When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS Absolutely Maybe

Nutrient X Prevents Disease? Sorting the Wheat from the Bran

Cartoon with slogan a promising treatment is the larval stage of a disappointing one

It started, as many issues do, because we didn’t get enough roughage in our diets. Before dietary fiber gained currency in the ‘70s as a way to protect against serious disease, people who believed we were eating ourselves into early graves weren’t taken very seriously. We’ll meet the interesting Irishman who turned that around shortly.

In the years since then, the claim that getting more fiber can reduce our chances of colorectal cancer has taken a solid scientific battering. On the side of the unconvinced are the National Cancer Institute and an old meta-analysis of randomized trials – with later trials from Japan and from England (on resistant starch) strengthening that position.

At the same time, though, others believe the case for dietary fiber from cereals is getting ever more convincing – and they point to a meta-analysis of prospective epidemiological studies.

Dueling meta-analysts don’t make life easy, do they? There’s a new interesting piece of “meta-epidemiology” that shows just how common this situation is when it comes to claims of nutrients’ ability to prevent disease.

Researchers commissioned by the U.S. Agency for Healthcare Research and Quality recently published an analysis of 34 cases of studying whether or not “nutrient X prevents disease Y.” For these, they compared the conclusions of randomized trials (where groups of people randomized to consuming more nutrient X were compared with groups who weren’t) and epidemiological studies (no randomized comparison group).

In only 6 of these 34 cases did both the randomized and epidemiological study pools give “thumbs up” or “thumbs down” to nutrient X, with a statistically significant result from both. Whereas in 11 cases, there was at least a maybe “thumbs up” from one (usually the epidemiological pool) but more of a “thumbs down” from the other – including dietary fiber to prevent colorectal cancer. The researchers concluded that to be sure about the effects of nutrients, “plausible epidemiological support is probably not enough.”

They were looking for signs to suggest when a body of evidence may be getting to a definitive answer. But mapping the research showed no clear pathway. There wasn’t a translational science trajectory from non-randomized studies to randomized trials, for example. Researchers are doing non-randomized studies at the same time – and after – strong trials.

Image of poster

National Institutes of Health poster (NIH Medical Arts & Photography, via National Library of Medicine)

Trials are the only way to remove other dietary, lifestyle and environmental differences from the mix of factors that could be muddying the waters, making nutrient X look better (or worse) than it really is. They will sometimes point to possible harm from tinkering with something as complex as the diet.

But the evidence from trials on nutrients is criticized for a range of reasons. For example, people didn’t get nutrient X in the right form or enough of it, or the trial didn’t go on long enough or at the right time in people’s lives.

The National Cancer Institute points out that even if there is an influence from dietary fiber on its own, it’s unlikely that it could be a large difference.

The idea that you could prevent yourself getting colorectal cancer by piling on the bran is appealing, though. It captured people’s imaginations in the ‘70s, and I don’t suppose it’s going to let go easily.

The fiber theory was the child of an earlier “nutrient X” hypothesis by an English naval surgeon, Captain Thomas Latimer Cleave. His little sister had died of appendicitis when they were young. He later became convinced over-refined diets were responsible for this disease and much more. He was nicknamed “the bran man” during World War II, for all the bran he had put on naval vessels to prevent constipation.

It wasn’t bran that he thought was the really big culprit, though. He believed it was refined sugar, publishing a book in the ‘60s on “the saccharine disease.” Cleave had convinced another physician who became involved, Hubert Trowell, but wasn’t breaking through on a large scale.

Sir Richard Doll – the epidemiologist famous for his work on establishing the link between smoking and disease – said although the data behind the theory could be torn into strips, there might nevertheless be something to the idea that western diets were contributing to diseases more common in western countries.

Photo
Denis Burkitt (click for YouTube interview)

Doll introduced Cleave to former surgeon turned researcher, Denis Burkitt. That was the turning point. Burkitt, like Trowell, had worked in Africa for years, and was then at the Medical Research Council. His credibility, network and energy were immense. He was responsible for making the connection between a virus and a kind of non-Hodgkin’s lymphoma which was named after him (Burkitt lymphoma and leukemia).

Burkitt explains on YouTube what happened next: “I can remember walking along the sands one time in Mombasa, and the thought coming to me: you can’t have sugar unless you take the fiber out in the first case, to refine it and give you sugar. And could it be the fiber that is taken out that we are short of, rather than the sugar we are feeding…And this sent me worldwide on measuring stools, because the amount of stool people pass is related to their fiber intake.” So they got people weighing stools, in Africa and India – and in Seventh Day Adventists who had high-fiber diets.

Newspaper clipping
The Pittsburgh Press, 26 April 1971 (via Google News archive)

Burkitt published his seminal paper linking lack of dietary fiber to the development of colorectal cancer in 1971, having an immediate major impact. A best seller, “Don’t forget fibre in your diet,” followed in 1979.  Burkitt’s advocacy for diet’s contribution to health, along with Linus Pauling’s advocacy for mega-doses of vitamin C, helped set off a spate of theories and beliefs about dietary supplements that shows no signs of letting up.

