it seems that most nutrition research is justified by the devastating effects of chronic disease on the economy rather than on reducing suffering, especially because there is a certain amount of freewill in lifestyle choice.
Will genetics and –omics help? They will certainly help to improve precision, stratifying those 500,000+ 25-70yr olds into different genetic pots. Perhaps it will also be better to accept that DNA damage is a better endpoint, in many ways, than “cancer”.
There are many studies where the genetics have been crucial in sorting out the associations and here are a few:
Palli et al, carcinogenesis – DNA damage was reduced by high fruit and veg diet but ONLY when the 600 subjects were grouped according to GSTM1+ or GSTM1null genotypes
Lampe et al showed that the effect of cruciferous on GST activity was only seen in GSTM1nulls
Brennan et al in the Lancet reported that a protective effect of cruciferous veg on lung cancer was only seen when stratyfying according to GSTM1 and GSTT1 genotype. A meta-analysis showed more or less the same thing
Cornelis et al in JAMA – coffee consumption had no effect on heart attack risk until the study population until CYP1A2 genotype was accounted for (CYP1A2 metabolises caffeine)
Li et al, in cancer research - antioxidants had little effect on prostate cancer unless stratified according to SOD2 genotype.
And so on. Nutrigenomics will help by giving us tools to analyse & monitor better the biochemical pathways, to go beyond simple HDL and LDL measurements and to get away from using end points such as cancer, heart attack, stroke, etc.
Here is an optimistic quote – I hope it comes true (very last para of the article):
Without genetics & nutrigenomics, epidemiological nutritional research will remain “mostly harmless”. Or to paraphrase a less amusing person maybe it’s like trying to govern the Italians - “not difficult, just a waste of time”
See also reports by American Institute for Cancer Research (AICR) and NHS Choices
