Something is not working – is it the vitamins or the trials?

This article in today’s Independent prompted me to finish off and post this – as far as good nutritional advice is concerned confusion reigns.

1. 1. Sat fats are OK after all
2. 2. Recently we were informed that the “5 a day” advice did not prevent cancer (see previous post)
3. 3. We read that >400 IU / day vitamin E increases mortality thanks to a 2005 meta-analysis (Conclusion: “High-dosage (≥400 IU/d) vitamin E supplements may increase all-cause mortality and should be avoided”)
4. 4. In 2007, a widely cited meta-analysis proved “the myth of antioxidant supplements” – they simply do no good and may even do harm
5. 5. The subject of this post – 2 major studies show that vitamin E and Selenium, despite early promise, do not protect against prostate cancer (SELECT and PHS)
6. 6. According to one author of the SELECT study –  “The prospects for cancer prevention through micronutrient supplementation have never looked worse”
   

So how much of all this is true? Probably some, none or all of it – we really don’t know, none of the work cited can be used to establish as fact such sweeping statements. We know that nutrition research is tricky – we need to improve it but we also need to be very careful what we conclude from studies. There is the temptation to overstate what the results really mean, maybe in order to justify the vast amounts spent on the studies (and with an eye on future grants).

There were a couple of high profile studies published in JAMA last year (SELECT and PHS) on vitamin E and selenium in cancer prevention. Very large studies, expensive, well designed randomised clinical trials (the “gold standard”) that showed no beneficial effect of either vit E or selenium leading to headlines like:

I don’t have a great problem with the studies themselves, they must have seemed like a good idea at the time and we have the advantage of hindsight. But I do have problems with some of the rather strident conclusions from what were basically negative (and no doubt disappointing) results. These lead to headlines that get ingrained into common beliefs. For example with vitamin E mortality (see 3 above)  – the actual studies cited in the meta-analysis showed that it required much higher levels (1-2,000IU) before there was any increase in mortality (this is mentioned in the SELECT study paper – (“…the finding of increased mortality was driven by studies using doses far higher than 400 IU/d”). Regarding the “antioxidant myth” (point 4), The 2007 Bjelakovic magnum opus “proving” that antioxidants don’t work is a very highly cited paper but very contentious with over half a page of errata and a lively debate on the Cochrane site.

I’m not “for” or “against” supplements (opinions are very polarised on this subject – just like with personal genetics, and as usual the poles are not the most useful places to be). But I am against some of the dogma (like RCT being the gold standard - it’s fine for single agents with powerful effects on well defined pathways but useless for anything much more complex), and against making conclusions that are not justified, especially in high profile studies where they will become the headlines without proper scrutiny of the work.

These recent studies on prostate cancer, vitamin E and selenium highlight several problems with current approaches. The caveats were thoroughly discussed but the conclusions were far too emphatic and anyway, probably wrong.

Brief background. Several studies had shown evidence that selenium and/or Vitamin E may reduce prostate cancer risk in particular the Nutritional Prevention of Cancer (NPC – see also this NPC study) study and the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) study. This lead to 2 very large randomised clinical trials the results of which were published in JAMA: SELECT Lippman et al, and the Physicians Health Study – Gaziano et al

SELECT:
35533 men from 2001 – 2008 (stopped early for lack of benefit)
1. Placebo
2. Selenium (200 µg/day)
3. Vitamin E (400 IU/day)
4. Selenium (200 µg/day) + Vitamin E (400 IU/day)
Conclusion: Selenium or vitamin E, alone or in combination at the doses and formulations used, did not prevent prostate cancer in this population of relatively healthy men.

Physicians Health Study:
14 641 male physicians in the United States initially aged 50 years
Mean follow up – 8 years
1. Placebo
2. Vitamin E (400 IU/every other day) plus Vitamin C (600 mg/day)
Conclusions: No benefit

What they said:
SELECT:
1.    Furthermore, the large sample size, inclusion of a substantial proportion of non-white men, and equal distribution of known risk factors across all trial groups make the conclusions drawn from SELECT especially robust and generalizable.
Problem: As even the authors say, African American men have among the highest prostate cancer risks in the world. The authors are proud of their ethnic mix, but it was actually 80% white and only 13% African American. Not really generalizable.

2.    In conclusion, SELECT has definitively demonstrated that selenium, vitamin E, or selenium + vitamin E (at the tested doses and formulations) did not prevent prostate cancer in the generally healthy, heterogeneous population of men in SELECT
Problem: “definitively demonstrated” – this is false, the many limitations of both trials (see below) do not allow such a bold statement to be made

3.    These findings also compel the medical research community to continue the search for new, effective agents for prostate cancer prevention.
Problem: Really? The situation is the same as before, these results have not excluded vitamin E or selenium, they cannot be discarded quite yet

Physicians Health Study:
”In this large, long-term trial of male physicians, neither vitamin E nor C supplementation reduced the risk of prostate or total cancer. These data provide no support for the use of these supplements for the prevention of cancer in middle-aged and older men.”
Problem: A very carefully worded conclusion. “These data” indeed “do not support…”. But they don’t support anything at all really, it’s quite a meaningless statement but of course the “meaning” that comes across is that the vitamins don’t work