The recent analysis showing the difficulty of reconciling epidemiological and randomized studies suggests the debate over evidence isn’t going to let up any time soon either.

~~~~

You can read more about prevention of colorectal cancer here.

More from me on prevention in my post on The Prevention Illusion: A Tragedy in Five PartsBrowse through all my posts at SciAm on related themes.

You might also be interested in my post on 5 things to know about meta-analysis.

My thanks to @Profmicro who pointed out an imprecise use of “mosquito-borne” with Burkitt Lymphoma in the original text (via Twitter). More on this.

 

Painting of harvesting wheatThe painting at the top of this post is Piantagione di grano, Tacuinum sanitatis Casanatense from the 14th century, via Wikimedia Commons.

The “promising treatments” cartoon is by the author, under a Creative Commons license.

 

 

* The thoughts Hilda Bastian expresses here are personal, and do not necessarily reflect the views of the National Institutes of Health or the U.S. Department of Health and Human Services.

Discussion
  1. Nice post, with lots of amusing points. However the problem with randomised intervention trials on fibre and colorectal cancer (CRC) is that they cannot really address the hypothesis. The development of colorectal cancer takes decades, and the protective effect of fibre, if there is one, probably acts over a similar timescale. Intervention trials usually test particular mechanistic hypotheses (e.g. a beneficial effect of butyrate derived from fermentable carbohydrate) and utilise short-term treatments in polyp patients who, by definition, are already vulnerable to neoplasia. Burkitt’s hypothesis could have been falsified by a consistent lack of evidence from large prospective studies but it has never quite happened. Clearly fibre is nothing like as important as the early enthusiasts believed, but on balance I think public health advice to consume about around 30g fibre per day is justifiable. Slightly more detail from me on this here:

    http://www.futuremedicine.com/doi/full/10.2217/crc.12.4

  2. Thanks, Ian! While I agree it would have been helpful to have better studies than we’ve got, I think the trials are indeed helpful for whether or not high amounts of fiber can prevent colorectal cancer. Testing in people at high risk of colorectal cancer has limitations, but that doesn’t mean it wasn’t a good strategy to address the issue. While I agree lots of fiber in the diet can have other health benefits, cancer prevention isn’t the reason to do it. And the uncontrolled studies haven’t been able to account for bigger issues that often accompany healthy diets – in particular, not smoking. The hypothesis that it might reduce cancer isn’t a good enough reason to encourage major dietary change – and the potential for people blaming themselves when they do get cancer and didn’t follow that kind of diet.

    It’s not that easy to get that much fiber in the diet regularly every day, decade in, decade out. Which means this ends up rather inextricably linked with encouraging people to use dietary supplements. And those definitely need better evidence of value before encouraging them.

  3. Interesting coming back to this after 6 years and nothing has changed!
    Cleave thought that refined carbs drove CRC, and today we can say that hyperinsulinaemia (or its proxy obesity, but insulin is a better marker) is definitely a strong risk factor for CRC that is corrected by the removal of refined carbs – or all carbs.
    Burkitt got the wrong end of the stick, and ended up championing the intervention with the weaker potential for benefit, eating additional fibre, which will decrease glucose absorption and insulin a bit but is not so unidirectional (or necessarily tolerated).
    Actor Luke Perry cut down on red meat and increased grain fibre after a polyp scare, quite understandably, but perhaps he also took on other “healthy eating” advice like limiting saturated fat. He died of a massive stroke 2 years later. Of course this sad story says nothing about causality either way – it is just a bleak reminder that following the results of epidemiological meta-analysis, interpreted in the light of a popular theory and your doctor’s wishful thinking, may not be associated with the outcome everyone wants.

  4. About the cancer time-scale problem in RCTs – the Polyp Prevention trial had no benefit from overall increasing grain fibre by 70%, reducing red meat by 20%, and a few other changes – including less fat, more fruit and vege – (that would make it harder to assign causality to the main two), in fact things were non-significantly worse.
    https://www.nejm.org/doi/full/10.1056/NEJM200004203421601
    So you could say that the population was wrong (too high risk) and the time too short, which makes sense.
    But then the PPT authors salvaged their trial with a subgroup analysis. It seems that participants that doubled down on their advice across at least 4 categories in more than one FFQ were protected. No one category of advice was associated with benefit on its own.
    The only problem there is, they didn’t case-match the controls but just took the total control population as the comparator for this compliant sub-group. tut tut.
    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2732986/
    But in any case, you can’t have your cake and eat it. If the PPT subgroup analysis was valid, then the PPT overall was a valid test of the hypothesis too. (and a much better test of the advice, as advice).
    https://www.ncbi.nlm.nih.gov/pubmed/17855692
    FWIW, another PPT subgroup analysis found very strong benefit associated with flavanoid foods, which is a bit more palatable and less likely to do harm than the rest of the recommendations
    http://cebp.aacrjournals.org/content/17/6/1344.short

Leave a Reply

Your email address will not be published. Required fields are marked *


Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top