Limitations of both trials
1.    the studies were conducted in a well-nourished population -  “SELECT men generally were replete in selenium”
2.    single dosages (also, curious that the ATBC used 50 IU and these trials used 400 IU vitamin E)
3.    synthetic α-tocopherol was used in the study (proper vitamin E = α, β, γ, δ-tocopherol and α, β, γ, δ-tocotrienol)
4.    Adherence  - did they really not take supplements?
5.    Regular observation for early detection of prostate cancers
6.    No genetic or diet analysis

The adherence is a really difficult problem. Both studies claimed that the subjects behaved themselves. In SELECT a subset were tested for blood levels once during the trial which showed adherence. But not really, if you are buying your own supplements you would stop taking them a few days before the blood test appointment. Also, you are part of a study the hypothesis of which is that antioxidants are good for you – you are reminded of this every day when you open the specially labelled clinical trials supplement bottle. This can make you a) not bother, b) improve diet and lifestyle in general, c) increase antioxidants in the diet, d) buy non vitamin E / selenium supplements or e) buy your own vitamin E and selenium.

Something really interesting was pointed out in the associated commentary – there was just 1 death from prostate cancer in the SELECT trial when up to 100 would have been expected, just being part of the clinical trial seemed to significantly reduce the risk (which could be reasonable - but that’s now an epidemiological observation, it needs testing, by RCT?!). Also most of the cancer cases were non-aggressive and there is other evidence that antioxidants have more protective effect against aggressive cancers.

The limitation about the well nourished population was very convincingly put in follow-up letters by Margaret Rayman and Sue Fairweather-Tait who both pointed out that actually the selenium evidence before the trials (and since) is that it is effective only in people who have low baseline levels of selenium, far below those in the SELECT trial who were “selenium replete”. I thought that the authors reply was rather shirty and dismissive (SELECT was a very large study and anyway “At present, the possibility that selenium may prevent prostate cancer in men with low plasma selenium levels is an unproved hypothesis that has not been tested as a primary end point of any clinical trial”. Well yes, precisely – and of course in the paper they don’t suggest any such trial, instead “These findings also compel the medical research community to continue the search for new, effective agents…”)

Further problems with the study population: PHS was 100% white and well-off (they were physicians). SELECT was 80% white and the majority (>80%) had higher education. This is the distribution of prostate cancer risk:

 

Prostate cancer mortality rates, by education level and race, for men aged 25-64 years in the United States, 2001
Albano, J. D. et al. J. Natl. Cancer Inst. 2007 99:1384-1394

So both trials were looking at the group with already the lowest risk (forget the 13% African American in SELECT, not enough to make a difference). Neither of the trials have shown anything regarding the usefulness or uselessness of vitamin E and selenium because, among the many other limitations, the wrong population was chosen.

PHS concluded:
”In this large, long-term trial of male physicians, neither vitamin E nor C supplementation reduced the risk of prostate or total cancer. These data provide no support for the use of these supplements for the prevention of cancer in middle-aged and older men.”

A less misleading conclusion should be:
“These data provide no support for the use of synthetic α-tocopherol and a single vitamin C dose in a group of healthy, wealthy, intelligent, upper middle-class male health professionals…maybe…”

I hope that these are the tail end of these types of studies. I also hope that something useful can be teased out of them through sub-group analysis (e.g. concentrate on those with low baseline selenium) and genotyping the DNA samples (SELECT authors say that this is underway). Meanwhile we need to rethink this dogmatic belief in RCT as gold standard for complex research questions and we should question the use of complex endpoints, concentrating on pathways and risk factors. There is a promising approach discussed by one of the Select authors at Cancer Epidemiology, Biomarkers & Prevention (Alan Krystal - Are Clinical Trials the “Gold Standard” for Cancer Prevention Research?)

Conclusions

1.    I think that one of the most important things to do, given the “genie out of the bottle” era we live in, is that accurate, useful and balanced information is given when translating medical research into something that can be understood by the general public – we all have to eat something and it would be helpful to have good information about ourselves (inc from genetics) and our environment so that we can make our own mind up. This for us at Eurogene is crucial, our job is to provide the tools and the means to translate research findings into meaningful decision making tools for healthcare professionals and individuals.
2.    These are 2 very large, well designed (in method at least), expensive, long (10+ years with planning, recruitment etc) randomised clinical trials – the gold standard – that have told us almost nothing about the questions they were supposed to address. SELECT cannot conclude that it has “definitively demonstrated” that vit E / selenium are not effective. There was only 1 death and very low rates of aggressive cancer – the subjects were mainly from the lowest risk groups and the trials were not long enough to see if there was any protection. Why choose these groups? Presumably for convenience – ease of recruitment, follow-up, adherence, etc. Convenient yes, but not useful.
3.    Studies should target the highest risk groups for all the obvious reasons (except of course convenience)
4.    These, any many others (HOPE2 and NORVIT for vitamins B) have maybe “definitively demonstrated” that the RCT will not work for optimising diet/lifestyle/behaviour for primary prevention. They have not produced any useful evidence, either way, in 20 years of trying, now is the time to try something different
5.    The hope, of course this is what I hope, is that genetics, genomics, other –omics, will bring us the tools to for better pathway analysis and better monitoring of health status

STOP PRESS: Just found this and this – subgroup analysis of the physicians health study showing benefits of selenium but only in some:

“This study provides evidence that SEP15 genetic variation may influence PCa mortality. Additionally, the association of selenium with PCa mortality was modified by a variant, suggesting the possibility that some men with PCa may benefit more from selenium than others, depending on their genotype”. Cancer Prev Res; 3(5); 604–10